acid synthase targeting the b-ketoacyl reductase reaction Marı´a Jesu´s Va´zquez1, William Leavens2, Ronggang Liu3, Beatriz Rodrı´guez1, Martin Read2, Stephen Richards2, Deborah Winegar4
Trang 1acid synthase targeting the b-ketoacyl reductase reaction Marı´a Jesu´s Va´zquez1, William Leavens2, Ronggang Liu3, Beatriz Rodrı´guez1, Martin Read2,
Stephen Richards2, Deborah Winegar4and Juan Manuel Domı´nguez1
1 GlaxoSmithKline R&D, Biological Reagents and Assay Development Department, Centro de Investigacio´n Ba´sica, Tres Cantos, Spain
2 GlaxoSmithKline R&D, Analytical Chemistry Department, Medicines Research Center, Stevenage, UK
3 GlaxoSmithKline R&D, Cardiovascular and Urogenital Centre of Excellence for Drug Discovery, King of Prussia, PA, USA
4 GlaxoSmithKline R&D, Metabolic Centre of Excellence for Drug Discovery, Research Triangle Park, Durham, NC, USA
Fatty acids are essential to all living cells and have a
wide diversity of biological functions As components
of phospholipids, triglycerides and other complex
lip-ids, fatty acids provide structural integrity to cellular
membranes and cell walls and serve as a source of
energy In the free state, fatty acids can act in the
transmission of cellular signals or can be used to
mod-ify proteins during post-translational processing The
de novosynthesis of fatty acids is accomplished by the
complex, multicomponent enzyme fatty acid synthase
(FAS; EC 3.2.1.85) FAS catalyzes the formation of
long-chain fatty acids from acetyl-CoA and
malonyl-CoA in a cyclic sequence of reactions, adding two
car-bon units per cycle, each reaction being catalyzed by a
different enzymatic activity The elongating chain is
covalently linked to an acyl carrier protein (ACP),
which transports it through the active sites where each reaction is catalyzed, namely malonyl-CoA⁄ acetyl-CoA-ACP-transacylase (MAT), b-ketoacyl synthase (KS), b-ketoacyl reductase (KR), dehydratase (DH), and b-enoyl reductase (ER) Once the synthesized fatty acid has reached the desired length, it is released from the ACP by means of a thioesterase (TE) activity FAS exists in two basic forms in nature Type I FAS is a multienzyme that integrates all the active sites catalyz-ing the individual reactions into one scatalyz-ingle polypeptide chain Type I FAS can be further classified as animal type I FAS, present as a homodimeric protein, and microbial type I FAS, which is present as oligomers of higher order (hexamers or dodecamers), including two types of polypeptide [1] In contrast, type II FAS, pres-ent in bacteria, plants, and eukaryotic mitochondria, is
Keywords
breast cancer; fatty acid synthase;
GSK837149A; ketoacyl reductase; obesity
Correspondence
J M Domı´nguez, GlaxoSmithKline R&D,
Centro de Investigacio´n Ba´sica, C ⁄ Santiago
Grisolı´a 4, 28760-Tres Cantos, Spain
Fax: +34 91 807 4062
Tel: +34 91 807 4000
E-mail: juan.m.dominguez@gsk.com
(Received 28 November 2007, revised 21
January 2008, accepted 30 January 2008)
doi:10.1111/j.1742-4658.2008.06314.x
GSK837149A has been identified as a selective inhibitor of human fatty acid synthase (FAS) The compound was first isolated as a minor impurity
in a sample found to be active against the enzyme in a high-throughput screening campaign The structure of this compound was confirmed by NMR and MS studies, and evaluation of the newly synthesized molecule confirmed its activity against FAS The compound and other analogs synthesized, all being symmetrical structures containing a bisulfonamide urea, act by inhibiting the b-ketoacyl reductase activity of the enzyme GSK837149A inhibits FAS in a reversible mode, with a Ki value of
30 nm, and it possibly binds to the enzyme–ketoacyl-ACP complex Although initial results suggest that cell penetration for these compounds is impaired, they still can be regarded as useful tools with which to probe and explore the b-ketoacyl reductase active site in FAS, helping in the design of new inhibitors
Abbreviations
ACP, acyl carrier protein; DH, dehydratase; ER, b-enoyl reductase; FabG, b-ketoacyl reductase enzyme of type II FAS; FAS, fatty acid synthase; HTS, high-throughput screening; KR, b-ketoacyl reductase; KS, b-ketoacyl synthase; MAT, malonyl-CoA ⁄
acetyl-CoA-ACP-transacylase; Q-TOF, quadrupole TOF; TE, thioesterase.
