FTO catalyzes the Fe(II) and 2-oxoglutarate (2OG)-dependent modification of nucleic acids, including the demethylation of N6-methyladenosine (m6A) in mRNA. FTO is a proposed target for anti-cancer therapy. Using information from crystal structures of FTO in complex with 2OG and substrate mimics, we designed and synthesized two series of FTO inhibitors, which were characterized by turnover and binding assays, and by X-ray crystallography with FTO and the related bacterial enzyme AlkB. A potent inhibitor employing binding interactions spanning the FTO 2OG and substrate binding sites was identified. Selectivity over other clinically targeted 2OG oxygenases was demonstrated, including with respect to the hypoxia-inducible factor prolyl and asparaginyl hydroxylases (PHD2 and FIH) and selected JmjC histone demethylases (KDMs). The results illustrate how structure-based design can enable the identification of potent and selective 2OG oxygenase inhibitors and will be useful for the development of FTO inhibitors for use in vivo.
FTO catalyzes the Fe(II) and 2-oxoglutarate (2OG)-dependent modification of nucleic acids, including the demethylation of N6-methyladenosine (m6A) in mRNA. FTO is a proposed target for anti-cancer therapy. Using information from crystal structures of FTO in complex with 2OG and substrate mimics, we designed and synthesized two series of FTO inhibitors, which were characterized by turnover and binding assays, and by X-ray crystallography with FTO and the related bacterial enzyme AlkB. A potent inhibitor employing binding interactions spanning the FTO 2OG and substrate binding sites was identified. Selectivity over other clinically targeted 2OG oxygenases was demonstrated, including with respect to the hypoxia-inducible factor prolyl and asparaginyl hydroxylases (PHD2 and FIH) and selected JmjC histone demethylases (KDMs). The results illustrate how structure-based design can enable the identification of potent and selective 2OG oxygenase inhibitors and will be useful for the development of FTO inhibitors for use in vivo.
The fat mass and obesity
associated protein (FTO) is an Fe(II)
and 2-oxoglutarate (2OG)-dependent oxygenase that catalyzes the oxidation
of N-methyl groups in nucleic acids, and, in particular,
of N6-methyladenosine (m6A)
in RNA.[1−3] FTO is a potential medicinal chemistry target due
to its roles in obesity and cancer. Single nucleotide polymorphisms
in an intron of the fto gene were reported to be
associated with body weight and type II diabetes in genome-wide association
studies.[4,5] More recent studies have reported roles
for FTO in m6A regulation and cancers, including acute
myeloid leukemia and glioblastoma.[6−11] Precisely how RNA oxidation by FTO is linked to disease is uncertain.
There is thus an unmet need to develop small-molecule probes to complement
genetic studies and explore its potential as a therapeutic target.FTO is one of 60–70 human 2OG oxygenases, some of which
are medicinal chemistry targets, for example, the hypoxia-inducible
factor prolyl hydroxylases (PHD 1–3) and selected JmjC Nε-methyllysine histone demethylases (KDMs).[12,13] PHD inhibitors are approved for the treatment of anemia in chronic
kidney disease, though there have been concerns regarding their safety.[14] Human nucleic acid oxygenases (NAOXs), including
FTO, are a substantial subset of 2OG oxygenases (>10 enzymes).
NAOXs
include eight AlkB homologues (ALKBH 1–8), some of which are
involved in DNA damage repair (e.g., ALKBH2) and RNA oxidation (e.g.,
FTO and ALKBH5) and the Ten Eleven Translocation enzymes (TET1-3),
which catalyze sequential oxidations of the methyl group of 5-methylcytosine.[15,16] FTO catalyzes hydroxylation which can lead to demethylation of N-methyl groups on single-stranded RNA and DNA substrates
(Figure A), including N3-methylthymidine (m3T), N3-methyluridine (m3U), m6A, and N6,2′O-dimethyladenosine
(m6Am).[1−3,17] The lifetime of the
nascent N-hydroxymethylated products/intermediates
varies depending on the nucleobase undergoing oxidation, leading to
the proposal that in some circumstances they may be biologically relevant.[17−19]
Figure 1
(A)
Schematic representation of FTO-catalyzed reactions. (B) Reported
FTO inhibitors with IC50[26] values
against FTO (determined using an LC assay with the m3T
nucleoside as a substrate).[26]
(A)
Schematic representation of FTO-catalyzed reactions. (B) Reported
FTO inhibitors with IC50[26] values
against FTO (determined using an LC assay with the m3T
nucleoside as a substrate).[26]There are some reports of FTO inhibition, including by known
inhibitors
of 2OG oxygenases, such as PHD inhibitors (Figure S1).[6,7,20−26] We have reported crystallographic and in vitro inhibition
studies of FTO with known 2OG oxygenase inhibitors including 2OG competitors/analogues
(2–4) and rhein (1) (Figure B).[26,27] The most potent FTO inhibitor identified, FG-2216 (2, IC50 3 μM), is a PHD inhibitor.[28] Treatment of mice with FG-2216 (2) led to
reduced bone mineral density and altered adipose tissue density distribution
without affecting body weight or respiratory exchange ratio.[29] Structures of FTO in complex with 3 (FTO IC50 3.3 μM) and 4 (FTO IC50 15 μM) were also obtained.[26]3 was reported as a broad spectrum inhibitor of the
JmjC histone lysine demethylases (KDM2A, 3A, 4A-E, 5C, 6A, and B,
7B: IC50 0.2–25 μM) and inhibits the bacterial
NAOX AlkB (IC50 3.4 μM).[26,30,31] The reported FTO inhibitors thus likely
do not have the potency or selectivity required for in vivo target validation of FTO. Here, we report the use of structural
knowledge of how FTO binds 2OG and substrate to design potent and,
at least partially selective, FTO inhibitors.
Results
During
2OG oxygenase catalysis, binding of 2OG in a bidentate manner
to the active site Fe(II) is followed by that of the substrate, and
then dioxygen.[12] One strategy in 2OG oxygenase
inhibitor design has been to covalently link a non-reactive 2OG mimetic
with a group occupying the substrate binding site, so yielding more
potent and selective inhibitors than the independent fragments.[32,33] Unlike 2–5, which bind in the 2OG binding pocket
of FTO, rhein (1) occupies the nucleobase binding site.[26] A comparison of the FTO and 4 (a
glycyl-hydroxypyridine 2OG mimic) and FTO and rhein (1)[26] structural complexes indicates that
the hydroxypyridinyl ring of 4 is proximate to the anthraquinone
ring of rhein (1), which binds competitively with the
substrate nucleobase (Figures A and S2). Based on this analysis,
we designed and synthesized a series of derivatives of 4 (Table ) predicted
to extend into the substrate binding site by functionalization at
the C-6 position (13a–d) (Figure A, Scheme ). 3-Hydroxypicolinonitrile was used as the starting
material for C-6 derivatization of 4. C-6 halogenation
of the pyridine ring of hydroxypicolinonitrile, followed by hydrolysis
of the nitrile gave 6-bromo-3-hydroxypicolinic acid (7), which was coupled with a protected glycine amide (glycine tert-butyl ester hydrochloride or glycine phenyl ester hydrochloride).
The resulting compounds were coupled to different aryl rings by Suzuki
or Buchwald couplings, then deprotected to give the desired acids
(13a–d and 14a–i).
Figure 2
Structure-based FTO inhibitor design. (A) View of the active site
from a crystal structure of FTOΔ31 (white sticks) in complex
with 4 (magenta sticks) (PDB ID 4IE5),[26] which was used as a basis for inhibitor design, with the
structure of FTOΔ31-1 (rhein, yellow sticks) (PDB
ID 4IE7)[26] superimposed. The inset indicates how the derivatization
of 4 at C6 could enable inhibition by combining fragments
of 2OG and nucleobase mimetics. (B) View of the active site from a
structure of FTOΔ31 (white sticks) in complex with 13c (cyan sticks surrounded by OMIT mFo-DFc electron
density displayed as green mesh, contoured to 3.0σ). (C) Superimposition
of a structure of FTOΔ31-13c with that of FTOΔ31-4 (PDB ID 4IE5; 4 in magenta) and FTOΔ31-1 (PDB
ID 4IE7, 1 in yellow). (D) View of the active site of FTO in complex
with 14a. The FTOΔ31 active site (white sticks)
showing OMIT electron density (mFo-DFc, contoured
to 3.0σ in a green mesh) around 14a (salmon sticks).
(E) View of superimposed structures of FTOΔ31-13c (PDB ID 4QHO, 13c cyan sticks), FTOΔ31-14a, and
a FTO-ssDNA substrate (lime green sticks) complex (PDB ID: 5ZMD). Only FTO residues
(white sticks) from the FTOΔ31-14a structural complex
are shown. (F) Superimposed structures of FTOΔ31-13c (PDB ID 4QHO, 13c cyan sticks) and FTOΔ31-14a. Residues from FTOΔ31-14a are white sticks; the
substrate binding Glu234 side chain from FTOΔ31-13c is highlighted in purple sticks. Note: the nitro-phenyl group of 14a displaces the substrate binding Glu234 side chain. Non-carbon
atoms colored: O (red), N (blue), Zn (gray sphere), and H-bond, electrostatic
and metal coordination interactions are indicated as black dashes.
