Programmed Negative Allostery with Guest-Selected Rotamers Control Anion-Anion Complexes of Stackable Macrocycles

Article abstract of DOI:10.1021/jacs.8b02993

A new rotamer-based strategy for negative allostery has been used to control host-host interactions and product yield upon anion complexation. Coassembly of anion dimers as guests inside two cyanostar macrocycles drives selection of one rotamer in which all ten steric groups get directed outward to destabilize triply stacked macrocycles. A large entropy penalty (ΔS) is quantified upon anion binding when the multiple dynamic rotamers collapse down to one.

Full text of DOI:10.1021/jacs.8b02993

Communication
Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
pubs.acs.org/JACS
Programmed Negative Allostery with Guest-Selected Rotamers
Control Anion−Anion Complexes of Stackable Macrocycles
Edward G. Sheetz, Bo Qiao, Maren Pink, and Amar H. Flood*
Department of Chemistry, Indiana University, 800 E. Kirkwood Avenue, Bloomington, Indiana 47405, United States
*
S Supporting Information
ABSTRACT: A new rotamer-based strategy for negative
allostery has been used to control host−host interactions
and product yield upon anion complexation. Coassembly
of anion dimers as guests inside two cyanostar macrocycles
drives selection of one rotamer in which all ten steric
groups get directed outward to destabilize triply stacked
macrocycles. A large entropy penalty (ΔS) is quantified
upon anion binding when the multiple dynamic rotamers
collapse down to one.
otamers are ubiquitous. They are conformational states
R
related by simple rotations about covalent bonds that have
1
found relevance in protein folding, in the rotary motors of
2
3,4
both biological and artificial machines, and in molecular
5
switches. In supramolecular chemistry, the effects of rotamers
have been detailed in foldamers, molecular tweezers, and
allosteric macrocycles. With their ability to dynamically
6
7
8
respond to input stimuli, however, rotamers have yet to be
investigated as a deliberate element of design for controlling
molecular self-association and recognition. Rather, rotational
isomers are typically avoided given that restricting their
9
rotations produces entropy penalties that impair affinity. We
Figure 1. (a) Parent cyanostar forms (b) a mix of 3:2 and 2:2
complexes as visualized on an energy landscape. Block arrow indicates
the principle of negative design. (c) New cyanodimer design with
substituents (blue) that can rotate to (d) sterically block faces and shut
down access to the 3:2 complex.
address these issues by investigating the benefits of rotamers as
key elements for producing an effect similar to negative
allostery. But instead of inhibiting the binding of other guests,
we use allostery to inhibit association of more hosts. Here we
10
show how to optimize the yield (aka fidelity) of a new self-
1
1−14
assembly system
involving anion−anion dimers and stacks
of cyanostar macrocycles (Figure 1a). These planar macrocycles
are exemplary of many others
coordination chemistry, and soft matter,
uncontrolled self-association and thus we offer a new way
to control stacking in this privileged class of building blocks.
Anion-driven assembly
of dimers of bisulfate,
cooperatively stabilized by stacks of macrocyclic cyanostars.
Yet, uncontrolled mixtures of complexes are formed (Figure
b) with 3:2 and 2:2 cyanostar:bisulfate stoichiometries
15−17
18,19
used in guest binding,
requirement. This noncovalent approach bypasses the syntheti-
cally challenging covalent alternative of preparing a single
rotamer with all five substituents on the same side, like a single
20
21,22
which undergo
19,23
24
28
atropisomer seen in picket-fence porphyrins. Using dynamic
2
5−27
is exemplified by the formation
rotamers allows sterically bulky groups to randomly distribute
above and below the macrocycle’s plane up until the moment
an anion is bound. Thereafter, the π-stacked dimer is expected
to form by rotation of the steric groups away from the center of
the macrocycles. Once directed outward, they would block
1
1,13
2−
[HSO ···HSO ] , which are
4
4
27
1
2
9
further stacking. Though use of sterics to control stacking,
possible that differ by the number of stacked macrocycles.
We attribute this outcome to uncontrolled π stacking of
macrocycles. This idea inspired the design strategy of using
guest-selected rotamers to preferentially form a doubly stacked
30
and the idea that guest binding freezes conformations is not
new, this is the first time the two ideas have been unified
together in a deliberate design that mimics negative
31−33
allostery.