Trang 2composed of individual monofunctional proteins, each
of which catalyzes one individual reaction Whereas
type I FAS produces only palmitate or stearate, type II
FAS is able to synthesize fatty acids of different
lengths, degrees of saturation, branching, and
hydroxy-lation [2]
The homodimeric animal type I FAS was initially
believed to be arranged in an antiparallel, head-to-tail
mode with a large interdomain region that served as
an area of contact between the two monomers This
region was thought to be critical for the correct
associ-ation of the two monomers required for functional
activity [3,4] Such an arrangement would give rise to
two identical ‘catalytic chambers’, each formed by the
active sites of the two polypeptide chains The detailed
structure of mammalian type I FAS at 4.5 A˚
resolu-tion has recently become available [5] and has revealed
a very different organization The two monomers are
arranged in a parallel conformation, forming an
inter-twined dimer with two catalytic chambers, each
cham-ber being composed of active sites from the same
monomer except for the KS site The shape of this
dimer is rather asymmetric, and hence the sizes of the
catalytic chambers are significantly different from one
another This fact, together with the considerable
dis-tances among the active sites within each chamber,
suggests that the enzyme structure must be flexible and
that domain motions must occur during each catalytic
cycle in order to ensure the accessibility of the
elongat-ing fatty acid to the individual active sites
Clinical studies conducted over the last decade have
shown that a biologically aggressive subset of
carcino-mas constitutively expresses high levels of type I FAS
and that upregulation of FAS gene expression is an
early event in the development of certain types of
can-cer [6] Moreover, two inhibitors of FAS, can-cerulenin
and C-75, have demonstrated significant antitumor
activity Cerulenin has been shown to inhibit the
growth of neoplasic cells in vitro, with the inhibition
being reversed only with high, supraphysiological
con-centrations of palmitate [7] C-75, on the other hand,
has demonstrated efficacy in vivo against xenografts of
breast cancer cells in nude mice [8] Although the
mechanisms behind the antitumor effect of FAS
inhibi-tion are still under discussion, it is thought that the
intracellular accumulation of toxic levels of
malonyl-CoA and⁄ or impairments in cellular membrane
struc-ture may contribute to the inhibition of cell growth
Recent data obtained with breast cancer and
endome-trial adenocarcinoma cells showing that FAS
differen-tially modulates the sensitivity of the cells to estrogen
have led to the proposal that FAS inhibition may
provide an alternative breast cancer therapy that may
avoid the onset of endometrial hyperplasia associated with current tamoxifen-based therapies [9]
In addition to this role in oncogenicity, FAS has also been considered as a possible target for the treat-ment of obesity Both cerulenin and C-75 have been shown to reduce food intake and cause profound weight loss in mice [10] These effects correlate with reciprocal changes in the expression of orexigenic and anorexigenic neuropeptides in the hypothalamus The mechanisms responsible for these effects are thought to
be related to increased malonyl-CoA levels in the hypothalamus [11,12]
Despite the increased interest in FAS as a therapeu-tic target for cancer and obesity, there remain few selective human FAS inhibitors The two best charac-terized FAS inhibitors, cerulenin and C-75, are affected by several drawbacks that limit their use, the main one being their irreversible behavior Cerulenin is
a natural antibiotic produced by the fungus Cephalos-porium ceruleans, and is known to inhibit the KS reac-tion by covalently binding to the Cys residue in the corresponding active site [13] However, this com-pound cannot be considered to be a selective inhibitor
of human FAS, as it inhibits both type 1 and type 2 FAS [14], and possesses high chemical reactivity, which
is responsible for its low specificity Indeed, interfer-ence of cerulenin with other cellular processes besides fatty acid synthesis (e.g protein acylation, cholesterol synthesis, and proteolysis) has been described [15] Moreover, this reactivity is also responsible for the low chemical stability of the compound In addition, the observed effects of cerulenin in in vivo experiments have been poorly reproducible [16], suggesting that strain differences between mice may influence the response to this compound C-75 is a weakly potent (Ki16 mm) synthetic inhibitor designed through molec-ular modeling based on the mechanism of ketoacyl synthesis at the KS active site [17] Despite this tar-geted design, it has been shown to interact at several sites in FAS; that is, it is not a selective KS inhibitor [18] Certain activities ascribed to C-75 appear to be paradoxical (increase in malonyl-CoA and stimulation
of carnitine palmitoyltransferase-1) [19], hence compli-cating the interpretation of the effects observed with this compound The antiobesity drug Orlistat has been reported to inhibit FAS by interacting with its TE domain Crystal structures of Orlistat covalently bound
to FAS at the TE domain have recently become avail-able [20] Like the other molecules described above, Orlistat is not a selective inhibitor of FAS, and indeed the antiobesity effects of Orlistat are thought be pri-marily related to the irreversible inhibition of gastric and pancreatic lipases [21] Some flavonoids have been
Trang 3reported to inhibit FAS, but it is unclear whether this
family of compounds act in the KS or in the KR
domain [22] The discovery of hydroxyquinolinones
[23] as FAS inhibitors has also been reported recently,
but details of the mechanisms of action of these
mole-cules at the enzyme level have not been described
In this article, we describe the discovery of
GSK837149A, the first selective human FAS inhibitor
known to act specifically and selectively on the KR
activity of the enzyme
Results
High-throughput screening (HTS) to identify
human FAS inhibitors
Aproximately 550 000 compounds belonging to the
GlaxoSmithKline collection were screened in the
search for inhibitors of human FAS As the enzyme
consumes 14 NADPH molecules during the synthesis
of one molecule of palmitate, an assay monitoring
NADPH consumption, based on substrate-induced
quenching technologies [24], was used in the HTS
cam-paign The compounds were tested at a single
