Substrate Fragmentation for the Design of M. tuberculosis CYP121 Inhibitors

Article abstract of DOI:10.1002/cmdc.201600248

The cyclo-dipeptide substrates of the essential M. tuberculosis (Mtb) enzyme CYP121 were deconstructed into their component fragments and screened against the enzyme. A number of hits were identified, one of which exhibited an unexpected inhibitor-like binding mode. The inhibitory pharmacophore was elucidated, and fragment binding affinity was rapidly improved by synthetic elaboration guided by the structures of CYP121 substrates. The resulting inhibitors have low micromolar affinity, good predicted physicochemical properties and selectivity for CYP121 over other Mtb P450s. Spectroscopic characterisation of the inhibitors′ binding mode provides insight into the effect of weak nitrogen-donor ligands on the P450 heme, an improved understanding of factors governing CYP121–ligand recognition and speculation into the biological role of the enzyme for Mtb.

Full text of DOI:10.1002/cmdc.201600248

DOI: 10.1002/cmdc.201600248
Full Papers
Substrate Fragmentation for the Design of M. tuberculosis
CYP121 Inhibitors
Madeline E. Kavanagh,[a] Janine L. Gray,[a] Sophie H. Gilbert,[a] Anthony G. Coyne,[a]
Kirsty J. McLean,[b] Holly J. Davis,[a] Andrew W. Munro,[b] and Chris Abell*[a]
The cyclo-dipeptide substrates of the essential M. tuberculosis
(Mtb) enzyme CYP121 were deconstructed into their compo-
nent fragments and screened against the enzyme. A number
of hits were identified, one of which exhibited an unexpected
inhibitor-like binding mode. The inhibitory pharmacophore
was elucidated, and fragment binding affinity was rapidly im-
proved by synthetic elaboration guided by the structures of
CYP121 substrates. The resulting inhibitors have low micromo-
lar affinity, good predicted physicochemical properties and se-
lectivity for CYP121 over other Mtb P450s. Spectroscopic char-
acterisation of the inhibitors’ binding mode provides insight
into the effect of weak nitrogen-donor ligands on the P450
heme, an improved understanding of factors governing
CYP121–ligand recognition and speculation into the biological
role of the enzyme for Mtb.
Introduction
Enzymes have evolved to bind substrates specifically. However,
the binding energetics of enzyme–substrate interactions are
optimised for catalytic efficiency, and therefore implicitly, they
should not bind too tightly. Consequently, the contribution to
the free energy of binding of individual structural motifs will
be unevenly distributed over the enzyme–substrate complex
and the relative importance of these interactions are difficult
to assess when combined in a large molecule.[1,2] In this way,
large (>250 Da) substrates are analogous to large molecule
drug leads or hits that are identified from a high throughput
screen. These compounds typically bind in the low micromolar
range by making multiple suboptimal interactions, and as
such, may not represent the best starting points for elabora-
tion.[3–7]
binding of a large substrate or a hit from a high throughput
screen. Fragment-based methods are now firmly established as
valuable techniques for identifying ligand efficient small mole-
cules that can act as leads for drug development.[13,14] The ap-
plication of fragment screening as a tool to assess target
druggability, identify binding hotspots, and interrogate biologi-
cal systems is also being increasingly appreciated.[15–17] The util-
ity of fragments for these applications is a consequence of
their small size and inherent ability to probe macromolecular
surfaces more effectively than larger compounds.[4,6]
The question as to whether fragments of a larger ligand
maintain the same binding interactions when they are discon-
nected has been addressed in recent years.[8–12,18–20] The size
and structures of the disconnected fragments, the choice of
disconnection, properties of the target macromolecule, and re-
quirements for cofactors or binding cooperativity, are all fac-
tors that have contributed to the variation in the results ob-
served. What is consistently apparent though, is that fragments
derived from large ligands bind preferentially to energetic hot-
spots and, regardless whether this binding mode recapitulates
that of the larger ligand, characterisation of these interactions
provides valuable insight for subsequent ligand design.[21,22]
Furthermore, lead deconstruction or fragmentation has result-
ed in the identification of novel chemical scaffolds for use as
drug leads and previously unknown binding sites, as well as
contributing to understanding the mechanisms of biomolecu-
lar recognition or catalysis; outcomes which are ultimately, of
more value.[8,9]
The deconstruction of large molecules into their representa-
tive fragments can provide an assessment of the binding con-
tributions of individual structural motifs, identify the minimal
pharmacophore required for biological activity and provide in-
sight into the fundamental basis of enzyme–substrate recogni-
tion.[8–12] This principle can be applied to deconvoluting the
[a] M. E. Kavanagh, J. L. Gray, S. H. Gilbert, Dr. A. G. Coyne, H. J. Davis,
Prof. C. Abell
Department of Chemistry, The University of Cambridge,
Lensfield Road, Cambridge, CB2 1EW (UK)
E-mail: ca26@cam.ac.uk
[b] Dr. K. J. McLean, Prof. A. W. Munro
Centre for Synthetic Biology of Fine and Specialty Chemicals
(SYNBIOCHEM), Manchester Institute of Biotechnology, Faculty of Life-
Sciences, The University of Manchester, Manchester M1 7DN (UK)
Although the retrospective deconstruction of successful
drug candidates, lead compounds and enzyme substrates has
been detailed in the literature,[9–12,19,23,24] there are few exam-
ples where the fragments derived in this way have been pur-
sued for subsequent ligand development.[25] Here, the dipep-
tide substrates of the cytochrome P450 enzyme CYP121 from
Supporting information for this article can be found under http://
dx.doi.org/10.1002/cmdc.201600248.
ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA.
This is an open access article under the terms of the Creative Commons
Attribution License, which permits use, distribution and reproduction in
any medium, provided the original work is properly cited.
ChemMedChem 2016, 11, 1924 –1935
1924 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Mycobacterium tuberculosis (Mtb) have been deconstructed as
a route to identifying novel scaffolds for inhibitor develop-
ment. The CYP121 gene is essential for Mtb survival in vitro
and the enzyme is believed to be the molecular target respon-
sible for the anti-tubercular efficacy of azole antifungal com-
pounds.[26–30] CYP121 catalyses an unusual biosynthetic C-H ac-
tivation reaction that cyclises the dipeptide cyclo-l-Tyr-l-Tyr
(cYY) to form mycocyclosin (Figure 1).[31] The aromatic dipepti-
des cyclo-l-Tyr-l-Trp (cYW) and cyclo-l-Tyr-l-Phe (cYF), which
are the minor products of the cyclo-dipeptide synthetase
Rv2275 that is encoded directly upstream of the CYP121 gene
(rv2276) in Mtb H37Rv,[32] were also recently reported to bind
to CYP121 and catalytic turnover was detected for cYW.[33]
We have previously reported the development of potent li-
gands against CYP121, however, these compounds had subop-
timal cellular activity, owing to either poor permeability or
efflux.[25] In the current study, it was reasoned that the decon-
struction of the CYP121 substrates into their component frag-
ments might provide access to novel scaffolds and insight into
the interactions governing enzyme-ligand recognition. Sub-
strate fragments were preferentially identified to bind to
CYP121 when they were screened as part of a library of chemi-
cally similar compounds by thermal shift and UV/vis spectros-
copy assays. A substrate fragment that exhibited an unexpect-
ed inhibitor-like mode of binding was identified, and syntheti-
cally combining the inhibitory pharmacophore of this novel
fragment scaffold with inspiration from the structures of sub-
strates allowed for fragment binding affinity to be improved.
