Application of fragment screening and merging to the discovery of inhibitors of the mycobacterium tuberculosis cytochrome P450 CYP121

Article abstract of DOI:10.1002/anie.201202544

Pieces of the puzzle: The first fragment-based approach was used to target cytochrome P450 enzymes (CYPs) for drug development (see scheme). The experiments provide new insights into the binding site of the essential Mycobacterium tuberculosis CYP121 enzyme, and resulted in a promising novel lead compound based on fragment merging.

Full text of DOI:10.1002/anie.201202544

Angewandte
Chemie
DOI: 10.1002/anie.201202544
Drug Discovery
Application of Fragment Screening and Merging to the Discovery of
Inhibitors of the Mycobacterium tuberculosis Cytochrome P450
CYP121**
Sean A. Hudson, Kirsty J. McLean, Sachin Surade, Yong-Qing Yang, David Leys, Alessio Ciulli,
Andrew W. Munro, and Chris Abell*
The emergence of drug-resistant Mycobacterium tuberculosis
(Mtb) drives a critical need for new front-line tuberculosis
(TB) drugs with a novel mode of action.[1] It is estimated that
there are over 650000 cases of multidrug-resistant tuber-
culosis emerging every year, and that 1.3 million cases will
need to be treated by 2015, at a budgeted cost of over 16
billion US dollars.[2] The World Health Organization has
declared this epidemic a global health emergency.[3]
The success of drugs that inhibit biosynthetic cytochrome
P450 enzymes (CYPs), such as abiraterone, letrozole, and
voriconazole, has propelled research towards understanding
the unusually high number of CYPs (20) found encoded in the
Mtb H37Rv genome.[4] Of particular interest is the essential
CYP121 isoform, which has recently come into focus as an
enticing new anti-TB drug target.[4a–c,5] This biosynthetic CYP
appears to be exclusive to Mtb, and construction of an Mtb
chromosomal CYP121 knock-out mutant was only possible
when a complementary vector carrying CYP121 was pre-
sent.[4b,5b] CYP121 has also recently been shown to catalyze an
À
unusual intramolecular C C bond-forming reaction between
the ortho-positions of two tyrosines in cyclodityrosine (cYY)
to form mycocyclosin.[5a] While the physiological roles of cYY
and mycocyclosin remain to be determined, we believe that
the unique catalytic action of CYP121 will lead to selective
inhibitors. Specific inhibitors could also be used as chemical
probes to show how this pivotal enzyme relates to Mtb
infection, growth and persistence.
The only high-affinity ligands of CYP121 currently known
are azole antifungals (traditional fungal CYP51 inhibitors,
which act by way of type-II azole–heme coordination).[5b,6]
These compounds exhibit potent in vitro/in vivo antimyco-
bacterial activity, where their MIC (minimal inhibitory
concentration) values for Mtb H37Rv correlate with their
binding affinities to CYP121.[5b,7] However, the large flexible
antifungals also have broad overlapping CYP inhibition
profiles and are thus poor scaffolds for developing specific
inhibitors and front-line TB drug candidates.[4a,6b,8] Those
antifungals that are administered clinically, for example,
fluconazole and voriconazole, also only bind weakly to
CYP121.[5b,6a,8a] Furthermore, resistant Mtb mutants have
already been isolated that show upregulation of a transmem-
brane transporter protein believed to act as an azole efflux
pump.[9]
[*] S. A. Hudson, Dr. Y.-Q. Yang,[+] Dr. A. Ciulli, Prof. C. Abell
Department of Chemistry, University of Cambridge
Lensfield Road, Cambridge, CB2 1EW (UK)
E-mail: ca26@cam.ac.uk
Homepage: http://www-abell.ch.cam.ac.uk/
Dr. K. J. McLean, Prof. D. Leys, Prof. A. W. Munro
Manchester Interdisciplinary Biocentre, Faculty of Life Sciences,
University of Manchester
131 Princess Street, Manchester, M1 7DN (UK)
Fragment-based approaches represent a new method in
the field of developing small-molecule ligands as chemical
tools and leads for drug development.[10] This powerful
method involves the structure-guided design and synthesis
of potent ligands from weaker-binding low-molecular-weight
fragment molecules (typically < 250 Da).[10] Herein, we report
a fragment-based approach to targeting Mtb CYP121 in an
attempt to identify new inhibitory molecules and to explore
the active-site properties of the enzyme. Through an initial
fragment-screening cascade involving thermal shift, NMR
spectroscopy, and X-ray crystallography, four fragments were
found to bind within the CYP121 active site, in two over-
lapping groups. A direct fragment–fragment merging strategy
was implemented, leading to the discovery of a novel type-II
aminoquinoline inhibitor with high ligand efficiency (LE =
ÀDG of binding/non-hydrogen atoms (NHA) in the ligand)
and fourfold greater affinity than the natural CYP121
substrate cYY. This lead provides a pattern for CYP121-
specific inhibition and confirms the potential druggability of
CYP121. This study represents the first successful application
of fragment-based approaches to a cytochrome P450.
