news & views follow from a single transition state. They show that to understand the selective formation of the natural product it is necessary to account for the inherent dynamical tendencies of the reactive intermediate, a carbocation, as it traverses the complex potential energy surface connecting the reactant to several possible products via intervening transition states. It is the motion, or dynamics, of the reactive species along this surface that defines the mechanism of the reaction, and considering transition-state energies alone cannot account for the biosynthetic selectivity. A plausible biosynthetic mechanism for formation of the diterpenoid miltiradiene involves the initial loss of a diphosphate leaving group to form a carbocationic intermediate, which, following successive rearrangements, may then be subsequently deprotonated to yield the natural product (Fig. 1b). The potential energy surface for this reaction as calculated by Hong and Tantillo is rather remarkable. Multiple, sequential bifurcations of the reaction path are found to occur following an initial dynamical bottleneck, which limits the overall rate of rearrangement. It is, however, when traversing the myriad downhill routes towards different products that selectivity is determined. Successive post-transition-state bifurcations enable the first transition state, in which interconversion of a tertiary to secondary carbocation takes place, to be connected to several isomeric intermediates via steepestdescent pathways on the potential energy surface without any intervening minima. Deprotonation of these intermediates yields different natural products. As mentioned above, the selectivity in this reaction results not just from the
location and stability of these various transition states, but also from inherent dynamic effects of the reactive intermediate in moving across the energy surface. To understand the post-transition-state dynamics of the carbocationic reactive species, Hong and Tantillo use so-called quasiclassical direct dynamics calculations5. In this approach the classical equations of motion are solved to produce dynamic trajectories along a potential energy surface obtained from an electronic structure calculation — here using density functional theory ‘on the fly’. Surprisingly, hundreds of trajectory calculations initiated in the region of the dynamical bottleneck indicate that, in spite of the complexity of the underlying energy surface, only two constitutionally distinct carbocation structures can be formed in appreciable amounts. One of these structures corresponds to the natural product’s direct precursor; however, the other possesses a diterpene skeleton previously unknown in nature. Thus, to ensure selective formation of miltiradiene the enzyme must exert control over the inherent (post-transition-state) dynamical preferences for traversing the underlying potential energy surface, tipping the balance in favour of forming cation A (Fig. 1c). The precise nature of how the enzyme achieves this control remains to be determined. It is, however, intriguing to consider whether the very existence of multiple bifurcations along a relatively flat energy surface in a biosynthetic mechanism could have a role in enabling nature to exert control over molecular diversity from a small number of starting materials6. These results from Hong and Tantillo4 illustrate how selectivity in organic and bioorganic reaction mechanisms may arise
from inherent dynamical tendencies along complex potential energy surfaces, and the necessity of considering much more than transition-state structures in such cases. The biosynthesis of miltiradiene is characterized by multiple, sequential reaction-path bifurcations, and it is likely that similar scenarios may now present themselves in the mechanisms of related carbocationic rearrangements, and perhaps more generally in organic and bioorganic reactions. The influence of post-transition-state dynamics on biosynthetic selectivity means we must consider a broader perspective of how enzymes may control, or perhaps more aptly, steer reactions towards a desired product. At present these calculations are restricted to the innate dynamic preferences of the reactive intermediate, and so we must await further studies that include interactions with the surrounding enzyme active site and/ or pyrophosphate counterion7. However, the tantalizing prospect that evolution of enzyme activity and selectivity has been influenced by such phenomena is raised by this work, and will no doubt inspire future exploration of this hypothesis. ❐ Robert S. Paton and Charles E. Hornsby are in the Chemistry Research Laboratory, University of Oxford, Mansfield Road, Oxford OX1 3TA, UK. e-mail: [email protected] References 1. Ess, D. H. et al. Angew. Chem. Int. Ed. 47, 7592–7601 (2008). 2. Rehbein, J. & Carpenter, B. K. Phys. Chem. Chem. Phys. 13, 20906–20922 (2011). 3. Thomas, J. B., Waas, J. R., Harmata, M. & Singleton, D. A. J. Am. Chem. Soc. 130, 14544–14555 (2008). 4. Hong, Y. J. & Tantillo, D. J. Nature Chem. 6, 104–111 (2014). 5. Paranjothy, M., Sun, R., Zhuang, Y. & Hase, W. L. WIREs Comput. Mol. Sci. 3, 296–316 (2013). 6. Hong, Y. J. & Tantillo, D. J. Nature Chem. 1, 384–389 (2009). 7. Sheppard, A. & Acevedo, O. J. Am. Chem. Soc. 131, 2530–2540 (2009).
