Nitroheterocyclic drug resistance mechanisms in Trypanosoma brucei

Cell lines and culture conditions

T. brucei bloodstream-form ‘single marker’ S427 (T7RPOL TETR NEO) and drug-resistant cell lines were cultured at 37°C in HMI9-T medium¹² supplemented with 2.5 μg/mL G418 to maintain expression of T7 RNA polymerase and the tetracycline repressor protein. Cultures were initiated with 1 × 10⁵ cells/mL and subcultured when cell densities approached 1–2 × 10⁶ cells/mL.

In order to examine the effects of inhibitors on the growth of these parasites, triplicate cultures containing the inhibitor were seeded at 1 × 10⁵ trypanosomes/mL. Cell densities were determined after culture for 72 h as previously described.¹³ EC₅₀ values were determined using the following two-parameter equation by non-linear regression using GraFit:

where the experimental data were corrected for background cell density and expressed as a percentage of the uninhibited control cell density. In equation (1), [I] represents the inhibitor concentration and m is the slope factor.

Drugs

Nifurtimox used in this study was a kind gift provided by Bayer, Argentina. Fexinidazole was synthesized in-house as previously described.⁹

Generation of drug-resistant parasites

An NfxR line was generated by subculturing WT T. brucei in the continuous presence of nifurtimox.⁹ Starting at a sublethal concentration of 1.5 μM, the drug concentration in the culture media was increased in a stepwise manner, usually by 2-fold. After a total of 140 days in culture, when trypanosomes were able to survive and grow in 50 μM nifurtimox, the resulting T. brucei line (designated NfxR) was cloned by limiting dilution in the absence of nifurtimox. Three clones (NfxR1, NfxR2 and NfxR3) displaying the highest resistance to nifurtimox were selected for further studies. Resistance to fexinidazole was generated in T. brucei in a similar manner, starting at 1.0 μM fexinidazole. Fexinidazole-resistant (FxR) clones were derived from the resulting FxR line after 137 days in culture and clones FxR1, FxR2 and FxR3 were selected for further analysis.

Generation of knockout and overexpression constructs

The primers used are summarized in Table S1 (available as Supplementary data at JAC Online) and were designed using the T. brucei NTR sequence in TriTrypDB (Tb427.07.7230) as a template. The accuracy of all assembled constructs was verified by sequencing. NTR gene replacement cassettes were generated by amplifying a region of DNA encompassing the 5′UTR, ORF and 3′UTR of T. brucei NTR from genomic DNA with primers 5′UTR-NotI_F and 3′UTR-NotI_R, using Pfu polymerase. An endogenous NotI site present within the 3′UTR was silenced using primers 3′UTR_mut_G89C_F and 3′UTR_mut_G89C_R and the QuikChange Lightning Site-Directed Mutagenesis Kit (Stratagene) as per the manufacturer's guidelines. This mutated sequence was then used as a template for the amplification of the individual regions used in the assembly of replacement cassettes containing the selectable drug resistance genes puromycin N-acetyl transferase (PAC) and hygromycin phosphotransferase (HYG) as previously described.¹⁴ The gene replacement construct (PAC) of the T. brucei gene Tb927.1.1050 was generated in an identical manner. The gene replacement construct (HYG) of the T. brucei gene Tb927.7.7410, encompassing the 450 bp upstream and downstream of the ORF, was commercially synthesized (GeneArt, Invitrogen).

To generate an overexpression construct of the T. brucei hypothetical gene annotated Tb927.7.7410, the ORF was amplified from genomic DNA using sense and antisense primers and cloned into pCR-Blunt II-TOPO (Invitrogen). The pCR-Blunt II-TOPO-Tb927.7.7410 construct was then digested with HindIII and PacI and the resulting fragment ligated into the HindIII/PacI cloning site of a modified pLew82 tetracycline-inducible expression vector,^14,15 resulting in the introduction of a C-terminal triple haemagglutinin (HA) epitope tag into the expressed protein. Similarly, the NTR gene and a mutated version of the NTR gene, encoding a mutation of residue 106 from Val to Ile, was generated by site-directed mutagenesis with the NTR^MUT primers (Table S1) as described above. The mutated gene was then ligated into the HindIII/PacI cloning site of the original pLew82 tetracycline-inducible expression vector.¹⁵

Generation of transgenic cell lines

Mid-log bloodstream trypanosomes were transfected with either knockout or recovery constructs using the Human T-Cell Nucleofector kit and nucleofector (Amaxa, program X-001). Following transfection, cells were allowed to grow for 16–24 h prior to appropriate drug selection (hygromycin and puromycin at 4 and 0.1 μg/mL, respectively). Cloned cell lines were generated by limiting dilution, maintained in selective medium and removed from drug selection for one passage prior to experiments.