Dr. S. Surade
Department of Biochemistry, University of Cambridge
80 Tennis Court Road, Cambridge, CB2 1GA (UK)
[+] Present address: Shanghai ChemPartner Co. Ltd.
Shanghai, 201203 (China)
[**] We acknowledge funding from the EC (as part of the NM4TB
project) and the BBSRC (grants BB/I019227/1 and BB/I019669/1 to
C.A. and A.W.M., respectively). S.A.H. was supported by a Sir Mark
Oliphant Cambridge Australia Scholarship awarded by the Cam-
bridge Commonwealth Trust & Cambridge Overseas Trust, Uni-
versity of Cambridge. We give thanks to: Dr. J. Goodman (Cam-
bridge) for assisting with the in silico calculations; Dr. J. E. Davies
(Cambridge) for determining the small molecule crystal structure of
the 1,5-diphenoltriazole 7; Dr. A. Boodhun (Cambridge) for
performing mass spectrometry on recombinant His6-tagged
CYP121; Dr. C. Levy (Manchester) for help with synchrotron data
collection; Prof. T. L. Blundell (Cambridge) for helpful discussions.
Supporting information for this article (experimental details) is
available on the WWW under http://dx.doi.org/10.1002/anie.
201202544. Crystal structures of the CYP121-ligand complexes are
deposited in the Protein Data Bank (http://www.rcsb.org/pdb/)
under the following accession codes: 1: 4G44; 2: 4G45; 3: 4G46; 4:
4G47; 7: 4G2G; 10: 4G48; 14: 4G1X.
Angew. Chem. Int. Ed. 2012, 51, 1 – 7
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

.
Angewandte
Communications
A rule-of-three-compliant library of 665 commercially
available fragments (see Supporting Information) was ini-
tially screened for binding to recombinant CYP121 in a high-
throughput 96-well plate format, using fluorescence-based
thermal-shift analysis (Table 1 and Figure 1a). The melting
point/denaturing temperature (Tm) of CYP121 alone (10 mm)
was 50.1 Æ 0.18C (n = 20), and cYY used as a control (1 mm)
gave a DTm of 1.0 Æ 0.18C (n = 6). Fragment hits were defined
as those that increased the melting point (DTm) by more than
0.88C at a concentration of 5 mm; 66 hits were identified in
total (9.9% hit rate).
Fifty-six of the most chemically diverse hits from the
thermal shift screen were then screened by ligand-detected
1D 1H NMR spectroscopy using saturation transfer difference
(STD) and WaterLOGSY experiments (Table 1 and Fig-
ure 1b). Of these hits, 26 showed prominent interactions with
CYP121 and were partially displaced by the addition of cYY
(46% validated hit rate), indicating that they may bind within
the active site. The solubility of the displaced NMR hits was
then tested in the mother liquor used for CYP121 crystal-
lization at 1 mm, and eight fragments showing the greatest
solubility were studied by protein X-ray crystallography. We
also conducted an identical fragment screen in parallel against
a second Mtb CYP, the CYP125 cholesterol/cholestenone
carbon-27 monooxygenase, which gave nine hits in total.[16]
There was only a single hit common to both CYPs, indicating
a surprisingly high degree of isoform selectivity even at the
fragment-screening level.
Table 1: Fragment-screening statistics.
