Lewis or Br?nsted? A Rectification of the Acidic and Aromatic Nature of Boranol-Containing Naphthoid Heterocycles

Article abstract of DOI:10.1021/jacs.1c02462

Boron-containing heterocycles are important in a variety of applications from drug discovery to materials science; therefore a clear understanding of their structure and reactivity is desirable to optimize these functions. Although the boranol (B-OH) unit of boronic acids behaves as a Lewis acid to form a tetravalent trihydroxyborate conjugate base, it has been proposed that pseudoaromatic hemiboronic acids may possess sufficient aromatic character to act as Br?nsted acids and form a boron oxy conjugate base, thereby avoiding the disruption of ring aromaticity that would occur with a tetravalent boronate anion. Until now no firm evidence existed to ascertain the structure of the conjugate base and the aromatic character of the boron-containing ring of hemiboronic "naphthoid"isosteres. Here, these questions are addressed with a combination of experimental, spectroscopic, X-ray crystallographic, and computational studies of a series of model benzoxazaborine and benzodiazaborine naphthoids. Although these hemiboronic heterocycles are unambiguously shown to behave as Lewis acids in aqueous solutions, boraza derivatives possess partial aromaticity provided their nitrogen lone electron pair is sufficiently available to participate in extended delocalization. As demonstrated by dynamic exchange and crossover experiments, these heterocycles are stable in neutral aqueous medium, and their measured pKa values are consistent with the ability of the endocyclic heteroatom substituent to stabilize a partial negative charge in the conjugate base. Altogether, this study corrects previous inaccuracies and provides conclusions regarding the properties of these compounds that are important toward the methodical application of hemiboronic and other boron heterocycles in catalysis, bioconjugation, and medicinal chemistry.

Full text of DOI:10.1021/jacs.1c02462

pubs.acs.org/JACS
Article
Lewis or Brønsted? A Rectification of the Acidic and Aromatic Nature
of Boranol-Containing Naphthoid Heterocycles
M. Zain H. Kazmi,# Jason P. G. Rygus,# Hwee Ting Ang, Marco Paladino, Matthew A. Johnson,
Michael J. Ferguson, and Dennis G. Hall*
Cite This: J. Am. Chem. Soc. 2021, 143, 10143−10156
Read Online
sı
*
Metrics & More
Article Recommendations
Supporting Information
ACCESS
ABSTRACT: Boron-containing heterocycles are important in a
variety of applications from drug discovery to materials science;
therefore a clear understanding of their structure and reactivity is
desirable to optimize these functions. Although the boranol (B−OH)
unit of boronic acids behaves as a Lewis acid to form a tetravalent
trihydroxyborate conjugate base, it has been proposed that
pseudoaromatic hemiboronic acids may possess sufficient aromatic
character to act as Brønsted acids and form a boron oxy conjugate
base, thereby avoiding the disruption of ring aromaticity that would
occur with a tetravalent boronate anion. Until now no firm evidence
existed to ascertain the structure of the conjugate base and the
aromatic character of the boron-containing ring of hemiboronic
“naphthoid” isosteres. Here, these questions are addressed with a
combination of experimental, spectroscopic, X-ray crystallographic, and computational studies of a series of model benzoxazaborine
and benzodiazaborine naphthoids. Although these hemiboronic heterocycles are unambiguously shown to behave as Lewis acids in
aqueous solutions, boraza derivatives possess partial aromaticity provided their nitrogen lone electron pair is sufficiently available to
participate in extended delocalization. As demonstrated by dynamic exchange and crossover experiments, these heterocycles are
stable in neutral aqueous medium, and their measured pKa values are consistent with the ability of the endocyclic heteroatom
substituent to stabilize a partial negative charge in the conjugate base. Altogether, this study corrects previous inaccuracies and
provides conclusions regarding the properties of these compounds that are important toward the methodical application of
hemiboronic and other boron heterocycles in catalysis, bioconjugation, and medicinal chemistry.
by recent developments in the use of 1,2-azaborine as an
isosteric replacement of benzene rings in various preclinical
candidates (Figure 1B).10−12 Pseudoaromatic hemiboronic
acids constitute another important and diverse subclass that
has been studied for the past five decades. Among the most
studied are the benzoxazaborine and benzodiazaborine
scaffolds represented by compounds 1−3, which are isosteric
and isoelectronic to isoquinoline (e.g., 4), along with the
“inverted” naphthoids 5 and 6 and the phenanthroids 7 and 8.1
For example, 1-hydroxy-1,2-dihydro-2,3,1-benzodiazaborines
have been proposed as components of molecules mimicking
estrogen and other steroid hormones,13 whereas N-sulfonyl
derivatives were identified as potent inhibitors of bacterial
NAD(P)H-dependent enoyl acyl carrier protein reductase14,15
INTRODUCTION
■
Once viewed as a chemical curiosity, boron-containing
heterocycles have taken a growing place in chemistry as
remarkable compounds with various uses.1 They display a wide
range of properties and applications from functional
luminescent materials and sensors to pharmaceutical drugs.
The replacement of alkene CC bonds with the isoelectronic
and isosteric B−N/B−O bonds has been shown to significantly
alter the photophysical and electronic properties of a
molecule.2,3 In drug discovery, the use of boron−heteroatom
bonds is an emerging strategy to unveil novel bioisosteric
pharmacophores with unique properties and opportunity to
explore uncharted patent space.4,5 For example, guided by new
fundamental advances,6 three cyclic nonaromatic hemiboronic
acids were developed and commercialized as therapeutic
agents in the past decade (Figure 1A).7−9 These boranol
(B−OH) compounds offer desirable in vivo properties as mild
acids, with a unique ability to undergo reversible exchange of
their boron−hydroxy bonds with alcohols and carboxylate
groups on target biomolecules. Similarly, pseudoaromatic
boron heterocycles display great potential, as demonstrated
Received: March 6, 2021
Published: June 24, 2021
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156
© 2021 American Chemical Society
10143

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Figure 2. (A) Resonance and pseudoaromaticity in oxazaborine and
diazaborine rings. (B) Questions of stability in aqueous medium and
Lewis or Brønsted acidic nature addressed herein. (X = O, NR)
questioned, with reports claiming they act as Lewis acids and
other reports suggesting Brønsted acidic behavior akin to
alcohols (see Background section below). In the former, the
corresponding conjugate base I disrupts the ring’s putative
aromatic character through forming a tetravalent hydroxyboryl
structure similar to that of the borate anion and reminiscent of
a ketone hydrate (Figure 2B).
This sort of indirect Brønsted acidity via boron’s Lewis
acidic behavior is well established for acyclic boronic acid
derivatives. Indeed, although they can engage in hydrogen-
bonding interactions in both the solid state and solution,18,19
boronic acids form a tetravalent trihydroxyborate anion as their
conjugate base in water.20 In the case of Brønsted acidic
behavior, the compounds would behave like alcohols and
undergo deprotonation of the boranol (B−OH) unit to give a
boron oxy anion II that may preserve the ring’s pseudoar-
omatic character (Figure 2B).
No comprehensive studies exist to confirm the nature of the
acidity of pseudoaromatic hemiboronic heterocycles in
aqueous solutions, and without strong evidence such as X-
ray crystallographic structures of the conjugate base of
heterocycles 1−3 and 5−8, this question remains unanswered.
To exploit boron heterocycles to their full potential, it is
critical to gain a definitive understanding of their aromatic
character, their stability in water, and the acidic nature of the
boranol unit (Figure 2B), especially in aqueous media relevant
to biological applications. Herein, using benzoxaborine
compound 1 and N-alkyl/aryl benzodiazaborines 2a−c and
N-sulfonyl derivative 3a as models (Figure 1B), we address
these questions with a combination of X-ray crystallographic,
spectroscopic, and computational studies.
Figure 1. (A) Cyclic nonaromatic hemiboronic acid drugs. (B)
Classes of pseudoaromatic boron heterocycles featuring CC/B−X
isosterism, with model compounds 1, 2a−c, and 3a studied herein.