concen-tration of 10 lm in the assay mixture, using a final
assay volume of 3 lL In total, 12 735 compounds
yielding significant inhibition of FAS (detected by the
impaired ability of the enzyme to consume NADPH in
the assay) were selected These compounds were
grouped in different chemical clusters, prioritized
according to their chemical features, and evaluated in
a dose–response manner for their potency against the
enzyme Among the compounds selected, SKF-100601
was identified as the most promising one because of its
biological and chemical properties (structure shown in
Fig 1) This compound appeared to inhibit FAS
reversibly with a pIC50 of 6.0, and acted exclusively
on the KR activity, as demonstrated by experiments such as those described later in this article In addition, its chemical structure looked suitable for the intro-duction of a diversity of modifications that could contribute to the modulation of its pharmacological properties A modest chemical effort was therefore ini-tiated, consisting in searching for analogs in corporate databases as well as in affording the synthesis of the molecule All of the 222 analogs selected failed to inhi-bit FAS when they were tested at 100 lm in the assay Furthermore, the newly synthesized molecule also failed to inhibit the enzyme It was concluded that the activity observed in the original sample used in HTS was probably due to an impurity that was absent in the newly synthesized molecule If this hypothesis was correct, the potency of the active component should be
at least one order of magnitude higher than that found for the HTS sample, as it was determined that the HTS sample was 85% pure With this in mind, and considering the interesting biological properties described above, it was decided to undertake the puri-fication and identipuri-fication of the FAS inhibitor present
in the SKF-100601 sample used in the HTS campaign
Identification of the active component
The impure SKF-100601 sample (103 mg) was sub-jected to preparative reverse-phase chromatography in
a Supelco ABZ Plus column As shown in Fig 2, four
S
N
O
O
N
N
O
F
F F
S N O
N
S
N
O
O
N
N
O SKF-100601
GSK837149A
Fig 1 Chemical structure of SKF-100601 and GSK837149A.
Time (min)
0 5 10 15 20
0 50 100
Fig 2 Preparative chromatography of SKF-100601 The chromato-gram corresponds to the injection of a 0.5 mL sample containing
30 mg of the impure SKF-100601 material that had been identified
as inhibiting FAS in the HTS Samples were collected, dried, dis-solved in dimethylsulfoxide, and tested for their inhibition in the FAS assay from 50 lgÆmL)1to 0.8 ngÆmL)1 The line corresponds
to the elution profile as the summed response of all wavelengths from 190 to 600 nm in arbitrary units, whereas the bars show the inhibitory activity of the fractions collected: their ability to inhibit FAS is expressed as the reverse of the IC50 in lgÆmL)1for graphi-cal purposes No value is reported when no inhibition was found even at the highest concentration tested.
Trang 4major peaks were identified Samples corresponding to
these peaks, as well as to the rest of the elution profile,
were pooled, dried, and tested for their ability to
inhi-bit FAS Only the fractions corresponding to the peak
eluting at 7.56 min were able to inhibit the enzyme
Notably, the major component of the mixture, which
eluted at 10 min, did not show any inhibition of FAS
The active component was subjected to liquid
chroma-tography high-resolution MS, confirming the presence
of a single entity, which yielded the mass spectrum
shown in Fig 3A The parent ion with an m⁄ z of
555.1238 Da was observed, as well as masses of
446.0598 Da and 398.0923 Da corresponding to
fragments of the parent ion Elemental composition
analysis was performed on the parent ion, giving a
molecular formula of C23H22N8O5S2 with a 0.2 p.p.m
error
The purified active sample was finally subjected to
NMR analysis The spectrum obtained is shown in
Fig 3B Comparison of this spectrum with that of
purified SKF-100601 (obtained from the preparative
chromatography described above) suggested that the
trifluorophenyl moiety present in SKF-100601 was
absent in this active component Simple cleavage of
the urea to yield the corresponding aniline was ruled
out Indeed, the relatively sharp peak for the amino
group of urea pointed to a dimeric structure with the
urea moiety bridging between the two halves of the
molecule This, combined with the molecular formula and mass values deduced above, led us to propose the structure shown in the lower part of Fig 1 as the one corresponding to the compound inhibiting FAS
Confirmation of the activity
On the basis of these analytical studies, 70 mg of the compound corresponding to the structure shown in Fig 1, which was registered as GSK837149A, was pre-pared The compound was tested in the FAS assay in parallel with the SKF-100601 sample used in the HTS campaign As shown in Fig 4, the new compound was able to inhibit FAS and showed a significantly higher potency than the SKF-100601 sample Fitting of these concentration–response curves to Eqn (1) (see Experi-mental procedures) yielded pIC50 values of 7.8 and 6.0 for GSK837149A and for the sample of SKF-100601 respectively These differences in potency suggest that GSK837149A represents 1–2% of the impure
SKF-100601 sample The new compound was also able to inhibit the FAS activity of crude cytosolic extracts from rat liver homogenates (data not shown), demon-strating that the ability to inhibit FAS is genuine and
is not an artefact generated by the use of recombinant enzyme, and also demonstrating that such inhibition is not restricted to the human enzyme but extends to FAS enzymes from other mammals
m/z
0 100 200 300 400 500 600 700 800 900 1000
0
20
40
60
80
100
398.0923
555.1238
446.0598
Chemical shift (p.p.m.) 0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
A
B
Fig 3 Analytical data on the active component identified (A) Mass
spectrum (B) Proton NMR spectrum.
log [compound] ( M )
0 20 40 60 80 100
Fig 4 Concentration–response curves of the newly synthesized GSK837149A and the impure SKF-100601 sample Evaluation of FAS activity was performed as described in Experimental proce-dures at different concentrations of compounds Data were fitted
to Eqn (1) to calculate the pIC50 value for GSK837149A and the apparent pIC50 for SKF-100601 as if it was pure and responsible for the inhibition caused The data correspond to the average of triplicate determinations ± SEM d, GSK837149A; s, SKF-100601.