ENZYMATIC C–H BOND ACTIVATION
Using push to get pull
Cytochrome P450 enzymes are able to oxidize substrates that are more inert than their own surrounding protein framework. Now, a quantitative understanding has emerged as to how the enzymes accomplish this remarkable feat.
John T. Groves
H
ow does nature break strong C–H bonds? By making stronger O–H bonds. This formalism is the central paradigm in our understanding of enzymatic hydroxylation of saturated hydrocarbons. These metal-catalysed reactions allow organisms to grow on alkanes as a carbon source. Also, C–H hydroxylations mediated
by enzymes of the cytochrome P450 family are central to phase I drug metabolism and steroid hormone biosynthesis1,2. The catalytic haem iron of cytochrome P450 enzymes — able to oxidize notoriously inert C–H bonds — have long been known to bear an unusual cysteine thiolate3 (Fig. 1a, left). Why sulfur? This chemically counterintuitive feature has
NATURE CHEMISTRY | VOL 6 | FEBRUARY 2014 | www.nature.com/naturechemistry
© 2014 Macmillan Publishers Limited. All rights reserved
been an enigma: thiolate is a strong electron donor to iron, and is easily oxidized, thus making for an unlikely candidate to generate an enzyme-bound oxidant strong enough to break the 100 kcal mol–1 C–H bond of a typical hydrocarbon (Fig. 1a). Finding an answer to this most intriguing question in oxidative catalysis has its origins 89
news & views a R O FeIV Cys
R
Pull
H
‡
O
+
Cys
S
H O
HAT
Fe
S
R
H
Push
FeIV Cys
P450-I
S
H
Rebound
O
R
FeIII Cys
S
P450-II
b
c
Figure 1 | Oxidation of C–H bonds by P450 enzymes, emphasizing the role of the thiolate ligand. a, Oxygen rebound reaction scheme showing push–pull assistance of hydrogen atom transfer (HAT) mediated by the haem centre of a P450 enzyme, P+–(Cys–S)Fe(iv)=O (P450-I) to afford the Cys–S– Fe(iv)O–H species (P450-II) and a substrate radical, ∙R. The porphyrin macrocycle P coordinating the iron centre is represented by an oval for clarity purposes. b, Active-site structure and protein scaffold of CYP119 (PDB 1IO7). The haem iron(iii) (orange), cysteine thiolate sulfur (yellow) and axial water oxygen (red) are highlighted. c, DFT-calculated, doubly occupied HOMO orbital chosen to illustrate the sulfur electron push to the ferryl oxygen in a haem model H3C-S-Fe(iv)=O. We thank Dina Sharon for assistance with the DFT.