Verification of single gene knockouts (SKO) by PCR

PCR primer pairs 5′UTR_−595_F and 3′UTR_+600_R (outer primers) and 5′UTR_−107_F and 3′UTR_+152_R (inner primers) (Table S1) were used to verify the removal of a single allelic copy of NTR from bloodstream parasites. Genomic DNA from putative NTR^SKO cell lines was prepared and PCR was carried out using Pfu polymerase. Similar confirmatory studies were carried out with primer pairs 5′UTR_+500 and HYG R for Tb927.7.7410^SKO cell lines (Table S1).

Southern blot analyses of transgenic T. brucei cell lines

The 5′UTR of T. brucei Tb927.1.1050 was amplified by PCR [using the primers previously described for the cloning of the single knockout (SKO) construct] and the PCR DIG Probe Synthesis Kit (Roche). The resulting DIG-labelled product was used as a probe. Samples of genomic DNA (5 μg) from WT and transgenic cell lines were digested with appropriate restriction endonucleases; then, the digestion products were separated on a 0.8% agarose gel and transferred onto a positively charged nylon membrane (Roche). The membrane was hybridized overnight in DIG Easy Hyb solution (Roche) at 42°C with the DIG-labelled 5′UTR probe (2 μL of PCR product). Following hybridization, membranes were washed twice in low-stringency conditions (25°C, 5 min, 1× SSC with 0.1% SDS) and twice in high-stringency conditions (68°C, 15 min, 0.5× SSC with 0.1% SDS), where 1× SSC comprises 150 mM sodium chloride and 50 mM sodium citrate (pH 7.0). Bound probe was detected using the DIG immunological detection kit (Roche) as per the manufacturer's instructions.

Western blot analysis of T. brucei cell lysates

Overexpression of the protein encoded by the gene Tb927.7.7410 was induced by the addition of tetracycline (2 μg/mL) for 72 h prior to analysis by western blotting. T. brucei whole-cell extracts (1 × 10⁷ parasites per lane) were separated by SDS–PAGE and subsequently transferred onto nitrocellulose. After blocking with 5% skimmed milk in PBS for 1 h, blots were incubated with anti-HA mouse monoclonal antibody (Roche, 1 : 500), washed in PBS containing 0.1% (v/v) Tween 20 and then incubated with a secondary antibody (horseradish peroxidase-conjugated antimouse IgG, R&D Systems, Minneapolis, MN, USA; 1 : 5000 dilution). Immunoblots were developed using the ECL Plus (enhanced chemiluminescence) system from GE Healthcare (Piscataway, NJ, USA).

Quantitative PCR (qPCR)

Levels of NTR transcript in resistant and WT parasites were analysed by qPCR using telomerase reverse transcriptase (TERT, Tb11.01.1950) as a reference gene. RNA was prepared from ∼1 × 10⁷ parasites using the RNeasy Kit (Qiagen) as per the manufacturer's instructions. cDNA was then generated from these samples using the iScript cDNA Synthesis Kit (Bio-Rad). All qPCR reactions were performed with PerfeCTa SYBR Green FastMix for iQ using the primer sets described in Table S1 and a Bio-Rad CFX96 Real-Time PCR Detection System. All statistical analyses are unpaired, two-tailed t-tests with a 95% CI and were performed using GraphPad Prism software.