Trang 5Analysis of the individual activities catalyzed
by FAS
There are several methods reported in the literature for
analysis of the individual FAS reactions [18,25], but
not all of them are adequate for the rapid analysis of a
large number of samples such as those derived from an
HTS campaign We decided to investigate the
possi-bility of using suitable substrates for some of the
individual reactions and monitoring them
spectro-photometrically through the consumption of NADPH
as described in Dodds et al [26] for the bovine
enzyme Therefore, acetoacetyl-CoA was used to
moni-tor KR, b-hydroxybutyryl-CoA was used to monimoni-tor
the activity of DH, and crotonoyl-CoA was used to
monitor the activity of ER NADPH consumption was
observed in all the cases (data not shown), confirming
that human FAS is able to utilize these CoA
deriva-tives as substrates The fact that NADPH is also
con-sumed when b-hydroxybutyryl-CoA is used also
confirms that the enzyme is able to catalyze all the
reactions following the one corresponding to the
sub-strate used The reactions were also monitored by
HPLC, using an appropriate method to isolate and
quantify the intermediate substrates and products
Figure 5A shows that when acetoacetyl-CoA and
NADPH were used as substrates, butyryl-CoA was
formed in parallel to the consumption of
acetoacetyl-CoA, whereas traces of b-hydroxybutyryl-CoA and of
crotonoyl-CoA (the latter was present in very small
amounts and was difficult to quantify properly) were also observed No traces of free CoA were detected
On the other hand, when acetyl-CoA and malonyl-CoA were used as substrates, their disappearance ran
in parallel with the release of free CoA, whereas no other intermediate of the FAS reaction was detected (Fig 5D) When b-hydroxybutyryl-CoA was used (Fig 5B), the formation of butyryl-CoA occurred simultaneously with the consumption of b-hydroxy-butyryl-CoA, and a steady-state concentration of the intermediate crotonoyl-CoA was also detected Finally, when crotonoyl-CoA was used, slow formation of butyryl-CoA was seen as well as significant levels of b-hydroxybutyryl-CoA (Fig 5C), suggesting that DH
is able to promote the reverse reaction quite efficiently,
as already reported [27] Therefore, the use of these substrates can be exploited to analyze the effect of FAS inhibitors on the KR, DH and ER reactions in a highly efficient manner simply by observing the effect
on NADPH consumption
Identification of the inhibited FAS activity and selectivity studies
GSK837149A was used in the reactions described above to identify which of the enzymatic reactions was inhibited by the compound Concentration–response studies were conducted on these reactions: as shown in Fig 6, the compound is able to inhibit KR, as well as the global FAS reaction, but neither DH nor ER
Time (min) 0
0 0.5 1 1.5 2
Time (min)
0
0.5
1
1.5
2
Time (min)
0 0.2 0.4 0.6 0.8 1
Time (min)
0
0.5
1
1.5
2
Fig 5 Analysis of the reactions catalyzed
by FAS in the presence of different sub-strates The reactions were performed in duplicate as described under Experimental procedures, except for the reaction volume, which was 100 lL, with 30 l M NADPH and
10 n M FAS and the substrates described below At the desired times, the reactions were stopped in an ice bath, and 50 lL was injected in the HPLC system to analyze the components of the reaction mixture follow-ing the method described in Experimental procedures The time courses correspond to the reactions run with 40 l M acetoacetyl-CoA (A), 40 l M b-hydroxybutyryl-CoA (B),
40 l M crotonoyl-CoA (C), and 2 l M acetyl-CoA and 20 l M malonyl-CoA (D) The follow-ing products were monitored: acetyl-CoA (d), malonyl-CoA ( ), CoA (.), acetoacetyl-CoA (s), b-hydroxybutyryl-acetoacetyl-CoA (h), croto-noyl-CoA (D), and butyryl-CoA ( ).
Trang 6Because the other activities involved in FAS (i.e.