in work by Green, Dawson and Gray on the haem–thiolate enzyme chloroperoxidase, with the recognition that its oxidized iron(iv) state compound II — reduced by one electron from the enzyme’s active intermediate (compound I) — was basic4. This meant that the Fe(iv) ferryl species was not the iron–oxo Cys–S–Fe(iv)=O, as had been presumed, but was instead a hydroxide, Cys–S–Fe(iv)–OH. The rationale for the basicity of the oxygen is that it derives from strong electron ‘push’ from the axial thiolate ligand. But how basic is the ferryl oxygen? And what does that have to do with the catalysis of C–H bond cleavage? A more complete and quantitative answer has emerged from recent work5 by Michael Green and co-workers, as described in Science. The researchers used a bacterial cytochrome P450, CYP158, which harbours a large, solvent-accessible active site near a tyrosine residue. Several staple bioinorganic spectroscopic techniques were used in concert to probe the ferryl protonation, Cys–S–Fe(iv)=Oδ– → Cys–S–Fe(iv)–OH. Rapid-mixing pH-jump experiments allowed the exploration of compound II of CYP158 (CYP158-II) over a wide range of →
90
pH. Concurrent changes in the UV–visible absorption and 57Fe Mössbauer spectral signatures of the haem indicated that there were two forms of CYP158-II, interconnected by a remarkably basic pKa = 12. A similar exploration was also carried out with a second P450 enzyme, a CYP119-II variant (Fig. 1b). Although the two enzymes have different active-site environments and substrate affinities, they both showed similarly high values of pKa for their compound II, pointing to the generality of this feature for P450 enzymes’ thiolate–haem groups. Both forms of CYP158-II displayed the ‘split’ Soret absorption band characteristic of thiolate binding in their UV–vis absorption spectra. Mössbauer spectra and the X-ray absorption edge data indicated that both forms were in the iron(iv) oxidation state. The protonated form, Cys–S–Fe(iv)–OH, showed an unusually large quadrupole splitting (2.0 mm s–1), similar to that of a well characterized dimethoxyiron(iv) porphyrin (MeO– Fe(iv)–OMe; ref. 2). This similarity indicates that the Fe=O bonding has decreased. Notably, there was a prominent
pre-edge feature in the XAS data for the deprotonated Fe=O state that was much weaker in the protonated form at lower pH, consistent with a more centrosymmetric structure on protonation. Indeed, fitting of the X-ray absorption fine structure data showed that protonation of Cys–S– Fe(iv)=O lengthened the Fe–O bond and shortened the Fe–S bond. Why is this basic Cys–S–Fe(iv)O–H pKa so informative of the mechanistic strategy for C–H bond cleavage by cytochrome P450? Because it is this FeO–H bond that is created during the scission of the substrate’s C–H bond, and it is this FeOH group that captures the resulting substrate radical in the product-forming oxygen rebound step (Fig. 1a). Finally, it is this FeO–H bond strength (D(OH)) that determines the thermodynamic driving force for hydrogen atom transfer from the substrate. Strong C–H bonds have D(CH) ~100 kcal mol–1. Recently, another enzyme family — the haem–thiolate aromatic peroxygenases (APO) — has been discovered to rival P450 enzymes in their ability to cleave strong C–H bonds. Kinetic analysis of hydrogen atom transfer rates in our group6,7 has shown that the bond strength D(OH) for APO-II must also be ~100 kcal mol–1. A similar value for P450 enzymes is thus expected. Thermodynamic considerations allow us to dissect the bond strength D(OH) into two components8: the Fe(iv)O–H pKa, and the one-electron reduction potential of P+–Fe(iv)=O (where P is the porphyrin macrocycle) in APO-I or P450-I (Fig. 1a). Thus, knowing the pKa of FeO–H allows the calculation of the reduction potential of compound I, which for APO-I is ~1.4 V (ref. 6). Returning to cytochrome P450-I, we can now understand how an electron ‘push’ effect from the trans-thiolate ligand can lead to greater ‘pull’ from the ferryl oxygen for C–H bond cleavage (Fig. 1a,c). The enzyme has engineered a clever strategy to increase the basicity of the ferryl oxygen of P450-II while sacrificing some of its redox potential. Insightful theoretical aspects of these effects have been discussed and reviewed by Shaik, Theil and colleagues9. A higher pKa in Fe(iv)O–H (that is, a more basic Fe=O) allows for a lower redox potential for the P+–Fe(iv)=O oxidant that is responsible for C–H bond cleavage. This arrangement will maintain the facile hydrogen atom transfer pathway, which can be seen as a kind of inner-sphere process (Fig. 1a), while discriminating against longer range, outer-sphere electrontransfer processes from the protein matrix. Tyrosine and tryptophan residues both have redox potentials around 1 V, or
NATURE CHEMISTRY | VOL 6 | FEBRUARY 2014 | www.nature.com/naturechemistry
© 2014 Macmillan Publishers Limited. All rights reserved