Table 1
IC50 Values of Compounds
against FTO and Selected Other Human 2OG Oxygenases
IC50 values (in μM)
obtained from the 5-mer m6A RNA LCMS-based demethylation
assay.
SPE-MS-based assay.[34]
A
histone demethylase luminescence-based
AlphaScreen assay.[35] For SPE-MS and AlphaScreen
assays, the data represent IC50 values plus lower and upper
95% confidence limits (in parentheses). Enzyme concentrations employed
are stated in parentheses on the first row. The high IC50 for 14c (IC50 5.1 μM) may in part
be due its relative lack of solubility at the tested concentrations.
Note: n.d., not determined.
Scheme 1
Synthesis of FTO inhibitors
Suzuki
or Buchwald–Hartwig
Pd-couplings were used to couple the C-6 position of the pyridinyl
ring with different aryl rings/aryl amines. Reagents and conditions:
(i) Br2 (1.1 equiv), H2O, 3 days; (ii) NaOH
30%, MeOH, reflux, 12 h; (iii) GlyOC(CH3)3 or
GlyOCH3, or GlyOCH2Ph (1.5 equiv), propylphosphonic
anhydride (1.5 equiv), diisopropylethylamine (4 equiv), microwave,
4 h; (iv) RB(OH)2 (1.3 equiv), K2CO3 (4 equiv), Pd(PPh3)4 (10 mol%), microwave,
130 °C; (v) ArNH2 (1.2 equiv), BrettPhos (0.15 equiv),
BrettPhosLigand G1 (0.01 equiv), K3PO4 (1.2
equiv); (vi) KOH (excess), THF; (vii) Pd/C (10 mol%), THF, r.t.
Structure-based FTO inhibitor design. (A) View of the active site
from a crystal structure of FTOΔ31 (white sticks) in complex
with 4 (magenta sticks) (PDB ID 4IE5),[26] which was used as a basis for inhibitor design, with the
structure of FTOΔ31-1 (rhein, yellow sticks) (PDB
ID 4IE7)[26] superimposed. The inset indicates how the derivatization
of 4 at C6 could enable inhibition by combining fragments
of 2OG and nucleobase mimetics. (B) View of the active site from a
structure of FTOΔ31 (white sticks) in complex with 13c (cyan sticks surrounded by OMIT mFo-DFc electron
density displayed as green mesh, contoured to 3.0σ). (C) Superimposition
of a structure of FTOΔ31-13c with that of FTOΔ31-4 (PDB ID 4IE5; 4 in magenta) and FTOΔ31-1 (PDB
ID 4IE7, 1 in yellow). (D) View of the active site of FTO in complex
with 14a. The FTOΔ31 active site (white sticks)
showing OMIT electron density (mFo-DFc, contoured
to 3.0σ in a green mesh) around 14a (salmon sticks).
(E) View of superimposed structures of FTOΔ31-13c (PDB ID 4QHO, 13c cyan sticks), FTOΔ31-14a, and
a FTO-ssDNA substrate (lime green sticks) complex (PDB ID: 5ZMD). Only FTO residues
(white sticks) from the FTOΔ31-14a structural complex
are shown. (F) Superimposed structures of FTOΔ31-13c (PDB ID 4QHO, 13c cyan sticks) and FTOΔ31-14a. Residues from FTOΔ31-14a are white sticks; the
substrate binding Glu234 side chain from FTOΔ31-13c is highlighted in purple sticks. Note: the nitro-phenyl group of 14a displaces the substrate binding Glu234 side chain. Non-carbon
atoms colored: O (red), N (blue), Zn (gray sphere), and H-bond, electrostatic
and metal coordination interactions are indicated as black dashes.
Synthesis of FTO inhibitors
Suzuki
or Buchwald–Hartwig
Pd-couplings were used to couple the C-6 position of the pyridinyl
ring with different aryl rings/aryl amines. Reagents and conditions:
(i) Br2 (1.1 equiv), H2O, 3 days; (ii) NaOH
30%, MeOH, reflux, 12 h; (iii) GlyOC(CH3)3 or
GlyOCH3, or GlyOCH2Ph (1.5 equiv), propylphosphonic
anhydride (1.5 equiv), diisopropylethylamine (4 equiv), microwave,
4 h; (iv) RB(OH)2 (1.3 equiv), K2CO3 (4 equiv), Pd(PPh3)4 (10 mol%), microwave,
130 °C; (v) ArNH2 (1.2 equiv), BrettPhos (0.15 equiv),
BrettPhosLigand G1 (0.01 equiv), K3PO4 (1.2
equiv); (vi) KOH (excess), THF; (vii) Pd/C (10 mol%), THF, r.t.IC50 values (in μM)
obtained from the 5-mer m6A RNA LCMS-based demethylation
assay.SPE-MS-based assay.[34]A
histone demethylase luminescence-based
AlphaScreen assay.[35] For SPE-MS and AlphaScreen
assays, the data represent IC50 values plus lower and upper
95% confidence limits (in parentheses). Enzyme concentrations employed
are stated in parentheses on the first row. The high IC50 for 14c (IC50 5.1 μM) may in part
be due its relative lack of solubility at the tested concentrations.
Note: n.d., not determined.We initially tested the in vitro inhibitory activity
of 13a–d against FTO (1.2 μM) using a reported
liquid chromatography (LC)-based assay employing the m3T nucleoside as a substrate.[26]13a (IC50 35 ± 1 μM), with a phenyl ring at C-6
was slightly less potent than 4 (IC50 15 μM)
against FTO. 13a and 13b (IC50 15 ± 1 μM) had potencies similar to that of 4 against PHD2 and FTO; however, 13a showed little or
no inhibition (IC50 > 100 μM) toward the 2OG-dependent
JmjC KDMs, KDM3A, KDM4C, and KDM6B and showed limited activity (IC50s 34–88 μM) toward KDM5A, B, C, and KDM2A (Tables and S2). The introduction of a biphenyl ring at C-6
of the pyridine ring (13b) resulted in moderate FTO inhibition
(IC50 15 ± 1 μM), while manifesting an increased
inhibition of the other tested 2OG oxygenases (PHD2, FIH, KDM3A, KDM4C,
KDM6B, KDM5A, KDM5B, KDM5C, KDM5D, and KDM2A) compared to 13a (Tables and S2). Substitution at C-6 of the pyridine with
1-naphthalene (13c) gave more potent inhibition (IC50 4 ± 1 μM), similar to that of FG-2216 (2) (IC50 2.8 ± 0.9 μM). 13c was, however, more selective for FTO, compared to 2 and the other compounds in this series over the JmjC KDMs. Quinoline
derivative 13d was prepared as a more basic and water-soluble
analogue of 13c but manifested a 6-fold decrease in potency
against FTO (13d, IC50 25 ± 1 μM) relative to 13c.To investigate the binding mode of 13c, we obtained
a structure of FTOΔ31 with Zn substituting for Fe in the complex
with it (2.6 Å resolution, PDB ID 4QHO). The structure was solved by molecular
replacement using an FTO-4 complex structure (PDB ID 4IE5) as a search model.
There was clear electron density for 13c in the active
site revealing that the hydroxypyridinyl glycyl scaffold manifests
a similar binding mode to that previously observed for 4 (PDB ID 4IE5),[26] with hydrogen bonding interactions
occurring between the 3-hydroxy group on the pyridine ring and the
Asn205 side chain and additional electrostatic and hydrogen bonding
interactions between the glycyl carboxylate and the side chains of
Arg316, Ser318, and Tyr295 (Figure B). As predicted, the 13c 1-naphthyl side
chain extends to occupy the substrate binding site (Figure B), with the plane of its naphthyl
group being near orthogonal relative to its pyridine ring. The 1-naphthyl
group of 13c appears to form partial π–π
interactions with the indole side chain of His231. The superimposition
of the structures of FTOΔ31 in complex with 4, 1, and 13c, show a substantial overlap, in particular
the naphthyl group of 13c is positioned similar to two
of the three anthraquinone rings of rhein (1) (Figure C).Because 13c showed enhanced potency and higher selectivity
compared to reported FTO inhibitors, we used the FTOΔ31-13c structure as a basis for the design of further inhibitors.
