Synthesis, in vitro and in vivo activity of thiamine antagonist transketolase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2007.11.101

Tumor cells extensively utilize the pentose phosphate pathway for the synthesis of ribose. Transketolase is a key enzyme in this pathway and has been suggested as a target for inhibition in the treatment of cancer. In a pharmacodynamic study, nude mice with xenografted HCT-116 tumors were dosed with 1 ('N3′-pyridyl thiamine'; 3-(6-methyl-2-amino-pyridin-3-ylmethyl)-5-(2-hydroxy-ethyl)-4-methyl-thiazol-3-ium chloride hydrochloride), an analog of thiamine, the co-factor of transketolase. Transketolase activity was almost completely suppressed in blood, spleen, and tumor cells, but there was little effect on the activity of the other thiamine-utilizing enzymes α-ketoglutarate dehydrogenase or glucose-6-phosphate dehydrogenase. Synthesis and SAR of transketolase inhibitors is described.

Full text of DOI:10.1016/j.bmcl.2007.11.101

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 2206–2210
Synthesis, in vitro and in vivo activity of thiamine
antagonist transketolase inhibitors
Allen A. Thomas,a,* Y. Le Huerou,a J. De Meese,a Indrani Gunawardana,a
Tomas Kaplan,a Todd T. Romoff,a Stephen S. Gonzales,a Kevin Condroski,a
Steven A. Boyd,a Josh Ballard,a Bryan Bernat,a Walter DeWolf,a May Han,b
Patrice Lee,a Christine Lemieux,a Robin Pedersen,b Jed Pheneger,a
Greg Poch,a Darin Smith,a Francis Sullivan,a Solly Weiler,b
S. Kirk Wright,b, Jie Lin,b Barb Brandhubera and Guy Vigersa
aArray BioPharma Inc., 3200 Walnut Street, Boulder, CO 80301, USA
bAVEO Pharmaceuticals, Inc., 75 Sidney Street, 4th floor, Cambridge, MA 02139, USA
Received 19 September 2007; revised 21 November 2007; accepted 27 November 2007
Available online 3 December 2007
Abstract—Tumor cells extensively utilize the pentose phosphate pathway for the synthesis of ribose. Transketolase is a key enzyme
in this pathway and has been suggested as a target for inhibition in the treatment of cancer. In a pharmacodynamic study, nude mice
with xenografted HCT-116 tumors were dosed with 1 (‘N30-pyridyl thiamine’; 3-(6-methyl-2-amino-pyridin-3-ylmethyl)-5-(2-hydro-
xy-ethyl)-4-methyl-thiazol-3-ium chloride hydrochloride), an analog of thiamine, the co-factor of transketolase. Transketolase activ-
ity was almost completely suppressed in blood, spleen, and tumor cells, but there was little effect on the activity of the other
thiamine-utilizing enzymes a-ketoglutarate dehydrogenase or glucose-6-phosphate dehydrogenase. Synthesis and SAR of transke-
tolase inhibitors is described.
Ó 2007 Elsevier Ltd. All rights reserved.
Activation of the nonoxidative branch of the pentose
phosphate pathway for ribose-5 phosphate synthesis
has been demonstrated in tumor cells.1 The enzyme
transketolase (TK) has a high control coefficient toward
ribose synthesis in the pathway, suggesting a potential
role in cancer. Mobilization of the transketolase co-fac-
tor thiamine from normal cells to tumors, leading to thi-
amine deficiency symptoms in cancer patients (e.g.,
severe cardiac failure), has been described.2 This reorga-
nization of cell metabolism offers one explanation of
how cancerous cells maintain a continuous proliferation
rate in the presence of decreased glucose metabolism
and hypoxic conditions in the weakening host. Further
support for the importance of TK in tumor cell mainte-
nance was provided by Rais et al.3 They demonstrated
that a TK inhibitor oxythiamine induces G1 arrest in
Ehrlich’s ascites tumor cells implanted in mice.
Thiamine exists in four possible forms: free alcohol
(Vitamin B1), thiamine monophosphate (TP), diphos-
phate also called pyrophosphate (TPP), and triphos-
phate (TPPP). Upon ingestion, thiamine phosphates
are hydrolyzed in the gastrointestinal tract by phospha-
tases and then the free alcohol is actively transported by
specific transporters ThTr1 and ThTr2 (Km = 2.6 lM).4
Thiamine is pyrophosphorylated in the cytoplasm by
thiamine pyrophosphokinase (TPPK) or other less
well-characterized kinases.5 Because binding to thia-
Keywords: Transketolase;
Pyrophosphate.
Thiamine;
Cancer;
Metabolism;
*
Corresponding author. Tel.: +1 303 386 1523; fax: +1 303 386
1420; e-mail: athomas@arraybiopharma.com
Present address: Novartis Institutes for BioMedical Research, Inc.,
250 Mass Ave, Cambridge, MA 02139, USA.
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.11.101