MAT, KS, TE, and the binding to ACP) were not
tested, it was not possible to rule out any effect on
these at this point However, the similar potencies
found for the compound in the global FAS reaction
(i.e the reaction promoted in the presence of
acetyl-CoA and malonyl-acetyl-CoA besides NADPH) and in the
KR reaction suggest that KR is the only activity being
inhibited Moreover, several analogs of GSK837149A
were prepared: as shown in Fig 7, a good linear
corre-lation (r2= 0.97) was observed between the pIC50 values obtained when the global FAS and the KR activities were measured Therefore, the possibility that these compounds act on FAS by inhibiting another activity in addition to KR seems unlikely
To check the selectivity of these inhibitors, their effect on the b-ketoacyl reductase enzyme of type II FAS (FabG), was tested, following the procedures described in Patel et al [28] Neither GSK837149A nor any of the analogs caused any inhibition of FabG when they were present at 100 lm, suggesting that the compounds are selective in inhibiting the human KR
of type I FAS The reversible nature of the inhibition caused by GSK837149A on FAS was also demon-strated: as shown in Fig 8, FAS could be rescued from the inhibitory effect of GSK837149A by simply removing the unbound compound by filtration and diluting to reconstitute the initial volume The drop-off
in the effect of the compound corresponded to the dilution made, and as the enzymatic assay was run immediately after diluting the filtered sample, the dis-sociation step seems to be rapid The same result was obtained when other analogs of GSK837149A were tested in the global FAS assay and in the KR activity assay (data not shown)
0
20
40
60
80
100
120
log [GSK837149A] ( M ) –10 –9 –8 –7 –6 –5 –4
Fig 6 Concentration–response curves of GSK837149A on the
indi-vidual activities catalyzed by FAS The reactions were performed as
described in Experimental procedures with 30 l M NADPH, 10 n M
FAS, and the following substrates: 2 l M acetyl-CoA and 20 l M
malonyl-CoA (d); 40 l M acetoacetyl-CoA (s); 40 l M
b-hydroxy-butyryl-CoA (h); and 40 l M crotonoyl-CoA (D) Data were fitted to
Eqn (1) to calculate the pIC50 values The data presented here
correspond to the average of triplicate determinations ± SEM.
pIC50 on global FAS
5.5
6
6.5
7
7.5
8
8.5
9
Fig 7 Correlation among pIC50 values obtained in the assays
monitoring the global FAS reaction and KR.
log [GSK837149A] ( M )
–20 0 20 40 60 80 100
Fig 8 Reversibility of inhibition by GSK837149A The experiment was performed by concentration and subsequent dilution (25-fold)
of the enzyme and inhibitor mixtures, using filters of a suitable pore size to allow the separation of the free compound from the enzyme, as described in Experimental procedures All samples were compared with their corresponding controls treated similarly
in the absence of compound The values in the horizontal axis cor-respond to the concentration of compound in the sample before any concentration and dilution cycle The lines represent the results
of fitting the data to Eqn (1) d, untreated sample (no dilution);
h , sample concentrated and diluted once (25-fold dilution); D, sam-ple concentrated and diluted twice (625-fold dilution).
Trang 7Enzymatic mode of inhibition
With the aim of finding the mode of inhibition of
GSK837149A on the KR activity of FAS substrates,
titrations were performed at different concentrations of
the inhibitor, varying the concentration of one
sub-strate while keeping the concentration of the other
fixed at saturating (i.e 10 times the Kmvalue) or
non-saturating (i.e at the Kmvalue) levels The results are
summarized in Fig 9 and in Table 1 The inhibition
constants obtained in all cases suggest that the Ki
value for GSK837149A is between 25 and 35 nm, in
agreement with the observed pIC50 value This
com-pound showed a clear competitive inhibition with
respect to NADPH that was not altered by the use of
saturating concentrations of acetoacetyl-CoA,
suggest-ing that the compound binds to the same enzyme form
that NADPH does On the other hand, the inhibition
pattern with respect to acetoacetyl-CoA was clearly
uncompeitive In all, these results suggest that the KR
activity catalyzed by FAS follows a compulsory
ordered kinetic mechanism, acetoacetyl-CoA being the
first substrate to bind, and that the inhibitor binds to
the enzyme–acetoacetyl-CoA complex to eventually
form a ternary dead-end complex
enzyme–acetoacetyl-CoA–GSK837149A
Discussion
In the search for novel inhibitors of human FAS, we
have identified GSK837149A as a potent and selective
inhibitor of the enzyme The high potency of this
com-pound has allowed its detection in an HTS campaign,
even though it was present as a minor impurity in the
sample of a different compound, and such potency has
also enabled us to track it during the isolation and
purification process from the original sample NMR and high-resolution MS data were pooled, and draw-ing on experience from related compounds that had been fully assigned by means of NOE experiments, a putative dimeric structure was proposed Such a struc-ture, shown in Fig 1, satisfied the NMR data in all respects and also gave the required mass of 555.1233 Da The synthesis of the compound corre-sponding to this structure, GSK837149A, led us to confirm that it was responsible for the inhibition of FAS
The results obtained demonstrate that GSK837149A inhibits the b-ketoacyl reductase activity of the enzyme Our experimental data demonstrate that neither DH nor ER activities are inhibited by the compound The other activities involved in the FAS-catalyzed cycle (MAT, KS, TE, and the binding to ACP) were not tested; however, the similarity between the inhibition potencies on KR and on the global FAS reaction suggests that KR is the only activity of the enzyme being inhibited Moreover, this similarity in potency extends to other analogs of GSK837149A cov-ering three orders of magnitude in potency, and indeed
an excellent correlation is observed among the pIC50 values for KR and global FAS If another enzymatic reaction was inhibited, such a correlation would be
1 / [NADPH]
0 1 2 3 4 5
0 2 4 6 8 10
1 / [AcAcCoA]
0
0
0.5
1
1.5
1 / [NADPH]
Fig 9 Double reciprocal plots of the inhibition caused by GSK837149A of the KR reaction of FAS with different varied substrates The experiment was run as described in Experimental procedures at a fixed nonsaturating concentration of NADPH (A) and at fixed nonsaturating (B) and saturating (C) concentrations of acetoacetyl-CoA Initial rates are expressed in l M NADPH consumed per min and concentrations are
in l M The lines correspond to the results of fitting by nonlinear regression all the experimental data to Eqn (2) (B,C) or Eqn (3) (A) The resulting inhibition constants are summarized in Table 1 Concentrations of the inhibitor were 0 n M (s), 16 n M (d), 31 n M (h), 62.5 n M ( ),
125 n M (D) and 250 n M ( ).