No. of compounds Hit rate[a]
[%]
Fragments screened by thermal shift
665
66
Thermal shift hits (DTm >0.88C)
9.9
3.9
CYP121 crystals were individually soaked with each of the
eight most soluble NMR-validated hits, leading to the
determination of four structures of CYP121 bound to frag-
ments at 1.24–1.53 ꢀ (Figure 2, see Figure S2 in the Support-
ing Information for the soaked
Fragments rescreened by 1D 1H NMR 56
cYY-displaced NMR hits
26
[a] Hit rate as a percentage of the total fragments screened.
fragments that did not yield frag-
ment-bound structures). All of
the fragments (1–4) bound within
the large water-filled CYP121
distal active site (Figure 2b),
and a superimposition of the
structures reveals that the frag-
ments form two overlapping
groups that occupy a near con-
tinuous stretch between the
heme iron and the top of the
active site cavity (Figure 2c).
Their
overlapping
binding
modes also define the proposed
CYP121 substrate entry channel/
trajectory,[11] and provide a clear
possibility for synthetic elabora-
tion by fragment merging. There
is no significant change in
CYP121 conformation upon
binding of the fragments com-
pared to the native structure (Ca
RMSD = 0.12–0.14 ꢀ).[11]
Fragments 1 and 2 coordinate
the heme iron through their aryl-
amine nitrogen in the sixth distal
ligand position (Figure 2b). This
type-II binding mode has not
been reported before for
Figure 1. Fragment screening by thermal shift, NMR spectroscopy, and ITC. a) Histogram of fragments
found to increase the melting point of CYP121 (10 mm) by fluorescence-based thermal-shift assay.
Fragments were present at 5 mm and CYP121 thermal denaturation was monitored by SYPRO Orange
1
CYP121. However
a
similar
(2.5ꢀ) fluorescence emission intensity. b) Water-suppressed 1D H and STD NMR spectra for
a representative fragment hit (1.5 mm) in the presence and absence of CYP121 (15 mm) or CYP121
(15 mm) plus cYY (0.5 mm). Only the fragment resonances in the aromatic region are shown and arrows
indicate the signals from cYY. The strong positive fragment signals in the presence of CYP121 indicate
protein binding, and this interaction is displaced/reduced by the addition of cYY. c) ITC binding
isotherm for a representative fragment hit (20 mm) titrated into CYP121 (100 mm; lower trace) or buffer
only (upper trace) as a negative control. The integrated enthalpy change for each injection as a function
of the molar ratio of total injected ligand to CYP121 is shown (squares), with the one-site binding
model fitted for estimating the KD (line).
mode of coordination has been
observed for arylamines binding
to the orphan Mtb CYP130.[12]
One of the amine protons is
involved in hydrogen bonding
(H-bonding) to the adjacent
CYP121 Ser237. The aromatic
2
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1 – 7
These are not the final page numbers!

Angewandte
Chemie
Figure 2. 1.24–1.53 ꢁ resolution X-ray crystal structures of fragment hits 1–4 binding to CYP121. a) Fragment hit chemical formulas drawn in their
predominant protonation state at pH 5-6 (CYP121 crystallization pH) and their associated KD values and calculated LEs (kcalmolÀ1 NHAÀ1),
determined by ITC. b) Fragments 1–4: thick sticks; heme b: purple; heme iron: brown sphere; I-helix: yellow; side chains/sulfates and waters
(within 5 ꢁ of the bound fragments): thin sticks and red spheres, respectively; direct fragment–CYP121 H-bonds and fragment–heme iron
coordination: dashed yellow lines. The FoÀFc omit electron density map associated with each fragment is shown as a green mesh contoured at
3s. c) Overlay of fragments 1–4 within the active site from the crystal structures. The large 1350 ꢁ3 active site cavity is bent around the I-helix
(yellow) that runs closely above the heme.
moieties of both fragments occupy a near parallel position to
the heme, suggesting possible aromatic-stacking effects with
the protoporphyrin IX. The quinoline nitrogen of fragment 2
also forms a water-mediated interaction with a heme carbox-
ylate. The 1,2,4-triazole of fragment 1 is modeled with the
nitrogen in the 4-position H-bonding directly to Asn85, based
on structure-activity relationships (SARs; see below).