A. A. Thomas et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2206–2210
2207
mine-utilizing enzymes depends heavily upon the pyro-
phosphate group, pyrophosphorylation of the alcohol
is an essential step in obtaining active TPP. In light of
the importance of the pyrophosphate group, synthetic
thiamine analogs which maintain TPPK substrate activ-
ity would be desirable. Antagonists of thiamine have
been previously described.6 In this report, we summarize
our initial efforts to identify a potent inhibitor of human
transketolase as a novel anticancer agent (Fig. 1).
mercially available 3-cyano-6-methyl-2(1H)-pyridinone,
with an overall yield of 25%. The key step was SNAr
of 3 with ammonia, in a pressurized steel bomb at
210 °C, to provide aminopyridine 4 in good yield.
We carried out numerous in vitro and in vivo studies on
1 and a variety of novel analogs. 1 has an average EC50
value of 26 nM in a functional assay of human transke-
tolase inhibition utilizing HCT-116 cells, a human colon
cancer cell line.8 Pyrophosphorylation of 1 is necessary
for binding to transketolase, and this apparently takes
place intracellularly. The average Kd of 1-PP9 for bind-
ing to apo-TK (transketolase lacking bound thiamine) is
22 nM, while the free alcohol shows no significant bind-
ing. Synthesis of pyrophosphates is exceedingly tedious,
so we developed a coupled TPPK/apo-TK assay, in
which compounds are exposed to TPPK for an extended
period of time prior to testing against apo-TK.10 In this
assay, 1 has a Kd of 33 nM, in good agreement with the
pyrophosphate binding result. We extensively explored
the SAR around 1 and thiamine in our search for potent
and selective transketolase inhibitors.
The substitution of carbon for the N1 nitrogen of thia-
mine results in a potent in vitro thiamine antagonist, the
‘N30-pyridyl thiamine’ 1, originally described by Mat-
sukawa.7 We developed an improved synthesis of this
material (Scheme 1), in six steps starting from the com-
With the thiazolium B-ring of thiamine held constant, the
SAR of the four substitutable positions of the N3-pyri-
dine A-ring was explored(Table 1). Potentcellular activity
was seen only when the R4-amino group was present as well
as small substitution (Me, Et, Cl) at the R2-position.
With the N3-pyridine A-ring held constant, the SAR of
the three substitutable positions of the thiazolium B-ring
was explored (Table 2). There is a requirement for
methyl substitution at the 40-position in order to main-
tain cellular potency, although substitutions by ethyl,
hydroxymethyl, or hydrogen are still active in the cou-
pled assay. The 50-position appears to have a size restric-
tion. Substitution of a hydroxypropyl group for the
native hydroxyethyl group results in a 50-fold loss in cel-
lular potency, while substitution of a hydroxymethyl
group only results in a 6-fold loss. Substitution of a hy-
droxyl group at the a-position of the hydroxyethyl
group is tolerated, and interestingly, the S-enantiomer
is 40-fold more potent than the R-enantiomer. Protec-
tion of the free alcohol of 1 as an acetate did not cause
any significant change in cellular potency; it is very likely
that the acetyl group is hydrolyzed under the conditions
of the cellular assay.
Figure 1. Structures of thiamine, thiamine pyrophosphate (TPP), and
the transketolase inhibitors 1 and 1-PP.
The ability to substitute at the 20-position is crucial, be-
cause such compounds are unable to effect catalysis.
Moreover, appropriate groups at this position may be
able to extend into the substrate-binding site of TK
and perhaps make additional binding interactions. The
simple 20 substitutions of methyl, ethyl, and a-hydroxy-
ethyl all showed cellular potency essentially identical to
1. Substitution of larger groups, however, significantly
decreased cellular potency.
Figure 2 shows inhibitor 34 from Table 2 docked into
the thiamine pyrophosphate binding site of transketol-
ase with a 20-ethyl substituent extending toward the sub-
strate binding region. The protein crystallographic X-
ray structure of compound 1-PP bound in a transketol-
Scheme 1. Synthesis of 1. Reagents and conditions: (a) POCl3, 68%;
(b) NH3, EtOH, 210 °C, 300 psi, 77%; (c) aq KOH, reflux, 100%; (d)
LiAlH4, THF, 80%; (e) SOCl2, THF, 82%; (f) 2-(4-methylthiazol-5-
yl)ethanol (5 equiv), neat, 100 °C, 74%.

2208
A. A. Thomas et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2206–2210
Table 1. SAR for substituted N3-pyridine A-rings
Compound
R1
R2
R4
R6
TPPK/apo-TKa
HCT-116b
1
H
H
H
H
I
Me
CH3CONH
NH2
NH2
NH2
NH2
NH2
NH2
NH2
NH2
NH2
H
H
33
23
26
800
8
H
9
CF3
CF3
H
CF3
H
1460
104
109
57
—
160
10
11
12
13
14
15
16
17
18
19
20
21
H
1000
360
H
H
H
H
H
H
Me
H
H
H
CN
Et
H
H
10
34
8.7
8.3
Me
Cl
Me
H
46
9.0
470
CF3
H
H
310
64
H
NH2
H
24,000
38,000
1600
110
H
H
220
7.0
36
NH2
H
H
H
NH2
H
H
Phthaloyl
H
140
3100
a Average Kd, nM, for coupled assay from at least three experiments. Data were within 2-fold of the mean.
b Average EC50, nM, for cellular assay performed in triplicate. Data were within 4-fold of the mean.
Table 2. SAR for substituted thiazolium B-rings
0
0
0
Compound
R2
H
R4
R5
TPPK/apo-TKa
33
HCT-116b
26
1
Me
Me
Me
Me
CH2CH2OH
22
23
24
H
H
H
53
28
27
300
50
2000
25
26
27
28
29
H
H
H
H
Et
Me
CH2CH2OH
40
950
4.7
CH2CH2OCOCH3
CH2CH2CO2CH3
CH2CH2OH
ND
ND
25
Me
CH2OH
H
2300
570
490
Me
CH2CH2OH
36
30
31
Me
Me
CH2CH2OH
CH2CH2OH
ND
ND
840
25,000
32
33
34
35
36
37
H
Me
Me
Me
Me
Me
H
CH2CH2CN
CH2CH2CH2OH
CH2CH2OH
CH2CH2OH
CH2CH2OH
CH2CH2OH
ND
ND
75
2200
1300
21
H
Et
CH(CH3)OH
120
25
30
56
Me
H
29
9500
a Average Kd, nM, for coupled assay from at least three experiments. Data were within 2-fold of the mean. ND, no data.
b Average EC50, nM, for cellular assay performed in triplicate. Data were within 4-fold of the mean.