(C) Examples of biologically relevant naphthoid hemiboronic acid
heterocycles.
and the human neutrophil elastase (NHE) serine protease,16
whereas thiosemicarbazone derivatives demonstrated antifun-
gal activity (Figure 1C).17
Over the past 50 years, several important characteristics of
these compounds’ structural nature and properties have been
the subject of debate and confusion. Taking into account the
contribution of Bp−Xn resonance of the lone electron pairs
from the endocyclic O or N atom into boron’s vacant p orbital,
the planar heterocyclic ring contains six delocalized π electrons
RESULTS
■
Background. First prepared by Snyder and co-workers in
1958,21 the properties and reactivity of 1-hydroxy-1H-2,3,1-
benzoxazaborine (1), a borylated oxime, were further studied
by Dewar and Dougherty in 1962, along with the diaza
derivative 2a.22 Noting the compounds’ exceptional stability
through their resistance to strong aqueous acid and base,
and thus meets Huckel’s rule (Figure 2A). Although the boron-
̈
containing rings of compounds 1−3 and 5−8 are isoelectronic
and isosteric to phenol, claims pertaining to the extent of these
compounds’ aromatic character have varied widely. In part due
to their pseudoaromaticity, their acidic nature has also been
10144
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
conditions that would normally destroy boronic acids, the
authors concluded that the boron-containing ring of both 1
and 2a is aromatic. Moreover, by infrared spectroscopy, the
NH stretching band of solid 2a appears at a relatively low
frequency (3340 cm−1) that is supportive of Bp−Nn
resonance.23 When 1 was treated with ethanolamine (toluene,
Dean−Stark), Dewar also isolated a high-melting-point
product suspected to be the adduct 9 (Figure 3A) where any
Lewis base form I (cf. Figure 2B). With this new perspective of
11B NMR spectroscopy, Dewar concurred that 1 had been
mistakenly identified as a Brønsted acid (in previous UV
spectrophotometric studies23) and is instead a Lewis acid. In
contrast, 2a was now decisively characterized as being aromatic
and, to avoid the ionization of boron that would disrupt its
aromatic stabilization, a Brønsted acid.24 It is noteworthy that
no data exist of trivalent boron oxy reference compounds in
aqueous solution. Therefore, without such a comparator, the
absence of a large upfield shift cannot be firmly associated with
a Brønsted acidic behavior. The only available data for trivalent
borate oxy salts are in the solid state or in organic solvent, and
these species display chemical shifts around 15 ppm that are
significantly distinct from tetravalent boronyl anions.25,26
Subsequent to these NMR studies, Snyder and co-workers
reported the isolation of an adduct between 1 and catechol
with a composition corresponding to 10 (Figure 3A), in full
support of the Lewis acidic character and the inferable lack of
appreciable aromatic stabilization in the oxazaborine ring of
1.27
In 1997, benefiting from new methods, Groziak and co-
workers re-examined the acidic nature of heterocycles 1 and 2
as part of an elegant study to evaluate their suitability as
mimics of nucleic acids.28 Differences in the reactivity of the
compounds were assessed. The methyl ether of 1, derivative
11, was found to readily hydrolyze to 1 upon exposure to water
(Figure 3B). In contrast, 2a, 2b, and 3a resisted their
transformation into a B−OMe derivative when treated with
anhydrous methanol even under forcing conditions (65 °C, 24
h). Exchange of the boranol hydroxy group was also studied by
mass spectrometry in 18OH2. In contradiction of the outcome
of MeOH exchange experiments, diazaborines 2a and 2b were
found to exchange their OH group much more readily than
boryl oxime 1. Ignoring this dichotomy, the authors suggested
that the facile 18O incorporation into the diazaborines 2 is
indicative of their pronounced Lewis acid character compared
to 1. The compounds were also analyzed by NMR spectros-
copy in solution. The authors corroborated the observations of
Dewar,24 confirming that heterocycles 1 and 3a display a large
upfield shift in alkaline solution (2.2 ppm in aqueous NaOH).
No such upfield shift was observed for 2a and 2b; they both
displayed resonances of ∼26−27 ppm whether in neutral or
alkaline solution. Using potentiometric measurements, 1 was
found to be significantly more acidic (pKa 4.8) compared to
2a,b and 3a (pKa ∼8).28 Considering the 11B NMR resonance
of 2a and 2b provided no upfield shift in alkaline solutions,
such near-neutral pKa values are open to question. Surprisingly,
the authors concluded that 1 is a Brønsted acid and ascribed
the large upfield 11B NMR shift of its conjugate base (∼2 ppm)
to the oxy anion structure, 1-II, which was hypothesized to
provide such an upfield 11B resonance in spite of the lack of
data for this type of trivalent boron oxy salt in aqueous solution
(Figure 3C). This surprising assessment was deemed to best
account for the sluggish ability of 1 to incorporate 18OH2, as
this ability to exchange the hydroxy group on the boranol unit
was assumed to proceed through addition−elimination and full
ionization into the Lewis base 1-I that would temporarily
disrupt ring aromaticity (Figure 3D). Finally, to support their
claim that 2a and 2b also display Brønsted-type ionization
despite the insensitivity of their 11B NMR shift in alkaline
solutions, the authors proposed that little or no delocalization
of the oxyanion into boron can occur in the aza analogues due
to the likelihood of electrostatic repulsion with the less
Figure 3. (A) Putative adducts of 1 reported in the 1960s.23,27 (B−D)
Previous studies on the reactivity (B−O bond exchange) and
properties of heterocycles 1−3.
ring aromaticity is temporarily destroyed, and, because 2a
failed to form a similar adduct, concluded that the heterocyclic
ring of 1 is less aromatic than that of aza analogues 2.23 When
examined by UV spectroscopy, compound 1 displayed a small
bathochromic shift in alkaline solution. Because aromatic
boronic acids and benzoxaborole show a small hypsochromic
shift under similar conditions and were considered to be Lewis
acids, it was concluded that 1 and 2a must be Brønsted acids.23
The advent of commercial NMR spectroscopy allowed a new
and powerful means of examining the structure of the
conjugate base by observing their 11B chemicals shifts, which
tend to be highly diagnostic of the hybridization and valence of
boron. Whereas neutral trivalent boronic acid derivatives
display 11B chemical shifts in the range of 25−30 ppm, their
anionic tetrahedral adducts with Lewis bases generally appear
drastically upfield between 0 and 5 ppm. Thus, in 1967, Dewar
and Jones reported the 11B NMR analysis of 1 and 2a under
neutral (EtOH) and alkaline conditions (20% KOH in
EtOH).24 The resulting chemical shift of 5.0 ppm for the
potassium salt of 1, when compared to the shift of 30.0 ppm
for its neutral form in EtOH, is supportive of a tetravalent
10145
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
electronegative N atom, which was thought to preclude the
accumulation of charge on boron that would create an upfield
shift (Figure 3C).
Mass spectrometric analysis of these compounds tends to
provide B−O−B dehydration dimers; however, this phenom-
enon is typically observed in the gas phase or in anhydrous
organic solvents.31 Formation of these species is kinetically and
entropically unfavorable in the presence of water and should
not interfere in pKa measurements and other structural studies
in aqueous−organic media. For example, the dimer of 2c,
whose structure was resolved by X-ray diffraction analysis (see
Figure S3), was found to break down in a CD3CN solution
upon addition of a small amount of water (see Figure S2).
Behavior and Stability in Hydroxylic Solvents. The
possibility for heterocycles 1−3 to exchange their boranol
hydroxy group in aqueous medium may provide more
information about the nature of their acidic character and
the extent of their aromatic stabilization. We first set out to
reexamine 18OH2 exchange experiments described by Groziak
and co-workers with heterocycles 1 and 2a,b28 as well as
extending these investigations to model compounds 2c and 3a.
The heterocycles, dissolved in anhydrous CH3CN, were
exposed to 18OH2 for several hours and analyzed by ESI-
TOF mass spectrometry. Due to the large imprecision caused
by the occurrence of reverse hydrolysis from adventitious
16OH2 in the chamber, these experiments were found to be
unreliable. Similar issues may have affected the interpretation
of exchange experiments performed by Groziak and co-
workers. Instead, we examined the ability of hemiboronic
acids 1−3 to undergo methanolysis to form the corresponding
methyl hemiboronic esters. To this end, the model hetero-
cycles were incubated in dry methanol for 0.5−24 h at ambient
temperature, and the extent of BOH-to-BOCH3 exchange was
Although they reported X-ray crystallographic studies
showing that the acid forms of 1 and 2 can engage in H-
bonds in the solid state (with additional evidence by solution
NMR studies), Groziak and co-workers did not report crystal
structures of any conjugate bases yet concluded that “the B−
OH moiety in both 1 and 2 is characterized by a predominant
Brønsted acidity”.28 Recently, others have studied N-aryl29,30
and N-acyl31 derivatives of benzodiazaborines 2 for their
stability in aqueous solutions, but the question of their
conjugate base was not examined in detail.
Synthesis of Model Heterocycles and Spectroscopic
Characteristics (NMR, MS). Model heterocycles 1, 2a−c, and
3a were prepared according to literature procedures by
condensation of the ortho-formyl arylboronic acid and
hydroxylamine or hydrazines (see Supporting Information).
1
Key H and 11B NMR resonances in acetone-d6 are shown in
Figure 4 along with the corresponding resonances for
1
analyzed by H NMR spectroscopy. Additionally, heterocycles
1−3 were dissolved in CD3CN along with 5 equiv of dry
methanol to monitor the extent of exchange in a polar aprotic
solvent (Figure 5).