Fragments 3 and 4 reside further from the heme coordi-
nation sphere, with their five-membered aromatic rings in an
offset p-stack with Phe168 (Figure 2b). Only the furan
oxygen of 3 appears to make a direct H-bond with CYP121
through Gln385. Fragment 4 lies in two configurations
centered around its triazole (each modeled at half occu-
pancy). In one configuration the phenol forms an H-bond
with Asn85, and in the other the phenol stacks between
Phe168 and Trp182. Its triazole 2- and 4-position nitrogen
atoms may H-bond to either Thr77 or Gln385.
It is interesting that the azole moiety in two of the
fragments (1 and 4) does not coordinate the heme iron
(Figure 2b), as is typically observed for the azole antifungal
CYP inhibitors.[13] We have seen the same thing for CYP121
binding by the antifungal fluconazole, where indirect coordi-
nation of an interstitial water is the dominant binding mode in
the crystal structure.[14] Our hypothesis is that orthogonal
heme coordination by N-heterocycles is disfavored in
CYP121 because of the unusually close I-helix residues
above the heme.
The CYP121-binding affinity of fragments 1–4 (equilibri-
um dissociation constants, KD) and calculated LEs (Fig-
ure 2a), were determined by isothermal titration calorimetry
(ITC) using direct titrations of the fragment into a solution of
CYP121 (see Figure 1c for example) or, when large heats of
dilution obscured the signal, by a competitive ITC experiment
titrating cYY into a solution of CYP121 pre-mixed with the
fragment. These fragments have KD values from 400 mm to
3 mm. The highest-affinity heme-coordinating fragment 2 has
a corresponding LE of 0.39 kcalmolÀ1 NHAÀ1. Interestingly,
the weakest-binding fragment 3 is also the one appearing to
make only a single direct H-bond with CYP121.
To increase the binding affinity of fragments 1–4 through
merging, several merged fragments and their structural
analogues were synthesized or obtained commercially
(Figure 3 and Supporting Information; Figure S3). Their
binding affinities (Figure 3) were determined by either ITC
as described above, or for the heme-coordinating amino-
quinolines 14–19, by a spectrophotometric assay monitoring
type-II red shifts in the heme absorbance band.
Merging the weaker fragments from the upper region of
the active site (3 and 4, or the two configurations of 4 alone) to
form compounds 5–11 did not lead to significantly higher
Angew. Chem. Int. Ed. 2012, 51, 1 – 7
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

.
Angewandte
Communications
Figure 3. Binding affinity and SAR of merged fragments and analogues. The parent fragment pairs (from Figure 2c) from which the merged
fragments were derived are shown as thick sticks. KD and LE values (kcalmolÀ1 NHAÀ1) were determined by ITC or, for 14–19, by
spectrophotometric heme-absorbance shift assay. Single determinations are rounded to the nearest 100 mm and repeat measurements are given
as the meanÆs.e.m. (n=3 or 4). Compound formulas are drawn in their predominant protonation state at pH 7.4. Synthesis of 14 is shown; the
syntheses of other analogues are shown in the Supporting Information, Figure S3.
affinity (Figure 3). The cyclohexanone-containing analogues
5 and 6 gave no ITC binding isotherms and were not
competitive with cYY in the NMR binding assays. This loss
of binding may be because of the relatively poor super-
position of fragments 3 and 4, or the loss of binding
interactions from replacing the carboxylate of 3 with the
phenol of 4. The biaryl-substituted azoles 7–9 show binding,
but with no significant improvement over their parent
fragment 4. In an attempt to understand the poor binding of
7–9, we performed quantum mechanical (QM) calculations
for the conformational energetics of 7 (Supporting Informa-
tion, Figure S4). These calculations suggest that a binding
position identical to fragment 4 (Figure 2b) would be
strained, with the ortho hydrogen atoms of 7 on opposite
phenol groups pointing/clashing towards each other. A small-
molecule crystal structure of 7 alone however shows crystal
packing in a conformation near the calculated energetic
minima (Supporting Information, Figure S4). Based on these
findings, we propose that the weak affinity of 7–9 could be the
result of a high conformational energy barrier necessary for
binding to CYP121. Moreover, the best compound of the
weak series 5–11, the phenoxypyrazole 10 (KD = 500 Æ
200 mm), potentially avoids such internal strain owing to the
oxygen separating its phenoxy and pyrazole rings.