Table 1 Summary of inhibition patterns caused by GSK837149A
on the KR reaction catalyzed by FAS.
Varied substrate
Concentration of fixed substrate
Inhibition pattern Ki(n M )
Trang 8difficult to explain unless the structure–activity
rela-tionships sustaining these inhibitions were the same for
KR as for the other sites affected, which seems very
unlikely, especially in the light of the differences
among the individual active sites in FAS [5]
As stated previously, the human FAS inhibitors
most extensively characterized so far are not selective
for the human enzyme In fact, most of them act
irre-versibly through covalent bond formation due to the
high chemical reactivity of the molecules rather than
to their similarity with intermediates of the enzymatic
reaction, which would lead to a more selective
mecha-nism-based inhibition GSK837149A has shown to act
reversibly, as its inhibition disappeared rapidly after
dilution, suggesting that the kinetics of association and
dissociation are fast The selectivity of the compound
is endorsed by its inability to inhibit FabG, the KR
counterpart in type II FAS
The potency of GSK837149A is such that its Ki
value (25–35 nm) is similar to the concentration of
enzyme used in the experiments reported in this article
(10 nm) Under these conditions, inhibitor depletion is
significant, and therefore the treatment of enzyme
inhi-bition data through classic steady-state methods is
questionable The quadratic equation derived by
Morrison [29] was used to recalculate the pIC50 values
that had been obtained with Eqn (1) (see Experimental
procedures), and essentially identical values were
yielded Analogously, the considerations established by
Williams & Morrison [30] were taken into account for
Ki determination in the mode of inhibition studies
However, the results shown in Fig 9 are clear, and
there are no reasons to suspect that these are
errone-ous, as might have been the case if a noncompetitive
pattern had been obtained In any case, other simpler,
graphical methods considering inhibitor depletion,
such as those described in Henderson [31], have been
used and have yielded identical Ki values and
inhibi-tion modes (data not shown)
The fact that the compound behaves as a
competi-tive inhibitor with respect to NADPH and an
uncom-petitive inhibitor with respect to acetoacetyl-CoA
suggests a compulsory order in the binding of the
sub-strates, with the reduced nucleotide being the second
substrate to bind There are numerous precedents in
the literature for compulsory ordered mechanisms in
reductases and dehydrogenases depending on
nicotin-amide adenine dinucleotides, but these mechanisms
usually start with the binding of the dinucleotide to
the free enzyme [32] A random mechanism has
recently been described for the b-ketoacyl reductase of
Streptococcus pneumoniae [24] However, we are not
aware of many compulsory ordered mechanisms with
the nucleotide binding last in sequence At this point,
it is not possible to ignore the possibility that this result may be an artefact introduced by the use of a non-natural substrate such as acetoacetyl-CoA None-theless, the result may also have implications for understanding the mechanism that FAS utilizes to carry out each catalytic cycle, suggesting that the growing ketoacyl–ACP complex must reach the KR active site to allow NADPH to bind
GSK837149A and several related analogs have been tested for cellular activity in a whole cell fatty acid synthesis assay in HepG2 cells, using [14C]acetate as a precursor, and in human primary adipocytes by moni-toring triglyceride accumulation (data not shown) Unfortunately, all the compounds were inactive in these assays Subsequent evaluation in a cell perme-ability assay confirmed very low cell permeperme-ability for GSK837149A Several chemical modifications of the parent compound have been tried, including alkylation
of urea and sulfonamide, removal of the sulfonamide groups, replacement of urea by other linkers, and preparation of asymmetrical analogous ureas In total, about 150 compounds have been synthesized, but per-meability has not improved and no activity in whole cell has been detected Although this result prevents the immediate therapeutic application of this chemical family, the relevance of the inhibitors is still noticeable because of their high-affinity binding to the enzyme and the novelty of their mechanism of action It is conceivable that they could be utilized as probes to investigate the KR active pocket, to aid the rational design of new molecules inhibiting FAS by binding at such sites
Experimental procedures
Materials
Human FAS was prepared by the Biological Reagents group at GlaxoSmithKline as described below St pneumo-niae b-ketoacyl-ACP reductase (FabG) was provided by the same group, following the procedures described in Patel et al [28] Unless otherwise stated, all chemical reagents used were from Sigma-Aldrich (St Louis, MO, USA)
Expression and purification of human FAS
