3-Alkylthio-1,2,4-triazine dimers with potent antimalarial activity

Article abstract of DOI:10.1016/j.bmcl.2010.08.065

We report on the discovery of 3-alkylthio-1,2,4-triazine dimers that are potently toxic to Plasmodium falciparum, with single digit nanomolar activity, and up to several thousand-fold lower toxicity to mammalian cells. They are equipotent against chloroquine-resistant strains of P. falciparum.

Full text of DOI:10.1016/j.bmcl.2010.08.065

Bioorganic & Medicinal Chemistry Letters 20 (2010) 6024–6029
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
3-Alkylthio-1,2,4-triazine dimers with potent antimalarial activity
Kung Ban a, Sandra Duffy b, Yelena Khakham a, Vicky M. Avery b, Andrew Hughes c, Oliver Montagnat d,
Kasiram Katneni d, Eileen Ryan d, Jonathan B. Baell a,
⇑
a Walter & Eliza Hall Institute of Medical Research, Parkville, Victoria, Australia
b Eskitis Institute for Cell and Molecular Therapies, Griffith University, Nathan, Queensland, Australia
c La Trobe University, Bundoora, Victoria, Australia
d Centre for Drug Candidate Optimisation, Monash Institute of Pharmaceutical Sciences, Monash University (Parkville Campus), Victoria, Australia
a r t i c l e i n f o
a b s t r a c t
Article history:
We report on the discovery of 3-alkylthio-1,2,4-triazine dimers that are potently toxic to Plasmodium fal-
ciparum, with single digit nanomolar activity, and up to several thousand-fold lower toxicity to mamma-
lian cells. They are equipotent against chloroquine-resistant strains of P. falciparum.
Ó 2010 Elsevier Ltd. All rights reserved.
Received 22 July 2010
Revised 12 August 2010
Accepted 12 August 2010
Available online 19 August 2010
Keywords:
Plasmodium falciparum
Malaria
Triazenes
We have previously reported that WEHI 59743, which supposed
to be triazine dimer 1, as purchased from the supplier, was found
by us to be potently toxic to Plasmodium falciparum, a causative
agent of malaria, and that it is highly selective with little activity
against a range of other pathogenic parasites as well as mamma-
lian cells.1 These data are summarized in Table 1. We also re-
marked that preliminary LCMS analysis indicated that WEHI
59743 comprised predominantly an impurity of M+16. We here re-
veal the composition of WEHI 59743 and the results of an SAR
investigation around the most active component.
two of these were yellowish in color, with the highest Rf spot being
particularly strongly colored. It was hypothesized that any
Table 1
Antiparasitic activity profile of compound WEHI 59743
N
N
N
N
N
SMe
N
MeO
WEHI 59743 supplied as 3-methoxy-5-(3-(methylthio)-1,2,4-
triazin-5-yl)-1,2,4-triazine and the NMR spectrum appeared to
confirm this structure. As shown in Figure 1 (disregarding the
two signals at 7.26 ppm and 1.5 ppm which correspond to the sol-
vent peak (chloroform) and water peak, respectively) there appears
to be clear SMe and OMe signals at 2.78 and 4.33 ppm, respec-
tively, and each of these integrates cleanly to three protons. Simi-
larly, the aromatic protons downfield near 10 ppm integrate
cleanly to two protons as expected.
However, the LCMS trace of WEHI 59743 suggested a major
component of longer retention time with a molecular ion of
M+16 (Fig. 2). This was unexpected and it was difficult to reconcile
the NMR spectrum in Figure 1 with the LCMS trace in Figure 2.
Simple thin layer chromatography initially added further to the
confusion because the results suggested that the sample was in
fact a mixture of three major components. It was observed that
1
a
Parasite
EC50, (lM)
P. falciparumb
L. infantosumc
T. cruzid
0.18
8.6
16
5.7
33
T. bruceie
Cytotoxicityf
a
Values are means of three experiments, 50%.
Plasmodium falciparum 3D7 strain, erythrocytic
b
stage, chloroquine was used as a control, EC50 30
l
g/ml.
Leishmania infantosum, amastigote stage, miltefo-
sine was used as a control, EC50 2.4 g/ml.
Trypanosoma cruzi Tulahaen C2C4 strain, amasti-
gote stage, nifurtimox was used as control, EC50
0.24 g/ml.
Trypanosoma brucei rhodesiense strain STIB 900,
c
l
d
a
l
e
bloodstream form. Suramin (EC50 0.13 lg/ml) was used
as a control.
f
⇑
Rat skeletal myoblast cell L-6 strain, Tamoxifen
g/ml.
Corresponding author.
E-mail address: jbaell@wehi.edu.au (J.B. Baell).
was used as a control, EC50 4.9
l
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.08.065