Detailed spectroscopic characterisation of the ligand binding
mode and analysis of the SAR of these compounds provides
insight into the requirements of CYP121–ligand binding inter-
actions, the possible biological role of CYP121 for Mtb and
contributes to understanding the effect of weak nitrogen
donor ligands on the P450 heme.
Results and Discussion
Substrate deconstruction and fragment screening
The CYP121 substrates cYY, cYW and cYF were con-
ceptually deconstructed and a collection of commer-
cially available fragments that represented their dif-
ferent structural motifs was compiled (Figure 1).
These substrate fragments were combined into a li-
brary of fragments that had broadly similar chemical
properties, specifically the combination of one aro-
matic ring with a polar aliphatic or heterocyclic motif,
in order to assess whether the substrate fragments
were preferentially identified as hits. This library of 65
fragments was screened against Mtb CYP121 using
a thermal shift assay. Seven hits were identified that
increased the denaturation temperature (Tm) of
CYP121 by more than 18C (Table 1). Five of the hits
1a–e were derivatives of tyrosine, tryptophan or phe-
nylalanine, while the remaining two fragments 1 f
and 1g contained morpholine rings bound to
phenyl- or benzyl-amine groups. Three weakly bind-
ing (DTm =18C) fragments were also identified, that
had more diverse scaffolds, including a cyclic sulfone,
methylamine-tetrahydropyran and phenylcyclopro-
pane carboxylic acid (Table S1, Supporting Informa-
tion). None of the l-amino acids tyrosine, tryptophan
or phenylalanine were identified as hits, nor were
a range of amide derivatives of the amino acids,
which had been included in the fragment library to
mimic cleavage of the diketopiperazine ring of the
CYP121 substrates. However, the methyl esters of d-
tryptophan and d-tyrosine increased in the Tm of
CYP121 by +4.5 and +38C, respectively. The methyl
ester of l-phenylalanine was also identified as a hit,
but produced a smaller DTm of +1.58C.
The seven fragment hits were subsequently ana-
lysed by UV/vis spectroscopy to detect binding inter-
Figure 1. Cyclo-dipeptide CYP121 substrates cyclo-l-Tyr-l-Tyr (cYY), cyclo-l-Tyr-l-Trp
(cYW) and cyclo-l-Tyr-l-Trp (cYW) are deconstructed into a library of representative frag-
ments (selected fragments shown). CYP121 catalyses the CÀH oxidation of cYY to form
mycocyclosin.[31,33] The dissociation constants (Kd) and ligand efficiency (LE) values of sub-
strates are noted.
actions with the heme prosthetic group of CYP121.
Ligands that bind to the P450 heme to form a more
strongly coordinated complex that the resting state
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1925 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
peptides cYW and cYF are produced in only small quantities
when the cyclo-dipeptide synthetase Rv2275 is expressed in
E. coli.[32] However, their binding affinities (Kd values) for
CYP121 are in the same order of magnitude (Kd =39–46 mm) as
the Kd value of cYY (Kd =19 mm) and thus, the preference ob-
served here for tryptophan could indicate a broader substrate
profile for CYP121 than previously recognised. The type II shift
in the CYP121 optical spectrum that was generated by trypto-
phan methyl ester 1a were contrary to expectations for a sub-
strate-like compound. Type II optical spectra are characteristic
of ligands that have “inhibitor-like” interactions with the heme
prosthetic group and are considered to indicate stabilisation of
the inactive, low-spin state of P450s.[34] In contrast, the cyclo-
dipeptides cYY, cYW and cYF all produce type I, or “substrate-
like” shifts in the optical spectrum of CYP121, which are consis-
tent with a change in spin state of the heme.[33,34] It was pro-
posed that combining the inhibitor-like binding mode of frag-
ment 1a with structural features that contribute to the binding
affinity of the cyclo-dipeptides substrates might result in
a novel class of CYP121 inhibitors. In addition, fragment 1a
was found to bind to CYP121 on an unusually slow time scale,
causing a progressively greater Dlmax of the Soret of +2 nm to
+4.5 nm over a 3 min period (see Figure 4c below). Slow bind-
ing interactions have not been observed for any other class of
CYP121 ligand (aniline, imidazole and triazole structures) that
have been previously studied. These unusual optical properties
and the structural similarity of fragment 1a to the CYP121 sub-
strates prompted further investigation of the SAR contributing
to fragment 1a-CYP121 binding interactions.
Table 1. Fragments identified by thermal shift and UV/vis spectroscopy
to bind to Mtb CYP121.
Fragment
DTm [8C][a]
Dlmax [nm][b]
1a
1b
1c
+4.5
+4.5
+3
0
0
+1.5
1d
1e
+2.5
+1.5
0
0
1 f
+2.0
+2.0
+4
1g
0
[a] Change in the denaturation temperature (Tm) of CYP121 relative to
a DMSO (5% v/v) control (Tm =49.08C); fragments were screened at 5 mm
against 5 mm CYP121. [b] Magnitude of change in the maximum wave-
length of the Soret band of the CYP121 (5 mm) optical spectrum, calculat-
ed relative to a DMSO (1% v/v) control (Dlmax =416.5 nm); fragments
were screened at 2 mm.
Binding mode and minimal pharmacophore of tryptophan
methyl ester
A selection of analogues structurally related to fragment 1a
were screened by UV/vis spectroscopy in order to identify the
heme interacting functional group and to establish the initial
SAR contributing to CYP121–1a binding interactions (Table S1,
Supporting Information). P450 inhibitors that exhibit type II op-
tical spectra typically interact with the heme cofactor using the
lone pair electrons of a nitrogen atom, with which they either
directly coordinate to the heme iron or indirectly bind the
heme iron by stabilising the axial heme water ligand. As such,
fragment 1a was predicted to interact with the heme iron
using either the a-amino group or indole nitrogen.[36–38]
Screening results from this initial set of fragment analogues in-
dicated that the methyl ester of fragment 1a was a key feature
of the binding pharmacophore. Replacement of the ester with
a range of functional groups, including carboxylic acids, pri-
mary (thio)-amide, alcohol, amine or nitrile substituents dis-
rupted heme binding interactions (Dlmax =0 nm). As such, the
synthesis of specific analogues that retained an ester substitu-
ent was required in order to clarify whether the a-amine or
the indole nitrogen atom of fragment 1a was essential for
heme binding. The design of these analogues was guided by
the size and structure of the substrate cYW, which when linear-
ised by cleavage of the diketopiperazine ring would yield ela-
borated analogues of fragment 1a (Figure 2).
enzyme, either by direct coordination of the ferric iron, or indi-
rect coordination via the axial water ligand, cause a red shift
(type II) in the maximum wavelength (lmax) of the Soret band
of the P450 optical spectrum. In contrast, ligands which bind
in the vicinity of the heme and displace the axial water ligand,
but do not ligate the ferric iron, cause a blue shift (type I) in
the Soret lmax.