Human FAS was cloned from a human testis cDNA library created from testis RNA purchased from Clontech (Moun-tain View, CA, USA) The gene was cloned into the Sal1⁄ Not1 sites of pSPORT1 and subsequently engineered into a pFastBac-1 expression system (Invitrogen, Carlsbad, CA,
Trang 9USA) with a C-terminal His6 tag The protein was
expressed in baculovirus-infected Spodoptera frugiperda Sf9
cells under the polyhedron promoter For large-scale
pro-duction, FAS was expressed at 27C in a 10 L or 100 L
cell bag using a Wave bioreactor (GE Healthcare, Giles,
Buckinghamshire, UK) Sf9 cells were infected using a
mul-tiplicity of infection of 1.0 The cells were harvested 48 h
after infection by centrifugation at 10 000 g, washed twice
with NaCl⁄ Pi, and flash frozen in a dry ice ethanol bath
They were subsequently thawed and mixed at a 5 : 1
(mLÆg)1) ratio with lysis buffer (100 mm NaCl in 25 mm
Hepes at pH 7.5) containing 20 mm imidazole
Homogeni-zation was carried out by mechanical disruption using a
Brinkmann polytron (Brinkmann Instruments Inc.,
West-bury, NY, USA), removing debris by centrifugation for 1 h
at 25 000 g The soluble fraction was filtered through a
1.2 lm filter (PALL, East Hills, NY, USA) and then loaded
onto a nickel-chelating Sepharose column (GE Healthcare)
equilibrated in lysis buffer with 20 mm imidazole All
chro-matography steps were performed at 4C After loading,
the column was washed with five column volumes of lysis
buffer with 20 mm imidazole followed by five column
vol-umes of 30 mm imidazole in lysis buffer FAS was eluted
with a 30–300 mm imidazole gradient in lysis buffer over
two column volumes followed by an eight column volume
hold at 300 mm imidazole in lysis buffer Fractions
contain-ing FAS were pooled and concentrated uscontain-ing PALL
JumboSep filters The sample was then applied to a
Super-dex 200 column (GE Healthcare) equilibrated with 100 mm
NaCl, 2 mm dithiothreitol, 1 mm EDTA and 15% glycerol
in 50 mm Tris⁄ HCl at pH 8.0 Fractions were collected and
checked for FAS activity, and active samples were pooled
and concentrated using JumboSep filters, aliquoted, and
frozen at)80 C
Measurement of FAS activity
Fatty acid synthesis catalyzed by FAS was followed by
monitoring NADPH consumption During the HTS
cam-paign, this was done by measuring the release of the
NADPH-induced quenching of resorufin as detailed in
Vazquez et al [24], as this procedure is suited for
minia-turization and thus enables the analysis of a large number
of samples, minimizing costs and protein expenses In all
other cases, the consumption of NADPH was monitored
spectrophotometrically through the decrease in the
absor-bance at 340 nm at 25C in 384-well clear-bottomed
polystyrene plates (Corning, Corning, NY, USA), using
a 384-well plate spectrophotometer (SpectraMax Plus;
Molecular Devices, Sunnyvale, CA, USA) The reactions
were carried out in a final volume of 50 lL containing
2 lm acetyl-CoA, 20 lm malonyl-CoA, 30 lm NADPH
and 10 nm enzyme in 50 mm sodium phosphate buffer at
pH 7.0 The extinction coefficient of NADPH for 50 lL
in these plates has been determined to be 3120 m)1 Initial rates were calculated from the slope of the progress curves during the first 3 min When the inhibition caused by compounds was being determined, these were added in neat dimethylsulfoxide, ensuring that the final dimethyl-sulfoxide concentration was 1% in all cases When dealing with compounds of a known molecular mass, their poten-cies were determined by means of the negative logarithm
of the IC50 (i.e the pIC50), as the experiments were car-ried out by performing serial dilutions of the compounds, and hence the range of concentrations was evenly distrib-uted in a logarithmic rather than a linear scale pIC50s were calculated by fitting the data to Eqn (1), using the nonlinear regression function of grafit 5.0.8 (Erithacus Software Limited) Equation (1) is a modified version of the classic isotherm equation
% Inhibition¼ Min þ Max Min
1þ 10ðlog½I þ pIC50Þn ð1Þ where % inhibition was calculated from the ratio of the ini-tial velocities in the presence and absence of compound,
‘Min’ and ‘Max’ are the lower and higher asymptotes of the sigmoid curve obtained, [I] is the concentration of inhibitor in molar units, and n is the Hill coefficient
Measurement of KR, DH and ER activities
The activities of the individual reactions catalyzed by FAS were monitored spectrophotometrically through the decrease in the absorbance caused by NADPH at 340 nm
at 25C in 384-well clear-bottomed polystyrene plates (Corning), using a 384-well plate spectrophotometer (Spec-traMax Plus; Molecular Devices) The reactions were car-ried out in a final volume of 50 lL containing 30 lm NADPH and 10 nm enzyme in 50 mm sodium phosphate buffer at pH 7.0, as well as either 40 lm acetoacetyl-CoA (for determining KR), 40 lm b-hydroxybutyryl-CoA (for DH), or 40 lm crotonoyl-CoA (for ER) The concentra-tions of these substrates have been determined previously
to correspond to their apparent Km values Initial rates were calculated from the slope of the progress curves dur-ing the first 3 min When the inhibition caused by com-pounds was being determined, these were added in neat dimethylsulfoxide, ensuring that the final dimethylsulfoxide concentration was 1% in all cases