K. Ban et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6024–6029
6025
Figure 1. 1H NMR spectrum of WEHI 59743.
C:\Xcalibur\data\GAH\WEHI-0059743
RT: 0.00 - 14.98
12/9/2008 10:04:53 AM
RT: 7.65
MA: 2942837
NL:
5.15E5
Channel B
500000
450000
400000
350000
300000
250000
200000
150000
100000
UV
WEHI-
0059743
RT: 6.91
MA: 1432756
14.17
RT: 6.16
MA: 72726
0.32
12.02
12
11.25
50000
10.22
10
9.63
8.56 9.10
9
5.59
0.94 1.28 1.74 2.12 2.60 3.02 3.74
4.58 5.27
5
0
0
1
2
3
4
6
7
8
11
13
14
Time (min)
237.0
NL: 1.53E5
100
80
WEHI-0059743#175-
178 RT: 6.82-6.94
AV: 4 T: + c ESI Full
ms [80.00-1000.00]
60
40
100.8
100.7
239.0
240.0
20
221.1
668.0
652.0
277.4
136.0
179.0
609.7
684.3
570.3
532.8 550.4
335.9
379.2
414.1
447.1 471.3 493.7
284.9
0
100
NL: 1.01E5
WEHI-0059743#192-
200 RT: 7.50-7.82
AV: 9 T: + c ESI Full
ms [80.00-1000.00]
80
60
40
20
0
252.9
255.0
256.1
414.0
424.8
486.1
481.3
237.0
151.9
191.2
200
652.1
650
129.9
644.9
519.4
574.7
672.9
293.7
602.9
600
560.0
550
369.0
342.8
397.1
400
300.7
100
150
250
300
350
450
500
m/z
Figure 2. Standard LCMS trace of WEHI 59743.
compound with a strong chromophore might give rise to an artifi-
cially intense signal at 254 nm in the LCMS and therefore not be
representative of the real concentration in the mixture. Therefore,
WEHI 59743 was re-analyzed at a wavelength of 230 nm in order
to better detect all impurities. Quite remarkably, as shown in
Figure 3, WEHI 59743 now came up as mixture of several

6026
K. Ban et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6024–6029
Figure 3. Fuller LCMS analysis of WEHI 59743. Top segment from top to bottom: UV absorption at 230 nm, ES scan for M+H 221, ES scan for M+H 237, ES scan for M+H 253,
UV absorption at 254nm, total ion current trace. Bottom segment from top to bottom: mass spectrum (ES+) for the mw 252 peak, the mw 236 peak, and the mw 220 peak.
compounds. The total ion current, like the TLC, was suggestive of
three major components, of molecular ion peaks MÀ16, M and
M+16, respectively. Analysis at different wavelengths supported
the view that the shorter retention time compound (MÀ16) had
the weakest chromophore at 254 nm and the longer retention time
compound (M+16) had the strongest chromophore at 254 nm.
N
N
(iii)
N
N
NH2NH NH
S
HI.NH2NH NH (ii)
SMe
(i)
N
N
SMe
OMe
2
3
Scheme 1. Reagents and conditions: (i) MeI [EtOH], 60 °C, 78%; (ii) 40% aq Glyoxal,
NaHCO3, 94%; (iii) NaOMe [MeOH], 55%.

K. Ban et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6024–6029
6027
N
N
N
N
N
N
N
(i)
+
2
3
+
+
N
N
N
SMe
N
N
N
OMe
N
N
SMe
N
N
N
1 (19%)
4 (22%)
MeO
5 (21%)
MeO
MeS
Scheme 2. Reagents and conditions: (i) KCN, air [H2O], 64%.
Table 2
It was proposed that a logical interpretation was that the MÀ16
compound was the methoxy homodimeric counterpart and the
M+16 compound was the methylthio homodimeric counterpart to
the intended heterodimeric structure. A search of the literature re-
vealed that synthetic routes for these compounds could plausibly
produce unintended mixtures of compounds.2 It was further real-
ized that if the two homodimeric components were present in
approximately equal amounts (regardless of the concentration of
the heterodimer), an NMR spectrum such as that in Figure 1, which
superficially looks like pure heterodimer, could arise. Indeed, scru-
tiny of the peaks supports this and show that each of the methyl
peaks, for example, comprises two singlets rather than one. This
would make sense if one methoxy signal belongs to the heterodimer
and one to the homodimer. Likewise for the methylthio signals (and
the aromatic signals). It was briefly considered that the M+16 signal
could represent sulfoxide formation, however, this would be very
polar and would have a shorter retention time rather than a longer
retention time. Also, this does not explain the NMR spectrum.
To summarize, WEHI 59743 was supplied as a mixture of three
compounds. This is a good example of why LCMS analysis at
254 nm, however commonly undertaken, can be highly misleading
for mixtures containing compounds with chromophores of differ-
ing absorbance maxima and extinction coefficients. This is highly
pertinent to compound quality control when samples are submit-
ted to bioassays.
Biological activity of 1,2,4-triazine dimersa against P. falciparum
N
N
N
N
N
R2
N
R1
R2
b
Compd
R1
EC50
(l
M)
Pfc
Cytotoxd
1
4
5
6
7
MeO
MeO
MeS
EtS
PrS
BuS
MeS
MeS
Me2NCH2CH2S
MeS
Et2NCH2CH2S
Et
MeS
MeO
MeS
EtS
PrS
BuS
Me2NCH2CH2O
Me2NCH2CH2S
Me2NCH2CH2S
Et2NCH2CH2S
Et2NCH2CH2S
Et
0.15 (0.25)e
4.1 (4.4)e
18
36
68
0.026 (0.053)e
0.022 (0.024)e
0.017 (0.022)e
0.096 (0.16)e
0.080 (0.12)e
0.008 (0.004)e
0.47 (0.62)e
0.023 (0.012)e
0.44 (0.79)e
1.6f
84
>115
>115
24
8
14
15
16
17
18
19
21
25
9
16
NTg
a
b
c
Monomers are inactive.
Values are means of three experiments, 50%.
Plasmodium falciparum 3D7 strain, erythrocytic stage. Chloroquine was used as
M.
HEK 293 cells. Puromycin was used as a control, EC50 0.41
a control, EC50 0.004
l
d
l
M.
Plasmodium falciparum Dd2 strain, erythrocytic stage. Chloroquine was used as
a control, EC50 0.17 M.
e
l
Due to the limited availability of the commercial sample, the
synthesis and retesting of each proposed component was under-
taken as shown in Scheme 1.
f
Tested in singlicate only.
Not tested.
g
Firstly, the methylthiotriazine 2 was made by cyclocondensa-
tion of methylated thiosemicarbazide3 with glyoxal.4 The corre-
sponding methoxy counterpart 3 was then made by reaction of 2
with methoxide.2
N
N
(i)
N
(ii)
N
N
2
NMe2
X
N
SO2Me
Triazine monomers 2 and 3 were then dimerized through the
use of excess cyanide in water2 as shown in Scheme 2. This gave,
after purification, homodimers 4 and 5 heterodimer 1 each in
around 20% yield.
9
10 X=O
11 X=S
Scheme 3. Reagents and conditions: (i) mCPBA (2 equiv) [DCM], 0 °C–rt, 2 h; (ii) for
X = O, N,N-dimethylethanolamine, NaH [THF], 3% over two steps; for X = S, N,N-
dimethylethanethiolamine, NaH [THF], 15% over two steps.
Significantly, the 1H NMR spectrum of the crude project mixture
was indistinguishable from that in Figure 1, and LCMS data
matched identically. These compounds were then tested for activ-
ity against P. falciparum, using a novel assay developed by Avery
and coworkers5 and as shown in Table 2, it was the methylthio
homodimer 5 in which most of the biological activity resides with
an EC50 of 26 nM.
N
N
N
N
NH2NH NH
S
(ii)
SH
(i)
N
N
NR2
S
12
11 R=Me
13 R=Et
On this basis, a small set of alkylthio dimers 6–8 was assembled
using analogous methodology to that shown in Scheme 1 for the syn-
thesis of the methylthio monomer 2, followed by homodimerization
using the same conditions as those in Scheme 2. As shown in Table 2,
alldimers were highlypotent and activityonly started to decrease for
the butylthio-containingcompoundsuch that8 had anEC50 of96 nM.
We were next interested in investigating whether a tertiary
amine could be tethered to these compounds with maintenance
of activity. This was on the basis that such compounds would likely
to be more soluble and hence more drug-like but also that this ap-
proach has yielded potent antimalarial compounds in other hetero-
cyclic systems.6 The fact that extended alkylthio groups were
reasonably tolerated and readily synthesized highlighted this
feature as the logical choice for the incorporation of amino tethers.
Scheme 4. Reagents and conditions: (i) 40% aq glyoxal, 80 °C, 82%; (ii) alkyl halide
(N,N-dimethylchloroethylamine for 11 and N,N-diethylbromoethylamine for 13),
Hünig’s base [DMA], reflux 2 h, 20%.
N
N
N
N
HCl.NH2NH Et
NH
(ii)
(i)
N
N
N
N
Et
Et
N
Et
19
20
Scheme 5. Reagents and conditions: (i) NaHCO3 [ether] 30 min (RTP) then 40% aq
glyoxal, 4 h (RTP), 73%; (ii) KCN [water] 40 deg 3 h, 30%.

6028
K. Ban et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6024–6029
Table 3
Physicochemical evaluation and in vitro metabolism in human liver microsomes of representative triazines
a
a,b
c,f
Compd no. In vitro CLint
(l
L/min/mg protein) Microsome predicted EH
Metabolites detecteda MW PSAc (Å2) pKa
Log De
Solubilityd
(lM)
pH 3 pH 7.4 pH 2
pH 6.5
5
19
105.4
11.7
0.85
0.39
None detected
P+16
252
216
127.9
77.3
N/A
N/A
2.4
1.6
2.4
1.6
15.3
426
14.5
311
a
The relative loss of parent compound and formation of metabolic products following incubation in HLM’s was determined by LCMS. The concentration of test compound
versus time was fitted to an exponential decay function to determine the first-order rate constant for substrate depletion. This rate constant was used to calculate an in vitro
clearance (CLint) and a hepatic extraction ratio (EH) value.
b
In vitro metabolic stability groupings: very high predicted hepatic extraction ratio EH >0.95; high = 0.7–0.95; intermediate = 0.3–0.7; low <0.3.
Values calculated with ACD Log D suite (version 9.12).
Values measured under equilibrium solubility conditions at 25 °C.
Values measured via a chromatographic Log D estimation method.
c
d
e
f
No physiologically relevant pKa.
Initially, we investigated activation of the thiomethyl group in 2
towards leaving through oxidation to the sulfone7 followed by
nucleophilic displacement, as shown in Scheme 3.
While the yield over two steps for both 10 and 11 was extre-
mely low (later shown to be due to impure mCPBA), this reaction
provided sufficient compound to progress to the next step and so
was not optimized. Nevertheless, an alternative method was
sought as shown in Scheme 4, involving synthesis of the mercapto-
triazine 12 from thiosemicarbazide and glyoxal,8 followed by alkyl-
ation of the thiol group to give either 11 or 13.
Shown in Scheme 5 is the synthesis of the 3-ethyl dimer 19,
which was achieved by the reaction of propionimidohydrazide
with glyoxal to give monomer 20, followed by cyanide-induced
dimerization.
As shown in Table 2, this dimer lost significant activity against
P. falciparum with an EC50 of 1.6 lM and so it would seem that sul-
fur atoms are very important in the 3-position.
Finally, we were interested to investigate whether the small and
relatively polar nature of these compounds may confer favorable
metabolic properties to what would otherwise appear to be inher-
ently liable systems. Shown in Table 3 are the physicochemical
properties of 5 and 19, as well as stability data from incubation
with human liver microsomes. Here, it can be seen that these com-
pounds have great potential for optimization as they are very small
and simple. With their low Log D and high solubility at both acidic
and near neutral pH, one can envisage the potential for very high
systemic exposure.
However, somewhat unsurprisingly, our most active system in
compound 5 with the thioethers installed is relatively unstable in
the environment of the microsome with a relatively high predicted
hepatic extraction ratio (EH) of 0.85. Conversely, 19 was signifi-
cantly more stable with a corresponding EH of 0.39. Intriguingly,
then, it was only 19 for which metabolites were detected, this
being M+16 and therefore a likely product of oxidation (N oxide
or less likely ring epoxide or alkyl hydroxyl).
In summary, we have reported the discovery of dimers of 3-
substituted 1,2,4-triazines that are potently toxic to P. falcipa-
rum with single digit nanomolar activity and up to several
thousand-fold lower toxicity to mammalian cells. They are equi-
potent against chloroquine-resistant strains of P. falciparum.
They are small and highly optimizable with low Log D and excel-
lent water solubility and can tolerate side-chain extension with
a tertiary amino group. However, thioether groups are required
for potent activity and these are a metabolic liability, as assessed
in human liver microsomes. The major challenge for these
compounds going forward will therefore be the maintenance
of potent antimalarial activity while improving metabolic
stability.
This route proved more efficient for the synthesis of amine-
tethered triazines, although the yield for this unoptimized alkyl-
ation was still relatively low and could further be improved with
optimization.
Triazine monomers 10, 11 and 13 were each then dimerized with
2 through the use of cyanide as in Scheme 2 to give homo- and het-
ero-dimeric product mixtures, which were then separated and
tested for biological activity. As shown in Table 2, monoalkylamine
tethers were all reasonably tolerated and dimers 14, 15 and 17
had EC50 values of less than 100 nM. As before, the oxygen replace-
ment of the sulfur atom was less favorable and so while the EC50 of
14 was 80 nM, that of 17 was 23 nM and 15 displayed extremely
impressive activity in the single digit nanomolar range with an
EC50 of 8 nM. Conversely, bis tertiary amine-tethered compounds
16 and 18 were significantly less potent with EC50 values in the half
micromolar range. While some of these dimers are becoming
moderately cytotoxic with EC50 values in the mid micromolar range,
the selectivity for P. falciparum for compound 15 for example is
still more than 2500-fold. Dimer 15 is 3–4 times more potent than
5, though whether this is significant enough to imply an amine-in-
duced localized concentration increase as designed, is hard to
say.
It is tempting to speculate that compounds such as 15 could
clearly have an affinity with heme and since they also structurally
resemble aspects of chloroquine, that a chloroquine-like mecha-
nism of action is responsible for their observed antimalarial activ-
ity. If this is the case, then these triazine dimers may be less active
against chloroquine-resistant strains of P. falciparum. Shown in
parentheses in Table 2 are the EC50 values of the compounds when
tested against the chloroquine-resistant Dd2 strain of P. falciparum.
Of particular note was that these compounds were essentially
equally active against chloroquine-resistant P. falciparum, suggest-
ing either a distinctive mechanism of action or that they are not
similarly recognized by efflux mechanisms in Dd2 strains.
We were interested as to whether we could move away from
thioethers to more drug-like systems. We have shown that an oxy-
gen atom is not bioisosteric with a sulfur atom in this instance and
wondered whether better results might be obtained with methy-
lene replacement, which could be the case if a less electronegative
and more hydrophobic atom than oxygen in the 3-position was
favorable for activity.
Acknowledgements
Financial support was received from the Victorian State
Government OISS Grant and NHMRC IRIISS Grant No. 361646 and
the Alan Harris Fellowship and for this we are most grateful.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2010.08.065.

K. Ban et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6024–6029
6029
Fowble, J. W.; Forquer, I.; McGinley, P. L.; Castro, S.; Angulo-Barturen, I.; Ferrer,
S.; Rosenthal, P. J.; DeRisi, J. L.; Sullivan, D. J., Jr.; Lazo, J. S.; Roos, D. S.; Riscoe, M.
K.; Phillips, M. A.; Rathod, P. K.; Van Voorhis, W. C.; Avery, V. M.; Guy, R. K.
Nature 2010, 465, 311.
References and notes
1. Nhu, D.; Duffy, S.; Avery, V. M.; Hughes, A.; Baell, J. B. Bioorg. Med. Chem. Lett.
2010, 20, 4496.
2. Krass, D. K.; Chen, T.-K.; Paudler, W. W. J. Heterocycl. Chem. 1973, 10, 343.
3. Emilsson, H. J. Heterocycl. Chem. 1989, 26, 1077.
4. Paudler, W. W.; Chen, T.-K. J. Heterocycl. Chem. 1970, 7, 767.
5. Guiguemde, W. A.; Shelat, A. A.; Bouck, D.; Duffy, S.; Crowther, G. J.; Davis, P. H.;
Smithson, D. C.; Connelly, M.; Clark, J.; Zhu, F.; Jimenez-Dıaz, M. B.; Martinez, M.
S.; Wilson, E. B.; Tripathi, A. K.; Gut, J.; Sharlow, E. R.; Bathurst, I.; El Mazouni, F.;
6. Kelly, J. X.; Smilkstein, M. J.; Brun, R.; Wittlin, S.; Cooper, R. A.; Lane, K. D.;
Janowsky, A.; Johnson, R. A.; Dodean, R. A.; Winter, R.; Hinrichs, D. J.; Riscoe, M.
K. Nature 2009, 459, 270.
7. Taylor, E. C.; Pont, J. L.; Warner, J. C. Tetrahedron 1987, 43, 5159.
8. Heravi, M. M.; Nami, N.; Oskooie, H. A.; Hekmatshoar, R. Phosphorus, Sulfur
Silicon Relat. Elem. 2006, 181, 87.