[34] Only two of the seven hit fragments identified
in the thermal shift assay produced observable changes in the
CYP121 optical spectrum. One of these fragments, compound
1 f contained an aniline functional group, which we have previ-
ously reported as a preferred CYP121 heme binding motif.[25,35]
The second, was the d-tryptophan methyl ester 1a, which pro-
duced a red shift in the Soret lmax of +4.5 nm. The remaining
fragments, including the methyl ester of d-tyrosine 1b, did not
perturb the optical spectrum of CYP121. This suggested that
the binding interactions of these fragments must occur distal
to the heme, or be too weak to perturb the heme coordination
sphere of the resting state enzyme. The preference of CYP121
for d-tryptophan methyl ester 1a over its respective d-tyrosine
methyl ester analogue, which was demonstrated in both ther-
mal shift and UV/vis assays, was surprising because cYY is con-
sidered to be the preferred substrate of CYP121. The cyclo-di-
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1926 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 2. Substrate-inspired elaboration of tryptophan methyl ester 1a. Cleavage of the diketopiperazine ring of cyclo-l-Tyr-l-Trp (cYW) yields linear com-
pounds 2–9 that incorporate the scaffold of fragment 1a (blue). Modification of the amide motif (X), a-amino group (R2) and indole nitrogen atom (R1) pro-
vides insight into key heme binding motif and SAR.
The phenethyl ester 2, as well as its l-enantiomer
3,[39] were synthesised as initial compounds, because
they represented the simplest structural elaboration
of fragment 1a (Table 2). In addition, the Na-acetyl
protected 4 and 1-N-methyl protected 5 analogues
of 3 were synthesised in order to probe the impor-
tance of the nitrogen atoms for heme binding. The
general syntheses of these compounds is outlined in
Scheme 1, using compound 2 as an example. All
other (thio)-ester and amide compounds in this study
were prepared by a similar route (Detailed synthetic
schemes and characterisation all compounds can be
found in Supporting Information). The free a-amine
of the appropriate amino acid precursor was protect-
ed as the tert-butyl carbamate (Boc) derivative and
then coupled to the appropriate alcohol, thiol or
amine using carbodiimide chemistry. Deprotection of
the a-amino group under anhydrous acidic condi-
tions yielded the desired ligands 2, 3, and 5 as their
hydrochloride salts, while compound 4 was retained
as the Na-acetyl protected ester.
Table 2. Structures and binding affinity data for linearised CYP121 substrate ana-
logues and ligands designed to establish the heme binding mode of fragment 1a.
Compound
Dlmax [nm][a]
Kd [mm][b] LE[c]
1 mm 100 mm
1a
2
+4.0
+4.5
+4.0
–
840Æ88 0.26
+1.0
+1.0
200Æ17 0.22
200Æ19 0.22
3[d]
4
5
6
+4.5
+0.5*
À0.5
+0.5
140Æ22 0.22
Both compounds 2 and 3, and the 1-N-methylated
analogue 5, produced type II shifts of +4.0–4.5 nm in
the Soret lmax of the CYP121 optical spectrum. In
contrast, the Na-acetylated analogue 3 did not pro-
duce a significant change in the Soret (Dlmax <1 nm),
indicating that the heme binding interactions of frag-
ment 1a and analogues were mediated by the a-
amino group. Aliphatic amines are relatively uncom-
mon among reversible P450 inhibitors compared to
heteroaromatic nitrogen ligands. The inherently
stronger heme binding affinity of heteroaromatic li-
gands arises from their greater lipophilicity and pi-ac-
ceptor properties. However, alkyl amines are well
characterised to form quasi-irreversible mechanism-
based P450 inhibitors.[34] These compounds typically
exhibit greater isoform selectivity than reversible
P450 inhibitors and have previously been incorporat-
ed in substrate analogues designed to selectively in-
hibit steroid biosynthesis.[40–42] Heme co-ordination
by a primary aliphatic amine has not been previously
identified for CYP121 ligands and the use of this
heme binding group might allow the development
of a series of CYP121 inhibitors with good selectivity
over other P450 enzymes. To investigate this pros-
pect, fragment 1a was screened against a panel of
0
ND
ND
ND
ND
À0.5
7
8
9
+4.5
+4.5
+7
+1
260Æ14 0.20
270Æ22 0.21
110Æ6.0 0.21
+0.5
+1
[a] Magnitude of change in the maximum wavelength of the Soret band of the
CYP121 optical spectrum calculated relative to a DMSO (1% v/v) control (lmax
=
416.5 nm); compounds were screened at 1 mm/100 mm (*except 5, which was
screened at 500 mm/100 mm, owing to limited solubility). [b] Spectral dissociation con-
stant from the optical titration of CYP121 (5 mm) with compounds; data were fitted
using a standard hyperbolic (Michaelis–Menten) function as described in the Support-
ing Information. [c] Ligand efficiency (kcalmolÀ1 per heavy atom). [d] Synthesised as
described in ref. [28]. ND: not determined.
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1927 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Scheme 1. Example synthesis of ester, thioester and amide and amine analogues of fragment 1a. Reagents and conditions: a) BOC-ONꢁ, Et3N, 1,4-dioxane/
water (1:1), RT, 5 h; b) EDC·HCl, HOBt, Et3N, CH2Cl2, RT, 24–36 h; c) Red-Alꢁ, THF/toluene (2:3 v/v), 0–208C, 5 min, then 408C, 15 h, then NaOH, 58C, 1 h; d) 4m
HCl in 1,4-dioxane, RT, 1 h; e) DCC, DMAP, CH2Cl2, 08C!RT, overnight; f) thiourea, EtOH, reflux, 24 h, then 2m NaOH, EtOH, reflux, 4 h.
Ester bioisostere replacements
six other Mtb P450s and found to produce type II binding in-
teractions with only one other isoform, the cholesterol/choles-
tenone oxidase CYP142 (see Table 6 below).