Analyses of the FTOΔ31-13c structure suggests that
the introduction of a heteroatom between the pyridinyl ring and the
C-6 aryl substituent of the original series would provide additional
side chain flexibility and potentially position the aryl ring between
the side chains of Tyr106 and Tyr108.Buchwald–Hartwig
coupling was thus used to prepare the aniline
derivatives (14a–i), which were tested against
FTO and 10 other 2OG oxygenases (Tables and S2). We initially
employed an optimized LCMS-based assay using a 5-mer m6A RNA oligonucleotide as the substrate and 1.2 μM FTO (Figure S3) to investigate the IC50s of 13a–d, 14a–d, 14g–i, and 15 (a reported FTO inhibitor).[21] The IC50s of 14a–d and g–i range from 1.5 to 60.2 μM, with
the majority displaying increased potencies over 13a–d (2.3–35.2 μM) (Table ). The most potent of the tested FTO inhibitors was 14a (IC50 1.5 μM), which has a p-nitrophenyl side chain (Table ). Against the panel of 2OG oxygenases, 14a–d and g–i showed generally greater selectivity
than 13a–d toward FTO. For the majority of JmjC
KDMs screened, IC50 values were in the tens of μM
to >100 μM. The selectivity of the 14a–d and g–i series for FTO over PHD2 was less pronounced
(IC50 range 2.2–59.8 μM) but with the most
potent FTO inhibitor, 14a, being ∼30-fold more
selective for FTO over PHD2 under the tested conditions.With
the LCMS-based assays, the IC50 measurement of 14a (1.5 μM) was limited as it was near the FTO concentration
(1.2 μM). We thus developed a more sensitive hydroxylation assay
employing a lower FTO concentration (100 nM) with a solid-phase extraction
coupled to MS (SPE-MS) assay (Figure S4, Table ). Instead
of demethylation, this assay monitors the hydroxylated product (hydroxymethyladenine,
hm6A) of a 15-mer m6A-containing RNA oligonucleotide
(hm6A has been reported to be a relatively stable adduct).[17,19] We determined the IC50s of 14a–i using
SPE-MS.The SPE-MS results show that the two compounds with p-nitrophenyl/p-nitropyridinyl groups at
C-6 of the
core pyridine, that is, 14a/14b, are the
most potent of the FTO inhibitors (IC50s 80 and 330 nM,
respectively), followed by the N-methyl amide-phenyl
(14h, IC50 430 nM), amine-pyridyl (14d, 800 nM), ester (14g, 900 nM), N-dimethyl
amide-phenyl (14i, 1.4 μM), and amine-phenyl (14c, IC50 = 5.1 μM) C-6 substituents. The
compounds with benzoic acid 14f and naphthylamine 14e side chains were the weakest FTO inhibitors (IC50s 15.2 and 26 μM, respectively). Even though the C-6 nitro-pyridine
analogue (14b) has a 4-fold higher IC50 than
the nitro-phenyl analogue 14a, the amine-pyridine substituent 14d has a lower IC50 than the amine-phenyl derivative 14c, thus pyridines at C-6 can be productively accommodated
within the active site.Changing the para-nitro group of the
C-6 phenyl ring of 14a to an amide, as in the monomethyl
amide 14h (IC50 430 nM) or dimethyl amide 14i (IC50 1.4 μM) increases the IC50 values by ∼5–18
fold over 14a, respectively. In general, substituting
the nitro group of 14a (amine: 14c and 14d; amide: 14h and 14i; carboxylate: 14f; and ester: 14g) resulted in a reduction
in potency.The results of testing 13a–d, 14a–i, and a previously reported selective
FTO inhibitor 15,[21] against
a panel of human 2OG oxygenases
(Tables and S2) revealed that, except for hydrazine compound 15 (FTO IC50 12.5 ± 0.7 μM; LCMS-based
assay), the hydroxy pyridine-glycyl compounds were for the most part
selective for FTO and PHD2. Interestingly, 15 was identified
as a potent inhibitor of the histone H3 lysine 36 JmjC KDM2A (IC50 63 nM), which is a potential cancer target.[36,37]To confirm that the compounds inhibit FTO by binding at the
active
site, we performed a competition-based NMR binding assay[38] (using 2OG as a competitor) to obtain binding
constants for the selected compounds (13c, 14a, 14h, and 14i) (Figures S5 and S6). The rankings of the KD values were generally in accord with the rankings of the IC50 values: 14a (KD 0.34 ± 0.16 μM, IC50 1.5 ± 0.5 μM), 13c (KD 1.2 ± 0.24 μM,
IC50 2.3 ± 0.5 μM), 14h (KD 0.67 ± 0.29 μM, IC50 6.2 ± 0.4 μM), and 14i (KD 3.9 ± 1.6 μM, IC50 54.5 ±
1.9 μM).We co-crystallized FTOΔ31 with 14a and determined
its structure (2.6 Å resolution). The structure revealed that
the hydroxypyridyl-glycine core of 14a binds, as observed
for 4, in the 2OG binding site (Figures D and S7). 14a coordinates to the metal (Zn substituting for Fe) via
its pyridinyl nitrogen and the carbonyl oxygen, its glycyl carboxylate
forms a salt bridge with the side chain of Arg316 and hydrogen bonds
with the side chains of Ser318 and Tyr295. In the FTOΔ31-14a complex, the 14a carbonyl oxygen is positioned trans to the metal-coordinating residue Asp233 and the metal-coordinating
pyridine nitrogen atom is trans to the metal coordinating
residue His307. The plane formed by the pyridinyl ring of 14a is skewed with respect to the plane formed by its pyridinyl nitrogen
and carbonyl oxygen and the active site metal (Figure S7A,B). This orientation is unusual, especially when
compared to the more typical co-planar orientation seen in the FTOΔ31-4 structure (Figure S7B).To investigate the generality of the skewed binding mode of 14a as observed with FTO, we solved a crystal structure of
a truncated version of the related bacterial NAOX AlkB (AlkBΔN11)
in complex with it (Figure A). By contrast with the crystallographically observed FTO
binding mode of 14a, in the AlkB-14a structure,
the metal-coordinating atoms of 14a and the metal-ion
are co-planar with the pyridinyl ring (Figure B). Contrary to the binding mode observed
with FTO, the p-nitrophenyl group of 14a is directed way from the center of the AlkB active site without
significantly displacing key substrate binding side chains (Figure A,B). The binding
mode of 14a in AlkB is different from the previously
reported potent AlkB glycinamide inhibitors where their side chains
are typically positioned in a hydrophobic pocket close to Trp178,
Ile143, and Phe154 (Figure S8).[31]
Figure 3
Structure-based rationalization of selectivity of FTO
inhibitors.
(A) View of the active site of the AlkBΔN11-14a complex (PDB ID 7NRO); 14a (green sticks) with surrounding OMIT electron
density mFo-DFc shown as a green mesh and contoured
to 3.0σ. (B) View of the FTOΔ31-14a structural
complex (FTO in white sticks) (PDB ID 7E8Z) superimposed on a structure of AlkBΔN11-14a. Note: 14a (salmon sticks, FTO; green sticks,
AlkB) binds slightly differently to FTO and AlKB in particular, the
metal-coordinating atoms of 14a and the metal ion are
co-planar with AlkB. (C) View of the active site from a structure
of truncated PHD2 (yellow sticks and surface) superimposed with a
structure of FTOΔ31 in complex with 13c (PDB ID 4QHO) (derived from a
structure of the PHD2-2 complex PDB ID 4BQX,[39] ligand not shown for clarity) indicating how the naphthyl
side chain of 13c may clash with Tyr310 in the PHD2 active
site, so reducing is affinity (FTOΔ31 residues are not shown
for clarity). (D) View of the active sites of structures of FTOΔ31
and AlkBΔN11 in complex with 14a (FTO, salmon sticks;
AlkB, green sticks) (PDB ID 7E8Z and 7NRO) superimposed onto PHD2 (yellow sticks and surface). Note the nitrophenyl
side chain of 14a is predicted to occupy the PHD2 active
site opening with a potential for minor clashes, notably for the crystallographically
observed AlkB 14a binding mode (green sticks). (E) View
from a structure of FTO in complex with 13c superimposed
onto a structure of KDM4C in complex with 5 (PDB ID 2XML)[40] (KDM4 C shown as light blue sticks and surface; 5 in gray sticks) showing that the side chain of 13c may
clash with Lys243 of KDM4C. (F) View from complexes of FTOΔ31-14a (salmon sticks) and AlkBΔN11-14a (green
sticks) superimposed on the structural complex of KDM4C-5 (PDB ID 2XML)[40] showing how the C-6 napthyl side chain
of 14a may clash with Lys243 and/or Asp193 of KDM4C.
Non-carbon atoms colored: O (red), N (blue), Zn (gray sphere), Mn
(purple sphere), and Ni (light green); H-bond, electrostatic, and
metal coordination interactions are indicated as black dashes.
Structure-based rationalization of selectivity of FTO
inhibitors.
