Veterinary trypanocidal benzoxaboroles are peptidase-activated prodrugs

Abstract

Livestock diseases caused by Trypanosoma congolense, T. vivax and T. brucei, collectively known as nagana, are responsible for billions of dollars in lost food production annually. There is an urgent need for novel therapeutics. Encouragingly, promising antitrypanosomal benzoxaboroles are under veterinary development. Here, we show that the most efficacious subclass of these compounds are prodrugs activated by trypanosome serine carboxypeptidases (CBPs). Drug-resistance to a development candidate, AN11736, emerged readily in T. brucei, due to partial deletion within the locus containing three tandem copies of the CBP genes. T. congolense parasites, which possess a larger array of related CBPs, also developed resistance to AN11736 through deletion within the locus. A genome-scale screen in T. brucei confirmed CBP loss-of-function as the primary mechanism of resistance and CRISPR-Cas9 editing proved that partial deletion within the locus was sufficient to confer resistance. CBP re-expression in either T. brucei or T. congolense AN11736-resistant lines restored drug-susceptibility. CBPs act by cleaving the benzoxaborole AN11736 to a carboxylic acid derivative, revealing a prodrug activation mechanism. Loss of CBP activity results in massive reduction in net uptake of AN11736, indicating that entry is facilitated by the concentration gradient created by prodrug metabolism.

Fig 1. AN11736 resistance selection in T. brucei and T. congolense and cross-resistance phenotype of the T. congolense AN11736-resistant clones, revealing a common chemical feature.

(A) In vitro selection of resistance to AN11736 in independent T. brucei lines TbOX^R_A and TbOX^R_C (left) and resistance levels of two resistant clones obtained from lines TbOX^R_A (EC₅₀ 58 nM) and line TbOX^R_C (EC₅₀ 97 nM) as compared to the wild type line (TbWT, 0.3 nM) (right). (B) Stepwise in vitro selection of resistance to AN11736 in independent T. congolense lines TcoOX^R_B and TcoOX^R_C (left) and resistance levels of two resistant clones obtained from lines TcoOX^R_B (EC₅₀ 15 nM) and line TcoOX^R_C (EC₅₀ 54 nM) as compared to the wild type parent line TcoWT (EC₅₀ 0.3 nM) (right). (C) Cross-resistance of the T. congolense AN11736-resistant clones to other benzoxaboroles revealed the presence of a peptide-bond linker (highlighted in blue) in the highly cross-resistant compounds (>20-fold), whereas the same chemical feature was absent in non cross-resistant compounds (< 2-fold). See S2 Table for full data. Values in (A), (B) (right panels) represent means ± SEM of n ≥ 4 (A) or n = 3 (B) independent biological replicates, with data in (B) each generated from two technical replicates.

Fig 2. Serine carboxypeptidases are deleted in T. brucei and T. congolense resistant to AN11736.

(A) Coverage of whole genome sequencing data for wild type (TbWT) and AN11736-resistant (TbOX^R_A, TbOX^R_B) T. brucei at the genomic locus of CBP1 gene copies. The number of fragments mapping to 100 bp windows is shown as individual points, with a moving average of nine windows shown as a line plot. Coverage was normalised for depth of sequencing as described in Materials and Methods. The position of genes within this region is shown as blocks above the plot, with CBP1 genes in blue and other genes in black. (B) Coverage plot as in (A) but with T. congolense wild type (both parent, TcoWT, and high passage, TcoWT_HP) and AN11736-resistant lines (TcoOX^R_A, TcoOX^R_B).

Fig 3. Serine carboxypeptidase disruption confers resistance to AN11736.

