American trypanosomiasis, or Chagas disease, is a neglected parasitic illness widely spread throughout the Americas, from the Southern United States to Argentina and Chile. Trypanosoma cruzi currently infects at least 7,694,500 individuals and between 60 and 80 million remain at risk of T. cruzi infection in endemic countries 1 2 . There is no vaccine to prevent the infection and chemotherapy is restricted to two nitroheterocyclic compounds: nifurtimox (NFX (4[(5-nitrofurfurylidene)amino]-3-methylthiomorpholine-1,1-dioxide) and benznidazole (BZ (N-benzyl-2-nitroimidazole-1-acetamide) 3 4 .

BZ is a nitroheterocyclic compound that contains a nitro group linked to an imidazole ring. As a pro-drug, BZ undergoes activation by enzymatic activity to have cytotoxic effects within the parasite, which is catalyzed by nitroreductases (NTRs) 5 . Because there are two possible enzymes acting on it, there are two proposed hypotheses for its toxic action. The first one postulates the generation of reactive oxygen species (ROS) following a one-electron reduction caused by NTR type II enzymatic activity. Under aerobic conditions, this induces the production of superoxide anions and causes re-cycling of the drug 6 . However, it is now known that T. cruzi possesses both enzymatic and non-enzymatic antioxidant defenses, making it unlikely that ROS production affects the viability of the parasite, at least at the doses used to treat the disease 7 8 9 10 11 . The second hypothesis proposes a two-electron reduction of the drug by NTR type I. This reaction goes through a nitroso species, to a hydroxylamine derivative using NADPH as a source of electron donors. Hydroxylamine can react to produce a nitrenium cation, which induces DNA strand breaks. Moreover, the high electrophilic intermediaries may affect other molecules within the cell 12 . There are two trypanosomal enzymes with this type of activity: prostaglandin F2α synthase or old yellow enzyme (OYE), which mediates two-electron reduction in NFX under anaerobic conditions 13 and nitroreductase I (NTR). This second enzyme has already shown strong experimental evidence associated with cross-resistance to NFX and BZ 5 12 14 15 . Despite the great efforts made to understand BZ's mode of action, it is not yet completely clear; even less thoroughly studied are the initially acquired mechanisms of resistance.

Acquisition of drug-resistant phenotypes is a complex physiological process based on single or multiple changes of genes involved in its mechanisms of action 16 17 18 . Many studies have been based on differential gene expression analysis in high concentrations of BZ in resistant phenotypes 19 20 21 22 23 24 25 26 , but the commencement of this condition remains to be understood. For this reason, the identification of genes that are differentially expressed during progression through the susceptible drug population to a resistant phenotype in T. cruzi populations may help to further our grasp of acquired stable resistance mechanisms, including the basis of the drug's mode of action. Additionally, it is also important to identify gene expression and/or genetic alterations as markers of either sensitivity or resistance responses, which could be useful for treatment prognosis and as potential new therapeutic targets. Therefore, the aim of this study was to examine the initial expression changes of parasites submitted to a stepwise concentration of BZ. For this reason, we chose 19 genes suggested to be involved in escaping BZ cytotoxic effects. Their RNA expression was quantified by real-time PCR (RT-qPCR). We also investigated other issues concerning drug resistance, such as the stability of BZ-resistant phenotypes without drug pressure over a long period of time as well as the cross-resistance with other nitroheterocyclic and non-nitroheterocyclic drugs.

Chagas disease is a public health problem in many tropical countries resulting from the lack of effective treatments and vaccines 37 . Two drugs are available to treat the disease, NFX and BZ, which have been previously reported as being impaired treatments with 70% of patients considered recovered in the acute stage and only 20% in the chronic stage 37 . Additionally, use of these drugs is usually followed by the appearance of parasite resistance 5 . Consequently, this study focuses on identifying markers of BZ resistance at different drug concentrations.

In recent years, many studies have been primarily concerned with the development of new treatments 38 39 , with a growing tendency toward reports trying to understand how BZ and NFX work within the parasite and less toward the study of early establishment of resistance markers at different drug concentrations. For this reason, we used a high-BZ-sensitivity Gal61 strain originally isolated from a wild reservoir, and we then induced in vitro BZ resistance by increasing concentrations of BZ. Next, five different resistant populations were established (Table 2, Figure 1A). The BZ response is considered a complex multifactorial trait, which involves many genes of different biochemical pathways because its mode of action is not expected to implicate a specific therapeutic target 5 6 . Therefore, this study initially selected 12 genes and later seven more genes for a total of 19 genes, all of which were regarded as potential new markers of BZ-resistance phenotypes.

