A multifunctional catalyst that stereoselectively assembles prodrugs

Article abstract of DOI:10.1126/science.aam7936

The catalytic stereoselective synthesis of compounds with chiral phosphorus centers remains an unsolved problem. State-of-the-art methods rely on resolution or stoichiometric chiral auxiliaries. Phosphoramidate prodrugs are a critical component of pronucleotide (ProTide) therapies used in the treatment of viral disease and cancer. Here we describe the development of a catalytic stereoselective method for the installation of phosphorusstereogenic phosphoramidates to nucleosides through a dynamic stereoselective process. Detailed mechanistic studies and computational modeling led to the rational design of a multifunctional catalyst that enables stereoselectivity as high as 99:1.

Full text of DOI:10.1126/science.aam7936

RESEARCH
◥
multifunctional catalyst that achieves high stereo-
selectivity at phosphorus in P–O bond-forming
reactions.
REPORT
We studied the reaction of chlorophosphora-
midate 1 with nucleoside 2 to assemble a ProTide,
MK-3682 (3) (Fig. 2), a hepatitis C virus (HCV)
NS5b viral RNA polymerase inhibitor, currently
in late-stage clinical trials for the treatment of
HCV infection (19). A variety of achiral nucleo-
philic catalysts were examined to probe inher-
ent substrate bias in the stereochemical outcome
of the reaction. Under conditions originally re-
ported by McGuigan [N-methyl imidazole (NMI),
ORGANIC CHEMISTRY
A multifunctional catalyst that
stereoselectively assembles prodrugs
Daniel A. DiRocco,* Yining Ji, Edward C. Sherer, Artis Klapars, Mikhail Reibarkh,
James Dropinski, Rose Mathew, Peter Maligres, Alan M. Hyde, John Limanto,
Andrew Brunskill, Rebecca T. Ruck, Louis-Charles Campeau, Ian W. Davies
2
,6-lutidine], 49% yield of the desired phosphor-
amidated product was obtained, with a 52:48 (R):
S) ratio at the phosphorus center (Fig. 2A, entry 2).
(
The catalytic stereoselective synthesis of compounds with chiral phosphorus centers
remains an unsolved problem. State-of-the-art methods rely on resolution or stoichiometric
chiral auxiliaries. Phosphoramidate prodrugs are a critical component of pronucleotide
Given the lack of substantial stereochemical bias
in either chiral substrate, we reasoned that ste-
reoselectivity in this reaction could be funda-
mentally controlled by the catalyst. A survey of
chiral nucleophilic catalysts demonstrated most
to be ineffective at influencing the stereochem-
ical outcome of the reaction; however, several ana-
logs of the dihydropyrroloimidazole framework
reported by Zhang and co-workers imparted
promising levels of stereoselectivity at phospho-
rus (table S1) (12, 20–22). Catalyst (R)-A, which
contains a large tert-butyldimethylsilyl ether, fa-
cilitated the desired phosphoramidation in mod-
est yield (60%) and 79:21 selectivity favoring the
desired (R)-stereochemistry at phosphorus (Fig. 2A,
entry 3). A mismatched relationship was observed
using the opposite antipode of the catalyst (Fig. 2A,
entry 4), suggesting chiral recognition of the nu-
cleoside. This substrate-binding event was further
(ProTide) therapies used in the treatment of viral disease and cancer. Here we describe
the development of a catalytic stereoselective method for the installation of phosphorus-
stereogenic phosphoramidates to nucleosides through a dynamic stereoselective process.
Detailed mechanistic studies and computational modeling led to the rational design of a
multifunctional catalyst that enables stereoselectivity as high as 99:1.
lmost half of antiviral and anticancer drugs
currently on the market are nucleosides (1).