(A) Genome-wide RIT-seq screen in T. brucei identified the CBP1 genes on chromosome 10 as ‘hits’ for AN11736 resistance. In the lower panel, blue and red peaks are forward and reverse barcoded reads from individual RNAi target fragments; grey, all other reads. (B) Targeted CBP1 RNAi knockdown of the three serine carboxypeptidases in T. brucei conferred resistance to AN11736 under tetracycline induction (Tet+ EC₅₀ 7.7 nM, Tet- EC₅₀ 0.3 nM). (C) Cumulative cell growth in the presence of 10 nM AN11736 (arrows indicate addition of fresh drug) for uninduced T. brucei and two independent induced Cas9/sgRNA^CBP1A-C clones, following 24 h of Cas9 induced editing in the latter case. (D) A PCR assay of the Cas9/sgRNA^CBP1A-C clones revealed TbCBP1 editing in two independent drug-resistant clones (upper panel, duplicate samples are shown); the locus from the uninduced line yielded a product of approximately 5.7 kbp, while both clones yielded a product of approximately 2 kbp. B (blank), no genomic DNA; control, uninduced line genomic DNA; predicted wild type CBP1 fragment size, 5,696 bp. The maps in the lower panel indicate the edited loci, as determined by sequencing. The small red bars indicate the gRNA target-sites; the red arrowheads indicate the primers used for the PCR-assay; blue indicates >99% identical regions among multiple paralogues; grey, unique to Tb927.10.1050; green, unique to Tb927.10.1030. (E) Dose-response curves for AN11736 of the two CRISPR-Cas9 edited clones analysed, both displaying a drug-resistant phenotype: when CBP1 function was disrupted by Cas9 editing, T. brucei became, on average, 250-fold more resistant to AN11736. Data in (B), (E) represent means ± SD of n = 3 independent biological replicates.

Fig 4. Re-expression and heterologous expression of serine carboxypeptidases re-sensitise AN11736-resistant trypanosomes.

(A) Re-expression of TbCBP1B (Tb927.10.1040) in resistant line TbOX^R_A partially re-established sensitivity to AN11736 (EC₅₀ 1.6 nM), except when serine 179 was replaced with alanine in the catalytic triad (TbOX^R_A + TbCBP1B (S179A), EC₅₀ 65 nM). (B) Add-back of TcoCBP1H and, to a lesser extent, TcoCBP1A partially restored sensitivity to AN11736 in the resistant line TcoOX^R_C (EC₅₀ 5.7 nM and EC₅₀ 11.7 nM respectively, TcoOX^R_C EC₅₀ 16.3 nM). (C) Heterologous expression of TbCBP1B (EC₅₀ 6.5 nM) partially restored sensitivity to AN11736 in resistant TcoOX^R_C line (EC₅₀ 16.3 nM) compared to TcoWT (EC₅₀ 1.2 nM). (D) Expression of TvCBP1 (EC₅₀ 1.4 nM) partially restored sensitivity to AN11736 in resistant TbOX^R_A line (EC₅₀ 70.7 nM) compared to TbWT (0.46 nM). (E) Expression of TcoCPB1H (EC₅₀ 0.65 nM) but not TcoCBP1A (EC₅₀ 75.1 nM) restored sensitivity to AN11736 in resistant TbOX^R_A line (EC₅₀ 70.7 nM) compared to TbWT (0.46 nM). Data represent means ± SD of n = 3 independent biological replicates.

Fig 5. AN11736 is cleaved at the ester bond to a carboxylate derivative, which accumulates at high levels in wild type, but not resistant T. brucei or T. congolense.

(A) LC-MS analysis revealed presence of AN11736 in TbWT and resistant lines TbOX^R_A and TbOX^R_C after 6 h of incubation. (B) A peak of m/z 292.1347 was detected in positive mode in TbWT cells after 6 h of incubation but was barely detectable in resistant cell lines TbOX^R_A and TbOX^R_C. The peak was identified as the AN11736 carboxylate derivative AN14667, whose chemical structure is shown above the graph. (C) AN11736 was detected in TcoWT, the resistant lines TcoOX^R_C, TcoOX^R_B and the CBP1H-complemented TcoOX^R_C line after 6 h of incubation. (D) The peak of m/z 292.1347 was detected in positive mode in TcoWT cells after 6 h of incubation and at around half the intensity for re-sensitised CBP1H add-back line TcoOX^R_C + CPB1H, while the peak was detected at very low levels in resistant cell lines TcoOX^R_C and TcoOX^R_B, with intensities even lower than that measured for TcoWT at 0 h. (E) Mass spectrometry quantification of AN11736 (full bars) and the metabolite AN14667 (empty bars) in T. brucei wild type (black) and AN11736 resistant line TbOX^R_A (blue) over a period of 6 h. (F) as in (E) but for T. congolense wild type and AN11736 resistant line TcoOX^R_C. Data represent means ± SD of n = 3 (A-D) or n = 2 (E, F) independent biological replicates.