news & views theory calculations, which are particularly valuable in examining specifically how the adatoms affect the reaction rate. It turns out that at large distances of up to around 1 nm, the effect is mostly due to attractive van der Waals interactions. This also explains the preference for the configuration in which the hydrogen atoms are closer to the adatom. On the other hand, at short distances, surface-related repulsive interactions become important and induce a shift in the tautomerization equilibrium. The experiments undertaken are an extremely elegant, quantitative demonstration of basic chemical phenomena, and offer a platform for studying such effects further. This behaviour is highly unlikely to be limited to the tautomerization reaction of porphycene, and further studies will undoubtedly offer important insights into, for example, catalysis. It is also worthy to note that this type of tautomerization reaction has also been

proposed as a molecular-scale switch2,3, and the present results bring both good and bad news for the achievement of this goal. The bad news is that when the active components of electronic circuitry shrink down to the molecular scale, the disrupting effects from disorder in the atomic environment become more relevant. This is probably unavoidable, and strategies for tolerating the inherent variability in the properties of the components (here chemical reaction rate and equilibrium) have to be developed instead of simply concentrating on trying to eliminate this variability. The good news is that this sensitivity to the environment can be exploited and the first steps in this direction are already indicated by Grill and colleagues. They demonstrate that cooperative effects are important in assemblies of several porphycene molecules (Fig. 1c): the tautomerization reaction of a given molecule

is affected by the tautomerization state of the neighbouring molecules. Realization of more complicated molecular information processing units based on assemblies of molecular switches requires precisely this kind of interaction between the switching units. Actual devices based on such concepts are some way off in the future, but the work of Grill and colleagues gives us a solid understanding of the underlying chemical phenomena of such systems. ❐ Peter Liljeroth is in the Department of Applied Physics, Aalto University School of Science, PO Box 15100, FI-00076 AALTO, Finland. e-mail: [email protected] References Kumagai, T. et al. Nature Chem. 6, 41–46 (2014). Liljeroth, P., Repp, J. & Meyer, G. Science 317, 1203–1206 (2007). Auwärter, W. et al. Nature Nanotech. 7, 41–46 (2012). Heinrich, A. J., Lutz, C. P., Gupta, J. A. & Eigler, D. M. Science 298, 1381–1387 (2002). 5. Wintterlin, J., Völkening, S., Janssens, T. V. W., Zambelli, T. & Ertl, G. Science 278, 1931–1934 (1997). 1. 2. 3. 4.

BIOSYNTHESIS

Non-stick natural products

Fluorine imparts many drugs with beneficial properties, however, the synthesis of fluorinated complex natural products is challenging. Biosynthetic strategies and recent experimental precedents have paved the way for bioengineered fluorinated polyketides.

Peter A. Jordan and Bradley S. Moore

A

mong the brave, early pioneers of fluorine chemistry, few would have guessed that this perilous element — which is rarely found in natural molecules — would hold such a privileged place in modern drug design. Yet, the current pharmacopoeia consists of an ever-expanding array of organofluorine drugs. At recent counts, over 20% of all pharmaceuticals contain at least one fluorine atom1, and improvements in virtually all aspects of pharmacokinetics and pharmacodynamics have been demonstrated through the strategic placement of fluorine substituents2. Although significant progress has been made in developing safe and efficient methods of controlling organofluorine chemistry, many of the properties that make fluorine so attractive to medicinal chemists — such as electronegativity, sigma bond strength and high redox potential3 — also contribute to the difficulties of working with fluorine. Nonetheless, great strides in organofluorine chemistry have been made, and many labs have set their sights on synthesizing 10

fluorinated analogues of complex natural products. These non-fluorinated natural products are often discovered to be new lead compounds in drug-discovery programs4, and a simple method of incorporating fluorine into these structures could unlock its beneficial properties. Lab bench chemical synthesis can take advantage of transition metal catalysts, highly reactive fluorinating reagents and a waterfree environment4, whereas the biosynthetic strategies developed for this field so far depend on harnessing existing enzymatic reactions. With only a handful of fluorinecontaining natural products and only one enzyme that converts inorganic fluoride to an organofluorine species, the careful consideration of appropriate fluoride sources and enzymes is essential5. A recent report in Science by Michelle C. Y. Chang and coworkers is set to change this by introducing a general strategy for biosynthesizing fluorinated natural products via the incorporation of fluorinated building blocks6. Acetate is one of the principal building blocks in biology that gives rise to a diverse

range of metabolites — including fatty acids, polyketides (like the immunosuppressant rapamycin), and terpenoids (such as the anticancer agent taxol). In polyketide biosynthesis, acetate must first be loaded onto coenzyme A and carboxylated to form malonyl-CoA (Fig. 1). Simply substituting fluoroacetate for acetate at an early stage of biosynthesis could enable the design of fluorinated polyketide molecules. However, the extreme electronegativity of fluorine potentially compromises the stability of all biosynthetic intermediates. Despite these challenges, Chang and co-workers have described two distinct enzymatic approaches starting with fluoroacetate or fluoromalonate that converge to produce fluoromalonylCoA with high yields. A biosynthetic route to fluoromalonylCoA is only the beginning. In its most basic form, polyketide assembly involves sequential Claisen condensation reactions. This requires generation of a highly nucleophilic carbanion intermediate from a malonyl substrate bound to an acyl carrier protein, with the concurrent loss of CO2.