(A) View of the active site of the AlkBΔN11-14a complex (PDB ID 7NRO); 14a (green sticks) with surrounding OMIT electron
density mFo-DFc shown as a green mesh and contoured
to 3.0σ. (B) View of the FTOΔ31-14a structural
complex (FTO in white sticks) (PDB ID 7E8Z) superimposed on a structure of AlkBΔN11-14a. Note: 14a (salmon sticks, FTO; green sticks,
AlkB) binds slightly differently to FTO and AlKB in particular, the
metal-coordinating atoms of 14a and the metal ion are
co-planar with AlkB. (C) View of the active site from a structure
of truncated PHD2 (yellow sticks and surface) superimposed with a
structure of FTOΔ31 in complex with 13c (PDB ID 4QHO) (derived from a
structure of the PHD2-2 complex PDB ID 4BQX,[39] ligand not shown for clarity) indicating how the naphthyl
side chain of 13c may clash with Tyr310 in the PHD2 active
site, so reducing is affinity (FTOΔ31 residues are not shown
for clarity). (D) View of the active sites of structures of FTOΔ31
and AlkBΔN11 in complex with 14a (FTO, salmon sticks;
AlkB, green sticks) (PDB ID 7E8Z and 7NRO) superimposed onto PHD2 (yellow sticks and surface). Note the nitrophenyl
side chain of 14a is predicted to occupy the PHD2 active
site opening with a potential for minor clashes, notably for the crystallographically
observed AlkB 14a binding mode (green sticks). (E) View
from a structure of FTO in complex with 13c superimposed
onto a structure of KDM4C in complex with 5 (PDB ID 2XML)[40] (KDM4 C shown as light blue sticks and surface; 5 in gray sticks) showing that the side chain of 13c may
clash with Lys243 of KDM4C. (F) View from complexes of FTOΔ31-14a (salmon sticks) and AlkBΔN11-14a (green
sticks) superimposed on the structural complex of KDM4C-5 (PDB ID 2XML)[40] showing how the C-6 napthyl side chain
of 14a may clash with Lys243 and/or Asp193 of KDM4C.
Non-carbon atoms colored: O (red), N (blue), Zn (gray sphere), Mn
(purple sphere), and Ni (light green); H-bond, electrostatic, and
metal coordination interactions are indicated as black dashes.As predicted, the amine linker of 14a apparently enables
its p-nitrophenyl group to have a different orientation
in FTO compared to the orientation of the naphthalene of 13c (Figure F). With 14a, the nitrogen of the amine linker is positioned to potentially
form a hydrogen bond with the Arg322 side chain (2.9 Å); the
Arg322 side chain is also positioned to form a cation−π
interaction with the pyridine ring of 14a. The p-nitrophenyl group of 14a is positioned between
the Glu234 and Tyr108 side chains of FTO. Notably, the Glu234 side
chain is displaced ∼5 Å by the p-nitrophenyl
moiety toward the surface of the protein (Figures F and S9), an
unprecedented observation in available FTO inhibitor structures (Figure ).[7,20,21,23−26,41] Glu234 is important as it has
also been shown to interact with m6A in a structure of
an FTO variant in complex with a ssDNA substrate.[42] Analysis of the FTO variant:ssDNA structural complex suggests
that electrostatic repulsion between the negatively charged Glu234
side chain and the phenyl carboxylate of 14f (IC50 15.2 μM) may account for the large (190-fold) difference
in IC50 between the isosteres 14a and 14f. Removing the inhibitor carboxylate group by substitution
with a methyl ester 14g (IC50 860 nM) appeared
to partially rescue potency resulting in a 10-fold difference relative
to 14a. The structural analysis implies that bulky naphthalene
of 14e would encounter steric hindrance in the pocket
of FTO where the p-nitrophenyl group of 14a is situated, explaining its significant 325-fold lower potency (IC50 26 μM).
Discussion and Conclusions
Our structure-guided
approach to FTO inhibition was based on the
knowledge that although Fe(II) and 2OG binding by 2OG oxygenases occurs
by a conserved general process, that of substrate binding varies considerably,
not least because the nature of the substrates varies for different
sub-families of human 2OG oxygenases.[12,15] Following
analysis of structures of FTO in complex with a 2OG mimic (4) and rhein (1), which binds competitively with the
substrate nucleobase (Figure S2), this
approach resulted in compounds (e.g., 13b and c) which were more potent than 4. Following crystallographic
studies of FTO with 13c, derivatives of 4 were further optimized to highly potent FTO inhibitors such as 14a, the proposed general binding mode of which was also validated
by crystallographic analyses. The 2OG-competing inhibition mode of
these inhibitors was supported by NMR studies.Importantly,
although there is likely scope for further optimization
using a larger set of 2OG oxygenases in counter screening, some of
the compounds we developed show selectivity for FTO (Tables and S2), which in some cases can be rationalized on the basis of structural
information. For instance, the set of compounds with an aryl ring
directly linked to C-6 of the pyridine ring of the scaffold (13a–d) are generally more selective for FTO over PHD2,
compared to 14a–i where the aryl ring is linked
to the pyridine C-6 via an amine. The superimposition of our FTOΔ31-13c structural complex with that of PHD2-2[39] reveals that the C-6 aryl ring of 13c will likely clash with the PHD2 Tyr310 side chain (Figure C), thus rationalizing in part
why this set of FTO inhibitors is selective toward FTO over PHD2.
In contrast, the superimposition of the FTOΔ31-14a and AlkB-14a structures with the PHD2-2 structure reveals that the nitrophenyl group of 14a will likely project into an unoccupied space at the active site
entrance of PHD2 (Figure D). The lack of steric hindrance for the C-6 aryl ring of 14a (and by implication) when modeled in PHD2 is thus proposed
to lower its selectivity toward FTO over PHD2. The superimposition
of the structural complexes of FTOΔ31-13c, FTOΔ31-14a, and AlkBΔN11-14a with KDM4C-5 as a representative of JmjC oxygenases suggests that similar
steric clashes of 14a proposed for PHD2 may also occur
in the KDM4C active site (Figure E,F). The superimposed structures also reveal that
the hydroxypyridinyl compounds adopt a different metal-ion binding
mode as 5 does with KDM4C, potentially providing a structural
explanation for the selectivity of the C-6 substituted hydroxypyridine
compounds for FTO over the JmjC sub-family of 2OG oxygenases.It should, however, be noted that apparently clear crystallographically
based structural explanations for the selectivity of inhibitors for
particular 2OG oxygenases do not always manifest in experimental studies,
as shown by the results with PHD inhibitors[43−46] and some of our results for JmjC
KDM inhibition (Tables and S2). In this regard, it is important
to note that, both in terms of 2OG and substrate binding by 2OG oxygenases,
there is substantial evidence for conformational changes, as exemplified
with PHD2, where such changes help enable a mechanism employing an
ordered sequential active site binding of 2OG, substrate, and then
dioxygen.[43,44] Such conformational changes, at least in
some cases, are observed on inhibitor binding, as observed with the
PHDs.[45,46] Further, with the JmjC KDMs, in some cases
there is evidence for the inhibitor-induced Fe(II) movement.[30] In the case of FTO, our structure with 14a, provides evidence that the side chain of Glu234, which
is likely important in catalysis, is susceptible to conformational
changes (Figures F
and S8). In the case of the structure of
the complex of FTOΔ31-14a, a skewed inhibitor binding
mode was observed; the skewed binding was less apparent in a structure
of the bacterial NAOX AlkBΔN11 in complex with 14a. Although further detailed work is required to define the precise
binding mode of 14a to FTO in solution (perhaps employing
NMR), our observations illustrate why the crystallographic information
should be used with care and highlight the importance of solution
and empirical inhibition studies.Nonetheless, the combined
results illustrate how the knowledge
of the 2OG cosubstrate and prime substrate binding modes as revealed
by high-resolution crystallographic analyses can be combined to develop
potent inhibitors of FTO, which is a human 2OG oxygenase of current
therapeutic interest. Future work can now be directed toward the optimization
of the compounds described here for in vivo studies
aimed at defining how the reactions catalyzed by FTO are linked to
its roles in physiology and disease.
Experimental
Section
Recombinant FTO and FTOΔ31 Production and Purification
The production and purification of FTO and FTOΔ31 were carried
out as reported.[26]Escherichia
coli BL21 (DE3) cells (or E. coli Rosetta (DE3) pLysS cells for FTOΔ31-14a) were
transformed with the pET28a_FTO plasmid (encoding for N-terminally
hexahistidine-tagged full-length human FTO without the thrombin cleavage
site between the hexahistidine tag and the protein) or pET28a_FTO
FTOΔ31 (encoding for N-terminally hexahistidine-tagged human
FTO with residues 2–31 deleted from the N-terminus) and were
grown (37 °C; 180 rpm) to an OD600 of 0.6–0.8.