Figure 4. Select NMR spectroscopic data of heterocycles 1, 2a−c, and
3a with comparators (all shifts in ppm; solvent: acetone-d6).
comparator compounds. Although the 11B NMR resonances
show nearly no variance, the boranol proton resonances span a
range of ∼1 ppm, with 1 being the most deshielded and 2a the
least. Upon addition of a drop of D2O, a precaution to
suppress the formation of B−O−B anhydrides, the boranol
resonance disappears, thus confirming the exchangeability of
these protons. In the case of heterocycle 2a, only ∼20% of the
NH proton exchanged. In agreement with this reluctance to
exchange, previous 15N NMR studies of 2a likened this NH
moiety to that of a secondary amide.28
The bivalent aldimine-like nitrogen atom of heterocycles 1−
3 is not expected to be protonated in neutral aqueous
solutions.32 In support of this assertion, the chemical shift of
the aldimine proton was found to be insensitive to an increase
of pH. The aldimine resonance of all model compounds is
significantly lower than that of the CC/B−X isostere, 4-
hydroxyquinoline (4); however it is more downfield compared
to the acyclic oxime and hydrazone comparators. The effect is
larger for 2a and 2b, but it is unclear whether these small
differences can be attributed to the electron-withdrawing effect
of boron or to the possible occurrence of ring anisotropy.
Figure 5. Exchange of heterocycles 1, 2a−c, and 3a with methanol.
Under these conditions, heterocycles 1 and 2a showed
moderate boranol exchange to the corresponding methyl
esters. The equilibrium between the corresponding acids and
esters was relatively unchanged after the first 30 min and
showed comparable exchange in either neat methanol or
acetonitrile. Addition of D2O to these mixtures was found to
promote rapid hydrolysis of the methyl esters of 1 and 2a to re-
form the corresponding B−OD hemiboronic acids. In contrast,
N-alkyl heterocycle 2b was inert to methanol exchange. Rather
than direct BOH-to-BOCH3 exchange, heterocycles 2c and 3a
behaved differently and appeared to undergo significant B−N
10146
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
bond cleavage in methanol, a process that was not reversed
upon addition of D2O (see Supporting Information).
In principle, in aqueous solution these boron heterocycles
can break down to their starting materials via hydrolysis of
both the internal B−X and CN bonds (Figure 6A). The fact
compounds 1, 2b, and 2c showed no significant crossover in
either medium.
These experiments were corroborated by another series of
crossover experiments between sets of heterocycles made with
different 2-formyl arylboronic acids and heteroatom compo-
nents (Figure 6C). These experiments were performed starting
from both directions. Consistent with the benzaldehyde
exchange experiments, crossover could be observed only
when two distinct N-sulfonyl heterocycles are employed in
dry methanol (Figure 6D and Supporting Information).
Overall, these experiments demonstrate the stability of
heterocycles 1 and 2a−c in aqueous media, thus confirming
their suitability for applications in medicinal chemistry and
chemical biology.
pKa Measurements. The pKa values of all model
heterocycles were measured in buffered 1:1 H2O/CH3CN
solvent mixtures by way of 11B NMR titration, a method that
has proven reliable for similar hemiboronic and boronic
acids.6,33,34 Use of mixed aqueous organic solvent was
necessary to ensure all model heterocycles provide homoge-
neous solutions across a wide range of pH. The pKa’s of 1,
2c,d, and 3a were also validated by the UV method and
provided similar values (see Supporting Information).
Benzoxaborole was chosen as a standard because its pKa in
water was previously reported using two methods (11B NMR
spectroscopy and UV spectrophotometry).33,35 As they are
slightly soluble in water, the pKa’s of benzoxaborole and oxime
1 were also measured in water to provide an estimate of the
effect of the cosolvent. The resulting values point to an
increase of ca. 1.5−2.0 units in the mixed 1:1 H2O/CH3CN
system compared to water alone (Table 1).
Table 1. 11B NMR Chemical Shifts and pKa of Heterocycles
a
1, 2a−d, and 3a in Aqueous Media
Figure 6. (A) Possible hydrolytic equilibrium between compounds
1−3 and their o-formyl arylboronic acid precursor. (B−D) Behavior
and stability of compounds 1, 2a−c, and 3a via exchange and
crossover experiments in hydroxylic solvents (see Supporting
Information for full details).
compound
benzoxaborole
1
2a
2b
δ acid (ppm) δ conj. base (ppm)
pKa
b
32.8
27.7
27.8
27.5
28.0
28.8
28.4
8.5
2.5
9.2
c
7.1
d
−
>14
>14
12.2
9.4
d
_
2c
1.2
1.6
1.6
2d (R = C6H4-4-SO2Me)
3a
Measured by 11B NMR titration in 1:1 H2O/CH3CN (phosphate
that these components are not observed by NMR spectroscopy
cannot rule out a dynamic equilibrium that strongly favors the
heterocyclic products. Therefore, to assess their stability in
water, a series of crossover experiments were set up between
each of the five compounds 1, 2a−c, and 3a and a distinct
benzaldehyde containing an ortho-boronyl substituent, 12
(Figure 6B). These experiments, which were performed from
both directions in either dry methanol or H2O/CH3CN for 24
h, revealed that the N-sulfonylated derivative 3a is the most
prone to exchange in either solvent medium. The labile nature
of 3a is consistent with observations by Bane and co-workers
on similar N-acyl analogues, which were shown to be in
equilibrium with their open form in aqueous solution.31
Interestingly, 3a demonstrated less crossover in aqueous
acetonitrile compared to methanol, where virtually complete
equilibration was observed. Use of buffered H2O/CH3CN (pH
6.9) did not change this outcome, and additional control
experiments confirmed the distinct ability of methanol to
promote the breakdown of N-sulfonyl heterocycles (see
Discussion and Supporting Information). Hydrazine-derived
heterocycle 2a showed a small amount of crossover in
methanol but was unreactive in acetonitrile/water, whereas
5.5
a
b
c
buffer, referenced to CD3CN). 7.4 in H2O with 3−4% DMSO. 5.5
d
in H2O with 3−4% DMSO. A negligible change in the chemical shift
was observed up to a very high pH: 2a: 27.0 ppm at pH 13.4; 2b: 27.1
ppm at pH 13.6.
Titration of 1 and 3a afforded well-defined curves
transitioning from a downfield resonance at low pH to the
expected upfield resonance of the corresponding conjugate
base (see Table 1 and Figures S68, S70). Because 3a is slightly
labile in aqueous solvent (cf. Figure 6), its aldehyde precursor,
o-formylphenylboronic acid (pKa 7.3, water),36 was examined
at high pH and it yielded a distinct 11B NMR resonance,
confirming that the measured pKa is truly that of 3a (see Figure
S79). As noted by Groziak,28 even at pH 13.6 no downfield
shifts were observed for 2a and 2b that could be
unambiguously attributed to the conjugate base. In our
hands, back-titration of a high-pH sample to a neutral pH
and analysis by both 1H and 11B NMR spectroscopy
reproduced these findings and confirmed that 2a and 2b
10147
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
sustained these conditions unchanged. Consequently, the pKa
of 2a and 2b cannot be determined in aqueous medium (H2O/
CH3CN, 1:1), and they were assessed a value of >14. The H
the imino nitrogen atoms, whereas the NH unit of 2a enables
dimerization via additional NH/BOH H-bonding interac-
tions.28 We successfully analyzed monocrystals of 2c and 3a
(see Supporting Information), and some particular features of
the resulting structures are worth mentioning. The N-phenyl
unit of 2c shows a tilt of ∼30° relative to the planar
borazaheterocycle, while the structure of 3a displays a short
(2.15 Å) hydrogen bond between the BOH unit and one of the
two sulfonamide oxygens.
No crystal structures of any of the conjugate bases of
heterocycles 1−3 had been reported to date. To form suitable
crystals of the conjugate bases, a modified procedure for the
preparation of potassium trihydroxy borate salts of boronic
acids was employed.40 The resulting potassium salts of
heterocycles 1−3 were subjected to various recrystallization
attempts. To our delight, X-ray-quality crystals were obtained
and successfully analyzed for the salts of 1, 2d, and 3a, which
unambiguously revealed their Lewis basic structure: form I.
ORTEP representations of 1-I, 2d-I, and 3a-I are shown in
Figure 7, clearly showing the tetravalent dihydroxy-substituted
1
NMR spectra of 2a,b show broader resonances in this solvent
mixture at high pH (pH > 12), and the UV absorption spectra
undergo some changes (see Supporting Information). Because
there is little variation in the 11B NMR resonance (<1 ppm),
this behavior may be attributable to the intractable
dimerization (reversible B−O−B anhydride formation)
observed with related heterocycles.37 Remarkably, it was
found possible to isolate the tetramethylammonium salt of
the conjugate bases of 2a and 2b as hygroscopic solids from
organic solvent (toluene/methanol),38 and these anionic
species display upfield 11B NMR chemical shifts (DMSO-d6:
1.3 ppm for 2a, 2.4 ppm for 2b), in full agreement with a Lewis
basic structure (i.e., 2a-I/2b-I) (see Supporting Information).
Consistent with their predicted high pKa, according to 1H and
11B NMR these anionic species cleanly revert to their acid form
2a and 2b when exposed to a pH 13.5 solution (H2O/CH3CN,
1:1) (see Figures S88, S89). Moreover, because the hydrazone
NH resonance of 2a-I is observable, it is implausible that
heterocycle 2a displays Brønsted acidity through deprotona-
tion of its NH group.
In the case of 2c, the observed high-pH transition appeared
to result from a slower exchange, as both resonances were
observed around the transition point, eventually becoming a
single upfield resonance above pH 13 (see Supporting
Information, Section 7.4). Addition of KOH to a solution of
2c in CD3CN also provided the same 11B NMR resonance.