NATURE CHEMISTRY | VOL 6 | JANUARY 2014 | www.nature.com/naturechemistry

© 2014 Macmillan Publishers Limited. All rights reserved

news & views The carbanion intermediate then attacks a neighbouring thioester on a ketosynthase domain to construct a new C–C bond as part of an enzymatic assembly-line process; however, it was not known to what extent fluoride substitution disrupts this precisely choreographed event within the enzyme active site. Chang et al. report that a ketosynthase can not only perform this reaction on fluoromalonyl-CoA, but that it also occurs within individual modules of the polyketide synthase responsible for erythromycin biosynthesis — namely 6-deoxyerythronolide B synthase (DEBS). Furthermore, the resulting keto products are readily converted to the corresponding alcohol and cyclized, indicating that the ketoreductase and thioesterase activities within the module are maintained — even for the fluorinated analogues. The fact that this multistep process, which is characteristic of most polyketide synthases, is preserved for fluorinated analogues bodes very well for the engineered biosyntheses of complex, fluorinated natural products. The investigations of Chang and coworkers also revealed unique opportunities for site-specific incorporation of fluorine using the assembly line of enzymes that are used to synthesize polyketide natural products. Normally, the native selectivity of DEBS favours methylmalonyl-CoA over malonyl-CoA as a substrate. In the absence of methymalonyl-CoA, fluoromalonylCoA is incorporated, albeit less efficiently. When Chang et al. generated an inactive mutation within the acyltransferase (AT) domain — which is the gatekeeper of the synthase that selects the correct substrate — this reversed the selectivity of DEBS, enabling direct acylation with the more reactive fluoromalonyl-CoA. They also found that a discrete ‘trans’ acyltransferase with known substrate promiscuity could complement the biosynthesis and further improve fluoroketide yield. In this manner they demonstrated site-specific fluorine substitution using modules 2 and 3 of DEBS. The targeted deactivation of either AT domain, and complementation with a trans-AT, granted access to two distinct fluorotetraketides (Fig. 1b). The ability to perform similar acrobatics with the biosynthetic machinery within live cells would be extremely useful. Chang and co-workers predicted this could be achieved using Escherichia coli cells due to the absence of intracellular methylmalonyl-CoA pools and the natural exclusion of malonyl-CoA by DEBS. Therefore, DEBS modules should be able to load and use fluoromalonyl-CoA even in the presence of relatively high levels of malonyl-CoA. To test this hypothesis, E. coli cells were grown that expressed a

a

AT

O

O

1 R1 = H 2 R1 = CH3 3 R1 = F

TE

KR

KS

SCoA

O

ACP

O

R1 SH

O

SH

R2

R S

O

R1

O HO

AT

ACP

KS OH

O

R1

KS

HO

S

SH

AT

R2

Me

O

R

KS

SCoA Me

KS

SCoA

KR ACP TE

O HO

F

KS

AT

×

HO

F

Me

TE

AT trans

O

Me O

KR ACP

AT ACP

Me O

Me

DEBSMod2 + DEBSMod3 + TE + ATtrans

O

F

× AT

AT ACP

KS

O

O

R1

R2

Me

O

O O

DEBSMod2 + DEBSMod3 + TE

O S

O

S

S

b OH

TE

OH O

R1

ACP KR

S

SH

O HO

AT KS

TE

O

R1

O

R2

ACP KR

S

S

R2

OH

O

OH

OH

OH

TE

KR

O

R1

R2

KS

ACP

AT

TE

KR

O Me

O

Me

Figure 1 | The engineered biosynthesis of simple fluorinated polyketides is achieved with the introduction of fluoromalonyl-CoA as a substrate. a, Schematic showing the extension of a polyketide with malonylCoA derivatives (1, malonyl-CoA, R1 = H; 2, methylmalonyl-CoA, R1 = CH3; 3, fluoromalonyl-CoA, R1 = F). The domains participating in each step are highlighted in blue. KS = ketosynthase; AT = acyltransferase; KR = ketoreductase; ACP = acyl carrier protein; TE = thioesterase. b, Site-specific incorporation of fluorine into simple tetraketides using modules 2 and 3 of DEBS using selective AT domain mutation.

mini-polyketide synthase based on a single module of DEBS as well as MatB (an enzyme capable of generating fluoromalonyl-CoA from fluoromalonate). Feeding these cells with all of the necessary precursors — including fluoromalonate — led to exclusive production of a fluoroketide with no signs of the non-fluorinated analogues. Chang and co-workers have described several remarkable vignettes explaining how the fundamental stages of polyketide biosynthesis are amenable to programming