The difference in the binding affinity of ester 2 and amide 8 in-
dicated that the ester group was an important contributor to
CYP121–ligand interactions. Furthermore, it was noted that
only substrate fragments that contained an ester functional
group were identified as type II ligands when screened by UV/
vis spectroscopy (Table 1 and Table S1, Supporting Informa-
tion). To investigate the SAR of the ester group, a series of
compounds containing various bioisosteres were synthesised
(Table 3). All of these compounds contained phenyl or phe-
nethyl substituents to allow access to a wider variety of syn-
thetic precursors and to enable the binding contribution of
the bioisosteric group to be assessed independently of other
structure changes. Ester, thioester and amide containing com-
pounds were synthesised as described earlier for compounds 2
and 8 in Scheme 1. Thiol precursors that were not commercial-
ly available, such as those required for the synthesis of com-
pounds 17, 23, 23 and 25 (Table 4), were prepared from the re-
spective alkyl bromides, using thiourea followed by hydrolysis
with sodium hydroxide (Scheme 1). Thiazoles were prepared
from the condensation of Boc-protected primary thioamide in-
termediates with a-bromoketones, as described for the synthe-
sis of compound 18 (Scheme 2). Primary thioamides were syn-
thesised from their respective Boc-protected amino acid pre-
cursors via their primary amides, by substitution of the activat-
ed anhydride intermediates with aqueous ammonia, followed
by treatment with Lawesson’s reagent or P2S5 (Scheme 2). a-
Bromoketones that were not commercially available, were ob-
tained by the homologation of the appropriate carboxylic acid
precursor with TMS-diazomethane and then substitution of the
diazo group using hydrobromic acid. Imidazole 14 was syn-
thesised from the condensation of Boc-d-tryptophan with a 2-
bromoacetophenone, followed by cyclisation of the resulting
keto ester intermediate using ammonium acetate in xylene at
reflux. “Reverse” ester 15 was obtained from the reduction of
Boc-d-tryptophan using NaBH4, followed by Steglich esterifica-
tion with 3-phenylpropionic acid. Amine 16 was synthesised
The binding affinities of fragment 1a and phenethyl ester
analogue 2 were determined by optical titration to be 840Æ
88 mm and 200Æ17 mm, respectively, demonstrating that elab-
oration from the ester motif of 1a, guided by the size and
structures of CYP121 substrates produced a significant im-
provement in binding affinity. Furthermore, elaboration of the
ester to include a second aromatic motif overcame the enan-
tioselectivity of CYP121 for fragment 1a over its l-tryptophan
methyl ester isomer, which had not been identified as a hit in
thermal shift screening and did not perturb the Soret lmax of
CYP121 in UV/vis assays (Table S1, Supporting Information). In
contrast, the l-phenethyl ester 3 had an equivalent Kd of 200Æ
19 mm to that of compound 2.
Encouraged by the 4-fold improvement in the binding affini-
ty of 2 over 1a, a series of linearised substrate analogues were
designed, which contained an amide functional group in place
of the ester of 2 to more closely mimic the CYP121 substrates.
Analogues resembling acyclic versions of cYY-6, cYW-7, and
two hypothetical cyclo-dipeptides “cWF”-8 and “cWW”-9 were
synthesised according to the route outlined in Scheme 1. Com-
pound 6 did not show binding interactions with the heme co-
factor (Dlmax ~0 nm) in direct UV/vis experiments or competi-
tive binding interactions when tested in combination with
a known type II CYP121 inhibitor clotrimazole (Dlmax = +8 nm,
50 mm; Table 2 and Figure S1, Supporting Information). These
experiments indicate that 6 is unlikely to bind within the
CYP121 active site, and are consistent with the preference ob-
served for CYP121 to bind tryptophan/indole fragments over
phenol or phenyl structures. Compounds 7–9 all produced
type II shifts in the Soret lmax and were calculated to have
binding affinities of 7: 260Æ14 mm, 8: 270Æ22 mm and 9:
110 Æ6.0 mm.
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1928 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
clusion. Conversion of ester 2 to thioester 10 also re-
sulted in a threefold improvement in binding affinity.
Ligand docking simulations predicted that the ester
motif of 2 likely formed interactions with an arginine
residue (Arg386) that is positioned in close proximity
to the CYP121 heme cofactor (see Figure 5 below).
Arg386 has previously been identified to form
a water-bridged hydrogen bonding network with
either the phenol ÀOH or diketopiperazine carbonyl
group of the CYP121 substrates cYY and cYW, respec-
tively. These interactions have been proposed to
both stabilise the binding of the substrates and to be
essential for the CYP121 catalytic mechanism.[31,33]
The introduction of sulfur-containing (thioester, thia-
zole, and thioamide) bioisosteres produced a consis-
tent improvement in the binding affinity of all com-
pounds that were subsequently synthesised, relative
to their respective ester or amide counterparts. These
SAR are likely the result of both the different confor-
mational and electronic properties imparted by the
sulfur substitution. A significant loss in binding affini-
ty was observed when the thiazole ring of 13 was re-
placed with imidazole 14. This might be a conse-
quence of mixed binding modes, as both the imida-
zole and a-amino groups of 14 could form favoura-
ble interactions with the heme cofactor.
Table 3. Structures and binding affinity data of selected analogues that contain bio-
isosteric replacements for the ester group of compound 2.
Compound
Dlmax [nm][a]
1 mm 100 mm
Kd [mm][b] LE[c]
66Æ9.7 0.25
61Æ7.6 0.25
10
11
–
–
+2.5
+3.5
12
13
14
15
16
–
–
+1.5
+2
150Æ18
0.23
44Æ3.6 0.26
530Æ100 0.19
–
+0.5
+0.5
–
+1.5
0*
ND
ND
ND
ND
The racemic analogue 11 of thioester 10 was also
synthesised and found to have a similar Kd value of
61Æ7.6 mm, confirming the results obtained for the
d- and l-phenethyl ester analogues 2 and 3, respec-
tively. Ligand docking simulations were used to justi-
fy these results, demonstrating that both the d- and
l-enantiomers of compounds that contained two aro-
matic motifs could satisfy the CYP121 binding phar-
macophore (Figure 5b,c). For example, both enantio-
mers should be able to form interactions with the
[a] Magnitude of change in the maximum wavelength of the Soret band of the
CYP121 optical spectrum calculated relative to a DMSO (1% v/v) control (lmax
=
416.5 nm); compounds were screened at 1 mm/100 mm (*except 16, which was
screened at 500 mm, owing to limited solubility). [b] Spectral dissociation constant
from the optical titration of CYP121 (5 mm) with compounds; data were fitted using
a standard hyperbolic (Michaelis–Menten) function as described in the Supporting In-
formation. [c] Ligand efficiency (kcalmolÀ1 per heavy atom). ND: not determined.
by the reduction of amide 8 using Red-Alꢁ (sodium bis(2-me-
thoxyethoxy)aluminium hydride; Scheme 1). All final com-
pounds, and a selection of intermediates, were deprotected
under anhydrous acidic conditions to yield the desired hydro-
chloride salts.
heme cofactor and Arg386, as well as aromatic stacking inter-
actions with Phe168 and Trp182 residues. These aromatic inter-
actions have been previously identified as a binding hotspot in
the CYP121 active site and are conserved in the X-ray crystal
structures of CYP121 substrates.[25,33] Confident that stereo-
chemistry did not affect the binding affinity of the ligands,
a range of racemic amino acid derivatives were employed
during subsequent iterations of fragment optimisation in order
to explore the binding contribution of the tryptophan indole.