news & views higher if they are buried in the protein. This means that the rate of electron transfer from those sites will depend on the reduction potential of P450-I and the distance from those residues. Green and co-workers estimate that this trade-off between redox potential and a very basic pKa allows the cysteine thiolateligated CYP158 to produce an oxidant that is wired for C–H bond cleavage, taking advantage of the basic ferryl species while suppressing long-range electron transfer rates — by as much as 10,000-fold
compared with neutral histidine ligation that is typical of electron-transfer haem proteins such as peroxidases. Insights like this are important to understand the range of subtle strategies the enzyme has exploited to accomplish these very difficult C–H oxidations and to apply those strategies to the development of new catalysts. ❐ John T. Groves is in the Department of Chemistry at Princeton University, Princeton, New Jersey 08544, USA. e-mail: [email protected]
References 1. Ortiz de Montellano, P. R. Chem. Rev. 110, 932–948 (2010). 2. Groves, J. T. in Cytochrome P450: Structure, Mechanism, and Biochemistry 3rd edn (ed. Ortiz de Montellano, P. R.) 1–44 (Klewer Academic/Plenum, 2005). 3. Poulos, T. L., Finzel, B. C. & Howard, A. J. J. Mol. Biol. 195, 687–700 (1987). 4. Green, M. T., Dawson, J. H. & Gray, H. B. Science 304, 1653–1656 (2004). 5. Yosca, T. H. et al. Science 342, 825–829 (2013). 6. Wang, X., Peter, S., Ullrich, R., Hofrichter, M. & Groves, J. T. Angew. Chem. Int. Ed. 52, 9238–9241 (2013). 7. Wang, X., Peter, S., Kinne, M., Hofrichter, M. & Groves, J. T. J. Am. Chem. Soc. 134, 12897–12900 (2012). 8. Warren, J. J., Tronic, T. A. & Mayer, J. M. Chem. Rev. 110, 6961–7001 (2010). 9. Shaik, S. et al. Chem. Rev. 110, 949–1017 (2010).
GRAPHENE SYNTHESIS
Nanoribbons from the bottom-up
The organic synthesis of graphene nanostructures requires exceptionally efficient chemistry and is made more challenging by difficulties in characterization and processing. Now, solution-dispersible graphene nanoribbons have been synthesized on the gram scale.
C. Scott Hartley
G
raphene has been the focus of intense research effort since it was first isolated and characterized in 20041, principally because of its potential in next-generation electronics. Substructures of graphene represent attractive targets for new materials, with properties intimately tied to their sizes and the arrangement of the six-membered rings along their edges (for example, so-called zigzag or armchair patterns). Fundamentally, graphene fragments can be constructed in two different ways. ‘Top-down’ approaches typically begin with exfoliation of graphite through physical or chemical means, and the resulting separated graphene sheets can be patterned by methods such as lithography. An alternative approach is to construct graphene structures from the ‘bottomup’, either through epitaxial growth of graphene sheets (by using techniques such as chemical vapour deposition)2 or by applying the principles and procedures of synthetic organic chemistry. Ideally, stepwise organic chemistry strategies would enable graphene-based materials to be produced in a massively parallel fashion with very high structural precision, and offer a complementary approach to other methods. Although the synthesis of nanometre-scale polycyclic aromatic hydrocarbons predates the current interest in graphene by decades (for example, the synthesis of hexa-peribenzocoronene in 1958), our capabilities
on this front remain fairly limited. Key challenges include (i) the development of sufficiently efficient chemistry for the creation of large numbers of C–C bonds in a single step, (ii) the management of the low solubility typical of large aromatics, which limits the processability of the resulting materials, and (iii) the characterization of nanometre-scale graphenes, for which even small numbers of undetected structural defects could have significant effects on their properties. Now, writing in Nature Chemistry, Müllen, Feng and co-workers describe3 the design, synthesis and characterization of soluble graphene nanoribbons (GNRs). Because the confinement of graphene to the nanometre scale in one dimension opens up a width-dependent bandgap — which is critically important in electronics applications4 — GNRs represent a privileged class of graphene-based structures. They can be made using various top-down routes, including the lithographic patterning of larger graphene sheets or the ‘unzipping’ of carbon nanotubes4. Although effective, these methods offer limited control over edge structures and ribbon widths. Bottom-up approaches have also been reported, including examples of surface-mediated planarizations of polyphenylene precursors to give GNRs with incredible structural precision5. The drawback of this method, however, is that only limited quantities of surface-bound products are produced.