Compounds 12–19, based on merging the two heme-
ligating fragments 1 and 2, proved significantly more success-
ful (Figure 3). All type-II-binding aminoquinolines 14–19
show improved affinity over their constituent azole fragment
1. The lead compound in the series is the 1,2,4-triazolylquino-
line 14 (Figure 3, boxed), which retains the original 1,2,4-
triazole subunit of fragment 1, has a KD of 28 Æ 4 mm (n = 4;
using both methods of affinity determination; see Figure 4b),
and high LE (0.39 kcalmolÀ1 NHAÀ1), equal to that of its root
fragments. Significantly, 14 has fourfold greater affinity than
the natural substrate cYY (KD = 110 Æ 12 mm by ITC, n = 7,[5a]
KD = 21.3 Æ 3.5 mm by spectral titrations) and inhibits CYP121
4
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1 – 7
These are not the final page numbers!

Angewandte
Chemie
and were synthesized based on the strategy used for 14.
Replacing the 2- or 4-position triazole nitrogen atoms of 14
À
with C H in the pyrazolyl-, imidazolyl- and 1,2,3-triazolyl
analogues 15–17 was not well tolerated. The larger ring sizes
of the thiazole or pyridine in 18 and 19 were also not
accommodated. These SAR data indicate that the 1,2,4-
triazole in fragment 1 and lead 14 may be optimal for the
CYP121 pocket defined by Asn85, Thr229, and Ala233,
because its 4-position nitrogen atom H-bonds to Asn85, and
À
its 2-position nitrogen atom (over C H) removes a steric
clash with Ala233. We have modeled the fragment 1–CYP121
structure accordingly (Figure 2b). The methyl from fragment
2 was intentionally excluded from analogues 12–19 for ease of
synthesis and because it appeared to point unfavorably into
the CYP121 water-filled pocket around the heme carbox-
ylates (Figure 2b). The analogue of fragment 2 without the
methyl group, 6-aminoquinoline (not included in the initial
fragment screen), was also tested and has a similar KD of
300 mm by ITC.
X-ray crystal structures of 7, 10, and 14 in complex with
CYP121 were obtained at 2.25, 1.50 and 1.30 ꢀ resolution,
respectively, showing that all three ligands adopt the binding
mode of their parent fragments (Figure 4a and Supporting
Information, Figure S5; compared to Figure 2b). Both the
weak 1,5-diphenoltriazole 7 and its higher affinity phenoxy-
pyrazole analogue 10 reside in the position distant from the
heme, like the two orientations of fragment 4. The phenol pair
of 7 appears to be sandwiched in the conformationally
strained state predicted by QM calculations (Supporting
Information, Figure S4c), that is, with its ortho hydrogen
atoms on opposite phenols pointing into each other to one
side of the triazole ring. The phenoxy sp3 oxygen atom of 10
clearly alleviates a clash of ortho hydrogen atoms by adding
a bond between the phenoxy and pyrazole rings, and also
allowing the phenoxy ring to twist out of the pyrazole plane.
Taken together, these findings form a pertinent illustration of
the importance of considering the conformational freedom of
Figure 4. Novel type-II binding to CYP121 by 14. a) The CYP121 active-
site complex with 14 (cyan) from a 1.30 ꢁ crystal structure. View and
colors are the same as those described in Figure 2b. b) Absorbance
difference spectra for CYP121 (5 mm) titrated with various concentra-
tions of 14. The red shift in the heme Soret absorbance band induced
by 14 (shown as a DAmax at 426.5 nm and DAmin at 395.5 nm in the
difference spectra) is indicative of inhibitory type-II CYP binding
associated with heme-iron coordination.[13] Inset: the change in heme
absorbance (quantified as DAmaxÀDAmin) as a function of 14 concen-
tration (squares) with the one-site binding equilibrium model fitted for
calculating the KD (line).