For determining the inhibition pattern caused by GSK837149A, initial velocity data were obtained using the same buffer conditions with variable concentrations of acetoacetyl-CoA (8–500 lm) at a fixed nonsaturating centration of NADPH (30 lm), as well as at variable con-centrations of NADPH (8–250 lm) at fixed nonsaturating and saturating concentrations of acetoacetyl-CoA (20 lm and 200 lm, respectively) Data analysis was performed by fitting the experimental data to the appropriate equations
Trang 10with the nonlinear regression function of grafit 5.0.8
(Erithacus Software Limited), using Eqns (2,3) for
compet-itive and uncompetcompet-itive inhibition respectively:
v0¼ V½S
Km 1 þ ½ IKic
v0¼ V½S
Kmþ ½S 1 þ Kiu½I ð3Þ where Kic denotes the inhibition constant for competitive
inhibition and Kiuthe inhibition constant for uncompetitive
inhibition, [I] and [S] are the concentrations of the inhibitor
and the substrate being varied respectively, Km is the
Michaelis constant for such substrate, and V is the
maxi-mum velocity
Reversibility experiments
The reversibility of the inhibition caused by GSK837149A
was tested by filtration followed by dilution of the
concen-trated retentate using Ultrafree MC filters (Millipore,
Biller-ica, MA, USA) with a cut-off of 10 kDa Samples of 10 nm
FAS in 50 mm sodium phosphate buffer at pH 7.0 were
incubated with a given concentration of compound for
30 min, and then filtered by centrifugation according to the
manufacturer’s instructions The retentate sample, whose
volume was 1⁄ 25th of that of the original sample, was
diluted 25-fold to recover the original volume Immediately
afterwards, it was tested for FAS activity or subjected to a
new filtration and dilution cycle (625-fold dilution) and
finally tested
Chromatographic procedures
Preparative chromatography was accomplished using a
system from Waters (Milford, MA, USA) with a Supelco
ABZ Plus column (100· 21.2 mm, 5 lm particle size)
Samples were injected in 0.5 mL containing 30 mg of
material A 15 min linear gradient was used, starting with
water with 0.1% (v⁄ v) formic acid and ending with 95%
acetonitrile with 0.05% (v⁄ v) formic acid at a flow rate of
20 mLÆmin)1
Analytical chromatography for liquid chromatography
high-resolution MS experiments was carried out in an
Agi-lent 1100 instrument (AgiAgi-lent Inc, Santa Clara, CA, USA)
with a Phenomenex Luna C18(2) reverse-phase column
(150· 2.1 mm, 3 lm particle size (Phenomenex, Torrance,
CA, USA) Samples of 5 lL were injected, and a linear
gra-dient elution was carried out at a flow rate of 0.4 mLÆ
min)1, starting from water containing 0.1% (v⁄ v) formic
acid and ending with acetonitrile containing 0.1% (v⁄ v)
for-mic acid in 21 min The system was left in these final
condi-tions for 5 min, and then decreased linearly to the initial
conditions over 1 min; this was followed by an equilibra-tion period of 3 min prior to the next injecequilibra-tion
Chromatographic analysis of the individual reactions cat-alyzed by FAS was also performed in an Agilent 1100 sys-tem, using a Kromasil C18 ODS-2 column (250· 4.6 mm,
5 lm particle size) (Eka Chemicals, Bohus, Sweden) Sam-ples of 50 lL were injected and subjected to a gradient of methanol⁄ ammonium acetate (50 mm in water at pH 7.4), progressing linearly from 10 : 90 to 45 : 55 in 20 min at a constant flow rate of 1 mLÆmin)1
MS
Positive ion mass spectra were acquired as accurate mass centroided data using a Micromass two hybrid quadrupole TOF (Q-TOF) mass spectrometer (Waters), equipped with
a Z-spray interface, over a mass range of 80–1100 Da, with
a scan time of 0.95 s and an interscan delay of 0.07 s Reserpine was used as the external mass calibrant ([M + H]+= 609.2812 Da) The mass spectrometer was operated in W reflectron mode to give a resolution (full width at half maximum) of 16 000–20 000 Ionization was achieved with a spray voltage of 3 kV, a cone voltage of
30 V, and cone and desolvation gas flows of 5–10 and
500 LÆmin)1 respectively The source block and desolvation temperatures were maintained at 120C and 250 C respec-tively The elemental composition was calculated using masslynxv3.5 (Waters) for the [M + H]+, and the mass error was quoted as p.p.m
NMR spectroscopy
NMR spectra were acquired in dimethylsulfoxide solution
on a Varian UNITY 400 MHz spectrometer (Varian Inc., Palo Alto, CA, USA) using a 5 mm inverse geometry probe Two or three milligrams of solid were dissolved in 0.6 mL of solvent Standard acquisition parameters were employed (20 p.p.m sweep width and 0.5 s relaxation delay)
Acknowledgements
We wish to thank Warren Rocque and William Burk-hart for their efficient help in providing with biological materials, as well as David Bickett, David Musso and Michael Moore for helpful discussions
References
1 Lomakin IB, Xiong Y & Steiz A (2007) The crystal structure of yeast fatty acid synthase, a cellular machine with eight active sites working together Cell 129, 319– 331