NATURE CHEMISTRY | VOL 6 | FEBRUARY 2014 | www.nature.com/naturechemistry
© 2014 Macmillan Publishers Limited. All rights reserved
An attractive route to GNRs would be to prepare them by analogy with more ‘traditional’ conjugated polymers — such as polythiophenes or poly(para-phenylene)s — beginning with small-molecule monomers and carrying out a bulk solution-phase synthesis. Indeed, some examples of this approach towards GNRs have been reported previously 6, but the resulting nanoribbons are relatively short (
location and stability of these various transition states, but also from inherent dynamic effects of the reactive intermediate in moving across the energy surface. To understand the post-transition-state dynamics of the carbocationic reactive species, Hong and Tantillo use so-called quasiclassical direct dynamics calculations5. In this approach the classical equations of motion are solved to produce dynamic trajectories along a potential energy surface obtained from an electronic structure calculation — here using density functional theory ‘on the fly’. Surprisingly, hundreds of trajectory calculations initiated in the region of the dynamical bottleneck indicate that, in spite of the complexity of the underlying energy surface, only two constitutionally distinct carbocation structures can be formed in appreciable amounts. One of these structures corresponds to the natural product’s direct precursor; however, the other possesses a diterpene skeleton previously unknown in nature. Thus, to ensure selective formation of miltiradiene the enzyme must exert control over the inherent (post-transition-state) dynamical preferences for traversing the underlying potential energy surface, tipping the balance in favour of forming cation A (Fig. 1c). The precise nature of how the enzyme achieves this control remains to be determined. It is, however, intriguing to consider whether the very existence of multiple bifurcations along a relatively flat energy surface in a biosynthetic mechanism could have a role in enabling nature to exert control over molecular diversity from a small number of starting materials6. These results from Hong and Tantillo4 illustrate how selectivity in organic and bioorganic reaction mechanisms may arise
from inherent dynamical tendencies along complex potential energy surfaces, and the necessity of considering much more than transition-state structures in such cases. The biosynthesis of miltiradiene is characterized by multiple, sequential reaction-path bifurcations, and it is likely that similar scenarios may now present themselves in the mechanisms of related carbocationic rearrangements, and perhaps more generally in organic and bioorganic reactions. The influence of post-transition-state dynamics on biosynthetic selectivity means we must consider a broader perspective of how enzymes may control, or perhaps more aptly, steer reactions towards a desired product. At present these calculations are restricted to the innate dynamic preferences of the reactive intermediate, and so we must await further studies that include interactions with the surrounding enzyme active site and/ or pyrophosphate counterion7. However, the tantalizing prospect that evolution of enzyme activity and selectivity has been influenced by such phenomena is raised by this work, and will no doubt inspire future exploration of this hypothesis. ❐ Robert S. Paton and Charles E. Hornsby are in the Chemistry Research Laboratory, University of Oxford, Mansfield Road, Oxford OX1 3TA, UK. e-mail: [email protected] References 1. Ess, D. H. et al. Angew. Chem. Int. Ed. 47, 7592–7601 (2008). 2. Rehbein, J. & Carpenter, B. K. Phys. Chem. Chem. Phys. 13, 20906–20922 (2011). 3. Thomas, J. B., Waas, J. R., Harmata, M. & Singleton, D. A. J. Am. Chem. Soc. 130, 14544–14555 (2008). 4. Hong, Y. J. & Tantillo, D. J. Nature Chem. 6, 104–111 (2014). 5. Paranjothy, M., Sun, R., Zhuang, Y. & Hase, W. L. WIREs Comput. Mol. Sci. 3, 296–316 (2013). 6. Hong, Y. J. & Tantillo, D. J. Nature Chem. 1, 384–389 (2009). 7. Sheppard, A. & Acevedo, O. J. Am. Chem. Soc. 131, 2530–2540 (2009).
ENZYMATIC C–H BOND ACTIVATION
Using push to get pull
Cytochrome P450 enzymes are able to oxidize substrates that are more inert than their own surrounding protein framework. Now, a quantitative understanding has emerged as to how the enzymes accomplish this remarkable feat.
