Artemisinin combination therapies (ACT) represent the current standard of care in
the treatment of uncomplicated malaria. The widespread adoption of ACT has been motivated
by a desire to minimize the emergence of drug resistance and to address the problem
of recrudescence associated with artemisinin monotherapy.1–4 We set out to explore
a single-molecule ‘fragmenting hybrid’ strategy in which an artemisinin-like peroxide
is employed to deliver a partner drug, only upon activation by ferrous iron in the
parasite. In principle, iron(II)-dependent drug delivery from a fragmenting hybrid
could alleviate unwanted off-target bioactivities of the partner drug, which would
be inactive in its hybrid form.
Our design for fragmenting hybrids was inspired by earlier work demonstrating that
antimalarial 1,2,4-trioxolanes, including the investigational agent arterolane (1,
Figure 1),5 undergo iron(II)-promoted ring opening to afford both reactive carbon-centered
radical species—the presumed agent responsible for parasite toxicity—and carbonyl
containing byproducts (e.g., 5, Scheme 1 a).6–8 In parasites, scission of the trioxolane
ring is thought to be initiated by free heme, a byproduct of hemoglobin degradation
in the parasite digestive vacuole (DV). Because heme concentrations are enormous (estimated
at 0.4 m)9 in the parasite DV and significantly lower (∼10−16
m)10 in human plasma, we reasoned that the presence of this reactive species in parasites
could be exploited for selective drug delivery. Thus, to achieve parasite-targeted
hybrid fragmentation, we embedded a masked retro-Michael linker within the 1,2,4-trioxolane
ring system (Figure 1). In the hypothesized activation–release sequence, free ferrous
iron heme in the parasite DV mediates opening of the trioxolane ring, which unmasks
the carbonyl function of the retro-Michael linker (6), leading to release of the second
drug species via a β-elimination reaction (Scheme 1 b). The release of free partner
drug from the linker distinguishes this approach from conventional covalent hybrids,11
while the introduction of the retro-Michael linker greatly expands the scope of possible
conjugation chemistry as compared to other peroxidic prodrugs that require drug conjugation
at a carbonyl function.12–14
Figure 1
Structure of the investigational antimalarial arterolane (1) and an iron(II)-targeted
fragmenting hybrid (2).
Scheme 1
a) Mechanism of iron(II)-promoted breakdown of the 1,2,4-trioxolane ring in arterolane
(1). b) Proposed unraveling of a fragmenting hybrid (2) to release an amine-bearing
drug species via β-elimination from a retro-Michael substrate (6).
Conclusive demonstration of dual action for a hybrid drug species is challenging since
simple potency comparisons are often ambiguous and difficult to interpret. Thus, although
fragmenting hybrids of mefloquine and primaquine could be readily prepared, it was
not trivial to establish that the quinoline species had been released from these hybrids,
since the trioxolane moiety itself is a potent antimalarial agent. Therefore, we turned
to studies of fragmenting hybrids containing a protease inhibitor, reasoning that
delivery of such a species could be confirmed using activity-based probes of protease
activity. Dipeptidyl aminopeptidase 1 (DPAP1)15 is an essential cysteine protease
involved in the latter stages of hemoglobin degradation in the parasite DV. Inhibitors
of this exo-protease are well suited for study of fragmenting hybrids since relevant
activity-based probes are available and inhibitor potency is dramatically attenuated
when the N-terminal amino group is acylated or otherwise blocked.16, 17 Herein, we
demonstrate proof of principle for the fragmenting hybrid approach by demonstrating
the successful release of the potent and irreversible DPAP1 inhibitor ML4118S15 from
fragmenting hybrid 8 (Figure 2). We anticipated that ML4118S in its hybrid form 8
should possess very weak DPAP1 inhibitory activity since the linkage is made at the
terminal amino function. Therefore, the observation of DPAP1 inhibition by hybrid
8 in parasites would indicate successful delivery of active ML4118S from the hybrid.
We could measure the inhibitory activity in parasite extracts using FY01 (Figure 3
a), a fluorescently labeled activity-based probe that specifically targets DPAPs.18
Although not a viable drug candidate for various reasons, hybrid 8 has proven to be
a useful tool for validating the fragmenting hybrid concept in live parasites.
Figure 2
Structures of the carbamate-linked fragmenting hybrid 8 and an amide-linked congener
9. Compound 9 is an important control compound that can be activated by ferrous iron
but cannot release free ML4118S. Both 8 and 9 contain the irreversible DPAP1 inhibitor
ML4118S, the α-keto position of which is not configurationally stable.15 R= CH2CH2CH2-N-pyrrolidinone.
Figure 3
Validation of the fragmenting hybrid concept in live parasites. a) Activities of ML4118S
and 8, 9, 13 towards DPAP1. Parasite lysates were treated for 30 min in acetate buffer
with different concentrations of hybrid 8, trioxolane 13, ML4118S, or the control
hybrid 9. Residual DPAP1 activity was labeled with 1 μm of FY01 for 1 h and visualized
by fluorescent scan of SDS-PAGE gels. b) Potency against P. falciparum parasites in
culture: ML4118S (○); 8 (⧫); 9 (▪); 13 (▵). Ring stage parasites were treated with
increasing concentrations of the indicated compound and cultured for ∼75 h. Parasitemia
was quantified by FACS analysis and fitted to a dose–response curve. The EC50,Pot
values are reported in Table 1. c) Kinetics of DPAP1 inhibition in vivo. A synchronous
culture of P. falciparum at trophozoite stage was treated with 50 nm of the indicated
compound or DMSO. After 0.5–6 h of treatment, parasites were separated from the erythrocytes
by saponin lysis, and the residual DPAP1 activity was labeled with 1 μm of FY01 in
acetate buffer containing 1 % nonidet P40.
Compounds 8 and 9 were prepared from the key trioxolane intermediate 13, which was
in turn prepared using established methods of trioxolane synthesis.19, 20 Briefly,
Griesbaum co-ozonolysis of 2-adamantanone methyl oxime and cyclohexa-1,4-dione afforded
ketone 11, which following Baeyer–Villiger oxidation provided lactone 12 on a gram
scale (Scheme 2). Ring opening of 12 with a primary amine provided alcohol 13, and
this material could then be joined to ML4118S via a carbamate linkage to provide hybrid
8 in 38 % yield. Similar coupling reactions with the more reactive and less hindered
amines present in the antimalarial drugs primaquine and mefloquine proceeded in yields
of >80 % and >55 %, respectively (unpublished results). The ability to form fragmenting
hybrids from a variety of primary and secondary amines is significant because most
of the agents currently used in malaria combination therapies possess such functional
handles. Also prepared from 13 was the crucial amide-linked control compound 9, which
can be activated by ferrous iron but cannot release free ML4118S because β-elimination
is precluded. Hence, two-step oxidation of alcohol 13 to the corresponding carboxylic
acid (74 % overall yield), followed by HATU-mediated coupling to ML4118S, afforded
conjugate 9 in 48 % yield.
Scheme 2
Synthesis of fragmenting hybrid 8 and nonfragmenting control compound 9. Reagents
and conditions: a) MeONH2, pyridine, RT, 48 h, 83 %; b) O3, cyclohexa-1,4-dione, pentane/CH2Cl2
(3:2), 0 oC, 1 h, 33 %; c) mCPBA, NaHCO3, CH2Cl2, RT, 48 h, 56 %; d) N-(3-aminopropyl)-2-pyrrolidinone,
toluene, 50 oC, 5 h, 64 %; e) p-NO2PhOC(O)Cl, Et3N, DMAP, CH2Cl2, RT, 16 h, 96 %;
f) ML4118S, DMF, DMAP, RT, 16 h, 38 %; g) Dess–Martin periodinane, CH2Cl2, RT, 30 min;
h) 1 m KMnO4, 5 % NaH2PO4, tBuOH, RT, 30 min, 74 % (two steps); i) ML4118S, HATU,
HOBt, DMF, DIEA, RT, 2 h, 48 %. Abbreviations: 2-(7-Aza-1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium
hexafluorophosphate (HATU); meta-Chloroperoxybenzoic acid (mCPBA); N,N-Diisopropylethylamine
(DIEA); 4-Dimethylaminopyridine (DMAP); N,N-Dimethylformamide (DMF); 1-Hydroxybenzotriazole
(HOBt).
With hybrid 8 in hand, validation of the proposed iron(II)-promoted fragmentation
chemistry was undertaken, employing LC/MS for the detection of reaction products (Scheme 1).
Using reaction conditions that are standard in the field,21 0.3 mmol of hybrid 8 was
treated with 100 equivalents of ferrous bromide in a 1:1 mixture of water and acetonitrile.
These conditions afforded rapid fragmentation of the trioxolane ring and clean formation
of the expected products: lactone 14 and retro-Michael substrate 15 (Scheme 3). The
subsequent β-elimination reaction of 15 to generate 16 and ML4118S was also observed
and was slower than the initial reaction with iron (see Supporting Information). In
control experiments without ferrous bromide, hybrid 8 remained intact throughout the
experimental time period. These results thus confirm that trioxolane scission and
hybrid fragmentation proceeds as predicted, effecting iron-dependent release of ML4118S.
Scheme 3
Observed reaction products (LC/MS) following treatment of fragmenting hybrid 8 with
excess ferrous bromide. R= CH2CH2CH2-N-pyrrolidinone.
With the fragmentation chemistry validated in vitro, we turned to the study of 8,
9, 13, and ML4118S in the context of cultured intra-erythrocytic Plasmodium falciparum
parasites. We first measured inhibition of DPAP1 by treating parasite lysates for
30 min with increasing concentrations of test compounds followed by labeling of residual
DPAP1 activity with FY01 (Figure 3 a). Free ML4118S fully inhibited DPAP1 with an
IC50 value of 70 nm. As expected, neither trioxolane 13 alone nor the hybrid control
9 blocked DPAP1 activity, even at the highest concentration studied (10 μm). Thus,
the DPAP1 activity of ML4118S is blocked in the hybrid form (9). Hybrid 8 partially
inhibited DPAP1 at the highest concentration (10 μm), but was at least 100-fold less
potent than ML4118S itself, presumably due to capping of its free amine function in
the hybrid form (Table 1). An authentic sample of the linker side product 16 (see
scheme S1 in the Supporting Information for the synthesis) had no effect on parasite
viability or DPAP1 inhibition (Table 1; figure S3 in the Supporting Information).