We found that OYE and NTR were the only two genes that showed differential expression in all the phenotypes including the TcR5x BZ-resistant population (Figure 3). Interestingly, Kubata et al. (2002) found that the OYE enzyme does not metabolize BZ, although it can metabolize naftoquinones and NFX under anaerobic conditions 13 . However, Hall et al. (2011) recently showed that the artificial overexpression of this gene in T. brucei did not induce NFX resistance 12 . Therefore, we believe that increased OYE levels could be a consequence of metabolic changes occurring for the down-regulation of the NTR gene without altering the resistance to BZ according to previous reports by Kubata et al. (2002) 13 . Murta et al. (2006) reported the OYE enzyme was associated with BZ resistance in induced in vitro T. cruzi-resistant lines due to the loss of three of its four gene copies. In this case, these authors associated its function with downstream steps after activation of the drug 25 . Here, we found that the OYE gene was overexpressed in resistant parasites. However, the BZ resistance level reached by the parasites could explain these differences in the transcription profiles found for Murta et al. (2006) and in the present study, where the maximum IC₅₀ obtained was lower (50 μM). Recently, Boiani et al. (2010) found that only concentrations higher than 400 μM of NFX generated reactive oxygen species (ROS) 40 . Although this generation of ROS by BZ has not been supported by many authors 6 41 , it is possible that a high concentration of the drugs has a different mode of action within the parasite, resulting in different biochemical pathways acting on it 21 26 .

There have been many nitroreductase enzymes reported in several organisms, such as bacteria and protozoans, that play a role in the activation of a wide spectrum of nitro-compounds such as pro-drugs 42 43 44 45 . Members of this protein family from Escherichia coli, Helicobacter pylori and Entamoeba histolytica have shown several alterations such as frameshift mutations, decreased expression, deletions, etc., and these alterations were correlated to nitrofurazone and metronidazole resistance 46 47 48 49 50 51 . Furthermore, experimental evidence suggests that NTR is the most important enzyme in the activation of NFX, and it is also thought that T. cruzi uses the disruption of this enzymatic catabolism to prevent its therapeutic action through a decreased generation of toxic nitro derivatives 12 15 . Furthermore, a similar observation was made by knock-down silencing through iRNA in T. brucei 52 . Thus, the present results showing the downregulation of NTR are not surprising and are supported by previous findings reported by Wilkinson et al. (2008), who proposed that the loss of a single gene copy favors the development of resistance for many heterocyclic compounds in T. cruzi 15 . Recently, it was found that the metabolites generated through reduction of NFX by NTR in T. brucei induce parasite and host cell death; however, if the activation is performed by the host cell it will show no cytotoxic effects 12 . In other words, the loss of cell viability determined only by NTR activation highlights the importance of this enzyme in the mechanism of action of both nitroheterocyclic drugs, BZ as shown here and NFX as reported previously 15 .

Surprisingly, the other genes evaluated did not change expression in the TcR5x phenotype, despite regulated changes in expression in the other resistant phenotypes. This could be explained as follows: the decrease of BZ activation caused by the NTR gene losing a copy in the most resistant population could prevent the formation of toxic products that altered expression of enzymes such as TS, P450 and MDR in other phenotypes. This could explain the same levels of expression found for the parental strain and the TcR5x phenotype for at least 17 other genes evaluated.

Approximately 93 genes have been reported to be involved in BZ response with different biological functions such as drug activation, transport, defense of ROS and metabolism, among others 19 20 21 22 23 25 26 . These genes are thought to act in molecular resistance pathways in in vitro-induced parasites, possibly as transitory and/or stable mechanisms to resist and survive the BZ-trypanocidal action. However, less than 5% of these genes have been confirmed in their role in the resistance mechanism and only one, the NTR gene, has been found to be directly involved in the NFX-induced resistance T. cruzi 12 15 52 . Interestingly, we found the same result in the BZ-induced resistance parasites.