The compounds were analysed for binding to CYP121 by
UV/vis spectroscopy and dissociation constants (Kd values)
were determined by optical titration for select compounds
(Table 3). The rigidity, polarity and position of the ester (or
equivalent functional group) relative to the a-amine, were
each shown to be important for CYP121 binding affinity. Ho-
mologation and reversal of the ester motif 15, or reduction of
the replacement of the ester with an amine 16, was detrimen-
tal, with neither compound 15 nor 16 producing significant
perturbation in the Soret lmax. Conversion of amide 8 to thioa-
mide 12 resulted in a twofold improvement in binding affinity,
which we hypothesised might be a conformational effect re-
sulting from the thioamide favouring an imine resonance
structure.[43] The further threefold improvement in binding af-
finity that resulted from constraining the thio-imine into a thia-
zole ring, as in compound 13, provided support for this con-
Aromatic group optimisation
The SAR of the phenylethyl substituent and indole ring of the
ligands were subsequently investigated (Tables 4 and 5). Thio-
ester 10/11 and thiazole 13 were selected as structural tem-
plates for this investigation, as they had provided the greatest
improvement in CYP121 binding affinity relative to the ester
group of fragment 1a and compound 2. However, a selection
of ester and amide compounds were also synthesised because
of availability of a wider variety of synthetic precursors.
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1929 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 4. Structures and binding affinity data of tryptophan thioester and
thiazole compounds substituted with a range of aromatic groups.
Table 5. Structures and binding affinity data of phenethyl-thioester and
phenyl-thiazole compounds substituted with a range of aromatic amino
acids.
Compound
Dlmax [nm][a] Kd [mm][b] LE[c]
Compound
Dlmax [nm][a] Kd [mm][b] LE[c]
17
18
+1.5
+1.0
45Æ4.4 0.23
52Æ3.1 0.22
26
+1.5
130Æ9.3 0.22
110 Æ9.1 0.23
27
28
29
30
+1.5
19
20
0
ND
ND
0
ND
ND
ND
ND
+1.5
120Æ21 0.22
91Æ31 0.22
+0.5
+3.5
21
22
+2.0
+0.5
30Æ3.8 0.29
16Æ2.4 0.27
ND
ND
ND
ND
ND
ND
31
32
+5.5
+3
23
+0.5
32Æ2.5 0.26
[a] Magnitude of change in the maximum wavelength of the Soret band
of the CYP121 optical spectrum calculated relative to a DMSO (1% v/v)
control (lmax =416.5 nm); compounds screened at 100 mm. [b] Spectral
dissociation constant from the optical titration of CYP121 (5 mm) with
compounds; data were fitted using a standard hyperbolic (Michaelis–
Menten) function as described in the Supporting Information. [c] Ligand
efficiency (kcalmolÀ1 per heavy atom). ND: not determined.
24
25
+0.5
+3.0
35Æ4.2 0.22
[a] Magnitude of change in the maximum wavelength of the Soret band of
the CYP121 optical spectrum calculated relative to a DMSO (1% v/v) con-
trol (lmax =416.5 nm); compounds were screened at 100 mm. [b] Spectral
dissociation constant from the optical titration of CYP121 (5 mm) with com-
pounds; data were fitted using a standard hyperbolic (Michaelis–Menten)
function as described in the Supporting Information. [c] Ligand efficiency
(kcalmolÀ1 per heavy atom). ND: not determined.
drophobic substituent by the introduction of a trans-alkene 20
or fused aromatic ring 21 produced a twofold improvement in
affinity relative to ester 2. The introduction of a range of aro-
matic substituents that contained polar heteroatoms, such as
in compounds 22, 23, and 24, was not tolerated and only the
phenyl-isoxazole-substituted compound 25 produced an im-
provement in binding affinity (Kd =35 mm) from this series of
compounds. Polar heteroatoms in close proximity to the (thio)-
ester group were postulated to interfere with Arg386 interac-
tions, while introducing shorter (thio)-ester substituents might
prevent compounds from accessing binding interactions with
active site residues. The relatively high ligand selectivity ob-
served for CYP121 is characteristic of biosynthetic P450s and
demonstrates the important contribution of active site interac-
tions to the binding affinity of this series of ligands.[34]
It was hypothesised that substitution of the phenylethyl
group with an indole or other heteroaromatic group might
produce a similar improvement in the binding affinity of scaf-
folds 10 and 13 as had been observed between the “cWY”-7/
“cWF”-8 analogues and the acyclic “cWW”compound 9. How-
ever, whereas indole-substituted thioester 17 and thiazole 18
maintained binding affinities similar to those of 10 and 13,
they provided no further improvement. Decreasing the chain
length of the phenyl ester substituent 19 was detrimental to
binding affinity, while increasing the length or rigidity of a hy-
A range of substituted tryptophan, phenylalanine, pyridine
or other aromatic amino acid analogues were selected to re-
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1930 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Scheme 2. Example syntheses of thiazole, imidazole and homologated ester compounds. Reagents and conditions: a) SOCl2, DMF, THF, 08C, 4 h, then TMS-di-
azomethane, THF/MeCN, 08C, 4 h; b) HBr(aq) (48%), AcOH, 08C, 40 min; c) EtOH, RT, overnight; d) 4m HCl in 1,4-dioxane, RT, 1 h; e) iBuCO2Cl, NMO, DME,
NH3(aq) (35%), 08C!RT, 2 h; f) NaHCO3, P2S5, DME, RT, overnight, or Lawesson’s reagent, THF, RT, overnight; g) 1. iBuOCOCl, NMM, DME, À158C, 2 min;
2. NaBH4, H2O, À158C, 5 min; h) 3-phenylpropionic acid, DCC, DMAP, CH2Cl2, 08C!RT, overnight; i) 1. Cs2CO3, EtOH, RT, 30 min, 2. PhCOCH2Br, DMF, RT, 4 h;
j) NH4OAc, m-xylene, reflux, 1 h.
place the indole ring of thioester 10/11 or thiazole 13
(Table 5). While substitution at C5 of the indole ring was toler-
ated, the introduction of a polar hydroxy group resulted in
a twofold increase in the Kd value of 26 (Kd =130 mm) relative
to 11. Similarly, the introduction of a second aromatic N atom
into the indole ring to give the aza-tryptophan analogue 27,
produced a similar loss of binding affinity (Kd (27)=110 mm).