2:2 assembly over the triply stacked complex.
The design strategy is based on creating a macrocycle with
one face open for stacking and the other blocked by sterically
Received: March 17, 2018
bulky groups (Figure 1d). Dynamic rotamers fulfill this
©
XXXX American Chemical Society
A
DOI: 10.1021/jacs.8b02993
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
2
9
Inspired by prior work, as well as experimental and
computational estimates of rotational barriers of monosub-
Scheme 1. Synthesis of Cyanodimer
3
4
stituted biaryl compounds, the isopropyl moiety was selected
as the bulky substituent for biaryl rotamers. Isopropyl achieved
a good balance between steric bulk and rotational kinetics (k ∼
5
−1
‡
−1
5
× 10 s , ΔG = 11 kcal mol at 300 K).
To estimate how well sterics could control stacking,
intermacrocycle distances of various rotamers (Figure 2a−c)
packing are unambiguously established, the structure cannot be
fully determined and refined on account of a lack of data.
Once dimerized, the macrocycles stop further stacking as
shown by the herringbone packing differing from slip-stack
27,35
packing observed for complexes with parent cyanostar.
Though the isopropyl groups are disordered (Figure S22),
most isopropyl sites are pointing away from the macrocycle
seam. At least one isopropyl site is not. In solution, we also see
the 2:1 stoichiometry around ClO4− with 0.5 equiv of added
anion (CD Cl , Figure S11). These findings are consistent with
Figure 2. (a−c) Molecular models (MMFF) and interplanar distances
of cyanodimer macrocycles with exterior groups in various
conformations: all-out, in-out, and all-in. Geometries optimized as
complexes of the bisulfate dimer. (d) Preliminary crystal structure
showing the herringbone packing of the 2:1 complex between
cyanodimer and perchlorate.
2
2
anion-driven selection of all-out rotamers.
Next, we undertook studies to verify that anion binding
turned off the triple stack. Consistently, titrations with bisulfate
1
in chloroform as monitored by H NMR spectroscopy (Figure
were determined using molecular mechanics. The closest
interplanar contact (3.8 Å, Figure 2a) is achieved when all of
the exterior isopropyl groups on both macrocycles are facing
outward. Larger distances are seen with inward facing
substituents (Figures 2b,c).
3a) show conversion of free macrocycle into 2:2 complex with
exclusion of the 3:2 complex across 0−1 equiv. By comparison,
the parent cyanostar shows the 3:2 complex as well as its
corresponding OH resonances at ∼13 ppm for the hydrogen-
bonded dimers of bisulfate in the 0.2−0.9 equiv region (red
boxes, Figure 3b). For cyanodimer, we also see a resonance for
the hydrogen-bonded dimer but at ∼14 ppm, which is
Cyanodimer was prepared (Scheme 1) using a modification
of the synthesis of the cyanosolo macrocycle in order to
29
11
1
enhance modularity. Macrocyclization proceeded in a one-pot
Knoevenagel cyclocondensation from aldehyde monomer in a
characteristic of the 2:2 complex. The H NMR titration
confirms that guest-selected rotamers control stacking with
exclusive formation of the 2:2 complex.
38% yield.
We first verified that stacking is suppressed in the absence of
We switched to dichloromethane to provide a more
challenging environment for the negative allostery on account
of the fact that this solvent is known to enhance formation of
the anionic guest as a result of random distributions of up−
down rotamers. Consistently, NMR and UV−vis spectra
13
(
Figures S1 and S2) show no change with concentration (1
the 3:2 complex. Though we see that the same control feature
is expressed, it is not as perfect, e.g., the 3:2 emerges at 0.2−0.6
equiv (Figure S4). Nevertheless, the 3:2 complex is completely
μM to 10 mM, dichloromethane) in contrast to the parent
1
3
cyanostar.
10
We confirmed that anionic guests trigger formation of a π−π
seam. A preliminary crystal structure of the 2:1 complex (Figure
d, see Supporting Information for details), which was formed
around perchlorate in lieu of bisulfate, confirms the dimer of
converted into the 2:2 by 1.0 equiv fulfilling the requirement
for high fidelity. With the parent cyanostar, however, the 3:2
persisted in solution out to 4.0 equiv (Figure S4). This
suppression of the 3:2 complex was also verified by mass
spectrometry (Figure S6).