FTO production was induced by the addition of a final concentration
of 0.5 mM isopropyl β-d-1-thiogalactopyranoside (IPTG);
growth was continued at 16 °C for 16 h (FTO) or 18 °C for
8 h (FTOΔ31). The resultant cell pellets were stored at −80
°C. The cell pellets were thawed and resuspended in 20 mM Tris,
pH 7.5; 500 mM NaCl; 10 mM imidazole, and 1 mg of DNaseI, 10 mM MgCl2, and a Roche cOmplete protease inhibitor tablet (or 17.8
μg/mL phenylmethylsulfonyl fluoride for FTOΔ31-14a) then lysed by sonication on ice. The lysates were cleared by centrifugation,
and the supernatant was loaded onto a 5 mL HisTrap column (Cytiva)
and purified using an AKTA FPLC system. The column was treated with
20 mM Tris, pH 7.5; 500 mM NaCl; and 40 mM imidazole, then eluted
with 20 mM Tris, pH 7.5; 500 mM NaCl; and 500 mM imidazole. The FTO
solution was then buffer-exchanged into 25 mM Tris, pH 7.5, and concentrated
to 20–40 mg/mL for storage. For crystallographic studies, the
eluted FTOΔ31 from the HisTrap column was treated with EDTA
(final concentration 200 mM) and incubated on ice for 30 min. The
FTOΔ31 solution was desalted using a PD-10 desalting column
with buffer A (25 mM Tris, pH 7.5), then further purified by a 5 mL
HiTrap Heparin column (Cytiva), followed by a MonoQ column (Cytiva);
both columns were equilibrated with buffer A (25 mM Tris–HCl,
pH7.5) and eluted with a gradient from 0 to 100% buffer B (25 mM Tris,
pH7.5; 1 M NaCl) for heparin chromatography and 0 to 60% buffer B
for MonoQ ion exchange chromatography. The purified protein was then
buffer-exchanged into buffer A and concentrated to 20 mg/mL for FTOΔ31-13c and 36 mg/mL for FTOΔ31-14a. The protein
was snap frozen in liquid nitrogen and stored at −80 °C.
Purification of the FTOΔ31-14a Complex
A final concentration of 8 mg/mL of FTOΔ31 was mixed with
a final concentration of 1 mM ZnSO4 and a final concentration
of 1 mM 14a in 25 mM Tris, pH 7.5, then incubated on
ice for 30 min. The solution was loaded onto a Superdex 200 increase
10/300 GL column (Cytiva), pre-equilibrated with 25 mM Tris, pH 7.5,
for purification using size exclusion chromatography. Relevant fractions
corresponding to the FTOΔ31-14a complex
were pooled and concentrated to 10 mg/mL.
AlkBΔN11 Expression
and Purification for Crystallographic
Studies
AlkBΔN11 was produced and purified as described.[31] In brief, E. coli BL21 (DE3) cells were transformed with pET24a_AlkBΔN11 and
allowed to grow at 37 °C until the OD600 reached 0.6
at which point AlkBΔN11 production was induced with a final
concentration of 0.5 mM IPTG. The cells were incubated at 18 °C,
180 rpm overnight (ca. 18 h). Cells pellets were thawed and resuspended
to homogeneity in 0.1 M MES pH 5.8 containing 1 mM MgCl2 with 1 mg of DNaseI and Roche cOmplete EDTA-free protease inhibitor
cocktail. Cells were lysed by sonication on ice and the supernatant
was obtained by centrifugation. The cell lysates were purified by
cation exchange chromatography using a 50 mL S Sepharose column, with
elution achieved by application of a gradient to 0.1 M MES buffer,
1 M NaCl, pH 5.8. The eluted protein was further purified using a
300 mL Superdex 75 column (Cytiva), pre-equilibrated in 50 mM HEPES,
pH 7.5.
In Vitro Demethylation/Hydroxylation Assays
for FTO
LC-Based Assays Using the m3T Nucleoside
LC-based in vitro m3T demethylation assays
were performed as reported.[26] For initial
assays using LC with the m3T nucleoside substrate, a 50
μL of reaction mixture containing final concentrations of 3
μM FTO; 70 μM m3T nucleoside; 160 μM
2OG; 500 μM l-ascorbate; 100 μM diammonium iron(II)
sulfate complex; 200 μM compound (or DMSO alone in control reactions);
and 50 mM 2-(N-morpholino)ethanesulfonic acid (MES),
pH 6.3, was incubated at room temperature for 1 h. Reactions were
quenched with methanol (50 μL), then centrifuged to remove the
precipitated protein. The supernatant was then dried using an Eppendorf
Speedvac concentrator and reconstituted with (50 μL) water.
The product (thymidine) and substrate (m3T) were separated
using a Waters Acquity Ultra Performance LC (UPLC) BEH C18 column
(130 Å, 1.7 μm, 2.1 mm × 50 mm) with a gradient of
95% A (H2O with 1% formic acid) to 80% B (methanol with
0.1% formic acid) over 5 min. The UV detection wavelength was set
to 266 nm. Elution was performed at room temperature; thymidine =
0.95 min, m3T = 1.48 min. The peaks were identified in
the positive ion mode using electrospray ionization-time-of-flight
(ESI-TOF) mass spectrometry (Waters LCT Premier XE machine). UV peaks
were integrated using MassLynx software (Agilent), and the percentage
conversion of m3T to thymidine was used to quantify the
activity of FTO in the presence of inhibitors. For IC50 determinations, assays were carried out in triplicate with a range
of compound concentrations (0, 10, 30, 100, and 300 μM and 1,
3, and 10 mM). IC50 values were calculated from dose–response
curves plotted using GraphPad Prism 5.
LCMS-Based Assay Using
the RNA m6A 5-Mer Oligonucleotide
For assays using
LC mass spectrometry (LC-MS) with a 5-mer ssRNA
oligonucleotide substrate [GG(m6A)CU] (Ella Biotech, Germany),
a 25 μL of reaction mixture containing final concentrations
of 1.2 μM FTO; 10 μM 5-mer ssRNA oligonucleotide [GG(m6A)CU]; 10 μM 2OG; 500 μM l-ascorbic acid;
10 μM diammonium iron(II) sulfate; and 25 mM Tris-HCl, pH 7.5,
was incubated at room temperature for 30 min. After incubation, the
reaction mixture was quenched with methanol (25 μL), then centrifuged
(14,000 rpm, 10 min) in Seahorse polypropylene 96-well filter microplates
with a 30 KDa molecular weight cutoff polyethersulfone membrane to
remove FTO. The product [GGACU] and substrate [GG(m6A)CU]
were separated using a Waters Acquity UPLC Oligonucleotide BEH C-18
column (130 Å, 1.7 μm, 2.1 mm × 50 mm) with a gradient
of 98% buffer A to 70% buffer B over 8 min, at room temperature. 1,1,1,3,3,3-Hexafluoro-2-propanol
(HFIP) and triethylamine were used as an ion-pairing buffer for better
separation of the 5-mer oligonucleotide substrate [GGm6ACU] and the demethylated product [GGACU]. Buffer A: 200mM HFIP,
8.15 mM TEA buffer, 5% methanol, buffer B: 20% buffer A + 80% methanol.
Separation of oligonucleotides was monitored by UV; the detection
wavelength was 265–292 nm. Retention times for [GGACU] and
[GG(m6A)CU] were 2.5 and 3.0 min, respectively. The masses
for each peak were confirmed by ESI-TOF mass spectrometry (Bruker
Daltonics, microTOF). The area under for the substrate ([GG(m6A)CU]) and product ([GGACU]) peaks were integrated using data
analysis software (Bruker) in the negative ion mode, and the percentage
conversion of [GG(m6A)CU] to [GGACU] was used to quantify
the demethylation activity of FTO in the presence of inhibitors. For
IC50 determinations, assays were carried out in triplicate
with a range of compound concentrations (0, 10, 30, 100, and 300 μM
and 1, 3, and 10 mM). IC50 values were calculated from
the dose–response curves plotted using GraphPad Prism 5.
Solid-Phase Extraction Mass Spectrometry-Based Assays
For
assays using solid-phase extraction mass spectrometry (SPE-MS)
with a 15-mer ssRNA oligonucleotide substrate [AUUGUGG(m6A)CUGCAGC], a 25 μL of reaction mixture was added 200 μM
ascorbic acid; 1 μM 2OG; 0.5 μM (NH4)2Fe(SO4)2·6H2O; 25 mM Tris,
pH 7.5; 0.6 μM 15-mer ssRNA substrate; and 0.1 μM FTO,
with a range of compound concentrations (3, 10, 30, 100, and 300 nM,
1, 3, 10, 30, and 100 μM). Reactions were incubated at 25 °C
for 10 min before quenching with 10 mM MgCl2. A RapidFire
RF 365 high-throughput sampling robot (Agilent) attached to an iFunnel
Agilent 6550 accurate mass quadrupole time-of-flight (Q-TOF) mass
spectrometer operated in the positive ionization mode was used for
sample analyses. The samples were aspirated under vacuum for 0.6 s
and loaded onto a C-8 reverse phase solid-phase extraction (SPE) cartridge.
The cartridge was washed with 6 mM octylammonium acetate (prepared
by mixing 0.1 M of octylamine and 0.1 M of acetic acid with 100 mL
diethyl ether on ice; this solution was cooled to −40 °C
resulting in OAA crystals, which were collected and washed with cold
cyclohexane.) in the LC-MS grade water treated with diethyl pyrocarbonate
(5.5 s, 1.5 mL/min). The hydroxylated product [AUUGUGG(hm6A)CUGCAGC] and the substrate [AUUGUGG(m6A)CUGCAGC] were
eluted from the SPE cartridge with 6 mM OAA 80/20v/v acetonitrile/water
into a mass spectrometer (5.5 s, 1.5 mL/min). The SPE cartridge re-equilibrated
with 6 mM OAA in water (0.5 s, 1.5 mL/min). Mass spectrometer parameters:
capillary voltage (4000 V), nozzle voltage (0 V), fragmentor voltage
(250 V), gas temperature (280 °C), gas flow (13 L/min), nebulizer
(40 psi), sheath gas temperature (350 °C), and sheath gas flow
(12 L/min). Signal intensities were quantified as the total ion count
and analyzed using RapidFire Integrator software (Agilent). Percentage
hydroxylation was calculated by comparing the ion intensities of m/z peaks corresponding to [M –
H]3– to [M + OH]3–: % hydroxylation
= 100 × (product)/(substrate + product). IC50 values
were calculated from the dose–response curves plotted using
GraphPad Prism 5.