Phosphorylation of the nucleoside in vivo
is required for biological activity. However,
nucleoside triphosphates are poor drug
ing the invention of a general protocol (16). In
addition to stereoselectivity at phosphorus, a
second crucial challenge associated with nucleoside
phosphoramidation in ProTide synthesis and anal-
ogous applications is chemoselectivity for 5′ over
3′ phosphoramidation, resulting from indiscri-
minate activation of both hydroxyls. Enzymes
have been shown to catalyze phosphorus-oxygen
bond formation through a series of concomitant
activation modes—including nucleophile activation
by general base catalysis, leaving-group activation
by general acid catalysis, and oxyanion hole–type
stabilization of the transition state (Fig. 1B)—
providing design principles for asymmetric ca-
talysis (17, 18). Here we report the design of a
A
candidates, owing to their instability and in-
effective transport across the cell membrane. The
McGuigan ProTide (pronucleotide) platform
markedly increases cell permeability and rate
of in vivo phosphorylation by introduction of the
1
5
′-aryloxy phosphoramidate moiety as a prodrug
corroborated by H nuclear magnetic resonance
strategy (Fig. 1A) (2, 3). The stereochemistry at
phosphorus in the prodrug can substantially alter
potency, rate of metabolism, and toxicity, making
control of that stereocenter critical to a viable syn-
thetic route. To date, standard methods for the
stereoselective synthesis of P-stereogenic prodrugs
have been limited to the use of chiral auxiliaries
or chiral nonracemic phosphorylating agents, and
critically, no general catalytic methods exist for
their synthesis (Fig. 1A) (4–8).
(NMR) spectroscopy, with intermolecular nuclear
Overhauser effects observed between these spe-
cies (fig. S27). During further development, car-
bamates were identified as a privileged catalyst
The synthesis of carbon stereocenters is now
the most mature subset in the field of asymmetric
catalysis. Translation of these concepts to the syn-
thesis of P-stereogenic compounds has remained
an unmet synthetic need (9–12). Generation of
phosphorus(V) heteroatom bonds in a catalytic
and stereoselective manner would require either
desymmetrization of an achiral species or a dynam-
ic kinetic asymmetric transformation (DYKAT)
that can stereoselect from an interconverting mix-
ture of chiral P(V) intermediates (Fig. 1A) (13–15).
Seminal reports of the formation of phosphor-
amidates via this strategy suffer from low catalytic
turnover and poor stereoselectivity (6, 12). Inher-
ent to this problem is the lack of a canonical
mechanism for P(V) bond formations, complicat-
Fig. 1. Synthetic strategies and concepts for phosphorus-oxygen bond formation. (A) Approaches
to the stereoselective synthesis of ProTides (pronucleotides). LG, leaving group; R, any atom; Ph, phenyl;
Pr, iso-propyl; DYKAT, dynamic kinetic asymmetric transformation. (B) Pentavalent transition state
Process Research and Development, Merck & Co., Inc.,
Rahway, NJ 07065, USA.
i
*Corresponding author. Email: daniel.dirocco@merck.com
proposed by Ryan et al. (17) for P–O bond formation in phosphatidylinositol-specific phospholipase C.
DiRocco et al., Science 356, 426–430 (2017) 28 April 2017
1 of 4

RESEARCH
| REPORT
motif. Catalyst (R)-B provided the product in
moderate diastereoselectivity [89:11 diastereomeric
ratio (d.r.) at 20 mol % loading] and excellent che-
moselectivity for the 5′ hydroxyl (Fig. 2A, entry 7).
During optimization, we discovered the ste-
reoselectivity of the reaction to be dependent on
catalyst concentration, with higher selectivity ob-
served at higher catalyst loadings (Fig. 2A, entries
that 1 is in rapid equilibrium with the activated
species 4 and 5 and that P–O bond formation is
the turnover-limiting step. The dynamic nature
supported the hypothesis that the rate-determining
step involves general base catalysis of the counter-
anion (25). We reasoned that a pathway reminis-
cent of an enzyme-mediated process involving
concomitant activation of the phosphorus donor
and general base catalysis of the 5′-hydroxyl could
lead to the observed rate and selectivity enhance-
ments. To test the validity of this hypothesis, we
designed a chiral pyridine that was structurally
analogous to catalyst A but was not sufficiently
nucleophilic to activate chlorophosphoramidate 1.
When stereochemically matched pyridine (R)-C
was used as a base in combination with catalyst
(R)-B, an improvement in P stereoselectivity was
observed with respect to the same experiment in
1
31
of the process was confirmed by H and P NMR
spectroscopy, as no substantial change in the
P-epimeric ratio of 1 was observed over the course
of the reaction. Furthermore, exchange spectros-
copy experiments unambiguously demonstrated that
catalyst-mediated interconversion of phosphoryl-
imidazolium chlorides 4 and 5 occurs rapidly un-
der the reaction conditions, precluding a change
in equilibrium as a source of stereocontrol (figs.
S8 and S9).