WGS and genomic variant analysis

Genomic DNA was prepared from T. brucei clones NR-B_427 (NfxR1), NR-D_427 (NfxR2), NR-G_427 (NfxR3), FR-A_427 (FxR1), FR-C_427 (FxR2) and FR-E_427 (FxR3). For each sample, 5 μg of genomic DNA was used to produce amplification-free Illumina libraries with fragment sizes of 350–450 bp.¹⁶ Sequencing was carried out on an Illumina HiSeq 2000 sequencer according to the manufacturer's standard sequencing protocol and yielded 33.6–43.8 million reads of 100 bp length per library for the drug-resistant lines. To allow direct comparison of genetic variant calls from these data with variant calls from shorter reads of an Illumina sequencing run of the T. brucei reference strain Lister 427 (lane 5189_8, 67.8 million reads of 76 bp length), the 100 bp reads were hard clipped to a length of 76 bp. The resulting datasets represented a nominal sequencing coverage of the 35 Mb T. brucei genome of ∼73- to 95-fold (drug-resistant strains) and 147.2-fold (Lister 427). The Illumina data were aligned against the T. b. brucei TREU927 reference genome assembly¹⁷ using SMALT v0.7.4 (http://www.sanger.ac.uk/resources/software/smalt/). For variant calling, the alignment was run employing an exhaustive search (−x) and with parameters wordlen = 13 (−k), skipstep = 1 (−s), minscor = 0.65 (−m) and insertmax = 1000 (−i). To assess relative read coverage and copy number variations, the alignment runs were repeated using the same parameters but additionally with repetitive mapping (−r) enabled. Variants were called using SAMtools v0.1.19 mpileup (−Q 15 for baseQ/BAQ filtering) and BCFtools.¹⁸ To exclude the hypervariable subtelomeric regions, only variants found in the following chromosomal core regions were included in the downstream analyses: Tb927_01_v4: 202 695–988 120; Tb927_02_v4: 259 723–1 161 408; Tb927_03_v4: 146 614–1 602 829; Tb927_04_v4: 80 380–1 467 268; Tb927_05_v4: 72 088–1 366 595; Tb927_06_v4: 111 409–1 414 033; Tb927_07_v4: 26 571–2 177 541; Tb927_08_v4: 135 192–2 476 033; Tb927_09_v4: 325 850–2 394 987; Tb927_10_v5: 55 698–3 993 940; and Tb927_11_01_v4: 36 585–4 482 610. SNP calls were further filtered for all of the following: for a minimum of eight ‘high-quality’ base calls (‘DP4’); for a minimum phred-scale QUAL score of 20; for a maximum phred-scale likelihood of the best genotype call of 5 (‘PL1’); for a minimum phred-scale likelihood of the second best genotype call of 10 (‘PL2’); for a minimum strand bias P value of 0.01 (first of ‘PV4’); for a maximum ratio of conflicting base calls for homozygous genotypes of 5% and, for positions with a minimum and maximum read depth of three times the median read depth observed for that chromosome: for a minimum mapping quality of 20 and for a minimum distance of 10 nucleotides from the nearest INDEL call. Files in Variant Call Format (VCF) listing all 190 064 genomic positions at which any one of the six drug-resistant and the parental parasite lines had a variant call are available at http://dx.doi.org/10.15132/10000107. The raw sequence data are available under the following accession numbers at the European Nucleotide Archive (http://www.ebi.ac.uk/ena): Tb427 WT (SM_S427): ERS012470; NR-B_427: ERS017528; NR-D_427: ERS017529; NR-G_427: ERS017530; FR-A_427: ERS017531; FR-C_427: ERS017532; and FR-E_427: ERS017533.

NfxR and FxR trypanosomes

In our previous studies we investigated the ease with which resistance to nifurtimox and fexinidazole could be generated in vitro in T. brucei (S427) bloodstream trypanosomes.⁹ In our current study, resistance was readily achievable with the resulting NfxR (NfxR1–2) and FxR (FxR1–2) clones being ∼6- and ∼11-fold resistant to nifurtimox and fexinidazole, respectively, compared with WT (Table 1). NfxR trypanosomes demonstrated cross-resistance in vitro and in vivo to fexinidazole, with FxR parasites similarly cross-resistant (Table 1). The EC₅₀ values determined for these cell lines in our current study are in good agreement with previously published data.⁹ However, in this study, the Hill slope values for the FxR cell lines (1.2–1.5) are half as steep in comparison with the WT parasite (2.6) and its genetically manipulated derivatives. This feature was not evident in our previous study (e.g. Hill slopes of 1.6 versus 1.4 for WT and FxR parasites, respectively) and is unique to FxR parasites. The significance of this observation is not clear, but may suggest some differences in the mechanisms of action of fexinidazole and nifurtimox. The reciprocal cross-resistance relationship observed here with NfxR and FxR parasites strongly supports our hypothesis that a predominantly similar mechanism of action is involved for these two compounds.⁹ However, it is notable that the level of resistance to fexinidazole in the selected lines is always greater than the resistance to nifurtimox, regardless of which drug was used in the selection. Here, the resistance mechanisms exploited by both resistant lines are studied.