When performed in toluene, the conjugate base crashes out
after 5 min of stirring, and the isolated solid was added to a pH
13.8 1:1 solution of H2O/CH3CN, providing a resonance at
1.3 ppm identical to that observed in the titration.
Furthermore, a back-titration followed by H and 11B NMR
1
analysis confirmed that the observed transition is a true,
reversible ionization equilibrium (see Figure S81). To further
support the validity of the pKa of 2c, another N-aryl derivative
was prepared with a strong electron-withdrawing substituent
that could depress the pKa. In the event, the 4-methylsulfonyl
derivative 2d provided a typical titration curve with a single
averaged resonance around a transition point corresponding to
a pKa of 9.4 (see Figure S69). A decrease of ∼2.5 units
compared to 2c is commensurate with the effect of para-
sulfonyl substitution on the pKa of phenol.39
X-ray Crystallographic Structural Analysis of Acids
and Conjugate Bases. Single-crystal X-ray diffraction
analysis can provide unequivocal insight on the solid-state
structure of the hemiboronic acids 1−3 and their correspond-
ing conjugate bases. Structural data such as the bond length of
key bonds can provide invaluable information on bond orders
and the extent of π electron delocalization in the neutral acid
species. X-ray crystal structures of heterocycles 1 and 2a were
reported and thoroughly discussed by Groziak and co-
workers.28 Salient details of these planar boron heterocycles
include bond lengths for the B−C, external B−O, and internal
B−X bonds of 1.533(6) and 1.530(3), 1.350(6) and 1.371(3),
and 1.388(6) and 1.432(3) Å, respectively, for 1 and 2a. Both
compounds display a planar trivalent boron atom, though 2a
shows significant distortion of the C−B−OH angle (128.1°, vs
122.5° for 1) and consequently also for the HO−B−N/O
angle (116.7°, vs 118.0° for 1) and for the C−B−N/O angle
(115.2°, vs 119.4° for 1). Both compounds show intermo-
lecular H-bonding motifs between the boranol hydrogen and
Figure 7. ORTEP representations of the Lewis conjugate bases of 1,
2d, and 3a: 1-I, 2d-I, and 3a-I. Note: the countercation and bound
solvent molecules were omitted for clarity.
boron atom. Moreover, 11B NMR analysis of isolated crystals
of 3a-I in H2O/CH3CN showed the same upfield resonance of
∼2 ppm obtained in the pKa titration, which provides strong
evidence that form I also represents the conjugate base
structure in solution. Compared to the crystal structure of 1, 1-
I shows significant lengthening of the C−B bond (1.533(6) to
1.613(3) Å) and especially of the internal B−O bond
(1.388(6) to 1.537(3) Å). Such a lengthening of the internal
B−O bond provides strong support for the olefinic character of
this bond in heterocycle 1, through extensive Bp−On
resonance. The structure of 3a-I shows a similar lengthening
of the C−B and B−N bonds (Å): 3a: C−B: 1.540(3), B−N:
1.461(3); 3a-I: C−B: 1.6084(18), B−N 1.6218(16).
It was not possible to form high-quality crystals of the
putative conjugate bases of 2a−c under the same reaction and
recrystallization conditions. We succeeded, however, in making
and resolving the structures of adducts of 1,3-diols under basic
conditions for these three heterocycles. The resulting single-
crystal X-ray diffraction structures of adducts 13−15 shown in
Figure 8 represent the first reported examples of derivatives of
2a−c with a tetravalent boron atom (11B NMR δ (CD3CN)
13: −0.3 ppm, 14: 1.0 ppm, 15: − 0.9 ppm). Like 1-I and 3a-I
(vide supra), a substantial lengthening of both C−B and B−N
bonds accompanies a change a rehybridization of boron from
10148
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
the endocyclic B−X bond was employed. For comparative
purposes, the corresponding CC isostere, 4-hydroxyisoqui-
noline (4), was also calculated. Some key properties are shown
in Table 2 along with a representation of limit resonance forms
with alternate single and double bonds. The calculated
structures are in excellent agreement with the crystal structures
of 1 and 2a described above. For example, key bond lengths for
both compounds such as the B−C, external B−O, and internal
B−X bonds differ by no more than 0.01 Å. Whereas key bond
angle values are identical within 1° for heterocycle 1, there is
substantial disagreement for 2a: C−B−OH: 128.1° vs 121.9°
(calcd); HO−B−N: 116.7° vs 123.3° (calcd); and for the C−
B−N angle: 115.2° vs 114.8° (calcd). The discrepancy is likely
due to the dimeric structure of 2a, in its crystal lattice, which
imposes substantial distortion of the orientation of the B−OH
bond to accommodate the geometry of this hydrogen-bonded
dimer. Such restrictions do not exist in the computed
monomer. It is also noteworthy that the calculated ground-
state structure of 3a displays the same internal BOH---(O)S
hydrogen bond observed in the crystallographic structure (vide
supra).
Figure 8. ORTEP representations of the tetravalent adducts 13−15 of
neopentyl glycol with 2a−c. Note: the tetrabutylammonium counter-
cation was omitted for clarity.
As the parent B−X/CC isostere, the aromatic naphthoid
compound 4 displays bond orders (Lowdin) that support the
existence of a highly delocalized structure represented by a
hybrid of both resonance limit forms 4A and 4B. For example,
all of the bond orders in the azacycle of 4 are fractional,
ranging between 1.23 and 1.76, which reflects a substantial
contribution from limit form 4B with alternating single and
double bonds. By comparison, bond orders of all B−X isosteres
1−3 hint at a relatively small contribution of resonance form B.
This observation is corroborated by the much shorter C1−C5
bond length in 4 compared to the crystallographic and
calculated C1−B bond lengths of 1−3, which are indicative of a
single-bond character. In contrast, the B−N bond orders for 2a
and 2b (1.323, 1.276) show that this bond is a reasonably good
isoelectronic mimic of the alkene bond of 4 (1.528). The B−O
sp2 to sp3 (Å), e.g., 2c: C−B: 1.551(2); B−N: 1.445(2); 14:
C−B: 1.6339(16); B−N: 1.5783(13). Unlike 14, and
consistent with the high pKa of 2a and 2b, adducts 13 and
15 are not stable in aqueous conditions and revert to their acid
form even upon exposure to a pH 13.5 H2O/CH3CN (1:1)
solution (see Table S2 and Figure S90).
Computational Studies. Density functional theory
(DFT) calculations of the ground-state equilibrium geometry
and molecular orbitals of model heterocycles 1, 2a−c, and 3a
were performed in the gas phase using the ωB97X-D
functional with the 6-31G* basis set.41 Because it is lower or
comparable in energy (see Supporting Information), the syn
conformers with the boranol O−H bond synperiplanar with
Table 2. Key Properties for DFT-Optimized Structures of Model Heterocycles 1−3 and the Parent CC/B−N Isostere, 4-
Hydroxyisoquinoline 4 (ωB97X-D/6-31G*, Gas Phase)
a
property
4 (B = C5, X = C4)
1
2a
2b
2c
3a
bond length C1−B (Å)
bond length B−OH (Å)
bond length B−X (Å)
bond length X−N (Å)
bond order C1−B
bond order B−OH
bond order B−X
1.422 (C1−C5)
1.357 (C5−OH)
1.375 (C5C4)
1.356 (C4−N)
1.231
1.280
1.528
1.461
1.765
1.273
0.323 (C5)
−0.085
−0.697
−0.104
−0.439
1.542
1.360
1.381
1.390
1.041
1.512
1.357
1.217
2.004
1.127
1.136
−0.390
−0.895
−0.569
−0.093
1.542
1.373
1.429
1.366
1.056
1.456
1.325
1.319
1.931
1.153
1.014
−0.372
−0.889
−0.678
−0.244
1.540
1.374
1.432
1.366
1.062
1.452
1.276
1.293
1.914
1.158
1.029
−0.371
−0.889
−0.497
−0.249
1.541
1.368
1.443
1.373
1.057
1.472
1.232
1.263
1.929
1.150
1.038
−0.372
−0.892
−0.498
−0.248
1.546
1.351
1.465
1.377
1.043
1.551
1.172
1.251
1.951
1.138
1.05
bond order X−N
bond order NC3
bond order C2−C3
atomic charge B
atomic charge C1
atomic charge OH
atomic charge X
atomic charge NC
−0.367
−0.894
−0.750
−0.266
a
Lowdin bond orders and charges.