Replacement of the tryptophan indole with phenol or pyridine
heterocycle was detrimental for ligand binding and neither the
pyridine thioester 28 nor thiazole 29 compounds produced
a change in the Soret lmax. This was consistent with observa-
tions previously noted for compound 6, the acyclic analogue
of cYY (Table 2). In contrast, analogues substituted with car-
boaromatic rings or halogens, such as 2-bromophenylanine de-
rivative 30, improved binding affinity (Kd (30)=30 mm) relative
to compound 11. The addition of bromine at C5 of the trypto-
phan indole ring similarly improved the binding affinity of
both the phenethyl-thioester 31 and phenyl-thiazole 32 series
of compounds. This modification resulted in the most potent
compound 31 to be developed from fragment 1a, which was
calculated to have a Kd value of 16 mm and an improved LE
over the original fragment (Figure 3).
tred at 420–422 nm for the majority of compounds, is small rel-
ative to that expected for heteroaromatic NÀFe3+ heme bind-
ing groups (lmax ꢀ424 nm)[36] and could indicate an indirect in-
teraction between the tryptophan compounds and heme via
a bridging axial water molecule, as has been observed in the
X-ray crystal structure of CYP121 in complex with fluconazole
(PDB ID: 2IJ7).[44] However, a recently characterised series of li-
gands that directly coordinate to the heme iron of CYP121
using a primary aniline as the nitrogen ligand (PDB ID: 5IBE)
also produce relatively small shifts in the Soret (lmax =422 nm
at saturating ligand concentrations), despite their having low
nanomolar affinity.[25]
Electron paramagnetic resonance (EPR) spectra of CYP121 in
complex with fragment 1a and fourteen of the elaborated
compounds, were collected to gain further insight into their
binding interactions (Table 3, Supporting Information). All com-
pounds generated a rhombic triplet of g values that are consis-
tent with a low spin, cysteinate-coordinated ferric P450 (Fig-
ure 4a).[36] The change in the g values of the ligand-free
CYP121 protein (gz =2.48, gy =2.25, gx =1.89) that were ob-
served for inhibitor-bound CYP121 complexes (e.g., compound
31, gz =2.46, gy =2.25, gx =1.90) were similar to those previ-
ously reported for both CYP121–fluconazole (gz =2.45, gy =
2.26, gx =1.90)[44] and CYP121–cYY (gz =2.46, gy =2.25, gx =
1.89/1.90)[31] complexes, both of which retain an axial heme
water ligand in X-ray co-crystal structures. However, the previ-
ously noted series of CYP121 aniline ligands that displace the
axial water ligand, also generate similar g values of 2.44/2.24/
1.90 (Table 4 and Figure S2, Supporting Information). The
direct coordination of heteroaromatic nitrogen ligands to ferric
P450 heme iron is well characterised in the literature and is ex-
pected to cause a widening of the low-spin g values (e.g., gz ~
Spectroscopic characterisation of inhibitor binding mode
and selectivity
The red shift in the Soret lmax of the CYP121 optical spectrum
that was observed for the enzyme bound to fragment 1a, or
its elaborated derivatives, is indicative of ligands that interact
with the ferric heme iron of a P450 and form a more strongly
coordinated complex than the water-bound resting state.[34]
The magnitude of the Dlmax of the Soret band, which was cen-
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1931 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
with cYY is likely a consequence of the low temperatures (~
10 K) required to collect P450 X-band EPR spectra. The greater
retention of high-spin CYP121 generated by compound 9 sug-
gests that the ligand forms more intimate contacts with the
distal face of heme group, possibly through indole-heme stack-
ing interactions, as observed in the X-ray crystal structure of
cYW (PDB ID: 4IQ9).[33] These stronger interactions are reflective
of the preferential binding of CYP121 to fragment 1a/indole
derivatives over the respective tyrosine/phenol analogues that
was revealed fragment screening by thermal shift and UV/vis
spectroscopy. The generation of high-spin CYP121 by com-
pound 9, in addition to the low micromolar Kd values deter-
mined for all three of the di-indole compounds 9, 17, and 18,
which are on the same order of magnitude as the Kd value of
CYP121 for cYY,[33] suggests a preference of CYP121 for trypto-
phan dipeptides. While the Mtb cyclic dipeptide synthetase
Rv2275 was not found to synthesise di-Trp (cWW) peptides
when expressed in E. coli,[32] cW-X natural products are preva-
lent in other microbial species, particularly fungi, and have
potent biological activities.[45] Although catalytic turnover of
compound 9 was not observed when CYP121 activity was re-
constituted with NADPH and ferredoxin/ferredoxin reductase
redox partners,[31] the preference of CYP121 for the di-indole
structural motif could indicate a broader substrate profile to
that previously identified or a possible role for CYP121 in mi-
crobial defence mechanisms.
Spectroscopic evidence that the tryptophan compounds
form relatively weak heme binding interactions suggests that
they instead obtain a significant proportion of their binding af-
finity from interactions made with enzyme active site residues.
This might engender the compounds with an improved selec-
tivity profile over inhibitors which have metal-coordination
driven modes of binding. In support of this conclusion, lead
compound 31 and cWW analogue 9 were both found to have
good selectivity for CYP121 over a panel of six other Mtb
P450s, when they were analysed by UV/vis spectroscopy.
Type II binding interactions were not identified for either 31 or
9 with any of the P450s other than CYP121. However, a weak
blue shift was observed in the spectrum of the cholesterol and
fatty acid oxidase CYP124 (Table 6).
Figure 3. Optical difference spectra (top) and ligand-induced change in
Soret absorbance (bottom) for the titration of CYP121 with compound 31.
Data were fitted using a standard hyperbolic (Michaelis–Menten) function to
calculate a Kd value of 16Æ2.4 mm and ligand efficiency (LE) of 0.27.
2.50–2.65), as is observed for CYP121 bound to econazole (gz =
2.48, gy =2.24, gx =1.90).[36] The opposite trend in the g values
of EPR spectra collected for CYP121 in complex with the cur-
rent series of tryptophan (aliphatic amine) ligands, and the pre-
vious series of aniline ligands, in addition to their characteristic
type II optical spectra and crystallographic evidence for direct
aniline-Fe3+ binding interactions,[25] provides novel insight into
the spectroscopic effect of weak donor ligands on the P450
heme environment.
Spectroscopic characterisation of CYP121–ligand binding in-
teractions and the SAR established during fragment elabora-
tion allowed the binding mode of analogues to be modelled
in ligand docking simulations (Figure 5). Compounds were pre-
dicted to either coordinate directly to the heme iron via the a-
amino group (Figure 5b), or indirectly co-ordinate via hydro-
gen bonding interactions to the axial water ligand (Figure 5a).
Possible progression between these two binding modes, that
The majority of the tryptophan compounds stabilised the
low-spin state of CYP121, decreasing the small amount (0.9– is, the formation of a hydrogen bridge complex initially, fol-
3.1%) of high-spin enzyme that is present in X-band EPR spec-
tra collected for the water-ligated, resting state enzyme. One
exception was the acyclic “cWW” analogue 9, which increased
the proportion of high-spin CYP121 to ~34% (Figure 4b, and
Table S3, Supporting Information), significantly more than that
generated by cYY in complex with CYP121 (~2.2% high-spin;
Figures S2 and S4, Supporting Information). The small propor-
tion of high-spin enzyme observed for CYP121 in complex
lowed by displacement of the axial water to form direct NÀ
Fe3+ coordination, might explain the slow onset of spectral
perturbations observed for these ligands (Figure 4c).