7
to 9), consistent with a non–first-order catalytic
pathway. Reaction progress kinetic analysis of
temporal concentration profiles revealed a second-
order dependence on catalyst concentration, first-
order dependence on nucleoside, and saturation
kinetics of 2,6-lutidine and chlorophosphorami-
date 1 (figs. S3 to S6) (23, 24). These data suggest
Mechanistic studies of DMAP (4-dimethyla-
minopyridine)–catalyzed alcohol acylation have
t
Fig. 2. Development of a stereoselective phosphoramidation. (A) Catalyst
development and mechanistic study. Yield, P(R):P(S), and 5′:3′ ratios were
determined by high-performance liquid chromatographic (HPLC) analysis. %
yield refers to assay yield determined by HPLC analysis; 5′:3′ refers to the molar
ratio of 5′- to 3′-hydroxyl phosphoramidation. Absolute stereochemistry of
catalysts and relative stereochemistry of MK-3682 (3) was confirmed by single-
crystal x-ray analysis (figs. S21 to S25). Cat, catalyst; Me, methyl; Bu, tert-
butyl; TBS, tert-butyldimethylsilyl; NMI, N-methyl imidazole; U, uracil; ND, not
determined. (B) Transition state model of the second-order catalytic pathway.
Calculations were performed at the M06L/6-31+G**//B3LYP-D3/6-31G** level
of theory. Orange, phosphorus; red, oxygen; blue, nitrogen; green, chlorine;
‡
gray, carbon; white, hydrogen. DDG , relative free energy activation barrier.
DiRocco et al., Science 356, 426–430 (2017) 28 April 2017
2 of 4

RESEARCH
| REPORT
n
which 2,6-lutidine functioned as a base (Fig. 2A,
entries 7 and 10). However, when stereochem-
ically mismatched pyridine (S)-C was employed,
a marked loss of stereoselectivity was observed
(Fig. 2A, entry 11). No substantial reaction occurs
in the absence of catalyst (R)-B, suggesting that
both species are involved in the stereochemistry-
determining event (26). Further evidence for a
general base mechanism is provided by a positive
H D H D
kinetic isotope effect (k /k = 1.45, where k /k
is the ratio of rate constants for hydrogen and
deuterium, respectively), observed by deuteration
of the 5′- and 3′-hydroxyl groups on the nucleoside.
To provide theoretical evidence for this distinc-
tive activation mode, we examined our proposal
by computational methods. Dispersion-corrected
B3LYP/6-31G**, M06L/6-31+G**, and M06-2X/
def2-TZVPP calculations in combination with a
conformational sampling routine estimated the
barrier of bond formation to be kinetically un-
feasible (~30 kcal/mol) in the absence of a general
base (tables S9 and S10) (27). In the presence of a
second molecule of catalyst (R)-B, a pentavalent
transition state is characterized by a specific in-
teraction in which the 5′-hydroxyl is hydrogen-
bonded to the exogenous imidazole (Fig. 2B). This
general base effect lowers the transition state
energy by ~10 kcal/mol and thereby presents an
accessible barrier to bond formation. Further in-
terpretation of the transition state structures re-
vealed a variety of interactions that stabilize the
polarized P–O bond in the favored transition state
(Fig. 2A). The preference for coplanarity of the
imidazolium cation and the apical P–O bond re-
sults in a distinct CH–O interaction, facilitated
by acidification of the a-C–H bond of the phos-
phorylated imidazole and bearing resemblance
to the oxyanion hole in the catalytic triad of phos-
pholipases (17, 18, 28). These systems share three
common features: leaving group activation, gen-
eral base catalysis, and oxyanion stabilization. Our
method estimates these effects to provide a tran-
sition state differentiation of 2.3 kcal/mol, favoring
the desired (R)-stereochemistry at phosphorus.
We hoped to develop a further-optimized cat-
alyst that would reinforce these effects to achieve
superior performance in both reaction rate and
stereoselectivity. Inspired by the work of Jacobsen
and co-workers, we envisioned a linking strategy
that would increase cooperativity by reducing the
entropic penalty of a highly ordered transition state
Fig. 3. Design of a dimeric phosphoramidation catalyst. (A) Linker optimization study. % yield, d.r.,
and % 3′ were determined by HPLC analysis. Yield refers to the combined yield of both 5′ diastereomers;
d.r. refers to the P(R):P(S) ratio. n, number of units; NA, not applicable. (B) Transition state model of the
stereoselective phosphorylation with dimeric catalyst I. Calculations were performed at the M06L/6-31
+G**//B3LYP-D3/6-31G** level of theory.
(29, 30). Our computational model suggested this
strategy to be feasible, given the relative proxim-
ity of the respective carbamate side chains in the
favored transition structure (Fig. 2B).
As a rational starting point for synthesis, a va-
riety of carbamate linkages were modeled, sug-
gesting the shortest feasible linkage between the
two carbamate nitrogens to be six carbon atoms.
Accordingly, dimers containing 6- to 10-carbon al-
kyl linkages were synthesized and compared with
catalyst (R)-B under identical reaction conditions
(Fig. 3A). The shorter linkages (six or seven car-
bons) provided a substantial improvement in ef-
ficiency while only modestly increasing selectivity
with respect to catalyst (R)-B at similar loading
(Fig. 3A, entries 3 and 4). A profound effect was
Fig. 4. Scope of the catalytic stereoselective phosphoramidation of nucleosides. Experiments were
performed on a 100-mg scale (nucleoside) using 2 mol % catalyst I, 1.2 equivalents of 1, and 1.2 equivalents
of 2,6-lutidine in dioxolane at –10°C for 24 to 36 hours. Yield refers to combined assay yield of both
1
5
′ diastereomers, as determined by H NMR spectroscopy versus the internal standard. Diastereose-
31
lectivity (d.r.) refers to the P(R):P(S) ratio, as determined by P NMR spectroscopy or HPLC analysis.
DiRocco et al., Science 356, 426–430 (2017) 28 April 2017
3 of 4

RESEARCH
| REPORT
observed at eight atoms (catalyst F), providing
the desired phosphoramidate in 90% yield and
could address this shortcoming, closing the gap
to “evolutionary perfection.”
18. L.-C. Lee, Y.-L. Lee, R.-J. Leu, J.-F. Shaw, Biochem. J. 397,
9–76 (2006).
6
1
9. Merck Sharp & Dohme Corp., Efficacy and Safety of
grazoprevir (MK-5172) and MK-3682 with elbasvir (MK-8742)
or ruzasvir (MK-8408) for chronic hepatitis C virus (HCV)
genotype (GT) 3, GT4, GT5, and GT6 Infection (MK-3682-012).
in ClinicalTrials.gov. [Internet] Bethesda (MD): National Library
of Medicine (US). 2000- [cited 21 November 2016]. Available
from: https://clinicaltrials.gov/ct2/show/NCT02332720; NLM
Identifier NCT02332720.
97:3 d.r. and validating our computationally driven
Although this catalyst was optimized for use
in a single ProTide therapeutic (MK-3682), the
key principles that govern selectivity proved gen-
erally applicable to other nucleoside analogs. A
variety of ribofuranose-based nucleosides could
be phosphorylated with high P stereoselectivity
by means of this method (Fig. 4). Neither the nu-
cleobase nor the C3′ hydroxyl is an important
driver for selectivity, exemplified in the highly
stereoselective phosphoramidation of dihydro-
furanones 5 and 9. Deoxyribofuranose derivatives,
lacking a C2′ substituent, are tolerated; notably,
the anti-HIV nucleoside derivative azidothymidine
(AZT) (7) could be converted to the prodrug in
93:7 d.r. In a seminal paper, McGuigan et al. re-
ported this nucleoside to yield only a 2:1 mix-
ture of diastereomers when phosphoramidated
in the presence of NMI (33).
Through a distinctive activation mode, we have
designed a practical small-molecule catalyst that
mimics the complex function of an enzyme, culmi-
nating in phosphoramidation that is highly ste-
reoselective at the phosphorus center. The general
applicability of the catalyst to a broad range of
therapeutically relevant nucleophiles invites its
rapid adoption in the preparation of these clin-
ically important molecules.
design principle (Fig. 3A, entry 5). Maximum se-
lectivity was achieved at nine atoms (catalyst G,
98:2 d.r.); further lengthening of the linker resulted
in erosion of both stereoselectivity and chemo-
selectivity, likely on account of undesirable flex-
ibility. Given this observation, we anticipated that
further conformational restriction of the linker
would result in increased selectivity through the
reduction of rotational degrees of freedom. As
such, catalyst I, which contains a central 1,3-phenyl
group, was synthesized and displayed optimal re-
activity [relative rate krel = 10 versus (R)-B (fig. S16)]
and further improved stereoselectivity (99:1 d.r.).
To better understand the effect of catalyst link-
age on stereoselectivity, we reoptimized the transi-
tion states with linked catalyst I (Fig. 3B). Relative
transition state energies predict the desired P-(R)
stereochemistry to be favored by 2.6 kcal/mol, in
agreement with the higher experimentally ob-
served selectivity (31). The same structural prefer-
ences are conserved and further amplified in this
structure (table S15).
Experimental validation of intramolecular co-
operativity was garnered by the observation of
a change to first-order dependence on catalyst
concentration in this system (fig. S14). We were
initially concerned that linking catalysts might
result in deactivation through competitive bis-
phosphoramidation of the dimer. Surprisingly,
P–O bond formation proved considerably faster
than chlorophosphoramidate activation, exem-
plified by first- and zero-order kinetics of chloro-
phosphoramidate 1 and nucleoside 2, respectively
20. P. M. Weintraub, P. L. Tiernan, J. C. Huffman, J. Het. Chem. 24,
61–563 (1987).
1. Z. Zhang, F. Xie, J. Jia, W. Zhang, J. Am. Chem. Soc. 132,
5939–15941 (2010).
2. A. S. De Miranda et al., J. Mol. Catal. B 91, 77–80 (2013).
5
2
2
1
23. D. G. Blackmond, Angew. Chem. Int. Ed. 44, 4302–4320
2005).
(
2
4. The less-selective catalyst (R)-A, based on the same bicyclic
framework but lacking the carbamate functionality, exhibited
a first-order dependence (fig. S10).
25. S. Xu et al., Chemistry 11, 4751–4757 (2005).
2
2
2
6. A traditional nonlinear-effect study was not meaningful in this
case because of the diastereomeric relationship of the products
and the complex equilibria associated with the chiral reactants.
7. Owing to the complexity in sampling van der Waals
precomplexes, we report relative activation free energies in
relation to a common interconverting lowest-energy species.
a a
8. Estimates of the pK (where K is the acid dissociation
constant) of this C–H moiety suggest that it is more acidic
than an amide or imidazole N–H, frequently implicated in
oxyanion hole–type stabilization.
29. R. G. Konsler, J. Karl, E. N. Jacobsen, J. Am. Chem. Soc. 120,
1
0780–10781 (1998).
0. S. E. Denmark, S. K. Ghosh, Angew. Chem. Int. Ed. 40,
759–4762 (2001).
3
4
31. The experimentally determined selectivities for catalysts B
and I correspond to energy differences at 263 K of 1.66 and
2.40 kcal/mol, respectively.
REFERENCES AND NOTES
3
2. W. J. Albery, J. R. Knowles, Biochemistry 15, 5631–5640
1976).
1.
L. P. Jordheim, D. Durantel, F. Zoulim, C. Dumontet, Nat. Rev.
Drug Discov. 12, 447–464 (2013).
(
33. C. McGuigan, R. N. Pathirana, N. Mahmood, A. J. Hay, Bioorg.
Med. Chem. Lett. 2, 701–704 (1992).
2.
D. Cahard, C. McGuigan, J. Balzarini, Mini Rev. Med. Chem. 4,
371–381 (2004).
3
4
.
.
M. J. Sofia et al., J. Med. Chem. 53, 7202–7218 (2010).
U. Pradere, E. C. Garnier-Amblard, S. J. Coats, F. Amblard,
R. F. Schinazi, Chem. Rev. 114, 9154–9218 (2014).
B. S. Ross, P. G. Reddy, H.-R. Zhang, S. Rachakonda,
M. J. Sofia, J. Org. Chem. 76, 8311–8319 (2011).
F. Pertusati, C. McGuigan, Chem. Commun. 51, 8070–8073 (2015).
K. Tran et al., J. Org. Chem. 80, 4994–5003 (2015).
J. M. McCabe Dunn et al., Chem. Sci. 8, 2804–2810 (2017).
B. R. Sculimbrene, S. J. Miller, J. Am. Chem. Soc. 123,
10125–10126 (2001).
ACKNOWLEDGMENTS
We thank Y. Yu for synthesis of catalyst libraries, J. Liu and M. Biba
for separations support, Y. Liu for NMR support, J. Forbes and
H. Gurukar for high-performance computing support, and
E. N. Jacobsen for helpful discussions. Structural parameters for
compounds 3, (R)-B, (R)-J, and (R)-G are available free of charge
from the Cambridge Crystallographic Data Centre under accession
numbers CCDC-1525797, CCDC-1523526, CCDC-1523527, and
CCDC-1523528, respectively. Full details of the kinetic, spectroscopic,
and computational studies are available in the supplementary
materials. Stationary points for all computed structures can be
located in data S1.