John T. Groves
H
ow does nature break strong C–H bonds? By making stronger O–H bonds. This formalism is the central paradigm in our understanding of enzymatic hydroxylation of saturated hydrocarbons. These metal-catalysed reactions allow organisms to grow on alkanes as a carbon source. Also, C–H hydroxylations mediated
by enzymes of the cytochrome P450 family are central to phase I drug metabolism and steroid hormone biosynthesis1,2. The catalytic haem iron of cytochrome P450 enzymes — able to oxidize notoriously inert C–H bonds — have long been known to bear an unusual cysteine thiolate3 (Fig. 1a, left). Why sulfur? This chemically counterintuitive feature has
NATURE CHEMISTRY | VOL 6 | FEBRUARY 2014 | www.nature.com/naturechemistry
© 2014 Macmillan Publishers Limited. All rights reserved
been an enigma: thiolate is a strong electron donor to iron, and is easily oxidized, thus making for an unlikely candidate to generate an enzyme-bound oxidant strong enough to break the 100 kcal mol–1 C–H bond of a typical hydrocarbon (Fig. 1a). Finding an answer to this most intriguing question in oxidative catalysis has its origins 89
news & views a R O FeIV Cys
R
Pull
H
‡
O
+
Cys
S
H O
HAT
Fe
S
R
H
Push
FeIV Cys
P450-I
S
H
Rebound
O
R
FeIII Cys
S
P450-II
b
c
Figure 1 | Oxidation of C–H bonds by P450 enzymes, emphasizing the role of the thiolate ligand. a, Oxygen rebound reaction scheme showing push–pull assistance of hydrogen atom transfer (HAT) mediated by the haem centre of a P450 enzyme, P+–(Cys–S)Fe(iv)=O (P450-I) to afford the Cys–S– Fe(iv)O–H species (P450-II) and a substrate radical, ∙R. The porphyrin macrocycle P coordinating the iron centre is represented by an oval for clarity purposes. b, Active-site structure and protein scaffold of CYP119 (PDB 1IO7). The haem iron(iii) (orange), cysteine thiolate sulfur (yellow) and axial water oxygen (red) are highlighted. c, DFT-calculated, doubly occupied HOMO orbital chosen to illustrate the sulfur electron push to the ferryl oxygen in a haem model H3C-S-Fe(iv)=O. We thank Dina Sharon for assistance with the DFT.
in work by Green, Dawson and Gray on the haem–thiolate enzyme chloroperoxidase, with the recognition that its oxidized iron(iv) state compound II — reduced by one electron from the enzyme’s active intermediate (compound I) — was basic4. This meant that the Fe(iv) ferryl species was not the iron–oxo Cys–S–Fe(iv)=O, as had been presumed, but was instead a hydroxide, Cys–S–Fe(iv)–OH. The rationale for the basicity of the oxygen is that it derives from strong electron ‘push’ from the axial thiolate ligand. But how basic is the ferryl oxygen? And what does that have to do with the catalysis of C–H bond cleavage? A more complete and quantitative answer has emerged from recent work5 by Michael Green and co-workers, as described in Science. The researchers used a bacterial cytochrome P450, CYP158, which harbours a large, solvent-accessible active site near a tyrosine residue. Several staple bioinorganic spectroscopic techniques were used in concert to probe the ferryl protonation, Cys–S–Fe(iv)=Oδ– → Cys–S–Fe(iv)–OH. Rapid-mixing pH-jump experiments allowed the exploration of compound II of CYP158 (CYP158-II) over a wide range of →
90
pH. Concurrent changes in the UV–visible absorption and 57Fe Mössbauer spectral signatures of the haem indicated that there were two forms of CYP158-II, interconnected by a remarkably basic pKa = 12. A similar exploration was also carried out with a second P450 enzyme, a CYP119-II variant (Fig. 1b). Although the two enzymes have different active-site environments and substrate affinities, they both showed similarly high values of pKa for their compound II, pointing to the generality of this feature for P450 enzymes’ thiolate–haem groups. Both forms of CYP158-II displayed the ‘split’ Soret absorption band characteristic of thiolate binding in their UV–vis absorption spectra. Mössbauer spectra and the X-ray absorption edge data indicated that both forms were in the iron(iv) oxidation state. The protonated form, Cys–S–Fe(iv)–OH, showed an unusually large quadrupole splitting (2.0 mm s–1), similar to that of a well characterized dimethoxyiron(iv) porphyrin (MeO– Fe(iv)–OMe; ref. 2). This similarity indicates that the Fe=O bonding has decreased. Notably, there was a prominent