2
macrocycles is accessible. Our crystal data lacks high-resolution
(
beyond 1.5 Å), which could not be mitigated. This is a known
Observation of the 3:2 complex with cyanodimer was a
surprise. Molecular models shows that the triple stack with the
least steric pressure (Figure S7) involves a central macrocycle
with three up and two down isopropyl units sandwiched by two
pathology for cyanostar structures that display high levels of
disorder
whole-molecule disorder. Though unit cell, space group and
1
1−13,27,29,35−39
from large, solvent accessible areas and
B
DOI: 10.1021/jacs.8b02993
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
complex (Figure S3) that forms between bisulfate and the
29
cyanosolo macrocycle.
The 1:1 complex is observed to a greater extent in
chloroform (34:100 for the ratio of 1:1 to 2:2 at 15 equiv of
added bisulfate) than in dichloromethane (7:100). This finding
is consistent with greater ion pairing in the less polar solvent as
11,13,36,40−43
+
seen elsewhere.
Thus, pairing of the TBA cation
with the 1:1 complex likely plays a role in the difference
between cyanostar and cyanodimer.
Observation of a 1:1 complex also indicates a loss in
cooperativity for the 2:2 complex with cyanodimer. This
9
outcome is attributed to a loss of entropy (Figure 4a) on
1
Figure 3. H NMR titration of (a) cyanodimer and (b) the parent
cyanostar with tetrabutylammonium bisulfate (TBAHSO , 1 mM,
4
CDCl , 298 K, 600 MHz). The signature for the 3:2 complex is
3
indicated by dashed red boxes. The signature for the 2:2 complex
bisulfate −OH peak is indicated by a dashed blue box. The signatures
assigned to the 1:1 complex with cyanodimer are indicated by blue
dots.
Figure 4. (a) Loss in conformational entropy upon complexation with
perchlorate. As a mimic for entropy losses upon macrocycle
dimerization with a bisulfate dimer. (b) Model of loss in torsional
−
entropy upon ClO4 binding that is based on the potential energy
surface calculated (B3LYP/6-31G*) using one biphenyl compound.
Binding selects one rotamer and narrows the rotational energy well.
(c) Van’t Hoff plots for cyanodimer (black) and cyanostar (red) for
all-out rotamers. With two partially congested seams,
intermacrocycle distances (3.9 Å) are consistent with weakened
π−π stacking that would favor 2:2 over 3:2, as observed. The
fact that the triple stack forms at all shows that the steric
repulsions inhibiting formation of a tight seam can be overcome
by interactions between the triple stack and the bisulfate dimer
dianion.
2:1 complex formation with TBAClO
CH ClCH Cl).
in dichloroethane
4
(
2
2
account of the restricted rotations of the biphenyl groups in the
2:2 complex. By contrast, the tert-butyl groups in cyanostar’s
44
complexes have been seen to rotate freely in its dimer stacks.
One additional behavior differing from the parent cyanostar
was the emergence of a new species beyond 1 equiv attributed
to the 1:1 complex. At saturation (∼15 equiv of bisulfate,
We considered two approaches to calculate the entropy loss
from restricting biphenyl rotations. The isopropyl groups were
not considered as they largely retain mobility. The first
calculation involves collapse of 32 rotational isomers in a
CDCl , Figure 3), this species has a signature matching the 1:1
3
C
DOI: 10.1021/jacs.8b02993
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
cyanodimer to one in the complex (57 J mol−1 K , Figures 4a,
−1
ASSOCIATED CONTENT
Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.8b02993.
■
Supporting Information). The second relies on Whitesides’
*
S
45
method of calculating torsional entropy. A biphenyl model
was used to plot potential energy versus dihedral angle (Figure
4
b) to predict torsional entropy loss (123 J mol− K ,
1
−1
Experimental procedures, compound characterization
and spectra, NMR titrations, UV−vis titrations, mass
spectrometry, and entropy calculations (PDF)
Supporting Information S8). Though both show significant
penalties, the discrepancy between them impacts their
usefulness for providing a deeper understanding of the
1
1,41,42,46−48
recognition phenomenon
and for their use in the
49
accurate computer-aided design of receptors.