NATURE CHEMISTRY | VOL 6 | JANUARY 2014 | www.nature.com/naturechemistry

© 2014 Macmillan Publishers Limited. All rights reserved

the incorporation of fluorine. Their strategy ultimately results in a polyketide backbone where the methyl substituents are replaced with fluorine. This strategy is potentially very powerful; however, it remains to be seen whether this is the preferred substitution for designing fluorinated polyketide drugs. At the moment it is also unclear whether this strategy could be applied to other biosynthetic pathways. MethylmalonylCoA is largely found in actinobacterial 11

news & views pathways. The ability to use fluoromalonylCoA in place of malonyl-CoA would greatly expand the generality of this approach. Unfortunately, the biochemical tools currently available do not permit this swapping of substrates. It is also unclear how many fluorine substitutions are possible, or if Chang and co-workers’ methodology is compatible with a full-sized antibiotic macrolide like erythromycin. However, this is only the beginning of what

is an exciting future for biosynthesis. There is a range of possibilities for developing complementary pathways or applying these techniques to synthesizing fluorinated molecules. The foundation has now been set for the site-specific biosynthetic incorporation of fluorine into a broad spectrum of secondary metabolites. ❐ Peter A. Jordan and Bradley S. Moore are in the Center for Marine Biotechnology and Biomedicine, Scripps

Institution of Oceanography, University of California at San Diego, La Jolla, California 92093, USA. e-mail: [email protected] References 1. Müller, K., Faeh, C. & Diederich, F. Science 317, 1881–1886 (2007). 2. Hagmann, W. K. J. Med. Chem. 51, 4359–4369 (2008). 3. O’Hagan, D. Chem. Soc. Rev. 37, 308–319 (2008). 4. Furuya, T., Kamlet, A. S. & Ritter, T. Nature 473, 470–477 (2011). 5. Dong, C. et al. Nature 427, 561–565 (2004). 6. Walker, M. C. et al. Science 341, 1089–1094 (2013).

SEPARATED TANDEM CATALYSIS

It’s about time

One-pot processes in which a single catalyst carries out several reactions are attractive, but typically promote the formation of by-products as well as the desired ones, and are not amenable to optimization of the individual transformations. Now, these issues have been overcome by separating the catalytic processes in time.

Sarah Abou-Shehada and Jonathan M. J. Williams

C

12

R

H

Anti-Markovnikov alkyne hydration

O

R

Transfer hydrogenation R

OH

Yield

atalysis is one of the overarching principles of green chemistry capable of providing substantial environmental and economic benefits1. The past three decades have seen significant research dedicated to the development of novel catalytic systems, particularly in the area of transition metal catalysis. However, the impact of these advances has been limited by the current application of chemical transformations as singular events, which gives rise to by-products associated not only with the desired reaction but also with reaction auxiliaries and the purification of each intermediate. An ideal alternative would be to place all of the starting materials and catalysts in one vessel, where a series of reactions could take place to give the final product, avoiding the need for excess reaction auxiliaries, work-up and purification stages. Such an approach is known as a ‘one-pot’ process and has generated much interest within the synthetic community in an effort to enhance the efficiency of chemical synthesis. Tandem catalytic transformations are a subset of such reactions, and have been described by Fogg and dos Santos as “coupled catalyses in which sequential formation of the substrate occurs via two (or more) mechanistically distinct processes”2. Writing in Nature Chemistry, Le Li and Seth Herzon now argue that unwanted side reactions associated with one-pot processes can be removed by a temporal separation between the different catalytic cycles involved3. This decoupling of the catalytic cycles allows for each of the

Time

Figure 1 | Temporal separation of two catalytic cycles. All of the starting material (red) is converted into intermediate (green), which is then converted into the final product (yellow).

different transformations to be individually analysed and optimized. Tandem catalysis can be divided into three sub-categories: auto, assisted and orthogonal. Auto tandem catalysis involves two or more mechanistically distinct reactions that are promoted by a single catalyst, with both catalytic cycles occurring spontaneously by cooperative interaction of the species that are present from the outset of reaction. Assisted tandem catalysis also uses a single catalyst but requires a change in reaction conditions to bring about a shift from one catalytic mechanism to another; these transformations are not concurrent.

In contrast, orthogonal tandem catalysis requires the use of two or more catalysts that have distinct mechanisms operating concurrently, but, in a similar way to assisted tandem catalysis, the substrate is transformed sequentially. Auto and assisted tandem catalysis, which use a single (pre)catalyst, are more efficient in their use of the catalyst than their orthogonal counterpart. However, auto and orthogonal tandem catalysis offers higher process efficiency in terms of time and energy, as both sets of reactions occur under the same reaction conditions. The drawback to these methods is the inability

NATURE CHEMISTRY | VOL 6 | JANUARY 2014 | www.nature.com/naturechemistry

© 2014 Macmillan Publishers Limited. All rights reserved