Conclusions
The deconstruction of CYP121 substrates into their component
fragments has been used to identify structural features of the
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1932 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 4. Spectroscopic characterisation of CYP121–ligand binding interactions. a) X-band EPR spectra collected over a narrow scan (2000–4000 G) and b) a
wide scan (500–4500 G), for ligand-free CYP121 (100 mm, black) and CYP121 (100 mm) bound to compound 13 (2 mm, red) or compound 9 (2 mm, blue). The
low-spin g values generated are annotated for all samples in panel (a). In panel (b), the high-spin g values generated by compound 9 are annotated. c) The
optical spectrum of ligand-free CYP121 (5 mm, black) and CYP121 bound to fragment 1a (2 mm) collected after 1 min (red) or 3 min (blue) incubation. The
Soret lmax value is labelled for each spectrum. d) Structures of compounds 1a, 9 and 13.
ported aliphatic amino group. Combining this inhibitory phar-
Table 6. Heme binding selectivity of fragment 1a, di-indole “cWW”ana-
logue 9 and optimised compound 31.[a]
macophore with the structures of CYP121 substrates allowed
rational synthetic elaboration to improve fragment binding af-
finity. The SAR established during synthetic optimisation sup-
port the importance of specific binding hotspots and the role
of certain active site residues for ligand recognition. These
CYP121 inhibitors represent a chemically distinct class of com-
pounds to those previously reported and interact via a different
mechanism. The compounds have low micromolar affinity, ap-
parent selectivity for CYP121 over other Mtb P450s and are
predicted to have good physicochemical properties.[46] Spectro-
scopic characterisation of the binding interactions of CYP121
with these compounds revealed that weak nitrogen donor li-
gands can have unusual effects on the heme microenviron-
ment and greater analysis of these interactions is warranted. A
preference of CYP121 for di-indole dipeptide mimetics was re-
vealed, provoking speculation on the biological role of the
enzyme. This study demonstrates the use of fragments to de-
convolute the binding interactions of large molecules, identify
novel binding modes and chemical scaffolds, and to under-
stand factors contributing to enzyme-ligand recognition.
Compd CYP121 CYP124 CYP125 CYP126 CYP142 CYP143 CYP144
1a
9
31
+4.5
+2.0
+5.5
0
À2
À2
À0.5
+0.5
+0.5
0
+2
0
0
0
À0.5
0
+0.5
+0.5
+0.5
0
0
[a] The change in the maximum wavelength of the Soret band (Dlmax
,
nm) of the optical spectrum of each enzyme in the presence of com-
pounds (1: 2 mm, 9: 100 mm or 31: 10 mm) relative to DMSO (1% v/v)
alone is shown. Ligand concentrations were selected based on com-
pound solubility. Type II and type I binding interactions are indicated by
a decrease or increase in the Soret lmax, respectively. Changes in the
(lmax <1 nm) were considered within experimental error.
larger compounds that can be used to drive binding affinity.
Substrate fragments were preferential identified to increase
the denaturation temperature of the enzyme when screened
as part of a library of chemically similar fragments. Validation
of the fragment hits using UV/vis spectroscopy resulted in the
identification of a substrate fragment that exhibited an unex-
pected inhibitor-like mode of binding using a previously unre-
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1933 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 5. Proposed binding mode of lead compound 31 to CYP121. The d- and l-enantiomers of compound 31 were docked into the prepared X-ray crystal
structure of CYP121 in complex with cYW (PDB ID: 4IQ9).[33] a) Proposed water-bridged heme binding mode of d-31 (green); b) direct heme binding mode of
d-31 (cyan); c) direct heme binding mode of l-31 (salmon). Proposed hydrogen bonding interactions with active site residues (grey) and metal coordination
to the heme cofactor (grey) are shown as yellow dashes. The I helix of the CYP121 protein backbone is shown as a grey helix for orientation. Docking was
performed using Glide ver. 6.5 (Schrçdinger LLC, NY, 2014-4), and the figures were prepared using the PyMOL Molecular Graphics System ver. 1.3.
[8] T. J. Stout, C. R. Sage, R. M. Stroud, Structure 1998, 6, 839–848.
[9] K. Babaoglu, B. K. Shoichet, Nat. Chem. Biol. 2006, 2, 720–723.
Additional data
[10] D. C. Fry, C. Wartchow, B. Graves, C. Janson, C. Lukacs, U. Kammlott, C.
Additional data relating to this publication are available at the
Belunis, S. Palme, C. Klein, B. Vu, ACS Med. Chem. Lett. 2013, 4, 660–
665.
University of Cambridge data repository: https://www.reposi-
tory.cam.ac.uk/handle/1810/256410.
[11] M. NazarØ, H. Matter, D. W. Will, M. Wagner, M. Urmann, J. Czech, H.
Schreuder, A. Bauer, K. Ritter, V. Wehner, Angew. Chem. Int. Ed. 2012, 51,
905–911; Angew. Chem. 2012, 124, 929–935.
[12] O. A. Andersen, A. Nathubhai, M. J. Dixon, I. M. Eggleston, D. M. F. va-
n Aalten, Chem. Biol. 2008, 15, 295–301.
Acknowledgements
[13] G. Bollag, J. Tsai, J. Zhang, C. Zhang, P. Ibrahim, K. Nolop, P. Hirth, Nat.
Rev. Drug Discovery 2012, 11, 873–886.
[14] A. J. Souers, J. D. Leverson, E. R. Boghaert, S. L. Ackler, N. D. Catron, J.
Chen, B. D. Dayton, H. Ding, S. H. Enschede, W. J. Fairbrother, et al., Nat.
Med. 2013, 19, 202–208.
[15] P. J. Hajduk, J. R. Huth, S. W. Fesik, J. Med. Chem. 2005, 48, 2518–2525.
[16] A. Ciulli, G. Williams, A. G. Smith, T. L. Blundell, C. Abell, J. Med. Chem.
2006, 49, 4992–5000.
[17] G. Dorman, F. Darvas in Chemical Genomics and Proteomics (Eds.: F.
Darvas, A. Guttman, G. Dorman), CRC Press, Taylor & Francis, Boca
Raton, 2013, pp. 1–48.
[18] S. Barelier, J. A. Cummings, A. M. Rauwerdink, D. S. Hitchcock, J. D. Fare-
lli, S. C. Almo, F. M. Raushel, K. N. Allen, B. K. Shoichet, J. Am. Chem. Soc.
2014, 136, 7374–7382.
[19] S. Barelier, J. Pons, O. Marcillat, J. M. Lancelin, I. Krimm, J. Med. Chem.
2010, 53, 2577–2588.
[20] P. Brandt, M. Geitmann, U. H. Danielson, J. Med. Chem. 2011, 54, 709–
718.
[21] D. Kozakov, D. R. Hall, S. Jehle, L. Luo, S. O. Ochiana, E. V. Jones, M. Pol-
lastri, K. N. Allen, A. Whitty, S. Vajda, Proc. Natl. Acad. Sci. USA 2015, 112,
E2585–E2594.