10149
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Figure 9. Frontier molecular orbitals, energies, and electrostatic potential maps for model boron heterocycles 1−3 and the parent isostere 4
(ωB97X-D/6-31G*, gas phase). The syn conformer with the boranol O−H bond synperiplanar with the endocyclic B−O/N bond was employed.
bond of 1 is an even better model, as indicated by its higher
bond order (1.357) and its shorter bond length, which is nearly
identical to that of the C4C5 bond in 4. However, a number
of indicators present significant differences between the
borylated oxime 1 and its aza congeners 2. Whereas compound
1 shows nearly no extended conjugation (e.g., C1−B and C3
N bond orders of 1.04 and 2.00), the boraza derivatives show
increased delocalization of their π electrons, as indicated by the
higher C1−B and lower C3N bond orders (optimal for 2b
with 1.062 and 1.914, respectively). Owing to boron’s vacant p
orbital, resonance of the hydroxy group of heterocycles 1−3
leads to a substantially higher B−(OH) bond order compared
to the corresponding C5−(OH) bond of 4 (∼1.5 vs 1.28), and,
in all cases it is higher than the endocyclic B−X bond order.
Finally, it is noteworthy that the boron atom appears to be
significantly more electron-deficient in 1 compared to the aza
analogues 2, which corroborates its relatively lower pKa.
Considering the higher B−X bond order in 1, this is suggestive
that the electrophilicity of the boron atom is largely
attributable to the inductive sigma effect of the endocyclic,
highly electronegative O atom.
Ground-state geometries and molecular orbitals of the
corresponding conjugate bases I of 1−3 were also computed
(see Table S1). Lengths of key bonds for theoretical structures
1-I and 3a-I compare closely, within 0.005−0.015 Å, to those
from the above crystallographic structures. As observed in the
solid-state structures, compared to the respective acids the
computed structures of conjugate bases show a large increase
of the B−X bond lengths that is indicative of the loss of of Bp−
Xn resonance (see Table S3). Reaction energies were
computed for conjugate base formation using single-point
calculation of equilibrium geometries at a higher level of theory
(ωB97XV/6-311+G(2df,2p)[6-311G*]). A linear relationship
was observed between the pKa and the computed reaction
energies for N-alkyl/aryl heterocycles 2a−d, which suggests a
pKa around 14 for 2a and 2b (see Supporting Information).
Heterocycles 1 and 3a, which are electronically distinct from
2a−d, did not provide a good fit in this correlation.
Calculated frontier molecular orbitals and electrostatic
potential maps of compounds 1, 2a−c, and 3a are shown in
Figure 9. Unlike isoquinoline 4 and compounds 2a−c, the
electrostatic potential maps of 1 and 3a depict no or little ring
current in the heterocyclic ring. Moreover, the shapes of
molecular orbitals (MOs) of all model heterocycles are
significantly different compared to 4, whose MOs show good
symmetry across both rings. The highest occupied molecular
orbital (HOMO) of 2a and 2b, however, shows partial
conjugation of the B−N bond with the benzenoid ring as
depicted with larger density in the C1−B bond. In contrast, the
HOMO of compounds 1, 2c, and 3a, which contain a less basic
O- or N-aryl/sulfonyl atom, is interrupted between the C1 and
B atoms. Calculated MO energies revealed that both 1 and 3a
have significantly lower-lying lowest unoccupied molecular
orbitals (LUMOs) than the other boron heterocycles, which is
consistent with their higher acidity.
Based on the above crystallographic structures and
calculations, boron heterocycles 1−3 appear to be valid
isoelectronic and isosteric analogues of isoquinoline 4.
However, based on calculated bond orders and MOs, only
the boraza compounds 2a−c display a small degree of
extended conjugation across rings, represented by resonance
limit form B, that is suggestive of a partial aromatic character.
To better assess the aromatic character of heterocycles 1−3,
nucleus independent chemical shift (NICS) calculations were
performed.42 For both NICS(0) and NICS(1) indexes,
negative values are indicative of aromatic character; however,
the former computes the average shielding at the center of the
ring and includes the contribution of the σ-framework,
including substituents, whereas the latter measures shielding
1 Å above the ring where π contributions are dominant. The
resulting NICS(0) and NICS(1) values for the boroheter-
ocyclic ring (B) of compounds 1−3 are significantly lower than
10150
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
that of the corresponding ring in isoquinoline 4 (Table 3), and,
while highest in 2a and 2b, they are substantially lower than
NMR titrations. The hemiboronic oxime 1 has a measured pKa
of 5.5 (in water), which is significantly lower than that of a
normal boronic acid (8−9) and even that of benzoxaborole
(7.4). Being nonaromatic, compared to 2a−c there is little
penalty for its ionization into a conjugate base that would
break π-conjugation. Conversely, owing to their partial
aromatic character, the boraza heterocycles 2a−c have pKa’s
well above that of boronic acids. For reasons mentioned above,
the N-sulfonyl derivative 3a behaves much like heterocycle 1.
When measured in 1:1 H2O/CH3CN, it displays a lower pKa
than 1 (5.5 vs 7.1). Collectively, the measured order of pKa’s,
3a < 1 < 2c < 2a,b, correlates with the anion-stabilizing ability
of the X group on the B−X unit that is consistent with the
pKa’s of the corresponding protic acids: H2NSO2Ph (pKa 10) <
H2O (pKa 15.7) < H2NPh (pKa 25) < H2NR (pKa ∼40).
Table 3. Nucleus-Independent Chemical Shift Calculations
for Heterocycles 1−3 and 4 (GIAO-B3LYP/6-
311+G(2d,p))
ring
4
1
2a
2b
2c
3a
A
NICS(0)
NICS(1)
B
−7.95
−7.41
−7.67
−7.66
−7.47
−7.59
−10.28 −10.25 −10.47 −10.45 −10.23 −10.46
1
Whereas 11B NMR is of no help in predicting acidity, the H
B−OH resonances of heterocycles 1 and 2 (cf. Figure 4)
match well with the above rankings, except for 3a, which is
biased by an internal H-bond (vide supra). The DFT-
computed ground-state structures can also help rationalize
the pKa’s of heterocycles 1−3. In this regard, the calculated
positive charge on boron is far greater in 1 than 2a−c; however
it is also much higher than 3a despite their similarly low pKa’s
(cf. Table 2). LUMO energy levels appear to correlate better
with the measured acidity. Indeed, heterocycles 1 and 3a
display similar LUMO energy levels that are significantly lower
than that of 2a−c (cf. Figure 9). Likewise, the HOMO energy
level of the corresponding Lewis conjugate bases correlates
neatly with their expected order of basicity based on the
measured pKa’s of their acid forms (see Table S3).
This study also puts to rest the question of the structure of
the conjugate base of heterocycles 1−3 that remained
unsettled for several decades. All of the heterocycles studied
here were confidently determined to act as Lewis acids in
aqueous solutions to afford a conjugate base of structural form
I characterized by a tetravalent dihydroxylated boron atom.
The results of 11B NMR titrations in alkaline solutions are
unequivocal; the presence of upfield species at 0−5 ppm
cannot be attributed to a Brønsted basic form II. Likewise, this
study corrects previous conclusions made in the literature and
confirms that the absence of such upfield resonances with N-
alkyl heterocycles 2a,b at high aqueous pH are simply due to
pKa values that are significantly above the normal range for
boronic acid derivatives. This conclusion is supported by the
relatively high pKa of 12.2 (1:1 H2O/CH3CN) measured for
the N-phenyl heterocycle 2c, which is expected to lie a few
units lower than the N-alkyl derivatives 2a,b. Moreover,
conjugate bases 2a-I and 2b-I are observable by NMR
spectroscopy, as their tetramethylammonium salts in dry
DMSO, and they predictably revert to their acid form even in
pH 13.5 aqueous conditions.
NICS(0)
NICS(1)
−7.87
−9.95
0.79
−2.75
−1.33
−4.35
−1.91
−4.86
−0.81
−3.75
−1.34
−3.88
that of 1,2-azaborine and 1,2-oxaborine.43 Together the DFT
and NICS calculations support the idea that the oxazaborine
ring of 1 possesses nearly no aromatic character, while 2a, 2b,
and 2c retain to a small degree the ring aromaticity of the
parent CC isostere, isoquinoline 4.
DISCUSSION
■
The aromatic character, the acidic nature (Brønsted vs Lewis),
and the pKa of individual hemiboronic heterocycles 1−3 are
strongly linked (cf. Table 1). Whether their conjugate base is
represented by form I or II, ionization is expected to affect the
electron distribution around the pseudoaromatic heterocycle.