elaborated
fragments
made
using
fragment-based
approaches. It is worth noting that the ortho-hydroxy group
À
and pyrazole nitrogen atom (or N H tautomer) of 10 find
a double H-bond interaction with Thr77, which may add to its
improved affinity over 7.
catalytic activity with a Ki of 50 mm in a coupled-absorbance
enzyme assay with heterologous redox partners. 14 also shows
no affinity for Mtb CYP125 by ITC (data not shown),
consistent with the isoform specificity observed in the frag-
ment screens, and has weak to no inhibitory effect on several
major metabolizing human hepatic CYPs (IC50 for
CYP2C19 > 100 mm; 2D6 > 100 mm; 3A4 > 100 mm (probe
substrate testosterone); 3A4 = 65 mm (probe substrate mid-
azolam)).
The analogues of lead 14 synthesized in the series 12–19
shed light on its SAR (Figure 3). Removal of the heme-
coordinating amino group from 14 (see 13) or its nitro
precursor (see 12) resulted in over 90-fold weaker affinity.
The lead aminoquinoline 14 coordinates the heme iron
through its arylamine nitrogen atom (not the azole) face-on in
the distal coordination site, identical to fragments 1 and 2
(Figure 4a compared to Figure 2b). The orientation of Asn85
and its 1,2,4-triazole has been modeled to match the SAR of
the aminoquinoline analogues 14–19 discussed above. This
À
places the acidic triazole 5-position C H group of 14 in-line to
form a second water-bridged H-bond with a heme carboxyl-
ate. QM calculations for this CYP121-bound pose of 14 and
also the phenoxypyrazole 10 indicate no significant internal
strain relative to their predicted ground-state conformation in
the gas phase (data not shown).
This gives the amine
a group efficiency of 2.6 kcal
molÀ1 NHAÀ1 (GE = ÀDDG of binding/DNHA), clearly indi-
cating the critical strength of the heme-coordinating inter-
action. Analogues 15–19 were developed to explore binding
within the azole pocket along the I-helix defined by fragment
1 (assuming they all bind to CYP121 in the same position as 1)
This study represents the first fragment-based approach to
targeting a biosynthetic cytochrome P450. The findings are
also the first significant advance towards the rational design of
any Mtb CYP inhibitor and CYP-targeting TB drug candi-
date. The binding modes of the developed ligands have given
Angew. Chem. Int. Ed. 2012, 51, 1 – 7
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!

.
Angewandte
Communications
[4] a) S. A. Hudson, K. J. McLean, A. W. Munro, C. Abell, Biochem.
Soc. Trans. 2012, 40, 573; b) H. Ouellet, J. B. Johnston, P. R.
Ortiz de Montellano, Arch. Biochem. Biophys. 2010, 493, 82;
c) K. J. McLean, J. Belcher, M. D. Driscoll, C. C. Fernandez, D.
Le Van, S. Bui, M. Golovanova, A. W. Munro, Future Med.
Chem. 2010, 2, 1339; d) S. T. Cole, et al., Nature 1998, 393, 537.
[5] a) P. Belin, et al., Proc. Natl. Acad. Sci. USA 2009, 106, 7426;
b) K. J. McLean, et al., J. Biol. Chem. 2008, 283, 33406.
[6] a) K. J. McLean, et al., J. Inorg. Biochem. 2002, 91, 527; b) F. C.
Odds, A. J. Brown, N. A. Gow, Trends Microbiol. 2003, 11, 272.
[7] a) S. T. Byrne, S. M. Denkin, P. Gu, E. Nuermberger, Y. Zhang, J.
Med. Microbiol. 2007, 56, 1047; b) Z. Ahmad, S. Sharma, G. K.
Khuller, FEMS Microbiol. Lett. 2006, 258, 200; c) Z. Ahmad, S.
Sharma, G. K. Khuller, FEMS Microbiol. Lett. 2005, 251, 19.
[8] a) J. R. Perea, B. S. Diaz De Rada, E. G. Quetglas, M. J. Juarez,
Clin. Microbiol. Infect. 2004, 10 Suppl 1, 96; b) J. A. Como, W. E.
Dismukes, N. Engl. J. Med. 1994, 330, 263.
[9] A. Milano, M. R. Pasca, R. Provvedi, A. P. Lucarelli, G. Manina,
A. L. Ribeiro, R. Manganelli, G. Riccardi, Tuberculosis 2009, 89,
84.