pre-edge feature in the XAS data for the deprotonated Fe=O state that was much weaker in the protonated form at lower pH, consistent with a more centrosymmetric structure on protonation. Indeed, fitting of the X-ray absorption fine structure data showed that protonation of Cys–S– Fe(iv)=O lengthened the Fe–O bond and shortened the Fe–S bond. Why is this basic Cys–S–Fe(iv)O–H pKa so informative of the mechanistic strategy for C–H bond cleavage by cytochrome P450? Because it is this FeO–H bond that is created during the scission of the substrate’s C–H bond, and it is this FeOH group that captures the resulting substrate radical in the product-forming oxygen rebound step (Fig. 1a). Finally, it is this FeO–H bond strength (D(OH)) that determines the thermodynamic driving force for hydrogen atom transfer from the substrate. Strong C–H bonds have D(CH) ~100 kcal mol–1. Recently, another enzyme family — the haem–thiolate aromatic peroxygenases (APO) — has been discovered to rival P450 enzymes in their ability to cleave strong C–H bonds. Kinetic analysis of hydrogen atom transfer rates in our group6,7 has shown that the bond strength D(OH) for APO-II must also be ~100 kcal mol–1. A similar value for P450 enzymes is thus expected. Thermodynamic considerations allow us to dissect the bond strength D(OH) into two components8: the Fe(iv)O–H pKa, and the one-electron reduction potential of P+–Fe(iv)=O (where P is the porphyrin macrocycle) in APO-I or P450-I (Fig. 1a). Thus, knowing the pKa of FeO–H allows the calculation of the reduction potential of compound I, which for APO-I is ~1.4 V (ref. 6). Returning to cytochrome P450-I, we can now understand how an electron ‘push’ effect from the trans-thiolate ligand can lead to greater ‘pull’ from the ferryl oxygen for C–H bond cleavage (Fig. 1a,c). The enzyme has engineered a clever strategy to increase the basicity of the ferryl oxygen of P450-II while sacrificing some of its redox potential. Insightful theoretical aspects of these effects have been discussed and reviewed by Shaik, Theil and colleagues9. A higher pKa in Fe(iv)O–H (that is, a more basic Fe=O) allows for a lower redox potential for the P+–Fe(iv)=O oxidant that is responsible for C–H bond cleavage. This arrangement will maintain the facile hydrogen atom transfer pathway, which can be seen as a kind of inner-sphere process (Fig. 1a), while discriminating against longer range, outer-sphere electrontransfer processes from the protein matrix. Tyrosine and tryptophan residues both have redox potentials around 1 V, or
NATURE CHEMISTRY | VOL 6 | FEBRUARY 2014 | www.nature.com/naturechemistry
© 2014 Macmillan Publishers Limited. All rights reserved
news & views higher if they are buried in the protein. This means that the rate of electron transfer from those sites will depend on the reduction potential of P450-I and the distance from those residues. Green and co-workers estimate that this trade-off between redox potential and a very basic pKa allows the cysteine thiolateligated CYP158 to produce an oxidant that is wired for C–H bond cleavage, taking advantage of the basic ferryl species while suppressing long-range electron transfer rates — by as much as 10,000-fold
compared with neutral histidine ligation that is typical of electron-transfer haem proteins such as peroxidases. Insights like this are important to understand the range of subtle strategies the enzyme has exploited to accomplish these very difficult C–H oxidations and to apply those strategies to the development of new catalysts. ❐ John T. Groves is in the Department of Chemistry at Princeton University, Princeton, New Jersey 08544, USA. e-mail: [email protected]
References 1. Ortiz de Montellano, P. R. Chem. Rev. 110, 932–948 (2010). 2. Groves, J. T. in Cytochrome P450: Structure, Mechanism, and Biochemistry 3rd edn (ed. Ortiz de Montellano, P. R.) 1–44 (Klewer Academic/Plenum, 2005). 3. Poulos, T. L., Finzel, B. C. & Howard, A. J. J. Mol. Biol. 195, 687–700 (1987). 4. Green, M. T., Dawson, J. H. & Gray, H. B. Science 304, 1653–1656 (2004). 5. Yosca, T. H. et al. Science 342, 825–829 (2013). 6. Wang, X., Peter, S., Ullrich, R., Hofrichter, M. & Groves, J. T. Angew. Chem. Int. Ed. 52, 9238–9241 (2013). 7. Wang, X., Peter, S., Kinne, M., Hofrichter, M. & Groves, J. T. J. Am. Chem. Soc. 134, 12897–12900 (2012). 8. Warren, J. J., Tronic, T. A. & Mayer, J. M. Chem. Rev. 110, 6961–7001 (2010). 9. Shaik, S. et al. Chem. Rev. 110, 949–1017 (2010).