Empirical observations based on chemical stability and UV
spectrophotometry made in decades-old pioneering studies can
provide a misleading assessment of the aromatic character of
these boron heterocycles. The analysis of aromatic stabilization
in pseudoaromatic heterocycles is a notorious challenge that
often requires a multipronged approach. The combination of
DFT and NICS calculations presented herein converges into a
reasonably consistent picture. Thus, according to calculated
bond orders, electrostatic potential maps, the shape of their
HOMO, and NICS values, only the boraza heterocycles 2a and
2b retain a significant degree of the aromaticity of the parent
isoquinoline B−X/CC isostere 4. A comparison of the five
model heterocycles highlights the importance of the availability
of lone electron pairs on the O or N heteroatom. According to
estimated bond orders and the values of bond lengths obtained
experimentally and by DFT calculations, the endocyclic B−O
unit of heterocycle 1 serves as an effective isoelectronic and
isosteric mimic of a CC bond, one that is slightly better than
the B−N bond in 2a and 2b. However, likely because of its
high electronegativity, the O atom does not appear to allow an
effective delocalization of its π electrons within the rest of the
pseudoaromatic system. As a result, there is little contribution
from the extended resonance form B (cf. Table 2) necessary
for developing significant aromatic character. The superior
ability of a nitrogen vs an oxygen atom to participate in
pseudoaromatic delocalization was also established in other
boron heterocycles.43,44 In this regard, compared to 2a−c, the
boraza heterocycle 3a contains a sulfonyl-conjugated nitrogen
lone electron pair that is relatively less available to delocalize
into the diazaboryl heterocycle. This qualitative assessment is
corroborated by the pKa measurements obtained from 11B
The structures of conjugate bases 1-I, 2d-I, and 3a-I
obtained by X-ray crystallography, which were shown by 11B
NMR to exist also in aqueous solution, provide indisputable
evidence of Lewis acidity. No such crystal structures could be
obtained for 2a−c; indeed difficulties in crystallizing these
conjugate bases underpins their higher pKa. On the other hand,
due to the entropic advantage of intramolecularity, anionic
adducts with diols form more easily and were successfully
crystallized. Thus, the structure of dialkoxy adducts 13−15 of
2a−c confirms that the partial aromatic character of hetero-
cycles 2 can be disrupted to form an anionic tetravalent
dialkoxyboronyl species consistent with Lewis acidic behavior.
10151
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
The reactivity of heterocycles 1, 2b, and 3a with methanol is
consistent with the higher Lewis acidity of 1 and 3a and the
partial aromaticity of 2b. The contradictory result of 2a, which
exchanges its hydroxy group with methanol despite showing a
high pKa and computed aromaticity similar with 2b, may be
rationalized without invoking the association−dissociation
mechanism and the corresponding anionic tetrahedral boron
intermediate. It can be proposed that H-bonding participation
of the N−H bond of 2a could help mediate a concerted
exchange of the B−O bonds without a formal ionization of the
boron atom (Figure 10A). This type of proton-transfer
alone does not predict the behavior or stability of these
hemiboronic acids in methanol.
While this study answers long-standing fundamentally
important questions, it raises new ones too. Even when
considering their lack of aromatic stabilization, it is surprising
that benzoxaborine 1 and 3a display such unusually low pKa’s
(high acidity), and its origin is intriguing. Both compounds
have pKa’s lower than that of benzoxaborole. Relief of angle
strain from the rehybridization of boron from sp2 to sp3, a
factor that was invoked to explain the relatively low pKa of
benzoxaborole,6 is not dominant with 1. Compared to
benzoxaborole and acyclic oxime and hydrazone analogues,
bond angles around the heterocyclic ring of 1 do not appear to
suffer from large deviations. A more plausible explanation
entails that when freed of its double-bond character with
boron, the internal O atom of conjugate base 1-I can benefit
from the stabilization of its partial negative charge by the
inductive effect of the neighboring N atom and through
resonance with the CN unit. As related above, this view is
consistent with the difference of protic acidity between
alcohols (pKa ∼16) and benzophenone oxime (pKa ∼11). As
demonstrated in a recent report by Raines and co-workers,
stereoelectronic factors can play an important role in the
reactivity and stability of tetravalent boronate intermediates.46
Thus, it is also possible that exo nO−σ*B−O/N and endo
nO/N−σ*B−OH interactions contribute to the differences in
stability between conjugate bases I. Specifically, charge
stabilization via donation of antiperiplanar lone pairs from
the B(OH)2 unit is expected to be more important with the
lower-energy σ* of the electron-poor endocyclic B−O bond of
1-I compared to the more electron-rich B−N bond in 2a-I. To
assess the potential contribution of secondary orbital
interactions to the relative stability of 1-I and 2a-I, we
performed a natural bonding orbital (NBO) analysis.47 Taking
into account the exo nO−σ*B−O/N overlap from both exocyclic
oxygen atoms on the optimized geometries (ωB97XD/6-
31G(d)), second-order perturbation energies (E2) are more
beneficial in 1-I by 2.6 kcal/mol (Figure 11 and Supporting
Information for additional data). However, the larger exo
nO−σ*B−O interactions are counterbalanced by the stronger
donor effect of N (despite having only a single lone pair) in the
endo nN-σ*B−OH interactions (1.5 kcal/mol favoring 2a-I),
which results in no net advantage for either structure (only 1.1
kcal/mol favoring 1-I). Thus although stabilizing secondary
orbital interactions do exist in these boronate conjugate bases,
they are unlikely to exhibit a significant influence on the
relative acidity of heterocycles 1−3.
Figure 10. (A) Proposed rationalization for the behavior of
heterocycles 2a in methanol and (B) for the different reactivity of
3a with water and methanol.
mechanism was shown to be kinetically plausible in the related
esterification of high-pKa boronic acids with aliphatic diols at
neutral pH.45 In contrast, N-methylated derivative 2b lacks the
requisite H-bond donor to promote a concerted mechanism
and is unlikely to undergo a stepwise mechanism through an
anionic tetrahedral boron intermediate due to its high pKa. The
higher propensity of 3a to break down in methanol, as seen in
both the methanol exchange (Figure 5) and aldehyde
crossover experiments (Figure 6), may originate from the
strong internal B−OH---OS hydrogen bond observed in the
solid state of both 3a and its conjugate base. In aqueous
solution, this H-bond is maintained in both the trigonal and
tetrahedral boron environments, bestowing increased stability
to heterocycle 3a. A transient exchange with methanol,
facilitated by the relatively high acidity of 3a, eliminates this
stabilizing internal H-bond and would facilitate hydrolysis of
the B−N bond (Figure 10B). While heterocycles 2c and 3a
both appear to undergo B−N bond cleavage in methanol, the
increased crossover observed with 3a may be due to its lower
pKa and the stronger electron-withdrawing effect of the
sulfonyl moiety that promotes the prerequisite hydrazone
CN bond cleavage. Altogether, the contrasting results of
heterocycles 1, 2a−c, and 3a in methanol exchange and
aldehyde crossover experiments clearly indicate that acidity
Overall, the substantially higher Lewis acidity of heterocycles
1 and 3a compared to 2a−c can be rationalized by a
combination of anion-stabilizing inductive effect from the B−X
heteroatom and its substituent in conjugate bases 1-I and 3a-I
and the larger loss of aromatic stabilization energy in 2a−c
relative to 1 and 3a.
The relatively low aromatic character of naphthoids 1−3 is
equally intriguing. Based on their pKa and computational data,
heterocycles 2a and 2b appear to be the most resonance-
stabilized among all five model compounds, yet they retain
only to a small degree the aromatic character of the parent
isoquinoline 4. This determination does not seem to arise from
a deficiency in the π bond character of the B−X bond, which
shows good mimicry of an alkene. Rather, it is likely to be
caused by the poor ability of Csp2−Csp2 π bonds to delocalize
into boron. Even in the most favorable case of 2a,b, a C−B
10152
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Figure 12. (A) Comparison of energies for tetravalent boronate form
I and boron oxy form II (ωB97X-D/6-31G*, water C-PCM
continuum solvent model, values not entropy-corrected). (B)
Rationalization of the preference for Lewis bases I.
This preference is reduced by roughly 6 kcal/mol in the case of
compound 2a. This difference may be attributable to the small
aromatic character of this heterocycle, which disfavors
structure I yet clearly not enough to favor the boron oxy
form II. It is noteworthy that the amido Brønsted base from
deprotonation of the NH group was found to be even less
stable than 2a-II according to DFT calculations (see
Supporting Information). Although the calculated B−O bond
order values of ∼2.0 are supportive of substantial charge
stabilization via resonance form II-B (Figure 12B), a simple
explanation to rationalize the preference of structure I resides
in the thermodynamic advantage of an additional σ B−O bond
over a π B−O bond and the sharing of the negative charge
between three electronegative heteroatoms instead of just two
such atoms in form II.
Figure 11. Stereoelectronic stabilization of conjugate bases assessed
by NBO analysis of key overlapping orbitals in 1-I and 2a-I with
second-order perturbation energies (E2) and the “pre-orthogonal”
NBO (PNBO) overlap matrix integral (S) (Oa: front-facing oxygen,
Ob: back oxygen).
CONCLUSION
■
This comprehensive study of naphthoid hemiboronic acid
heterocycles 1−3 rectifies a number of incorrect conclusions
and ambiguous results reported in the literature regarding the
acidic nature and aromatic character of these B−X/CC
isosteric compounds (Figure 13). The evidence obtained
herein, using a combination of experimental, spectroscopic, X-
ray crystallographic, and computational studies of model
boranol-containing benzoxa- and benzodiazaborines confirms
the Lewis acidic nature of these compounds in aqueous
conditions. Particularly compelling is the X-ray crystallographic
elucidation of several conjugate bases or tetrahedral adducts
that clearly highlights their Lewis basic structure and the
resulting disruption of electronic delocalization within the
boroheterocyclic ring. Unlike the highly acidic benzoxaborine
1 and N-sulfonyl benzodiazaborine 3a, which display no
evidence of heterocyclic ring aromaticity, boraza derivatives
bond order of 1.06 does not compare well with the
corresponding C−C bond order of 1.23 in isoquinoline 4,
which is reflective of a significant contribution from resonance
forms with alternating single and double bonds (cf. B, Table
2). Finally, although this study firmly establishes the large
preference for tetravalent conjugate bases (form I), the reason
for its preference over the boron oxy form (II) is still lingering.