PHD2 and FIH Inhibition Assays
PHD2,
FIH, and KDM inhibition
assays were performed using a solid-phase extraction–mass spectrometry-based
methods as described.[34] KDM inhibition
assays were performed using an AlphaScreen-based assay as described.[35]
NMR-Based Binding Studies
Apo-FTO
was made by treatment
with EDTA as follows: fractions containing purified FTO at a concentration
of >1 mg/mL in 200 mM EDTA and 15 mM ammonium acetate (pH 7.0 to
8.0)
was incubated overnight at 4 °C, then concentrated to 2 mL volume
using an Amicon Ultra-4 or Amicon Ultra-15 centrifugal filtration
units. Carr–Purcell–Meiboom–Gill (CPMG) displacement
experiments were carried out using an AVIII 700 Bruker instrument
equipped with a TCI inverse cryoprobe using 3 mm diameter MATCH NMR
tubes with 160 μL of sample volume at 298 K. The PROJECT-CPMG
sequence of Aguilar et al. was applied (90°–[τ–180°–τ–90°–τ–180°–τ]–acq)[47] using: total echo time
48 ms (τ = 2 ms, n = 6); acquisition time,
2.94 s; relaxation delay, 2 s; and number of scans 200–400.[38] Water suppression was achieved by presaturation.
Samples containing 10 or 20 μM of apo-FTO, 100 μM of Zn(II),
10 or 20 μM of 2OG, and 20–400 μM of the inhibitor
buffered in 50 mM Tris-D11/HCl (pH 7.5) and 0.02% NaN3 (w/v) in 90% H2O and 10% D2O (v/v)
were analyzed, and the 2OG displacement was measured via the reappearance
of free 2OG signals.The percentage of the 2OG displacement
was calculated using:where, I2OG =
integral of 2OG in the presence of both protein (FTO) and inhibitor; I2OG(0) = integral of 2OG in the presence of
the protein (FTO) and absence of the inhibitor; I2OG(blank) = integral of 2OG in the absence of both protein
(FTO) and inhibitor; and the apparent K (KDapp) of
the inhibitors (13a–d, 14a–i) were calculated using reported methods.[38] The KD of 2OG for FTO (1.32 ± 0.5
μM) was used as a reference value.
RNA Oligonucleotide Synthesis
and Purification
RNA
synthesis was performed using an Applied Biosystems 394 automated
DNA/RNA synthesizer on the 1.0 μmole scale using a standard
phosphoramidite cycle of detritylation, coupling, capping, and oxidation.
Solid supports were packed into a TWIST column (Glen Research) for
synthesis.For non-modified RNA, 2′-O-TC-protected RNA phosphoramidite monomers were used (A-bz, C-ac,
G-ib and U, Sigma-Aldrich). Monomers were dissolved at 0.1 M in anhydrous
toluene/acetonitrile (1:1 v/v) immediately prior to use. Coupling,
capping, and oxidation reagents were identical to those used in standard
DNA synthesis except a solution of ethylthiotetrazole (ETT) (0.25
M in acetonitrile, Link Technologies) was used as a coupling reagent
(activator). The coupling time for all monomers during RNA synthesis
was 3 min. Stepwise coupling efficiencies were determined by the automated
trityl cation conductivity measurement and in all cases were >97%.
For the cleavage of the RNA from the solid support and deprotection
of the bases and sugar, the solid support was exposed to dry ethylenediamine:toluene
(1:1 v/v) for 6 h at room temperature, washed with toluene (3 ×
1 mL) then acetonitrile (3 × 2 mL), and dried using argon. The
crude cleaved RNA was eluted with water.For N6-methyladenosine RNA (5-mer and
15-mer), 2′-O-TBDMS RNA phosphoramidites
(A-tac, C-tac, G-tac, and U, where tac = tert-butylphenoxyacetyl;
Sigma-Aldrich) were dissolved in anhydrous acetonitrile (0.1 M) immediately
prior to use. Coupling, capping, and oxidation reagents were 5-benzylthio-1H-tetrazole (0.3 M in acetonitrile; Link Technologies),
fast deprotection Cap A (5% tert-butyl phenoxyacetyl
acetic anhydride in tetra-hydrofuran)/Cap B (16% N-methylimidazole in tetrahydrofuran), and iodine (0.1 M in tetrahydrofuran,
pyridine, and water), respectively. The coupling time during RNA synthesis
was 10 min. Stepwise coupling efficiencies were determined by automated
trityl cation conductivity monitoring and in all cases were >97%.For the cleavage of the RNA from the solid support and deprotection
of the bases, the solid support was exposed to concentrated aqueous
ammonia:ethanol (3:1 v/v) for 2.5 h at room temperature followed by
2 h at 55 °C in a sealed vial. Ammonia was removed in vacuo.
To remove the 2′-TBDMS protecting group, the ammonia-free solution
was freeze-dried and re-dissolved in a 1:1 mixture of dry DMSO (300
μL) and triethylaminetrihydrofluoride (300 μL) and heated
for 2.5 h at 65 °C. After cooling down to room temperature, sodium
acetate (3 M pH 5.2, 50 μL) and butanol (3 mL) were added, and
the RNA was stored for 30 min at −80 °C. RNA was then
pelleted by centrifugation (12,000g, 30 min, 4 °C),
the supernatant discarded, and the pellet washed twice with 70% ethanol
(750 μL). The pellet was then dried in vacuo and dissolved in
water.Oligonucleotides were purified using a Gilson HPLC system
with
an ACE C8 reversed-phase column (10 mm × 250 mm, pore size 100
Å, particle size 10 μm) with a gradient of buffer A (0.1
M TEAB, pH 7.5) to buffer B (0.1 M TEAB, pH 7.5 containing 50% v/v
acetonitrile) and a flow rate of 4 mL/min. The gradient of acetonitrile
in triethylammonium bicarbonate (TEAB) was increased from 0 to 50%
buffer B over 30 min. Elution was monitored by ultraviolet absorption
at 298 nm. After HPLC purification, the RNA was freeze-dried and then
dissolved in water. Purified oligonucleotides were characterized by
ESI mass spectrometry using a XEVO G2-QTOF MS instrument. Data were
processed using MaxEnt and in all cases confirmed the integrity of
the sequences.
Protein Crystallization
Crystals
of the FTOΔ31-13c and FTOΔ31-14a complexes were grown
in hanging drops at 293 K by the vapor diffusion method. For FTOΔ31-13c crystallization, 2 μL of the protein solution containing
8 mg/mL FTOΔ31 solution, 1 mM ZnSO4, and 1 mM 13c was mixed with 1 μL of the reservoir solution containing
100 mM trisodium citrate pH 5.6, 8% PEG 3350, and 4% tert-butanol to create a crystallization drop. For FTOΔ31-14a crystallization, 1 μL of solution containing 8 mg/mL
of the FTOΔ31-14a complex and 1 mM NADH was mixed
with 1 μL of reservoir solution containing 100 mM trisodium
citrate pH 5.6, 13.5% PEG 3350, 4% tert-butanol to
create a crystallization drop. The drops were equilibrated with 200
μL of the reservoir solution. The resultant crystals were cryoprotected
by soaking crystals in the reservoir solution diluted with 20% (v/v)
glycerol before being flash cooled in liquid nitrogen.Crystals
of the AlkBΔN11-14a complex were grown using the
sitting drop method at 293 K. The volume ratio of the protein to reservoir
was 1:2. 1 μL of the protein solution containing 20 mg/mL AlkBΔN11,
2 mM MnCl2, and 2mM of 14a were mixed with
2 μL of the reservoir solution containing 0.1 M NaCl, 21% w/v
PEG 3350, 0.1 M HEPES to create a crystallization drop. The resultant
crystals were cryoprotected by soaking crystals in the reservoir solution
diluted with 20% (v/v) glycerol before being flash cooled in liquid
nitrogen.
X-ray Data Collection, Structure Determination,
and Refinement
Data were collected at Diamond Light Source
beamlines I04 or I24
equipped with a Pilatus 6M-F (FTOΔ31-13c and AlkBΔN11-14a) or Dectris EIGER2 9M (FTOΔ31-14a)
detector, respectively. Data for FTOΔ31-13c were
processed in-house using HKL2000[48] and
data for FTOΔ31-14a and AlkBΔN11-14a were processed by the beamlines autoprocessing pipeline using the
XIA2 strategy.[49] FTO structures were solved
by molecular replacement (MR) using PHASER[50] with a structure of FTOΔ31-4 as the search model
(PDB ID 4IE5), followed by iterative cycles of map fitting in COOT[53] and refinement using PHENIX.[51] The structure of AlkBΔN11-14a was solved
by MR using PHASER and PDB ID 2FDJ as the search model. Iterative cycles
of model building using COOT and refinement with PHENIX were performed
until the converging Rwork and Rfree no longer decreased. 14a in
the FTOΔ31-14a structure was refined with an occupancy
of 0.7. Residual positive difference densities near the active site
were present, likely derived from low occupancy citrate and/or glycerol
from the crystallization mother liquor and cryoprotectant and could
not be modeled with confidence. Data collection and refinement statistics
for all structures can be found in Table S1.