GRAPHENE SYNTHESIS
Nanoribbons from the bottom-up
The organic synthesis of graphene nanostructures requires exceptionally efficient chemistry and is made more challenging by difficulties in characterization and processing. Now, solution-dispersible graphene nanoribbons have been synthesized on the gram scale.
C. Scott Hartley
G
raphene has been the focus of intense research effort since it was first isolated and characterized in 20041, principally because of its potential in next-generation electronics. Substructures of graphene represent attractive targets for new materials, with properties intimately tied to their sizes and the arrangement of the six-membered rings along their edges (for example, so-called zigzag or armchair patterns). Fundamentally, graphene fragments can be constructed in two different ways. ‘Top-down’ approaches typically begin with exfoliation of graphite through physical or chemical means, and the resulting separated graphene sheets can be patterned by methods such as lithography. An alternative approach is to construct graphene structures from the ‘bottomup’, either through epitaxial growth of graphene sheets (by using techniques such as chemical vapour deposition)2 or by applying the principles and procedures of synthetic organic chemistry. Ideally, stepwise organic chemistry strategies would enable graphene-based materials to be produced in a massively parallel fashion with very high structural precision, and offer a complementary approach to other methods. Although the synthesis of nanometre-scale polycyclic aromatic hydrocarbons predates the current interest in graphene by decades (for example, the synthesis of hexa-peribenzocoronene in 1958), our capabilities
on this front remain fairly limited. Key challenges include (i) the development of sufficiently efficient chemistry for the creation of large numbers of C–C bonds in a single step, (ii) the management of the low solubility typical of large aromatics, which limits the processability of the resulting materials, and (iii) the characterization of nanometre-scale graphenes, for which even small numbers of undetected structural defects could have significant effects on their properties. Now, writing in Nature Chemistry, Müllen, Feng and co-workers describe3 the design, synthesis and characterization of soluble graphene nanoribbons (GNRs). Because the confinement of graphene to the nanometre scale in one dimension opens up a width-dependent bandgap — which is critically important in electronics applications4 — GNRs represent a privileged class of graphene-based structures. They can be made using various top-down routes, including the lithographic patterning of larger graphene sheets or the ‘unzipping’ of carbon nanotubes4. Although effective, these methods offer limited control over edge structures and ribbon widths. Bottom-up approaches have also been reported, including examples of surface-mediated planarizations of polyphenylene precursors to give GNRs with incredible structural precision5. The drawback of this method, however, is that only limited quantities of surface-bound products are produced.
NATURE CHEMISTRY | VOL 6 | FEBRUARY 2014 | www.nature.com/naturechemistry
© 2014 Macmillan Publishers Limited. All rights reserved
An attractive route to GNRs would be to prepare them by analogy with more ‘traditional’ conjugated polymers — such as polythiophenes or poly(para-phenylene)s — beginning with small-molecule monomers and carrying out a bulk solution-phase synthesis. Indeed, some examples of this approach towards GNRs have been reported previously 6, but the resulting nanoribbons are relatively short (