AUTHOR INFORMATION
■
*
Experimental determination of thermodynamic signatures
requires accurate estimates of all equilibrium constants. For
this reason, we selected a simpler system derived from
perchlorate, which has only two coupled equilibria rather
than the four seen with bisulfate. Entropy losses upon
macrocycle dimerization with ClO4 approximate those with
bisulfate. The equilibria involved with the macrocycles (MC; as
11
Corresponding Author
aflood@indiana.edu
2
7
ORCID
11
Amar H. Flood: 0000-0002-2764-9155
Notes
The authors declare no competing financial interest.
−
either cyanodimer or cyanostar) are as follows:
MC + ClO − ⇌ MC·ClO4− ΔG1:1
ACKNOWLEDGMENTS
4
■
We acknowledge support from the NSF (DMR 1533988).
2
MC + ClO − ⇌ MC ·ClO4
−
ΔG2:1
4
2
REFERENCES
The parent cyanostar forms a 2:1 perchlorate complex with
■
16
large cooperativity and this complex is stable even with excess
anion. With cyanodimer, however, after the 2:1 complex forms
in high fidelity at 0.5 equiv it is followed by formation of a 1:1
complex (∼5 equiv, Figure S11) indicative of a reduction in
cooperativity. This behavior matches bisulfate and is consistent
with the 2:1 perchlorate complex serving as a reasonable
approximation of the 2:2 bisulfate complex.
The binding thermodynamics were established from van’t
Hoff plots (Figure 4c, 303−350 K). The overall stability of
cyanostar’s 2:1 complex with perchlorate differs very little with
(1) Lovell, S. C.; Word, J. M.; Richardson, J. S.; Richardson, D. C.
Proteins: Struct., Funct., Genet. 2000, 40, 389−408.
2) Boyer, P. D. Nature 1999, 402, 247.
(
(
3) Kelly, T. R.; Bowyer, M. C.; Bhaskar, K. V.; Bebbington, D.;
Garcia, A.; Lang, F.; Kim, M. H.; Jette, M. P. J. Am. Chem. Soc. 1994,
16, 3657−3658.
4) Koumura, N.; Zijlstra, R. W.; van Delden, R. A.; Harada, N.;
Feringa, B. L. Nature 1999, 401, 152−155.
5) Su, X.; Aprahamian, I. Chem. Soc. Rev. 2014, 43, 1963−1981.
(6) Berl, V.; Huc, I.; Khoury, R. G.; Krische, M. J.; Lehn, J. M. Nature
1
(
(
2000, 407, 720−723.
−
1
temperature (ΔG = 76.0 ± 0.3 kJ mol ; 303−350 K)
(7) Petitjean, A.; Khoury, R. G.; Kyritsakas, N.; Lehn, J.-M. J. Am.
Chem. Soc. 2004, 126, 6637−6647.
2
:1
generating a small entropic effect on binding (ΔS = 12 ± 16 J
mol K ). The enthalpy of binding for cyanostar is ΔH
2
:1
−1
−1
(8) Rebek, J.; Trend, J. E.; Wattley, R. V.; Chakravorti, S. J. Am.
Chem. Soc. 1979, 101, 4333−4337.
=
2:1
−1
−
72 ± 5 kJ mol .
(
1
(
9) Chang, C.-E.; Gilson, M. K. J. Am. Chem. Soc. 2004, 126, 13156−
Consistent with our idea, cyanodimer has a large entropic
3164.
10) Todd, E. M.; Quinn, J. R.; Park, T.; Zimmerman, S. C. Isr. J.
Chem. 2005, 45, 381−389.
11) Fatila, E. M.; Twum, E. B.; Sengupta, A.; Pink, M.; Karty, J. A.;
−1
−1
penalty to 2:1 binding of −115 ± 18 J mol K . This value
−
1
−1
matches the torsional entropy (−123 J mol K ). Thus,
cyanodimer’s cooperativity is reduced by entropic costs of
(
restricting rotations of 10 biphenyl units and impeding their
Raghavachari, K.; Flood, A. H. Angew. Chem., Int. Ed. 2016, 55,
14057−14062.
−
±
40° rocking (Figure 4b). Finally, the enthalpy of ClO
4
1
−
binding with cyanodimer is ΔH = −98 ± 4 kJ mol
(12) Zhao, W.; Qiao, B.; Chen, C.-H.; Flood, A. H. Angew. Chem., Int.
Ed. 2017, 56, 13083−13087.
2
:1
We expect entropy penalties of forming a 2:1 complex with
ClO4 matches formation of a 2:2 complex with bisulfate; both
−
(13) Fatila, E. M.; Twum, E. B.; Karty, J. A.; Flood, A. H. Chem. - Eur.
J. 2017, 23, 10652−10662.
reactions freeze ten biphenyl rotamers. For the formation of the
:2 complex, however, one additional macrocycle is attached to
(
14) Dobscha, J. R.; Debnath, S.; Fadler, R.; Fatila, E.; Pink, M.;
3
Raghavachari, K.; Flood, A. H. Chem. - Eur. J. 2018, DOI: 10.1002/
the 2:2 complex by freezing out an additional five biphenyl
rotamers. Thus, the entropy penalty for the 3:2 complex is
expected to be 50% greater that of the 2:2 complex. This
entropy cost likely works together with the designed steric
effects. Thus, a more accurate formulation of the design
principle employed herein is of steric destabilization and
enhanced entropic destabilization of the 3:2 species relative to
the entropic destabilization of the 2:2 complexes that form
around a bisulfate dimer.
In conclusion, a new strategy for negative allostery involves
guest-selected rotamers capable of stabilizing π-stacked dimers
over multimers to select 2:2 bisulfate complexes with high
fidelity. The design mechanism produces lower overall stability
of π-stacked dimers in the 2:2 complex originating from losses
in torsional entropy.
chem.201800827.
(
15) Grave, C.; Schlu
098.
(16) Yuan, L.; Feng, W.; Yamato, K.; Sanford, A. R.; Xu, D.; Guo, H.;
̈
ter, A. D. Eur. J. Org. Chem. 2002, 2002, 3075−
3
Gong, B. J. Am. Chem. Soc. 2004, 126, 11120−11121.
(17) Guieu, S.; Crane, A. K.; MacLachlan, M. J. Chem. Commun.
2
(
011, 47, 1169−1171.
18) Qin, B.; Ren, C.; Ye, R.; Sun, C.; Chiad, K.; Chen, X.; Li, Z.;
Xue, F.; Su, H.; Chass, G. A.; Zeng, H. J. Am. Chem. Soc. 2010, 132,
564−9566.
19) Lee, S.; Hirsch, B. E.; Liu, Y.; Dobscha, J. R.; Burke, D. W.; Tait,
S. L.; Flood, A. H. Chem. - Eur. J. 2016, 22, 560−569.
9
(
(
20) Hui, J. K. H.; MacLachlan, M. J. Chem. Commun. 2006, 2480−
2
482.
(21) Venkataraman, D.; Lee, S.; Zhang, J.; Moore, J. S. Nature 1994,
371, 591.
D
DOI: 10.1021/jacs.8b02993
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
(
22) Ren, C.; Xu, S.; Xu, J.; Chen, H.; Zeng, H. Org. Lett. 2011, 13,
840−3843.
23) Ma, C. T. L.; MacLachlan, M. J. Angew. Chem., Int. Ed. 2005, 44,
178−4182.
24) Zang, L.; Che, Y.; Moore, J. S. Acc. Chem. Res. 2008, 41, 1596−
608.
25) He, Q.; Kelliher, M.; Bah
Am. Chem. Soc. 2017, 139, 7140−7143.
26) Mungalpara, D.; Valkonen, A.; Rissanen, K.; Kubik, S. Chem. Sci.
017, 8, 6005−6013.
27) Lee, S.; Chen, C.-H.; Flood, A. H. Nat. Chem. 2013, 5, 704−
10.
28) Collman, J. P.; Gagne, R. R.; Reed, C.; Halbert, T. R.; Lang, G.;
Robinson, W. T. J. Am. Chem. Soc. 1975, 97, 1427−1439.
29) Qiao, B.; Anderson, J. R.; Pink, M.; Flood, A. H. Chem.
3
(
4
(
1
(
̈
ring, S.; Lynch, V. M.; Sessler, J. L. J.
(
2
(
7
(
(
Commun. 2016, 52, 8683−8686.
(
(
30) Raker, J.; Glass, T. E. J. Org. Chem. 2002, 67, 6113−6116.
31) Lockman, J. W.; Paul, N. M.; Parquette, J. R. Prog. Polym. Sci.
2
(
(
005, 30, 423−452.
32) Changeux, J.-P. Annu. Rev. Biophys. 2012, 41, 103−133.
33) Shimoyama, D.; Yamada, H.; Ikeda, T.; Sekiya, R.; Haino, T.
Eur. J. Org. Chem. 2016, 2016, 3300−3303.
34) Mazzanti, A.; Lunazzi, L.; Minzoni, M.; Anderson, J. E. J. Org.
Chem. 2006, 71, 5474−5481.
35) Hirsch, B. E.; Lee, S.; Qiao, B.; Chen, C.-H.; McDonald, K. P.;
Tait, S. L.; Flood, A. H. Chem. Commun. 2014, 50, 9827−9830.
36) Qiao, B.; Hirsch, B. E.; Lee, S.; Pink, M.; Chen, C.-H.; Laursen,
B. W.; Flood, A. H. J. Am. Chem. Soc. 2017, 139, 6226−6233.
37) Benson, C. R.; Fatila, E. M.; Lee, S.; Marzo, M. G.; Pink, M.;
Mills, M. B.; Preuss, K. E.; Flood, A. H. J. Am. Chem. Soc. 2016, 138,
5057−15065.
38) Fatila, E. M.; Pink, M.; Twum, E. B.; Karty, J. A.; Flood, A. H.
Chem. Sci. 2018, 9, 2863−2872.
39) Qiao, B.; Liu, Y.; Lee, S. M.; Pink, M.; Flood, A. H. Chem.
Commun. 2016, 52, 13675−13678.
40) McDonald, K. P.; Qiao, B.; Twum, E. B.; Lee, S.; Gamache, P. J.;
Chen, C.-H.; Yi, Y.; Flood, A. H. Chem. Commun. 2014, 50, 13285−
3288.
41) Qiao, B.; Sengupta, A.; Liu, Y.; McDonald, K. P.; Pink, M.;
Anderson, J. R.; Raghavachari, K.; Flood, A. H. J. Am. Chem. Soc. 2015,
37, 9746−9757.
42) Liu, Y.; Sengupta, A.; Raghavachari, K.; Flood, A. H. Chem.
017, 3, 411−427.
43) Hirsch, B. E.; McDonald, K. P.; Tait, S. L.; Flood, A. H. Faraday
Discuss. 2017, 204, 159−172.
44) Liu, Y.; Singharoy, A.; Mayne, C. G.; Sengupta, A.;
Raghavachari, K.; Schulten, K.; Flood, A. H. J. Am. Chem. Soc. 2016,
38, 4843−4851.
45) Mammen, M.; Shakhnovich, E. I.; Whitesides, G. M. J. Org.
Chem. 1998, 63, 3168−3175.
46) Ramabhadran, R. O.; Hua, Y.; Li, Y.-j.; Flood, A. H.;
Raghavachari, K. Chem. - Eur. J. 2011, 17, 9123−9129.
47) McDonald, K. P.; Ramabhadran, R. O.; Lee, S.; Raghavachari,
K.; Flood, A. H. Org. Lett. 2011, 13, 6260−6263.
48) Liu, K.; Xing, R.; Zou, Q.; Ma, G.; Mohwald, H.; Yan, X. Angew.
Chem., Int. Ed. 2016, 55, 3036−3039.
49) Ramabhadran, R. O.; Liu, Y.; Hua, Y.; Ciardi, M.; Flood, A. H.;
Raghavachari, K. J. Am. Chem. Soc. 2014, 136, 5078−5089.
(
(
(
(
1
(
(
(
1
(
1
(
2
(
(
1
(
(
(
(
̈
(
E
DOI: 10.1021/jacs.8b02993
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX