Abstract
Trypanosoma cruzi, the agent of Chagas disease, probably infects tens of millions of people, primarily in Latin America, causing morbidity and mortality. The options for treatment and prevention of Chagas disease are limited and underutilized. Here we describe the discovery of a series of benzoxaborole compounds with nanomolar activity against extra- and intracellular stages of T. cruzi. Leveraging both ongoing drug discovery efforts in related kinetoplastids, and the exceptional models for rapid drug screening and optimization in T. cruzi, we have identified the prodrug AN15368 that is activated by parasite carboxypeptidases to yield a compound that targets the messenger RNA processing pathway in T. cruzi. AN15368 was found to be active in vitro and in vivo against a range of genetically distinct T. cruzi lineages and was uniformly curative in non-human primates (NHPs) with long-term naturally acquired infections. Treatment in NHPs also revealed no detectable acute toxicity or long-term health or reproductive impact. Thus, AN15368 is an extensively validated and apparently safe, clinically ready candidate with promising potential for prevention and treatment of Chagas disease.
Main
Chagas disease, caused by the protozoan parasite Trypanosoma cruzi, remains the highest-impact parasitic disease in Latin America and one of the major causes of infection-induced myocarditis worldwide1. For more than five decades, two nitroheterocyclic compounds, benznidazole and nifurtimox, have been available for treatment of the infection, but are relatively rarely used due to their inconsistent efficacy and high frequency of side effects. Recent trials of potential new therapies have yielded disappointing results2,3.
Among the challenges for drug development in T. cruzi infection is the parasite’s predominantly intracytoplasmic location in mammals and its ability to invade a wide variety of host cell types and tissues, although it shows a clear preference for muscle cells, including cardiac, skeletal and smooth muscle of the gut. The recent discovery of arrested ‘dormant’ intracellular forms of T. cruzi that are relatively and transiently resistant to otherwise highly effective trypanocidal compounds4 may partially explain why these therapeutics must be given for extended periods of time (60 d is common) but still have a high failure rate.
Previous work from our laboratories have identified a novel class of boron-containing molecules, the benzoxaboroles5,6 as having potent activity against protozoans including Trypanosoma brucei7, Leishmania donovani8 and Plasmodium falciparum9. Screening of the Anacor benzoxaborole compound library against T. cruzi revealed several hits, but initial assessment of structure-activity relationships (SARs) suggested limited opportunity for improvement of potency and/or selectivity, particularly in those subclasses previously found to have activity against T. brucei and L. donovani. Here we have taken advantage of this benzoxaborole scaffold and the multiple natural host species for T. cruzi to move rapidly from in vitro detection of trypanocidal activity in lead compounds into facile in vivo tests of efficacy in mice and ultimately in naturally infected non-human primates (NHPs). The result is identification of a class of benzoxaboroles that provide high rates of parasitological cure of T. cruzi infection. AN15368 from this class is the first, extensively validated and safe potential clinical candidate in over 50 yr for the prevention/treatment of Chagas disease.
Results
In vitro activity and SARs
The initial lead benzoxaborole 6-carboxamide AN4169 (Fig. 1) provided 100% cure of mice infected with the T. cruzi Brazil strain5; however, rodent tolerability studies suggested that an insufficient therapeutic margin existed for further progression of this compound. Profiting from a concurrent project evaluating analogues of AN4169 against Trypanosoma congolense10, several compounds with submicromolar activity against T. cruzi in vitro and good metabolic stability in an in vitro mouse S9 liver fraction assay were identified, among these an ester of the 6-valine ‘transposed’ carboxamide AN10443 (Fig. 1). Further manipulation of this compound, in particular inclusion of a methyl group at C(7) of the benzoxaborole ring as in AN11735, drastically increased in vitro activity against T. cruzi (IC50 < 10 nM), whereas substituents larger than methyl at C(7) ablated activity (Extended Data Fig. 1a).
SAR development of the ester region in the C(7)-methyl series showed that potency was not substantially impacted by substituents on the benzyl ester, with the exception of the 4-(3-pyrrolidinylethyloxy) analogue AN14502 and the 4-methylsulfonyl analogue AN14561, which were markedly less potent (Supplementary Table 1). Physicochemical properties were more affected, with most simple halogenated analogues being poorly soluble in aqueous media. Metabolic stability, as estimated by incubation with the mouse S9 liver fraction, was variable, and roughly tracked with lipophilicity (cLogD; Supplementary Table 1).
These observations prompted us to more substantially modify the ester region of the molecule through preparation of aliphatic and heterocyclic esters that would be expected to be less lipophilic, more water soluble and less susceptible to metabolism (Supplementary Table 2). Several interesting SARs emerged from this group of analogues: (1) esters containing basic amines (for example, AN15143, AN15144, AN15658, AN15678, AN15129, AN15192, AN15078, AN14504 and AN15159) were less active than neutral compounds and (2) small aliphatic esters were quite potent except for the t-butyl ester (AN15134). The relationship between lipophilicity and solubility or metabolic stability continued to exist for these compounds and provided reasonably wide latitude for modulation of such properties by choice of ester substituent.
In vivo activities
In addition to being very potent in vitro, the valine esters were also of generally good stability, including in mouse and human S9 liver fraction assays (Supplementary Table 3). Several valine esters also exhibited low clearance (<20% hepatic blood flow) following intravenous dosing and good bioavailability following oral administration to mice, achieving good to excellent exposure (area-under-the-curve (AUC) > 10 µg h ml−1) with low mg kg−1 doses (Supplementary Table 3). Concurrent testing results of these compounds in vivo for the ability of a single oral dose to reduce an established focal infection in the footpad of mice over 3 d11,12 were very encouraging, with AN14353 emerging as the lead on the basis of activity at reduced doses (Fig. 1b). Notably, AN14353 demonstrated rapid in vivo trypanocidal activity (Fig. 1c), had high in vitro potency for a range of T. cruzi isolates for different genetic lineages (discrete typing units, DTUs; Extended Data Fig. 1b) and could consistently resolve established T. cruzi infections at a dose of 25 mg kg−1 in a standard5,13 40 d treatment protocol in wild-type mice (Fig. 1d) as well as infections in immunodeficient mice (Fig. 1e).
Lead benzoxaboroles are prodrugs activated by a parasite serine carboxypeptidase
We next sought to understand the essentiality of the ester group for activity, as this functional group carries liabilities for hydrolytic and metabolic instability. An early indication of the importance of the ester function to anti-T. cruzi activity was evident from the already noted lack of potency of the t-butyl ester AN15134. Amide, N-methyl amide, ketone, ether and acylsulfonamide analogues of AN11735, as well the 1,2,4-oxadiazole ester bioisostere AN14562 all lacked activity (Supplementary Table 4). Furthermore, the expected carboxylic acid metabolite, AN14667, had ~1,000-fold reduced activity on both intra- and extracellular amastigotes (Figs. 1a and 2a). Thus, the ester functionality is absolutely essential for anti-T. cruzi activity….