Synthetic Chemistry Methods
Reagents were from Sigma-Aldrich,
Alfa Aesar, Cambridge Biotech, Fischer Scientific, or Link Technology,
unless otherwise stated. Anhydrous solvents used in reactions were
either analytical grade, as obtained commercially (Alfa Aesar), or
were freshly distilled. HPLC grade solvents were employed for work-up
and chromatography. For the chromatographic purification of phosphoramidites,
solvents were dried over P2O5 prior to use.
Reactions involving moisture-sensitive reagents were carried out under
an argon atmosphere; glassware was oven-dried and cooled under nitrogen
before use. Reagents were used as supplied (analytical or HPLC grade)
without prior purification. Anhydrous MgSO4 was used as
a drying agent. Microwave experiments were carried out using a Biotage
Initiator 8 machine. Thin-layer chromatography was performed using
aluminum plates coated with 60 F254 silica. Plates were visualized
using UV light (254 nm), or 1% (m/v) aq KMnO4 stain. Flash column chromatography was performed
using Kieselgel 60 silica in a glass column, or on a Biotage SP4 flash
column chromatography platform. Retention factors (Rf) are quoted to a precision of 0.05.Deuterated
solvents were from Sigma and Apollo Scientific Ltd. 1H
NMR and 13C NMR spectra were recorded using Bruker AVIII400,
AVIII500, AVIII600, and AVIII700 NMR spectrometers. Fields were locked
by external referencing to the relevant residual deuterium resonance.
Chemical shifts (δ) are reported in ppm; coupling constants
(J) are recorded in Hz to the nearest 0.5 Hz; when
peak multiplicities are reported, the following abbreviations are
used: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet,
br = broadened, dd = doublet of doublets, dt = doublet of triplets,
and td = triplet of doublets. Spectra were recorded at room temperature
unless otherwise stated.Low-resolution mass spectra (m/z) and high-resolution mass spectra (HRMS)
were recorded using an
LCT Premier XE (Waters, MA, USA) or a microTOF (Bruker, MA, USA).Melting points were recorded on a Gallenkamp Hot Stage apparatus.
Purity of synthesized compounds was ≥95% as determined by analytical
reverse-phase LC/MS. IR spectra were recorded using a Bruker Tensor
27 FT-IR spectrometer as thin films. The selected characteristic peaks
are reported in cm–1.
General Synthetic Procedures
General
Procedure A: Synthesis of Protected Glycinate Derivatives
(Suzuki Coupling) (Scheme S1)
To an overnight dried microwave vial were added the requisite aryl
halide (8, 9; 1 equiv), K3PO4 (3
equiv), and the requisite boronic acid (1.2 equiv) in CH3CN/H2O (3:1). The vial was evacuated and filled with N2 three times. To this degassed solid mixture was added Pd(PPh3)4 (10 mol %); the vial was again evacuated and
filled with N2 three times. The vial was then subjected
to microwave radiation (Biotage Initiator 8 machine, 135 °C,
3 h), pH was raised with 1 M NaOH. The mixture was diluted with ethyl
acetate and washed with 3 × H2O. The combined organic
extracts were dried over anhydrous MgSO4 and concentrated
in vacuo. The crude mixture was chromatographed (cyclohexane/ethyl
acetate 8:2 to 7:3) to provide the product.
General Procedure B: Synthesis
of Protected Glycinate Derivatives
(Buchwald Coupling) (Scheme S2)
To an overnight-dried microwave vial were added the appropriate aryl
halide (8–10; 1 equiv), K3PO4 (3 equiv), BrettPhos (16; 0.14 equiv), and Pd precatalyst
(17; 0.01 equiv). The vial was evacuated and filled with
N2 three times. To this degassed solid mixture were added
appropriate amine (1.2 equiv) and t-BuOH (2.5 mL).
The vial was again evacuated and filled with N2 three times.
The vial was then subjected to microwave heating (Biotage Initiator
8 machine, 8 h, 140 °C). The solvent was removed under reduced
pressure and the residue was diluted with ethyl acetate. The organic
layer was washed with 3 × H2O and crystallized from
CH2Cl2. Yields varied from 22-86%.
General
Procedure C: Benzyl Group Deprotection (Scheme S3)
To a stirred solution of
the appropriate compound (12c–d;
0.1 mmol) in THF (2 mL) was added Pd/C (10 mol %). The reaction mixture
was evacuated and filled with N2 three times. The reaction
mixture was stirred for 24 h under a H2 atmosphere. The
suspension was filtered through celite pad; the filtrate was concentrated
under reduced pressure. The oily residue was washed with methanol
to afford the desired acid.
General Procedure D: tert-Butyl Group Deprotection
(Scheme S4)
To a stirred solution
of compound (11a–d, 12a–b, 12e–i, 0.1
mmol) in THF was added excess KOH; the reaction was stirred for 2
days at room temperature. The crude product mixture was concentrated
under reduced pressure, then diluted with ethyl acetate and water.
The aqueous layer was acidified with 1 M HCl to pH 3. The resulting
precipitate was filtered and washed with cold water (5 mL).
6-Bromo-3-hydroxypicolinonitrile
(6)
To
a suspension of 3-hydroxypicolinonitrile (1.20 g, 10 mmol) in H2O, Br2 (0.6 mL, 11.5 mmol) was added portion wise
over 3 days at room temperature. The mixture was washed with 10% sodium
thiosulfate (50 mL). The product was crystallized from H2O, resulting in a white solid (670 mg, 53%). mp 135 °C. Rf 0.20; IR (neat): ν/cm–1 3150 (b, ArO–H), 2290 (−CN); 1H NMR (400
MHz, CD3CN): δ 7.37 (d, J = 9.0
Hz, 1H), 7.60 (d, J = 9.0 Hz, 1H), 11.56 (s, 1H); 13C NMR (125 MHz, CD3CN): δ 127.5 (ArC), 128.2
(ArC), 130.4 (ArC), 132.9 (ArC), 142.5 (ArC), 156.6 (CN); HRMS (ESI) m/z: calcd for C6H279BrN2O [M – H]−, 196.9356; observed, 196.9350.
6-Bromo-3-hydroxypicolinic
Acid (7)
To
a stirred suspension of 6-bromo-3-hydroxypicolinonitrile (6; 2.8 g, 14.1 mmol) in methanol (50 mL), 30% w/v NaOH (44 mL, 51.8
mmol) was added. The reaction mixture was heated at reflux for 12
h. The solvent was removed under reduced pressure and the suspension
was acidified with concentrated HCl to pH 2. The resulting precipitate
was filtered and washed with cold water (5 mL). Filtration afforded
product 7 (3.1 g, 100%) as white needles. mp 122–124
°C. IR (neat): ν/cm–1 3100 (ArO–H),
2850 (COO–H), 1716 (HOC=O); 1H NMR (400 MHz,
DMSO-d6): δ 7.05–7.10 (m,
1H), 7.33–7.75 (m, 1H), 11.43 (s, 1H); 13C NMR (125
MHz, DMSO-d6) 131.2 (ArC), 133.7 (ArC),
135.8 (ArC), 137.9 (ArC), 143.3 (ArC), 169.1 (C=O);
HRMS (ESI) m/z: calcd for C6H379BrNO3 [M – H]−, 215.9302; observed, 215.9298.
The desired compound was prepared according to
a modified
version of the reported procedure.[52] To
a stirred suspension of inosine (2.12 g, 8 mmol) in 40 mL of anhydrous
DMF at 0 °C, di-tert-butylsilylditrifluoromethanesulfonate
(3.0 mL, 8.8 mmol) was added dropwise under an N2 atmosphere.
After consumption of the starting material (30 min, as assessed by
TLC), the reaction was quenched immediately with imidazole (2.7 g,
40 mmol) at 0 °C. After 5 min, the reaction was warmed to room
temperature. tert-Butyldimethylsilyl chloride (1.5
g, 9.6 mmol) was then added portionwise, and the reaction mixture
was refluxed at 60 °C for 12 h. The suspension was then cooled
to room temperature, water was added, and the precipitate was collected
by suction filtration. The filtrate was discarded, and the white precipitate
was washed with cold methanol. The methanol layer was evaporated under
reduced pressure and the product was crystallized from CH2Cl2 to give a white solid (4.0 g, 98%). m.p 191–193.4
°C. Rf 0.45 (3:2 cyclohexane/ethyl
acetate); 1H NMR (600 MHz, CDCl3): δ 0.17
(s, 3H), 0.18 (s, 3H), 0.96 (s, 9H), 1.07 (s, 9H), 1.10 (s, 9H), 4.02–4.09
(m, 1H), 4.25 (td, J = 10.0, 5.0 Hz, 1H), 4.38 (dd, J = 9.5, 4.5 Hz, 1H), 4.45–4.58 (m, 2H), 5.96 (s,
1H), 7.87 (s, 1H), 8.11 (s, 1H), 12.56 (s, 1H); 13C NMR
(151 MHz, CDCl3): δ −5.0, −4.3, 18.3,
20.4, 22.8, 25.9, 27.0, 27.5, 67.8, 74.8, 75.89, 75.94, 92.3, 125.5,
138.3, 144.7, 148.1, 158.9; HRMS (ESI) m/z: calcd for C24H43O4N528Si2 [M + H]+, 523.2767;
observed, 523.2756.