Table 1
DPAP1 Inhibitory activities, antimalarial activities, and rates of hybrid fragmentation
in vitro and in parasite cultures.[a]
Compd
IC50
[b] [nm]
EC50,Pot
[c] [nm]
t
1/2 [h]
DPAP1
P. falciparum
in vitro[d]
in vivo[e]
ML4118S
70 (13)
5.2 (0.4)
n/a
n/a
8
10 000
4.0 (0.2)
9
1.5 (0.25)
9
>10 000
52 (7)
n/a
n/a
13
>10 000
29 (13)
n/a
n/a
16
>10 000
>10 000
n/a
n/a
[a] n/a: not applicable. The standard deviation for measured values is shown in parentheses.
[b] Half maximal inhibition of DPAP1 in parasite lysates after 30 min treatment with
inhibitor. DPAP1 activity was measured using the FY01 probe. [c] Antimalarial potency
measured by treating a culture of P. falciparum at ring stage with increasing concentrations
of compound. The decrease in parasitemia was quantified by FACS analysis and fitted
to a dose–response curve. [d] Half-life for the release of ML4118S from hybrid species
8 as measured in vitro by LC/MS spectrometry. [e] Half-life for the release of ML4118S
from hybrid 8 in living parasites. This value was estimated based on the kinetics
of DPAP1 inhibition observed in culture and the independently determined second-order
rate constant for inhibition of DPAP1 by ML4118S in vitro.
We next evaluated the antimalarial activities of the compounds using intra-erythrocytic
P. falciparum parasites (Figure 3 b and Table 1). Trioxolane intermediate 13 and hybrid
control 9 exhibited antimalarial activities in the mid-nanomolar range (EC50,Pot=29 nm
and 52 nm, respectively). The potent activity of 9 confirms action via the trioxolane
moiety, since no inhibition of DPAP1 is conferred by this compound in live parasites
(see below). The observation of trioxolane-based activity in 9 also rules out an alternative
decomposition mechanism involving acid-mediated Hock fragmentation. Hock fragmentation
would not produce cytotoxic carbon radicals, and thus the observation of potent activity
by 9 (and 13) suggests action via the canonical mechanism of trioxolane toxicity (Scheme 1).
Significantly, ML4118S and its hybrid form 8 were both active at single-digit nanomolar
concentrations (EC50,Pot=5.2 nm and 4.0 nm, respectively), approximately tenfold more
potent than 9 or 13. The enhanced potency of 8 relative to 9 suggests that active
ML4118S is indeed released from hybrid 8 within parasites; that hybrid 8 and ML4118S
have similar potencies suggests additive activity rather than synergistic or antagonistic
activity.
To further investigate the release of ML4118S from hybrid 8 in parasites, we measured
the kinetics of DPAP1 inhibition using the FY01 probe (Figure 3 c). No inhibition
of DPAP1 activity was observed upon exposure of parasites to compound 13 for 0.5–6 h,
which indicates that the toxic effects of the trioxolane moiety do not alter the levels
of DPAP1 activity for at least 6 h. As expected, ML4118S completely inhibited DPAP1
activity in parasites at all time points while control hybrid 9 was unable to inhibit
DPAP1 even after 6 h. Hybrid 8, on the other hand, inhibited DPAP1 in a time-dependent
fashion, with complete inhibition observed after 3 h. These results are consistent
with a slow release of ML4118S from fragmenting hybrid 8. Based on the inhibition
rate constant measured for ML4118S in vitro (k
i=10 200 m
−1 s−1; figure S2 a in the Supporting Information), we estimate the t
1/2 of ML4118S release from hybrid 8 in parasites to be ∼1.5 h (k
r=0.00013 s−1; figure S2 b in the Supporting Information), which is sufficiently rapid
to be useful in the context of antimalarial therapy.
In summary, we have shown that a prototypical fragmenting hybrid delivers multiple
antimalarial activities in a targeted fashion to the intra-erythrocytic P. falciparum
parasite. In principle, the slow release of a partner drug in such hybrids could complement
the rapid action of the trioxolane moiety, much as the ACT strategies seek to combine
a longer-acting partner drug with a rapid-acting artemisinin. As demonstrated with
hybrid 8, intrinsic bioactivities of the partner drug can be masked in the hybrid
form, raising the possibility that undesired on/off-target effects of known drugs
might similarly be attenuated using this approach. Antimalarial agents like primaquine
and amodiaquine that exhibit systemic toxicities might be more safely administered
in the form of a fragmenting hybrid, as this would limit systemic exposure to free
partner drug. More speculatively, agents conferring irreversible target inhibition
or polypharmacology might be safely delivered using a fragmenting hybrid approach
in which systemic exposure to free drug is avoided. Currently, we are exploring next-generation
fragmenting hybrids that overcome limitations of the initial prototypical systems
described herein.