[10] a) D. E. Scott, A. G. Coyne, S. A. Hudson, C. Abell, Biochem-
istry 2012, 51, 4990; b) C. W. Murray, M. L. Verdonk, D. C. Rees,
Trends Pharmacol. Sci. 2012, 33, 224; c) “Fragment-Based Drug
Discovery and X-Ray Crystallography”: Topics in Current
Chemistry, Vol. 317 (Eds.: T. G. Davies, M. Hyvçnen), Springer,
Heidelberg, 2012.
us unique insights into the active site properties of CYP121.
We were particularly fortunate to find fragments having the
correct overlapping binding poses to study using direct
fragment–fragment merging. This is not common in frag-
ment-based drug discovery, and usually hybrid merged
molecules are developed by combining fragment hits with
elements from a larger known substrate, cofactor, or inhib-
itor.[15] Our novel lead 14, with its high LE and selectivity,
provides an excellent scaffold for further type-II, reversible,
competitive CYP121 inhibitors. Linking 14 with the non-
heme coordinating fragments 3 and 4 presents an attractive
way forward, and we have begun assessing the anti-tubercular
efficacy of compounds developed in this series. The findings
lay the groundwork for the deployment of fragment-based
design to other members of the cytochrome P450 superfamily.
Characterization of fragment-binding modes to orphan CYPs
may even shed light on their likely functions at the substrate
level. For instance, it is intriguing that the CYP121 natural
substrate cYY is a biphenol and we also find the phenol
fragment 4 bound in two configurations (Figure 2b).
Received: April 1, 2012
Revised: June 10, 2012
[11] D. Leys, C. G. Mowat, K. J. McLean, A. Richmond, S. K.
Chapman, M. D. Walkinshaw, A. W. Munro, J. Biol. Chem.
2003, 278, 5141.
[12] L. M. Podust, H. Ouellet, J. P. von Kries, P. R. Ortiz de Montel-
lano, J. Biol. Chem. 2009, 284, 25211.
[13] P. R. Ortiz de Montellano, Cytochrome P450: Structure, Mech-
anism, and Biochemistry, 3rd ed., Kluwer Academic/Plenum
Publishers, New York, 2005.
[14] H. E. Seward, A. Roujeinikova, K. J. McLean, A. W. Munro, D.
Leys, J. Biol. Chem. 2006, 281, 39437.
Published online: && &&, &&&&
Keywords: cytochrome P450 · drug design ·
.
fragment screening · tuberculosis
[1] A. Koul, E. Arnoult, N. Lounis, J. Guillemont, K. Andries,
Nature 2011, 469, 483.
[2] a) World Health Organization, 2011/2012 Tuberculosis Global
Facts, 2011; b) World Health Organization, Multidrug and
extensively drug-resistant TB (M/XDR-TB): 2010 global
report on surveillance and response, 2010.
[3] World Health Organization, Global tuberculosis control: WHO
report 2011, 2011.
[15] a) D. A. Erlanson, R. S. McDowell, T. OꢁBrien, J. Med. Chem.
2004, 47, 3463; b) E. Edink, et al., J. Am. Chem. Soc. 2011, 133,
5363.
[16] S. A. Hudson, et al., unpublished results.
6
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1 – 7
These are not the final page numbers!

Angewandte
Chemie
Communications
Drug Discovery
Pieces of the puzzle: The first fragment-
based approach was used to target
cytochrome P450 enzymes (CYPs) for
drug development (see scheme). The
experiments provide new insights into the
binding site of the essential Mycobacte-
rium tuberculosis CYP121 enzyme, and
resulted in a promising novel lead com-
pound based on fragment merging.
S. A. Hudson, K. J. McLean, S. Surade,
Y.-Q. Yang, D. Leys, A. Ciulli, A. W. Munro,
C. Abell*
&&&& — &&&&
Application of Fragment Screening and
Merging to the Discovery of Inhibitors of
the Mycobacterium tuberculosis
Cytochrome P450 CYP121
Angew. Chem. Int. Ed. 2012, 51, 1 – 7
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7
These are not the final page numbers!