It can be addressed in a preliminary manner through DFT
calculations in a water solvation model that compares
compounds 1, 2a, and 16 with the isoatomic compensation
of a water molecule. The benzoxaborine 16 was selected for its
nonaromatic character and its unstrained structure. Although
this estimation ignores entropic contributions (expected to
favor form II), the huge preference for the tetravalent boronate
anion form I is evident in all three examples (Figure 12A).
10153
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
AUTHOR INFORMATION
■
Corresponding Author
Dennis G. Hall − Department of Chemistry, University of
Alberta, Centennial Centre for Interdisciplinary Science,
Edmonton, AB T6G 2G2, Canada; orcid.org/0000-
0001-8555-6400; Phone: 1-780-492-3141;
Email: dennis.hall@ualberta.ca
Authors
M. Zain H. Kazmi − Department of Chemistry, University of
Alberta, Centennial Centre for Interdisciplinary Science,
Edmonton, AB T6G 2G2, Canada
Jason P. G. Rygus − Department of Chemistry, University of
Alberta, Centennial Centre for Interdisciplinary Science,
Edmonton, AB T6G 2G2, Canada
Hwee Ting Ang − Department of Chemistry, University of
Alberta, Centennial Centre for Interdisciplinary Science,
Edmonton, AB T6G 2G2, Canada
Marco Paladino − Department of Chemistry, University of
Alberta, Centennial Centre for Interdisciplinary Science,
Edmonton, AB T6G 2G2, Canada
Matthew A. Johnson − Department of Chemistry, University
of Alberta, Centennial Centre for Interdisciplinary Science,
Edmonton, AB T6G 2G2, Canada
Figure 13. Summary of key conclusions for model naphthoid
compounds 1, 2a−c, and 3a.
Michael J. Ferguson − X-ray Crystallography Laboratory,
Department of Chemistry, University of Alberta, Edmonton,
AB T6G 2G2, Canada; orcid.org/0000-0002-5221-4401
2a−c with a basic nitrogen atom demonstrate an unusually
high pKa for a class of hemiboronic acids that is in agreement
with a small aromatic character. As demonstrated by dynamic
exchange and crossover experiments, most of these naphthoid
heterocycles are stable in neutral aqueous medium, and they
offer different and predictable boron configurations (ionized
tetravalent or neutral trivalent) at biological pH that can be
useful in drug discovery applications. Moreover, heterocycles 1
and 2a can undergo reversible exchange of their boranol group
with alcohols that makes these scaffolds attractive in the design
of covalent ligands for biomolecules or for catalyzing reactions
of alcohol substrates. On the other hand, the stability of N-
alkyl and N-aryl scaffolds represented by 2b,c along with the
ability to modify their nitrogen atom substituent is appealing
for applications in bioconjugation. Altogether not only does
this study resolve inaccuracies and ambiguities reported in the
literature in the past five decades, it also provides new
conclusions that are important toward the methodical
application of boron heterocycles in catalysis, medicinal
chemistry, and chemical biology.
Complete contact information is available at:
https://pubs.acs.org/10.1021/jacs.1c02462
Author Contributions
#M.Z.H.K. and J.P.G.R. contributed equally to the work.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was funded by the Natural Sciences and Engineering
Research Council (NSERC) of Canada (Discovery Grant to
D.G.H.). J.P.G.R. thanks NSERC for a PGS-D scholarship.
M.Z.H.K., J.P.G.R., and H.T.A. hold Alberta Graduate
Excellence Scholarships. We thank Mr. Bela Reiz for help
with mass spectrometry experiments.
REFERENCES
■
(1) Wang, B. J.; Groziak, M. P. Recent Advances in the Chemistry of
Boron Heterocycles. Adv. Heterocycl. Chem. 2016, 118, 47−90.
(2) Giustra, X.; Liu, S.-Y. The State of the Art in Azaborine
Chemistry: New Synthetic Methods and Applications. J. Am. Chem.
Soc. 2018, 140, 1184−1194.
ASSOCIATED CONTENT
■
sı
* Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/jacs.1c02462.
(3) Wang, X.-Y.; Wang, J.-Y.; Pei, J. BN Heterosuperbenzenes:
Synthesis and Properties. Chem. - Eur. J. 2015, 21, 3528−3539.
(4) Santos Fernandes, G. F.; Denny, W. A.; Dos Santos, J. L. Boron
in Drug Design: Recent Advances in the Development of New
Therapeutic Agents. Eur. J. Med. Chem. 2019, 179, 791−804.
(5) Baker, S. J.; Ding, C. Z.; Akama, T.; Zhang, Y.-K.; Hernandez,
V.; Xia, Y. Therapeutic potential of boron-containing compounds.
Future Med. Chem. 2009, 1, 1275−1288.
(6) Dowlut, M.; Hall, D. G. An Improved Class of Sugar-Binding
Boronic Acids, Soluble and Capable of Complexing Glycosides in
Neutral Water. J. Am. Chem. Soc. 2006, 128, 4226−4227.
(7) Baker, S. J.; Zhang, Y.-K.; Akama, T.; Lau, A.; Zhou, H.;
Hernandez, V.; Mao, M.; Alley, M. R. K.; Sanders, V.; Plattner, J. J.
Discovery of a New Boron-containing Antifungal Agent, 5-Fluoro-1,3-
Experimental details, analytical data, and spectral
reproductions for all new compounds; details of all
mechanistic and computational studies (PDF)
Accession Codes
CCDC 2065164−2065171 and 2081825 contain the supple-
mentary crystallographic data for this paper. These data can be
obtained free of charge via www.ccdc.cam.ac.uk/data_request/
cif, or by emailing data_request@ccdc.cam.ac.uk, or by
contacting The Cambridge Crystallographic Data Centre, 12
Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033.
10154
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
dihydro-1-hydroxy-2,1-benzoxaborole (AN2690), for the Potential
Treatment of Onychomycosis. J. Med. Chem. 2006, 49, 4447−4450.
(8) Akama, T.; Baker, S. J.; Zhang, Y.-K.; Hernandez, V.; Zhou, H.;
Sanders, V.; Freund, Y.; Kimura, R.; Maples, K. R.; Plattner, J. J.
Discovery and Structure-Activity Study of a Novel Benzoxaborole
Anti-Inflammatory Agent (AN2728) for the Potential Topical
Treatment of Psoriasis and Atopic Dermatitis. Bioorg. Med. Chem.
Lett. 2009, 19, 2129−2132.
(9) Hecker, S. J.; Reddy, K. R.; Totrov, M.; Hirst, G. C.;
Lomovskaya, O.; Griffith, D. C.; King, P.; Tsivkovski, R.; Sun, D.;
Sabet, M.; Tarazi, Z.; Clifton, M. C.; Atkins, K.; Raymond, A.; Potts,
K. T.; Abendroth, J.; Boyer, S. H.; Loutit, J. S.; Morgan, E. E.; Durso,
S.; Dudley, M. N. Discovery of a Cyclic Boronic Acid β-Lactamase
Inhibitor (RPX7009) with Utility vs Class A Serine Carbapenemases.
J. Med. Chem. 2015, 58, 3682−3692.
(10) Zhao, P.; Nettleton, D. O.; Karki, R. G.; Zécri, F. J.; Liu, S.-Y.
Medicinal Chemistry Profiling of Monocyclic 1,2-Azaborines.
ChemMedChem 2017, 12, 358−361.
(11) Rombouts, F. J. R.; Tovar, F.; Austin, N.; Tresadern, G.;
Trabanco, A. A. Benzazaborinines as Novel Bioisosteric Replacements
of Naphthalene: Propranolol as an Example. J. Med. Chem. 2015, 58,
9287−9295.
(12) Davies, G. H. M.; Mukhtar, A.; Saeednia, B.; Sherafat, F.; Kelly,
C. B.; Molander, G. M. Azaborininones: Synthesis and Structural
Analysis of a Carbonyl-Containing Class of Azaborines. J. Org. Chem.
2017, 82, 5380−5390.
(13) Robinson, P. D.; Groziak, M. P. A boron-containing Estrogen
Mimic. Acta Crystallogr., Sect. C: Cryst. Struct. Commun. 1999, C55,
1701−1704.
18 Independent Molecules in Its Asymmetric Unit. Inorg. Chem. 2003,
42, 1140−1150.
(27) Dunn, H. E.; Catlin, J. C.; Snyder, H. R. Arylboronic Acids.
Imino Derivatives from o-Formylbenzeneboronic Acid. J. Org. Chem.
1968, 33, 4483−4486.
(28) Groziak, M. P.; Chen, L.; Yi, L.; Robinson, P. D. Planar Boron
Heterocycles with Nucleic Acid-Like Hydrogen-Bonding Motifs. J.
Am. Chem. Soc. 1997, 119, 7817−7826.
(29) Dilek, O.; Lei, Z.; Mukherjee, K.; Bane, S. Rapid Formation of a
Stable Boron−Nitrogen Heterocycle in Dilute, Neutral Aqueous
Solution for Bioorthogonal Coupling Reactions. Chem. Commun.
2015, 51, 16992−16995.
(30) Stress, C. J.; Schmidt, P. J.; Gillingham, D. G. Comparison of
Boron-Assisted Oxime and Hydrazone Formations Leads to the
Discovery of a Fluorogenic Variant. Org. Biomol. Chem. 2016, 14,
5529−5533.
(31) Gu, H.; Chio, T. I.; Lei, Z.; Staples, R. J.; Hirschi, J. S.; Bane, S.
Formation of Hydrazones and Stabilized Boron−Nitrogen Hetero-
cycles in Aqueous Solution from Carbohydrazides and Ortho
Formylphenylboronic Acids. Org. Biomol. Chem. 2017, 15, 7543−
7548.
(32) Kalia, J.; Raines, R. T. Hydrolytic Stability of Hydrazones and
Oximes. Angew. Chem., Int. Ed. 2008, 47, 7523−7526.
(33) Bérubé, M.; Dowlut, M.; Hall, D. G. Benzoboroxoles as
Efficient Glycopyranoside-Binding Agents in Physiological Condi-
tions: Structure and Selectivity of Complex Formation. J. Org. Chem.
2008, 73, 6471−6479.
(34) Bhangu, J.; Whittal, R. M.; Hall, D. G. Design, Synthesis and
Structure of a Frustrated Benzoxaborole and its Applications in the
Complexation of Amines, Amino Acids, and Protein Modification.
Org. Biomol. Chem. 2020, 18, 3492−3500.
(35) Tomsho, J. W.; Pal, A.; Hall, D. G.; Benkovic, S. J. Ring
Structure and Aromatic Substituent Effects on the pKa of the
Benzoxaborole Pharmacophore. ACS Med. Chem. Lett. 2012, 3, 48−
52.
(36) Kowalska, K.; Adamczyk-Wozniak; Gajowiec, P.; Gierczyk, B.;
Kaczorowska, E.; Popenda, L.; Schroeder, G.; Sikorski, A.; Sporzynski,
A. Fluoro-substituted 2-Formylphenylboronic acids: Structures,
Properties and Tautomeric Equilibria. J. Fluorine Chem. 2016, 187,
1−8.
(14) Baldock, C.; Rafferty, J. B.; Sedelnikova, S. E.; Baker, P. J.;
Stuitje, A. R.; Slabas, A. R.; Hawkes, T. R.; Rice, D. W. A Mechanism
of Drug Action Revealed by Structural Studies of Enoyl Reductase.
Science 1996, 274, 2107−2110.
(15) Baldock, C.; de Boer, G.-J.; Rafferty, J. B.; Stuitje, A. R.; Rice,
D. W. The Mechanism of Action of Diazaborines. Biochem. Pharmacol.
1998, 55, 1541−1550.
(16) António, J. P. M.; Gonca̧ lves, L. M.; Guedes, R. C.; Moreira, R.;
Gois, P. M. P. Diazaborines as New Inhibitors of Human Neutrophil
Elastase. ACS Omega 2018, 3, 7418−7423.
(17) Hicks, J. W.; Kylea, C. B.; Vogels, C. M.; Wheaton, S. L.;
Baerlocher, F. J.; Decken, A.; Westcott, S. A. Synthesis, Character-
ization, and Antifungal Activity of Boron-Containing. Chem.
Biodiversity 2008, 5, 2415−2422.
(37) Harris, K. D. M.; Kariuki, B. M.; Lambropoulos, C.; Philp, D.;
Robinson, J. M. A. A New Hydrogen Bonding Motif Based on 10-
Hydroxy-10,9-Borazaphenanthrene. Tetrahedron 1997, 53, 8599−
8612.
(38) Mikhailov, B. M.; Kuimova, M. E. Organoboron Compounds.
CCCXVIII. On the Problem of the Aromatic Character of Boron
Heterocyclic Compounds. J. Organomet. Chem. 1976, 116, 123−133.
(39) Zhang, S.; Baker, J.; Pulay, P. A Reliable and Efficient First
Principles-Based Method for Predicting pKa Values. 2. Organic Acids.
J. Phys. Chem. A 2010, 114, 432−442.
(18) Hall, D. G. In Boronic Acids: Preparation and Application in
Organic Synthesis, Medicine, and Materials; Hall, D. G., Ed.; Wiley-
VCH: Weinheim, Germany, 2011; Chapter 1, pp 1−133 and
references therein.
(19) Martínez-Aguirre, M. A.; Yatsimirsky, A. K. Brønsted versus
Lewis Acid Type Anion Recognition by Arylboronic Acids. J. Org.
Chem. 2015, 80, 4985−4993.
(20) Lorand, J. P.; Edwards, J. O. Polyol Complexes and Structure of
the Benzeneboronate Ion. J. Org. Chem. 1959, 24, 769−774.
(21) Snyder, H. R.; Reedy, A. J.; Lennarz, W. J. Synthesis of
Aromatic Boronic Acids. Aldehydo Boronic Acids and a Boronic Acid
Analog of Tyrosine. J. Am. Chem. Soc. 1958, 80, 835−838.
(22) Dewar, M. J. S.; Dougherty, R. C. Boron-Containing Analogs of
Isoquinoline. J. Am. Chem. Soc. 1962, 84, 2648−6249.
(40) Cammidge, A. N.; Goddard, V. H. M.; Gopee, H.; Harrison, N.
L.; Hughes, D. L.; Schubert, C. J.; Sutton, B. M.; Watts, G. L.;
Whitehead, A. J. Aryl Trihydroxyborates: Easily Isolated Discrete
Species Convenient for Direct Application in Coupling Reactions.
Org. Lett. 2006, 8, 4071−4074.
(41) Shao, Y.; et al. Advances in molecular quantum chemistry
contained in the Q-Chem 4 program package. Mol. Phys. 2015, 113,
184−215.
(23) Dewar, M. J. S.; Dougherty, R. C. New Heteroaromatic
Compounds. XX. Derivatives of 4,3-Borazaroisoquinoline. J. Am.
Chem. Soc. 1964, 86, 433−436.
(42) Chen, Z.; Wannere, C. S.; Corminboeuf, C.; Puchta, R.;
Schleyer, P. V. R. Nucleus-Independent Chemical Shifts (NICS) as an
Aromaticity Criterion. Chem. Rev. 2005, 105, 3842−3888.
(43) Yruegas, S.; Patterson, D. C.; Martin, C. D. Oxygen Insertion
Into Boroles as a Route to 1,2-Oxaborines. Chem. Commun. 2016, 52,
6658−6661.
(44) Chen, J.; Bajko, Z.; Kampf, J. W.; Ashe, A. J., III. 1,2-Dihydro-
1,2-oxaborine: A Boron-Oxygen Heterocycle Isoelectronic with
Benzene. Organometallics 2007, 26, 1563−1564.
(45) Eisuke Watanabe, E.; Miyamoto, C.; Tanaka, A.; Iizuka, K.;
Iwatsuki, S.; Inamo, M.; Takagi, H. D.; Ishihara, K. Relative Kinetic
(24) Dewar, M. J. S.; Jones, R. New Heteroaromatic Compounds.
XXV. Studies of Salt Formation in Boron Oxyacids by 11B Nuclear
Magnetic Resonance. J. Am. Chem. Soc. 1967, 89, 2408−2410.
(25) Kroeker, S.; Stebbins, J. F. Three-Coordinated Boron-11
Chemical Shifts in Borates. Inorg. Chem. 2001, 40, 6239−6246.
(26) Fujdala, K. L.; Oliver, A. G.; Hollander, F. J.; Tilley, T. D.
Tris(tert-butoxy)siloxy Derivatives of Boron, Including the Boronous
Acid HOB[OSi(OtBu)3]2 and the Metal (Siloxy)boryloxide Complex
Cp2Zr(Me)OB[OSi(OtBu)3]2: A Remarkable Crystal Structure with
10155
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Reactivity of Boronic Acid and Boronate Ion Towards Tiron, 2,2′-
Biphenol, and Propylene Glycol. Dalton Trans. 2013, 42, 8446−8453.
(46) Graham, B. J.; Windsor, I. W.; Gold, B.; Raines, R. T. Boronic
Acid with High Oxidative Stability and Utility in Biological Contexts.
Proc. Natl. Acad. Sci. U. S. A. 2021, 118, e2013691118.
(47) Glendening, E. D.; et al. NBO 6.0; Theoretical Chemistry
Institute, University of Wisconsin: Madison, WI, 2013.
NOTE ADDED AFTER ASAP PUBLICATION
■
This paper was published on June 24, 2021. The compound
number in the second column header of Table 2 was incorrect.
This has been updated and the paper was re-posted on June
28, 2021.
10156
https://doi.org/10.1021/jacs.1c02462
J. Am. Chem. Soc. 2021, 143, 10143−10156