The desired compound was prepared according
to a modified version
of the reported procedure.[52] To a stirred
solution of 3′,5′-O-bis(tert-butylsilyl)-2′-O-(tert-butyldimethylsilyl)inosine
(20; 663 mg, 1.2 mmol) and BOP (0.64 g, 1.44 mmol) in
20 mL of THF, DBU (0.3 mL, 1.8 mmol) was added dropwise and the mixture
was heated at 40 °C. After the consumption of the starting material
(40 min, as assessed by TLC), the reaction mixture was cooled to room
temperature and methylamine (0.3 mL, 6.0 mmol) was added dropwise
and the reaction mixture was stirred overnight. The crude product
mixture was concentrated under reduced pressure and diluted with ethyl
acetate and was washed with water (3 × 10 mL). The organic layer
was dried (anhydrous MgSO4) and concentrated under vacuum.
The residue was purified by column chromatography (9:1 to 3:2 cyclohexane/ethyl
acetate) which resulted in oil (665 mg, 98%). Rf 0.20 (7:3 cyclohexane/ethyl acetate); 1H NMR (400
MHz, CDCl3): δ 0.00 (s, 3H) 0.02 (s, 3H) 0.78 (s,
9H) 0.90 (s, 9H) 0.94 (s, 9H) 3.05 (d, J = 1.0 Hz,
3H) 3.86–3.90 (m, 1H) 4.02–4.10 (m, 1H) 4.34 (dd, J = 9.0, 5.0 Hz, 1 H) 4.38–4.44 (m, 1 H) 4.47 (d, J = 4.5 Hz, 1 H) 5.76 (br s, 2 H) 7.62 (s, 1 H) 8.22 (s,
1 H); 13C NMR (101 MHz, CDCl3): δ −5.0,
−4.3, 18.3, 20.4, 22.8, 25.9, 27.1, 27.5, 27.6, 67.9, 74.6,
75.5, 75.8, 92.4, 120.5, 125.0, 138.0, 153.4, 155.5; HRMS (ESI) m/z: calcd for C25H46O4N528Si2 [M + H]+, 536.3082; observed, 536.3078. Analytical data are consistent
with those reported.[52]
The desired
compound was prepared according to the reported procedure.[52] To a stirred solution of 3′,5′-O-bis(tert-butylsilyl)-2′-O-(tert-butyldimethylsilyl)-N6-methyladenosine (21; 240 mg, 0.45 mmol)
in 4 mL of CH2Cl2 at −15 °C, a cooled
solution of (HF)·pyridine (0.06
mL, 2.3 mmol) in 365 μL of pyridine was added. The reaction
temperature was maintained at 0 °C and stirred for 12 h. The
reaction was diluted with CH2Cl2, then washed
first with sat. aq NaHCO3 solution and with water (3 ×
10 mL). The organic layer was dried (anhydrous MgSO4) and
concentrated under reduced pressure. The residue was purified by column
chromatography (9:1 to 3:2 cyclohexane/ethyl acetate) which resulted
in oil (160 mg, 90%). Rf 0.15 (2:3 hexane/ethyl
acetate); 1H NMR (400 MHz, CDCl3): δ 0.00
(s, 3H), 0.02 (s, 3H), 0.94 (s, 9H), 3.42 (d, J =
1.0 Hz, 3H), 3.89 (dd, J = 10.5, 9.0 Hz, 1H), 4.01–4.11
(m, 1H), 4.34 (dd, J = 9.0, 5.0 Hz, 1H), 4.41 (dd, J = 9.0, 5.0 Hz, 1H), 4.47 (d, J = 5.0
Hz, 1H), 5.76 (s, 2H), 7.62 (s, 1H), 8.22 (s, 1H); 13C
NMR (101 MHz, CDCl3): δ −5.4, −5.3,
17.9, 25.6, 25.8, 27.5, 63.5, 73.1, 74.4, 87.8, 91.3, 119.7, 140.0,
140.1, 152.9, 155.8; HRMS (ESI) m/z: calcd for C17H30O4N528Si [M + H]+, 396.2062; observed, 396.2068.
Analytical data are consistent with those reported.[52]
Authors: WeiShen Aik; Marina Demetriades; Muhammad K K Hamdan; Eleanor A L Bagg; Kar Kheng Yeoh; Clarisse Lejeune; Zhihong Zhang; Michael A McDonough; Christopher J Schofield Journal: J Med Chem Date: 2013-04-23 Impact factor: 7.446
Authors: Rasheduzzaman Chowdhury; Michael A McDonough; Jasmin Mecinović; Christoph Loenarz; Emily Flashman; Kirsty S Hewitson; Carmen Domene; Christopher J Schofield Journal: Structure Date: 2009-07-15 Impact factor: 5.006
Authors: Richard J Hopkinson; Anthony Tumber; Clarence Yapp; Rasheduzzaman Chowdhury; WeiShen Aik; Ka Hing Che; Xuan Shirley Li; Jan B L Kristensen; Oliver N F King; Mun Chiang Chan; Kar Kheng Yeoh; Hwanho Choi; Louise J Walport; Cyrille C Thinnes; Jacob T Bush; Clarisse Lejeune; Anna M Rydzik; Nathan R Rose; Eleanor A Bagg; Michael A McDonough; Tobias Krojer; Wyatt W Yue; Stanley S Ng; Lars Olsen; Paul E Brennan; Udo Oppermann; Susanne Muller-Knapp; Robert J Klose; Peter J Ratcliffe; Christopher J Schofield; Akane Kawamura Journal: Chem Sci Date: 2013-08-01 Impact factor: 9.825
Authors: Timothy M Frayling; Nicholas J Timpson; Michael N Weedon; Eleftheria Zeggini; Rachel M Freathy; Cecilia M Lindgren; John R B Perry; Katherine S Elliott; Hana Lango; Nigel W Rayner; Beverley Shields; Lorna W Harries; Jeffrey C Barrett; Sian Ellard; Christopher J Groves; Bridget Knight; Ann-Marie Patch; Andrew R Ness; Shah Ebrahim; Debbie A Lawlor; Susan M Ring; Yoav Ben-Shlomo; Marjo-Riitta Jarvelin; Ulla Sovio; Amanda J Bennett; David Melzer; Luigi Ferrucci; Ruth J F Loos; Inês Barroso; Nicholas J Wareham; Fredrik Karpe; Katharine R Owen; Lon R Cardon; Mark Walker; Graham A Hitman; Colin N A Palmer; Alex S F Doney; Andrew D Morris; George Davey Smith; Andrew T Hattersley; Mark I McCarthy Journal: Science Date: 2007-04-12 Impact factor: 47.728
Authors: Rasheduzzaman Chowdhury; José Ignacio Candela-Lena; Mun Chiang Chan; David Jeremy Greenald; Kar Kheng Yeoh; Ya-Min Tian; Michael A McDonough; Anthony Tumber; Nathan R Rose; Ana Conejo-Garcia; Marina Demetriades; Sinnakaruppan Mathavan; Akane Kawamura; Myung Kyu Lee; Freek van Eeden; Christopher W Pugh; Peter J Ratcliffe; Christopher J Schofield Journal: ACS Chem Biol Date: 2013-06-12 Impact factor: 5.100
Authors: Ye Fu; Guifang Jia; Xueqin Pang; Richard N Wang; Xiao Wang; Charles J Li; Scott Smemo; Qing Dai; Kathleen A Bailey; Marcelo A Nobrega; Ke-Li Han; Qiang Cui; Chuan He Journal: Nat Commun Date: 2013 Impact factor: 14.919
Authors: Thomas Gerken; Christophe A Girard; Yi-Chun Loraine Tung; Celia J Webby; Vladimir Saudek; Kirsty S Hewitson; Giles S H Yeo; Michael A McDonough; Sharon Cunliffe; Luke A McNeill; Juris Galvanovskis; Patrik Rorsman; Peter Robins; Xavier Prieur; Anthony P Coll; Marcella Ma; Zorica Jovanovic; I Sadaf Farooqi; Barbara Sedgwick; Inês Barroso; Tomas Lindahl; Chris P Ponting; Frances M Ashcroft; Stephen O'Rahilly; Christopher J Schofield Journal: Science Date: 2007-11-08 Impact factor: 47.728