(figs. S11 and S12). These data suggest a change in
5.
the turnover-limiting step to chlorophosphorami-
date activation. Given that P–O bond formation
is faster than activation of 1, it seemed plausible
that equilibration may be a stereo-determining fac-
6
.
7.
8
9.
.
31
tor in this system. However, P NMR studies re-
vealed a single observable species, consistent with
fast equilibration and a lower barrier to P inver-
sion through a cooperative displacement mech-
anism (figs. S32 and S33). As argued by Albery
and Knowles in their view of the “evolutionary
perfection” of an enzyme, the enzyme must con-
front the problem of finding a mechanism that
lowers the barrier to each elementary step (32).
We have demonstrated that this catalyst lowers
barriers to P–O bond formation and P(V) inver-
sion but is turnover-limited by the rate of catalyst
addition to 1. Arguably, further development
10. C. M. Longo, Y. Wei, M. F. Roberts, S. J. Miller, Angew. Chem.
Int. Ed. 48, 4158–4161 (2009).
11. J. W. Evans, M. B. Fierman, S. J. Miller, J. A. Ellman, J. Am.
Chem. Soc. 126, 8134–8135 (2004).
12. S. Liu et al., Tetrahedron Asymmetry 23, 329–332 (2012).
3. J. Steinreiber, K. Faber, H. Griengl, Chem. Eur. J. 14,
060–8072 (2008).
4. W. B. Farnham, R. A. Lewis, R. K. Murray Jr., K. Mislow, J. Am.
SUPPLEMENTARY MATERIALS
1
www.sciencemag.org/content/356/6336/426/suppl/DC1
8
Materials and Methods
Figs. S1 to S41
Tables S1 to S17
References (34–63)
NMR Spectra
1
Chem. Soc. 92, 5808–5809 (1970).
15. M. Dutartre, J. Bayardon, S. Jugé, Chem. Soc. Rev. 45,
5771–5794 (2016).
1
6. C. Brown, J. A. Boudreau, B. Hewitson, R. F. Hudson, J. Chem.
Soc. Chem. Commun. 1975, 504–505 (1975).
7. M. Ryan, T. Liu, F. W. Dahlquist, O. H. Griffith, Biochemistry 40,
9743–9750 (2001).
Data S1
1
17 January 2017; accepted 9 March 2017
10.1126/science.aam7936
DiRocco et al., Science 356, 426–430 (2017) 28 April 2017
4 of 4

A multifunctional catalyst that stereoselectively assembles
prodrugs
Daniel A. DiRocco, Yining Ji, Edward C. Sherer, Artis Klapars,
Mikhail Reibarkh, James Dropinski, Rose Mathew, Peter Maligres,
Alan M. Hyde, John Limanto, Andrew Brunskill, Rebecca T. Ruck,
Louis-Charles Campeau and Ian W. Davies (April 27, 2017)
Science 356 (6336), 426-430. [doi: 10.1126/science.aam7936]
Editor's Summary
Getting phosphorus into healthy shape
ProTide therapeutics play a trick on the body, getting nucleoside analogs where they need to be
by decorating them with unnatural phosphoramidates in place of ordinary phosphates. These
compounds pose an unusual synthetic challenge because their configuration must be controlled at
phosphorus; most methods have been refined to manipulate the geometry of carbon. DiRocco et al.
report a metal-free, small-molecule catalyst that attains high selectivity for nucleoside
phosphoramidation by activating both reaction partners. Kinetic studies with an early prototype
revealed a double role for the catalyst that inspired the rational design of a more active and selective
dimeric structure.
Science, this issue p. 426
This copy is for your personal, non-commercial use only.
Article Tools
Permissions
Visit the online version of this article to access the personalization and
article tools:
http://science.sciencemag.org/content/356/6336/426
Obtain information about reproducing this article:
http://www.sciencemag.org/about/permissions.dtl
Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week
in December, by the American Association for the Advancement of Science, 1200 New York
Avenue NW, Washington, DC 20005. Copyright 2016 by the American Association for the
Advancement of Science; all rights reserved. The title Science is a registered trademark of AAAS.