Role of NTR in NfxR and FxR drug resistance

Bioactivation of nifurtimox and fexinidazole is catalysed by type I NTR enzymes in both T. brucei and Leishmania.^3,7 The potential role of this enzyme in the nitro drug resistance mechanisms of NfxR and FxR trypanosomes was investigated. To determine whether any mutations within the NTR genes of these cell lines could account for nitro drug resistance, the ORF of TbNTR, flanked by 100 bp at the 5′-terminus and 150 bp at the 3′-terminus, was amplified by PCR from the genomic DNA of each cell line and analysed by in-house sequencing in Dundee. The NTR gene in all NfxR clones was found to contain a point mutation that encoded a single amino acid change (V106I). This mutation was not present in any of the FxR clones. To assess the potential impact of this amino acid change on the activity of NTR, a tetracycline-inducible ectopic copy of the mutant gene was introduced into WT cells. All tetracycline-induced clones, derived from the resulting TbNTR_V106I-overexpressing T. brucei cell line, were found to be ∼9-fold more susceptible to nifurtimox than the parental WT cells with a mean EC₅₀ value 0.38 ± 0.01 μM (Table 1). This is very similar to the 10.6-fold increase in susceptibility seen with cells overexpressing WT NTR (EC₅₀ value 0.32 ± 0.01 μM). Collectively, these data indicate that mutant TbNTR_V106I is capable of activating nifurtimox in bloodstream trypanosomes and that this mutation is unlikely to play any significant role in nitro drug resistance.

An initial Southern analysis of genomic DNA from the nitro drug-resistant trypanosomes appeared to suggest that a single allele of NTR has been lost in both NfxR1 and NfxR2 cell lines. However, FxR clones appeared to maintain both NTR alleles (Figure S1). The impact of a reduction in the NTR copy number on nitro drug susceptibility was investigated by generating NTR SKO cell lines. A single copy of the NTR gene was replaced with either the hygromycin resistance gene (HYG) or the puromycin resistance gene (PAC) by homologous recombination and drug selection. The successful removal of single copies of the NTR gene from these diploid parasites was confirmed by PCR (Figure 1). The resulting SKO^HYG and SKO^PAC cell lines showed no major growth phenotype with doubling times of ∼8 h, similar to the 7.3 h doubling time of WT parasites. However, SKO^HYG parasites were found to be 1.7- and 1.9-fold less susceptible than WT T. brucei to nifurtimox and fexinidazole, respectively (Table 1). Similarly, SKO^PAC cells were 1.9-fold less susceptible to nifurtimox. Given that NfxR cells demonstrate a ∼6-fold shift in their susceptibility to nifurtimox, it is clear from these data that the loss of a single allele of NTR may not be entirely responsible for the levels of nitro drug resistance seen with these cells and additional factors may be involved.

WGS of NfxR and FxR cell lines reveals large-scale losses of single alleles and numerous genetic variants

In an attempt to identify any additional factors that may be involved in the drug resistance phenotypes of NfxR and FxR cell lines, the genomes of each clone were sequenced. The sequence data readily confirmed the loss of one copy of the NTR locus for the NfxR parasites, rendering a region of >100 kb hemizygous (Figure 2a). This includes 32 putative genes, including a putative oxidoreductase (Tb927.7.7410) (Table S2). For the FxR parasites, the 3′-end of the chromosomal core region of chromosome 7 was likewise found to be present in only a single copy, but here the hemizygous region started ∼20 kb downstream of the NTR locus (Figure 2a) with hemizygosity for 22 genes (Table S2). In addition, the sequencing data revealed another 5.8 kb region on chromosome 1, encompassing Tb927.1.1050, that has been lost solely in the NfxR cell lines (Figure 2b). Comparing the genotype calls of the drug-resistant lines with those of the parental strain also uncovered many dozens of SNPs that are predicted to result in non-synonymous amino acid changes that could potentially contribute to the observed drug resistance phenotypes (Table S3). Unfortunately, because of the repetitive nature of the T. brucei genome, which contains many duplicated genes, such lists of genomic variant candidates are often too long to be experimentally verified one by one. Below, we report the experimental follow-up of the most prominent genetic variants.