The authors acknowledge Dr. Colin W. Levy, Dr. Muralidharan
Shanmugam, and Mr. Jude Chenge (Manchester Institute of Bio-
technology, The University of Manchester) for their technical help
running EPR and X-ray crystallography experiments. We also
thank Dr. Duncan E. Scott (Department of Chemistry, The Univer-
sity of Cambridge) for his contribution to the organic synthesis of
the ligands, and Dr. Alicia P. Higueruelo (The Cambridge Crystal-
lographic Data Centre, The University of Cambridge) for her guid-
ance with molecular docking. M.E.K. was supported by a Com-
monwealth (University of Cambridge) Scholarship awarded in
conjunction with the Cambridge Commonwealth Trust and Cam-
bridge Overseas Trust. A.G.C. and K.J.M. were supported by grants
from the Biotechnology and Biological Sciences Research Council
(BBSRC) (grant nos. BB/I019669/1 and BB/I019227/1). We ac-
knowledge the support of the Wellcome Trust (Translation Award
GR080083/Z/06).
[22] D. R. Hall, C. H. Ngan, B. S. Zerbe, D. Kozakov, S. Vajda, J. Chem. Inf.
Model. 2012, 52, 199–209.
[23] H. Chen, X. Zhou, A. Wang, Y. Zheng, Y. Gao, J. Zhou, Drug Discovery
Today 2015, 20, 105–113.
Keywords: drug discovery · enzymes · fragment-based
methods · substrate analogues · tuberculosis
[24] I. Krimm, J. M. Lancelin, J. P. Praly, J. Med. Chem. 2012, 55, 1287–1295.
[25] M. E. Kavanagh, A. G. Coyne, K. J. McLean, G. G. James, C. W. Levy, L. B.
Marino, L. P. S. de Carvalho, D. S. H. Chan, S. A. Hudson, S. Surade, D.
Leys, A. W. Munro, C. Abell, J. Med. Chem. 2016, 59, 3272–3302.
[26] K. J. McLean, P. Carroll, D. G. Lewis, A. J. Dunford, H. E. Seward, R. Neeli,
M. R. Cheesman, L. Marsollier, P. Douglas, W. E. Smith, I. Rosenkrands,
S. T. Cole, D. Leys, T. Parish, A. W. Munro, J. Biol. Chem. 2008, 283,
33406–33416.
[1] P. A. Rejto, G. M. Verkhivker, Proc. Natl. Acad. Sci. USA 1996, 93, 8945–
8950.
[2] W. P. Jencks, Proc. Natl. Acad. Sci. USA 1981, 78, 4046–4050.
[3] M. Congreve, G. Chessari, D. Tisi, A. J. Woodhead, J. Med. Chem. 2008,
51, 3661–3680.
[4] A. R. Leach, M. M. Hann, J. N. Burrows, E. J. Griffen, Mol. BioSyst. 2006, 2,
429–446.
[27] K. J. McLean, M. R. Cheesman, S. L. Rivers, A. Richmond, D. Leys, S. K.
Chapman, G. A. Reid, N. C. Price, S. M. Kelly, J. Clarkson, W. E. Smithe,
A. W. Munro, J. Inorg. Biochem. 2002, 91, 527–541.
[5] M. M. Hann, G. M. Keserü, Nat. Rev. Drug Discovery 2012, 11, 355–365.
[6] M. M. Hann, A. R. Leach, G. Harper, J. Chem. Inf. Comput. Sci. 2001, 41,
856–864.
[7] N. Teague, N. Davis, N. Leeson, N. Oprea, Angew. Chem. Int. Ed. 1999, 38,
3743–3748; Angew. Chem. 1999, 111, 3962–3967.
[28] Z. Ahmad, S. Sharma, G. K. Khuller, FEMS Microbiol. Lett. 2006, 261, 181–
186.
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1934 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[29] Z. Ahmad, S. Sharma, G. K. Khuller, FEMS Microbiol. Lett. 2005, 251, 19– [37] T. L. Poulos, A. J. Howard, Biochemistry 1987, 26, 8165–8174.
22.
[38] K. P. Conner, A. A. Cruce, M. D. Krzyaniak, A. M. Schimpf, D. J. Frank, P.
Ortiz de Montellano, W. M. Atkins, M. K. Bowman, Biochemistry 2015, 54,
1198–1207.
[30] Z. Ahmad, S. Sharma, G. K. Khuller, FEMS Microbiol. Lett. 2006, 258, 200–
203.
[31] P. Belin, M. H. Le Du, A. Fielding, O. Lequin, M. Jacquet, J.-B. Charbonni-
er, A. Lecoq, R. Thai, M. CourÅon, C. Masson, C. Dugave, R. Genet, J.-L.
Pernodet, M. Gondry, Proc. Natl. Acad. Sci. USA 2009, 106, 7426–7431.
[32] M. Gondry, L. Sauguet, P. Belin, R. Thai, R. Amouroux, C. Tellier, K. Tu-
phile, M. Jacquet, S. Braud, M. CourÅon, C. Masson, S. Dubois, S. Lautru,
A. Lecoq, S.-i. Hashimoto, R. Genet, J.-L. Pernodet, Nat. Chem. Biol.
2009, 5, 414–420.
[33] M. Fonvielle, M. H. Le Du, O. Lequin, A. Lecoq, M. Jacquet, R. Thai, S.
Dubois, G. Grach, M. Gondry, P. Belin, J. Biol. Chem. 2013, 288, 17347–
17359.
[34] Cytochrome P450: Structure, Mechanism and Biochemistry (Ed.: P. R. Ortiz
de Montellano), Kluwer Academic/Plenum Publishers, New York, 2005.
[35] S. A. Hudson, K. J. McLean, S. Surade, Y. Q. Yang, D. Leys, A. Ciulli, A. W.
Munro, C. Abell, Angew. Chem. Int. Ed. 2012, 51, 9311–9316; Angew.
Chem. 2012, 124, 9445–9450.
[39] D. E. Scott, A. G. Coyne, A. Venkitaraman, T. L. Blundell, C. Abell, M. Hy-
vçnen, ChemMedChem 2015, 10, 296–303.
[40] J. R. Halpert, Annu. Rev. Pharmacol. Toxicol. 1995, 35, 29–53.
[41] J. J. Sheets, E. Vickery, J. Biol. Chem. 1983, 258, 11446–11452.
[42] A. Nagahisa, T. Foo, M. Gut, W. H. Orme-Johnson, J. Biol. Chem. 1985,
260, 846–851.
[43] A. Choudhary, R. T. Raines, ChemBioChem 2011, 12, 1801–1807.
[44] H. E. Seward, A. Roujeinikova, K. J. McLean, A. W. Munro, D. Leys, J. Biol.
Chem. 2006, 281, 39437–39443.
[45] A. Borthwick, Chem. Rev. 2012, 112, 3641–3716.
[46] QikProp version 4.2, Schrçdinger LLC, New York, NY (USA), 2014.
Received: May 2, 2016
Revised: June 24, 2016
[36] J. H. Dawson, L. A. Anderssons, M. Sono, J. Biol. Chem. 1982, 257, 3606–
3617.
Published online on July 19, 2016
ChemMedChem 2016, 11, 1924 –1935
www.chemmedchem.org
1935 ꢀ 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim