Key words
acetal - dioxolane - radical - HAT - decarboxylation
1
Introduction
Dioxolanes appear often within the context of both classical and modern organic chemistry
in largely one of two simple ways, as either a reaction solvent or protecting group.
While commonplace in both synthesis and methodology, dioxolanes are often overlooked
in conversations centered around the discovery and development of advanced synthetic
strategies. Such a simple structure can be activated to yield a multifunctional carbon-bond
framework, poised for the direct introduction of oxygen-containing functional groups
such as aldehydes, ketones, and free diols, while circumventing traditional oxidation
chemistry (Figure [1]A).
Dr. Daniel K. Kim earned his BSc at Gettysburg College (USA) before earning his doctoral studies under
Professor Vy M. Dong, where he developed Co- and Rh-catalyzed hydroacylation reactions
at the University of California, Irvine (USA). He completed a postdoctoral fellowship
with Prof. David W. C. MacMillan at Princeton University (USA) where he helped develop
a site-selective tyrosine bioconjugation using photoredox catalysis. In 2020, he started
his independent career in Philadelphia at Temple University. His research recently
focuses on developing new radical methodologies and transition-metal catalysis focused
on the installation of industrially relevant scaffolds.
Figure 1 Dioxolane general uses and bond-dissociation energies (BDEs)
Historically, fully saturated cyclic ethers, such as furans, dioxolanes, pyrans, and
dioxanes, have been employed as aprotic solvents for organic reactions from as early
as the late 19th century.[1] These ethers are known for their low boiling points, low reactivity, and relative
polarity, and are an important class of nonaromatic heterocycles. Dioxolane, specifically
1,3-dioxolane, has been and continues to be employed as an environmentally friendly
and cost-conscious substitute to other ethereal solvents like THF or diethyl ether.[2]
[3]
Aside from its use as reaction media, dioxolane is most famously known to be employed
in protection strategies for labile oxygen-containing functional groups.[4]
[5]
[6] Diols, ketones, and aldehydes can be easily masked by inexpensive and commercial
reagents such as acetone and ethylene glycol, respectively, and are just as easily
deprotected in mild acidic conditions.[7,8] The capability of such a functional group to protect both carbonyls and glycols
in equal ease highlights its duality and wide applicability. Facile syntheses coupled
with low cost and minimal hazard make dioxolanes an effective and obvious choice for
shielding reactive species. Dioxolanes as protecting groups are long-established in
literature, ranging from protections of complex carbohydrates to key fragments in
natural product syntheses.[9–12]
Building upon their role in traditional protection strategies, it has been documented
that dioxolanes also have success in radical-mediated reactions, enabling the synthesis
of high-value masked carbonyls. These desired radicals can be generated through a
variety of means, such as hydrogen atom transfer (HAT) or oxidative decarboxylation
processes, which we discuss extensively in our review of acetal radicals and their
applications.[13]
In a study by Doyle and coworkers, the abstraction of C–H bonds on solvent-grade 1,3-dioxolane
was explored (Figure [1]B).[14] Intriguingly, radical formation at both the 2- and 4-positions was observed. The
C–H bond at the 2-position has a lower bond-dissociation energy (BDE) of 86.8 kcal/mol,
compared to the 4-position of 88.2 kcal/mol, and the authors note that there is a
preference to the more accessible BDE. This leads to the observed selectivity for
radical formation at one position over the other, mediated by a photoredox-generated
HAT reagent. This report provides a foothold to demonstrate potential functionalization
at each respective position.
The overall need for protection strategies stems from the fundamental issue of chemoselectivity
as molecules become more complex.[15] An approach that leverages both the intrinsic reactivity and polarity of diols and
carbonyls would be extremely valuable to the synthetic community. With this consideration
in mind, we postulate that 1,3-dioxolane motifs can act as highly oxidized carbon
constituents ideal for fragment coupling. The duality of dioxolanyl radicals lends
itself well to establishing a general reagent that can access both diols and carbonyls
through controlled chemoselective radical generation at either of the aforementioned
positions. We proposed employing these dioxolane skeletons as specially tuned synthetic
handles towards radical C–C bond construction. This method is uniquely situated to
aid in the installation of highly oxidized functional groups such as ketones, aldehydes,
and alcohols. Just as dioxolanes can be employed in dual protecting strategies, dioxolanyl
radicals can be thought of as dual-use coupling fragments that will facilitate the
construction of highly polar complex molecular scaffolds.
With this design strategy in hand, our lab has recently designed a fluorinated dioxolane-based
carboxylic acid that enabled an otherwise challenging coupling reaction, in this case,
at the dioxolane 2-position, accessed by oxidative decarboxylation, to construct highly
sought after trifluoromethyl ketones (TFMKs).[16] Following the success of our initial findings, we continued our exploration of dioxolanyl
radicals instead at the 4-positon to install highly polarized and valuable vicinal
diols through a direct photomediated hydrogen atom transfer (HAT) process.[17]
Masked Trifluoromethylacetylation
2
Masked Trifluoromethylacetylation
A cornerstone of our research program relies heavily on dioxolanyl-radical based strategies
to form C–C bonds. We had broadly aimed to develop a new methodology to incorporate
TFMKs into complex molecular scaffolds. Fluorine-containing coupling fragments are
becoming an essential component in small-molecule drug discovery, with fluorine appearing
in many of the best-selling agrochemicals and FDA-approved therapeutics.[18]
[19]
[20] As such, the synthetic community would greatly benefit from an arsenal of complementary
synthetic strategies to access a variety of fluorofunctional groups.
In particular, TFMKs are of interest due to their ability to enhance desired pharmacokinetic
properties such as increased lipophilicity, metabolic stability, and enzyme inhibition.[21]
[22]
[23]
[24]
[25] It is well-reported that TFMKs, in both their ketone and hydrate form, act as bioisosteres
to carboxylic acids, which again can help alleviate undesired biological responses.[26,27] Many traditional approaches to synthesizing TFMKs rely on two-electron chemistry,
adding various nucleophiles into fluorinated electrophiles such as trifluoroacetic
anhydride, ethyl trifluoroacetate, or fluorinated Weinreb amides.[28] We noticed that there was a need for an alternate disconnection and began developing
a nucleophilic TFMK reagent which would allow us to install TFMKs into medicinally
relevant electrophilic substrates such as strained bicycles, imines, and heteroaromatics.
While we were exploring this, a radical-based photochemical approach was disclosed
by Katayev and coworkers wherein trifluoroacetic anhydride was reduced to yield a
trifluoromethyl acyl radical by C–O fragmentation to put forth in coupling to electron-rich
aryl olefins.[29] We were able to expand upon their strategy with our own, in turn furnishing access
to electron-deficient partners, thus yielding a complementary radical approach towards
TFMK synthesis.[16]
Trifluoromethyl acyl radicals are prone to rapid decarbonylation and are thus difficult
to harness for coupling.[30] They are also exceptionally electrophilic when compared to other acyl radicals,
with a global electrophilicity value calculated to be 1.75 by Katayev and coworkers
(Figure [2]).[31] This is considerably more electrophilic than that of the methyl acyl radical with
a value of 1.08, calculated by Nagib and coworkers.[32] Considering the relative electrophilicity of trifluoromethyl acyl radicals, reactivity
towards electron-deficient partners would be challenging, even without the added issue
of decarbonylation. This premise is further supported experimentally by our inability
to couple trifluoropyruvic acid to electron-deficient olefins.[16]
Figure 2 Decarbonylation rates and electrophilicity values for common acyl radicals
With all of this in mind, we concluded that we could solve both issues, radical polarity
and stability, by installing some type of masking group. For simplicity, we envisioned
putting forth conventional carbonyl-protection strategies that would promote nucleophilic
radical generation, such as the introduction of a stabilizing ether. It is well-precedented
that C-centered radicals can be formed at the α,α-dialkoxyalkyl position of both cyclic
and acyclic acetals.[13] It was imperative to design an ethereal skeleton that would yield a radical that
was stable and persistent enough for coupling without further rearrangement, as many
of these radicals are known to promote β-fragmentation pathways.[33]
[34]
[35]
[36]
[37] Doyle and coworkers have done extensive studies on the generation of methyl radicals
from acyclic trimethyl orthoformate, which is promoted through excellent σ*–p orbital
overlap.[38] In addition to acyclic systems, Martin and coworkers disclosed an informative report
that highlights the orbital overlap and likelihood of β-scission in cyclic dioxolanes,
dioxanes, dioxepanes, and dioxocanes.[39] It was found that the conformational flexibility afforded to larger seven- and eight-member
cyclic acetals enabled the σ*–p orbital overlap to promote β-cleavage (Figure [3]), while the more inflexible five- and six-member rings yielded only the tertiary
acetal radical.
Figure 3 Circumventing β-fragmentation pathways in reagent design
With the ability to circumvent β-fragmentations, as well as enhanced stability and
in turn nucleophilicity, the five-membered dioxolane was targeted as an ideal scaffold
and thus inspired a variety of fluorinated carboxylic acids (Figure [4]), with varying alkyl and heterocyclic substituents, such as cyclobutane, oxazolidine,
dioxane, and dioxolane, which were all easily synthesized in two steps.
Figure 4 Initial design strategies and designer radical equivalent reagents
Interestingly, the dioxolanyl-based structure 1 proved best for coupling with 95% coupling yield with model substrate 2-vinylpyridine
through a traditional oxidative decarboxylation followed by addition into the olefin
and subsequent reduction and protonation. Following our initial report, the Katayev
group has also calculated the electrophilicity value of the resulting trifluoromethyl
dioxolanyl radical to be 1.01, again reinforcing the improved nucleophilicity of our
trifluoromethyl dioxolanyl radical.[31] We used this reagent to couple to forty electron-deficient alkenes and vinyl heterocycles,
with a selected scope highlighted here to showcase the reactivity of our dioxolanyl
radical (Figure [5]). Excitingly, these mild conditions tolerated complex substrates like loratadine,
menthol, and sclareolide.
Subsequently, the dioxolanyl-containing compounds could be deprotected to yield the
highly sought-after TFMKs using Lewis acidic boron tribromide which was demonstrated
on our model substrate (Figure [6]A). The unmasking was uniquely challenging and did not proceed using traditional
dioxolane-removal strategies. For sensitive substrates, like sclareolide, alternative
masking agents can be engaged, although with loss of yield in the coupling (Figure
[6]B).
Figure 5 Highlighted substrate scope for the incorporation of the trifluoromethyl dioxolanyl
radical reagent
Figure 6 Deprotection strategies towards accessing TFMKs
The data collected in our initial study supports that electron-donating atoms enable
more nucleophilic character of the resulting dioxolanyl radical, creating the desired
electron-rich ‘acyl’ radical. An interesting result from this study shows that the
oxazolidine-based reagent had decreased yields when coupling to olefins and other
electrophilic partners, while presumably possessing more electron donation from the
nitrogen atom. We are currently investigating the unique properties and reactivity
of the generated radical through further experimentation and computations.
The use of our trifluoromethyl dioxolanyl radical promoted chemical reactivity that
has been otherwise challenging through other more traditional hydroacylation strategies.
The twofold use of dioxolanyl radicals as both a stabilizer and inverse polarity facilitator
was essential to the success of this study. Since our seminal report, other groups
have employed our masked radical equivalent towards couplings to specialized heterocycles
and alkenes.[40]
[41] While this reinforces that we have developed a synthetically useful reagent which
promotes a highly desired transformation to new electrophiles, the deprotection still
necessitates the use of boron tribromide to access the resulting TFMK. In many contexts,
the use of boron tribromide as an unmasking reagent is a reasonable way to reveal
the desired TFMKs. However, in the case of Lewis acid sensitive functional groups,
or necessity on industrial scale, its use as an unmasking reagent can be too harsh.
Since our initial finding, we have been working towards milder and more general deprotection
strategies to access TFMKs from other electrophiles. Additionally, specifically regarding
Minisci-type reactivity, we successfully developed a second-generation masked radical
equivalent which lends itself to deprotection much more easily and without the need
for strong Lewis acids or bases.
Dioxolanyl Radicals: Vicinal Functionality
3
Dioxolanyl Radicals: Vicinal Functionality
Our group successfully employed a dioxolane-based designer reagent as both a masking
agent and inverse polarity inducer for the introduction of the trifluoromethyl ketone
moiety.[16] While concluding our initial study, we considered other possible uses for the dioxolane
core. We saw an opportunity for dioxolanyl radicals to be used to install another
highly oxidized functional group, 1,2-diols, which are often protected into dioxolanes
in traditional syntheses. Highly polarized fragments like vicinal diols can introduce
favorable hydrogen-bonding interactions in enzymatic binding pockets and improve aqueous
solubility, which in turn improves their overall pharmacokinetic profile.[42]
[43] Additionally, internal hydrogen bonding helps to preserve drug permeability, which
can rapidly deteriorate in the presence of other highly polar functional groups.[43] When targeting pan-BET bromodomain inhibitors, GSK scientists found that the introduction
of a diol motif in lieu of the traditional alcohol helped improve aqueous solubility
while maintaining suitable membrane permeability (Figure [7]).[44] As such, pharmaceutical companies like GSK, Novartis, Genentech, and Argenta have
used this to great effect in various enzyme inhibitors.[44]
[45]
[46] Additionally, 1,2-diols are synthetically useful intermediates to rapidly build
molecular complexity from simple building blocks.[47]
Figure 7 Vicinal diols in drug synthesis
In 1936, Nicholas Milas and Sidney Sussman reported the use of osmium tetroxide as
a means to oxidize alkenes to syn-diols.[48] Nearly a century later, olefins are still highly used to facilitate the desired
dihydroxylations. Since this seminal report, many synthetic groups have optimized
and refined this oxidation for milder reaction conditions and stereoselective control.[49] These efforts culminated in the Nobel Prize winning Sharpless dihydroxylation, demonstrating
how alkenes have become the premier precursor to access this substructure in both
natural product total synthesis and small-molecule drug design.[47]
[49] Further expanding upon this strategy, synthetic chemists have used alkenes as a
precursor to other vicinally functionalized motifs like 1,2-diamines, 1,2-amino alcohols,
and 1,2-thiol alcohols through oxidative pathways.[50–52] In the early stages of development in our dioxolanyl radical chemistry, we saw an
opportunity to install 1,2-diols in a complementary C–C bond-forming approach via
hydrogen atom transfer (HAT) catalysis (Figure [8]).[17]
Figure 8 Advancements in vicinal diol synthesis
In our initial study, we attempted to use ethylene glycol directly as a cheap and
readily available reagent for the desired C–C bond construction with model olefin
phenyl vinyl sulfone. Interestingly, this reaction proved to be very inefficient with
only 12% conversion to product after 16 h of irradiation with 370 nm light (2). We envisioned that, similar to our previous study involving the trifluoromethyl
dioxolane reagent, that a dioxolane scaffold could be used as an activator of ethylene
glycol to bolster this reactivity (Figure [9]).[53] We hypothesized that we could use C–H BDEs as a way to quickly evaluate masked diol
reagents. When calculating the C–H bonds of 1,3-dioxolane, we found that the 2-position
was the weakest (86.8 kcal/mol), followed by the 4-position (88.2 kcal/mol), which
is also in accordance with the report from Doyle and coworkers.[14]
Figure 9 Using computations to design reagents
These slightly differing values can be due to a hyperconjugation effect from the flanking
oxygens at the 2-position, which in turn help in stabilizing the radical intermediate.
Additionally, Doyle and coworkers confirmed that C–H abstraction does occur at the
2-position of the dioxolane over the 4-position at about a 9:1 ratio, a selectivity
in line with the difference in BDE between the two positions (1.4 kcal/mol).[14] Because of this insight, we designed our acetal reagent to block the 2-position
of the dioxolane for sole regioselectivity at the 4-position. Excitingly, moving to
simple ethylene carbonate improved our yield to 43% (3). We recognized that we could use thermodynamic BDEs as a measure to roughly estimate
reactivity. However, it was not obvious how distal changes at the 4-position might
impact the BDE at the 2-position. Thus, we turned to computational analysis for a
systematic approach to reagent design. We explored many possible contenders, computing
the BDE and subsequently exploring the coupling experimentally (3). Simple screenings of acetal reagents by swapping ketones in an acetal condensation
yielded great differences in C–H BDE and subsequently coupling yields (4). We found that simple 2,2-dimethyl-1,3-dioxolane, derived from readily available,
green, and affordable ethylene glycol and acetone, was the best reagent for our desired
transformation, affording a 72% yield (5).
Furthermore, we thought we could expand this logic to access not only 1,2-diols, but
also 1,2-diamines, 1,2-amino alcohols, and 1,2-thiol alcohols (Figure [10]). Symmetrical diamine reagent imidazolidinone coupled efficiently with phenyl vinyl
sulfone in 75% yield (6). Likewise, 2,2-dimethyl-1,3-dithiolane coupled in 72% yield (7). However, amino alcohol and thiol alcohol reagents provided a new challenge. Mixed
heteroatom reagents provide two unique C–H bonds for abstraction, in turn yielding
two regiomeric products. Using the same computational methods as for the symmetrical
reagents, we could predict regioselectivity based on BDE. After exploring a variety
of amino alcohol reagents, oxazolidinone proved to be the best reagent due to ease
of deprotection, cost-effectiveness, and favorable computational calculations. We
calculated a 3.3 kcal/mol difference between the α-amide C–H bond (84.1 kcal/mol)
and the α-acyloxy C–H bond (87.4 kcal/mol). Experimentally, we observed an 82% yield
and 100% regioselectivity towards the lower C–H bond (8). Using the same logic, we used 2,2-dimethyl-1,3-oxathiolane for our thiol alcohol
reagents because we calculated a large BDE difference between the α-thiol C–H bond
(83.5 kcal/mol) and the α-oxy C–H bond (86.4 kcal/mol). While coupling yield was competent
(60% yield), regioselectivity was modest (67% regioselectivity) towards the lower
C–H BDE (9). We turned to radical philicity calculations to potentially explain these trends.
We found that the α-thiol radical was slightly more nucleophilic than the α-oxy radical,
with electrophilicity values of 0.69 eV and 0.75 eV, respectively. A similar discrepancy
is seen regarding the amino alcohol reagent (0.88 eV vs 0.93 eV, respectively). This
indicates to us that the regioselectivity in the thiol alcohol case is not governed
solely by C–H BDE or radical philicity, launching a new opportunity for our lab to
explore other computationally and experimentally driven approaches to solve this interesting
selectivity challenge.
Figure 10 Other reagents to access vicinal functionality
Once our optimized designer-masked diol reagents were in hand, the reaction was implemented
on a variety of different electron-deficient olefins including vinyl phosophonates,
vinyl amides, vinyl pyridines, and α,β-unsaturated ketones and esters (10–13, Figure [11]). Excitingly, we can tolerate a coupling with substrates that also have labile C–H
bonds. For example, lactones are known to readily engage in α-oxy HAT abstraction.
Additionally, d-sclareolide is well-known to be susceptible to HAT processes at the C2 position using
TBADT as a catalyst (14).[54]
Figure 11 Selected substrate scope
However, in both cases, we observe high coupling without any observation of cross
selectivity. This highlights the chemoselectivity of our designer acetal reagent.
The low BDE of our acetal reagent outcompetes other weak C–H bonds that are ubiquitous
in organic molecules (90–105 kcal mol–1).[53]
Figure 12 Selected substituted acetal scope
We can also activate pre-oxidized diol motifs and synthesize complex diols in a C–C
bond-forming reaction (Figure [12]). Knowing that radical reactions can be sensitive to steric bulk, we were curious
about the competition between steric bulk and BDE. For example, methyl-substituted
acetal had slight regioselectivity towards the more substituted methine carbon (BDE
87.1 vs 85.8 kcal mol–1; 15). However, more sterically encumbered phenyl-substituted acetal had complete regioselectivity
towards the less sterically hindered methylene carbon despite large differences in
BDE (87.2 vs 75.0 kcal mol–1; 16). Excitingly, we could couple with complex sugar substrates as well with multiple
hydridic C–H bonds in a regioselective fashion. Additionally, because the dioxolane
here is bound within a bicyclical structure, radical addition occurs at only one face,
giving the diastereoselective product 17. This is in line with work done by Taylor and coworkers, wherein they use borinic
acids to activate sugar motifs for HAT processes.[55]
Lastly, we demonstrated the value of our coupling by engaging it in an in situ cyclization by coupling with arylates, propiolates, and ethylene malonates (Figure
[13]; 18 and 19). In a one-step procedure, we can access a wide variety of unique scaffolds. In all
cases, the five-membered lactone is synthesized rather than the six-membered, following
the expected reaction kinetics for ring-closing reactions. Excitingly, this is not
only limited to the synthesis of lactones. Chiral cyclic phosphonates can also be
synthesized very rapidly in two steps (20).
Figure 13 Selected lactonization scope
We have demonstrated a unified approach to vicinally functionalized motifs like 1,2-diols,
1,2-amino alcohols, 1,2-diamines, and 1,2-thiol alcohols through the use of acetal-like
reagents. We designed a variety of different reagents using computationally driven
BDE and radical polarity calculations.
With the results of this initial study in hand, we are excited to further enhance
the value of these acetal-like reagents by taking advantage of important asymmetric
C–C bond-forming frameworks. Furthermore, oxidation of alkenes to provide dihalogenated
motifs is a growing area of interest due to favorable pharmaceutical properties and
their use as a versatile synthetic intermediate. A complementary C–C bond-forming
approach to build vicinally functionalized dihalogen motifs via an α-halogen C–H process
would be extremely valuable to the broader synthetic community.
4
Conclusion and Outlook
In summary, this perspective highlights our recent developments in both generating
dioxolanyl radicals and implementing them in the syntheses of highly oxidized motifs
such as ketones and alcohols. Dioxolanyl radicals can be thought of as dual-use reagents,
as they are able to prevent decomposition like decarbonylation, as well as enabling
otherwise difficult chemical reactivities. Our fluorinated dioxolane-based carboxylic
acid enabled coupling towards electrophilic targets that are traditionally challenging
to access using previously established methods. In contrast, selective radical generation
on the dioxolane backbone gives opportunities to access 1,2-diol motifs in a C–C bond-forming
approach. We can expand this system to other saturated heterocycles to access other
vicinally functionalized motifs like diamines, amino alcohols, and thiol alcohols.
Nearly a century after their introduction as a protecting group, dioxolanes have been
realized as a powerful, synthetic building block that can be harnessed to install
highly oxidized motifs for the complex synthesis of drugs and natural products.