A. A. Thomas et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2206–2210
2209
Figure 3. Comparison of selectivity of thiamine-utilizing enzymes of
treated and control animals after treatment with 1. Error bars
represent standard deviation between mice (3–5/group).
Figure 2. Inhibitor 34 docked into human TK active site.
ase construct was determined (in preparation), and
inhibitor 34 was then docked with Schrodinger’s Glide
program using extended precision parameters. Prior to
docking, the 1-PP ligand was removed from the transke-
tolase X-ray structure coordinates, and the protein was
prepared by optimizing all asparagine, glutamine, histi-
dine, and side chain hydroxyl groups and then running a
short constrained refinement of the resulting structure.
Indeed, inhibitor 34 was 75 nM in the coupled TPPK/
apo-TK assay (Table 2), supporting the structural obser-
vations of the TK binding site that substitution at the 20
position should be tolerated. A xenograft model in
which human colon cancer HCT-116 cells were im-
planted in nude mice was employed to test the efficacy
of our transketolase inhibitors in vivo.11 In order to
boost chances of success, the mice were put on one of
two low thiamine diets (5 or 1 ppm, vs a normal diet
of 16.5 ppm). The tumors were allowed to grow to
50 mm2 before treatment was initiated. The mice were
treated with compound 100 mg/kg iv bid for two weeks,
after which time the tumors were assessed for size and
enzymatic activity. Control animals were dosed with
the vehicle.
Figure 4. Comparison of tumor sizes of treated and control animals
after treatment with 1. Error bars represent standard deviation
between mice (3–5/group).
late that there are alternative pathways to generate ri-
bose for DNA synthesis that are operating in these
tumor cell lines. To gain a full understanding of this is
beyond the scope of the current work, but worthy of fu-
ture investigation.
Supplementary data
Procedures and analytical data are available for com-
pounds 1 and 8–37. Supplementary data associated with
this article can be found, in the online version, at
doi:10.1016/j.bmcl.2007.11.101.
Figure 3 shows the pharmacodynamic effects of 112 in
these HCT-116 tumor-bearing nude mice, with a prefer-
ential reduction of transketolase activity in tumor over
that of another thiamine-utilizing enzyme, a-ketoglutar-
ate dehydrogenase (a-KGDH), and no effect on the
NADPH-containing glucose-6-phosphate dehydroge-
nase (G6PD). Figure 4 shows that despite the reduction
of transketolase activity, there was no apparent effect on
tumor size in either cohort.
References and notes
1. Boros, L. G.; Torday, J. S.; Lim, S.; Bassilian, S.;
Cascante, M.; Lee, W.-N. P. Cancer Res. 2000, 60, 1183.
2. Cascante, M.; Centelles, J. J.; Veech, R. L.; Lee, W.-N. P.;
Boros, L. G. Nutr. Cancer 2000, 36, 150.
3. Rais, B.; Comin, B.; Puigjaner, J.; Brandes, J. L.; Creppy,
E.; Saboureau, D.; Ennamany, R.; Lee, W. N.; Boros, L.
G.; Cascante, M. FEBS Lett. 1999, 456, 113.
4. Subramanian, V. S.; Marchant, J. S.; Parker, I.; Said, H.
M. J. Biol. Chem. 2003, 278, 3976.
5. Baker, L.-J.; Dorocke, J. A.; Harris, R. A.; Timm, D. E.
Structure 2001, 9, 539.
Thus, although we did prepare potent and selective thi-
amine antagonists that inhibited transketolase both
in vitro and in vivo, there was no apparent effect on tu-
mor cell growth, at least in this particular xenograft
model and cell line. This result is in contradiction to ear-
lier reports that inhibition of TK prevents tumor
growth.2 As for why this is the case, we can only specu-
6. Du, M. X.; Sim, J.; Fang, L.; Zheng, Y.; Koh, S.; Stratton,
J.; Pons, J.; Wang, J. J.-X.; Carte, B. J. Biomol. Screen

2210
A. A. Thomas et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2206–2210
2004, 9, 427; Solovjeva, O. N.; Kochetov, G. A. FEBS
8. Boyd, S. A. WO2005095391.
Lett. 1999, 462, 246; Boros, L. G. Cancer Res. 1998,
58, 3188; Boros, L. G.; Puigjaner, J.; Cascante, M.;
Lee, W.-N. P.; Brandes, J. L.; Bassilian, S.; Yusuf, F. I.;
Williams, R. D.; Muscarella, P.; Melvin, W. S.; Schirmer,
W. J. Cancer Res. 1997, 57, 4242; Kochetov, G. A.;
Izotova, A. E.; Meshalkina, L. E. Biochem. Biophys. Res.
Commun. 1971, 43, 1198.
9. Gutowski, J. A. Methods Enzymol. 1979, 62, 120.
10. Prior to performing the TPPK/apo-TK assay, each com-
pound was evaluated for its ability to be pyrophosphory-
lated by TPPK using thiamine as a control substrate. The
extent of substrate pyrophosphorylation relative to thia-
mine pyrophosphorylation was determined prior to com-
petition with TPP in inhibiting TK activity according to
methods described in WO2005/095391.
7. This compound has been previously reported (a) Matsuk-
awa, T.; Matsuno, T. Yakugaku Zasshi 1944, 64, 145; (b)
Dornow, A.; Hargesheimer, A. Chem. Ber. 1953, 86, 461;
(c) Ostrovskii, Y. M.; Chernikevich, I. P.; Voskoboev, A.
I.; Shellenberger, A. Bioorg. Khim. 1977, 3, 1083; (d)
Meshalkina, L. E.; Golbik, R.; Usmanov, R. A.; Neef, H.;
Schellenberger, A.; Kochetov, G. A. Dokl. Akad. Nauk
SSSR 1989, 307, 486.
11. Gyuris, J.; O’Hagan, R. C.; Han, M.; Robinson, M.;
Weiler, S., WO2005094803.
12. Though other compounds (e.g., 13 and 26) demonstrated
comparable or better cellular potency to compound 1, the
latter was tested in the xenograft model due to its synthesis
being amenable to scaleup at the time the xenograft
experiments were performed.