Does the putative oxidoreductase (Tb927.7.7410) play a role in nitro drug resistance?

Loss of genetic material from a single copy of chromosome 7 in NfxR and FxR parasites resulted in the deletion of a putative oxidoreductase. Since this class of enzymes has the potential to catalyse the reduction of nitro drugs, the enzyme's role in the activation of fexinidazole and nifurtimox was investigated. A triple HA epitope-tagged version of the putative oxidoreductase, encoded by gene Tb927.7.7410, was overexpressed in bloodstream trypanosomes. Expression of the putative oxidoreductase in these transgenic parasites was confirmed by western blotting using an anti-HA mouse monoclonal antibody (Figure 3a). An extra band at 42 kDa is visible in the overexpressing lines consistent with the predicted molecular mass of the tagged protein. Two clonal cell lines expressing the tagged oxidoreductase were then assessed for their susceptibilities to both nifurtimox and fexinidazole; these clones were found to be equally susceptible to the nitroaromatic compounds as WT parasites (Table 1). A proteomic study of the mitochondria of procyclic T. brucei¹⁹ established that this oxidoreductase may be associated with subcomplex Iα of the respiratory chain. However, preliminary immunofluorescence studies suggested that the HA-tagged enzyme was predominately localized to the cytosol of bloodstream trypanosomes. Since we cannot confirm whether or not our overexpressed, tagged oxidoreductase is correctly localized in the cell, or indeed functional, additional studies were carried out. A single copy of Tb927.7.7410 was replaced in WT trypanosomes (Tb927.7.7410^SKO) and also in NTR^SKO (NTR/Tb927.7.7410^SKO) parasites, thus mimicking the genetic composition of NfxR and FxR parasites (Figure 3b and c). Removal of a single copy of this gene was found to have no impact on the nitro drug susceptibility of either WT parasites or in cells lacking a single copy of NTR (Figure 3d). Collectively, these data do not support a role for this putative oxidoreductase in the mechanism of action of, or resistance to, fexinidazole and nifurtimox in T. brucei.

Hypothetical protein Tb927.1.1050

Genome-wide sequencing also revealed that one allelic copy of the gene Tb927.1.1050, which encodes a putative metallo-dependent phosphatase, was lost from the NfxR cloned cell lines. Changes to this uncharacterized gene drew our attention since it had previously been identified in genome-scale RNA interference target sequencing (RIT-seq) screens of nifurtimox in T. brucei.²⁰ To investigate any potential role in nifurtimox resistance, a single copy of Tb927.1.1050 was replaced in WT trypanosomes (Tb927.1.1050^SKO) and also in NTR^SKO parasites (Figure 4). Removal of a single copy of this gene was found to have absolutely no impact on the nitro drug susceptibility of either WT parasites or in cells lacking a single copy of NTR (Table 1). It is perplexing that this gene has been implicated in nitro drug resistance in both RIT-seq and WGS studies; however, our data do not support a role for this putative metallo-dependent phosphatase in the mechanism of action of, or resistance to, nifurtimox in T. brucei.

NTR transcript levels in NfxR and FxR cell lines

Confirming our preliminary Southern analysis (Figure S1), genome-wide sequencing revealed that both NfxR clones had lost a single copy of NTR while FxR clones maintained both copies. However, a large section of chromosome 7 downstream of NTR appeared to have been deleted from single alleles of both the NfxR and FxR cell lines. This includes a divergent strand-switch region, i.e. a probable transcription start site for RNA polymerase II,²¹ between Tb927.7.7480 (a putative trans-sialidase) and Tb927.7.7490 (a conserved hypothetical protein). The impact of this deletion on the transcript levels of NTR was determined by qRT-PCR. Relative levels of NTR cDNA in WT, NfxR1 and FxR1 cell lines were normalized using the reference housekeeping gene telomerase reverse transcriptase (TERT). NTR transcript levels in NfxR1 and FxR1 clones were reduced to 0.42 ± 0.06 and 0.6 ± 0.07 of the levels seen in the WT parasites, respectively. The ∼50% reduction in the levels of NTR mRNA is consistent with the loss of a single allele of NTR from NfxR cells, already demonstrated by genome sequencing. Further, it suggests that the loss of genetic material downstream of NTR leaves FxR cells unable to transcribe one allele of NTR and that these cells are functionally hemizygous for NTR despite maintaining both allelic copies of this gene.