With the aim of detecting genes with an early response to BZ and determining the possible chemotherapeutic targets of the drug, the levels of mRNA expression of the sensitive phenotype were examined when exposed to BZ at five times its IC₅₀ for different periods of time. However, a specific profile of expression in the ten genes studied was not found compared to the strain without drugs. In L. amazonensis exposed to arsenic, major changes in the expression of proteins involved in tryparedoxin pathways were detected in the first 7 h 53 . Therefore, shorter-term assessment may be necessary to detect changes related to the genes involved in detoxification. However, it is possible that other genes that were not studied in this research are affected with relation to other pathways such as heat shock proteins.

Additionally, we investigated the specific activity of two proteins, CPR and SOD. These proteins were chosen because they could participate in the BZ's metabolism role 6 , and there is some evidence of their role in resistance to BZ 26 34 . Our results indicate that enzymatic activities for both enzymes were similar among the parental strain and the TcR5x resistant phenotype in accordance with the mRNA level. Thus, recently the overexpression of two different CPRs did not confer resistance to NFX in T. brucei 12 . In conclusion, the level of BZ-resistant phenotypes obtained in this investigation did not support the ROS generation hypothesis.

After the down-regulated expression was found for both NTR and OYE genes, the number of gene copies was calculated using the qPCR approach. The NTR gene had two copies per diploid genome in the parental strain and just one in the TcR5x phenotype as a consequence of the loss of one copy in the acquisition of resistance to BZ, as shown in other studies of NFX conferring resistance to different nitroheterocyclic compounds 15 . These results are in accordance not only with the mRNA expression, but also with previous reports, showing that this technique is capable of detecting changes in the number of copies in an easier and less time-consuming way than other traditional techniques such as Southern blot or pulsed field electrophoresis. For the OYE gene, there were no changes in the number of gene copies, contrary to previous studies that found a loss of three gene copies out of four present in the haploid genome 25 . Therefore, it is also probable that high transcription levels are the result of physiological changes in the resistant parasites and not related to the resistant condition, as discussed above 12 .

On the other hand, we found that BZ-resistance parasites presented cross-resistance to NFX (Figure 1B), based upon previous experimental evidence where it was confirmed that resistance could be shared among similar nitroheterocyclic compounds of induced and natural resistance parasites 15 54 55 56 . However, G418 susceptibility did not show differences within all the phenotypes (Figure 1C). In fact, the molecular mode of action of G418 is different from both NFX and BZ 57 . This means that the resistance or the susceptibility trait shares pathways between both NFX and BZ. Therefore, in the event of clinical resistance, BZ cannot be changed alternatively for NFX. As a consequence of cross-resistance among nitroheterocyclic compounds, it is necessary to search for promising new compounds.

In addition to our early drug resistance model of T. cruzi, we also clearly demonstrate that BZ resistance in these parasitic populations did not necessarily depend on a constant presence of the drug, which can continue over the course of time (Figure 1A). The stability of the resistant phenotype has been confirmed in many in vitro and in vivo models through different parasite species 19 54 . This condition is the result of many cases of stable genetic alterations such as mutations, amplifications and deletions 16 18 58 .

Alves et al. (1994) observed alterations in the restriction fragment length polymorphism of the kDNA in T. cruzi submitted to a number of passages during exponential growth phase or after subcloning. This phenomenon was called transkinetoplastid 59 and it consists of fast changes in the kDNA minicircle population leading to different restriction profiles. Thus, LSSP-PCR based on the kDNA variable region can reveal not only changes in the kDNA, but also the profiles of different clones. In this manner, LSSP-PCR and RAPDs analysis have been shown to be good tools in evaluating the genetic variability within the clones belonging to the same strain. These approaches are able to identify multiple or simple variations on the DNA sequence and genetic characterization of T. cruzi 35 60 61 62 . Furthermore, the high variability of kDNA markers has been extensively used to identify many clones within a strain 63 . We induced resistance using a susceptible Gal61 strain as a parental line, which was also genetically stable through different induced BZ-resistant phenotypes using kDNA and SL genes as well as the RAPD approach. Therefore, there was no evidence of clonal selection of resistant parasites or even drastic changes in DNA sequences from resistant parasites compared to the parental sensitive line.

Finally, these results indicate that exposure to BZ leads to early irregular expression of different genes possibly involved in response to cell stress, but the loss of one copy of the NTR gene could be responsible for the acquisition and maintenance of a BZ-resistant line. However, it is necessary to study whether NTR is involved in natural resistance and to use other tools such as next generation sequencing to know whether or not there are other genes affected.

Diminished mRNA levels of NTR and a loss of one copy of this gene in the BZ-resistant phenotype of T. cruzi were documented. Therefore, we propose that the NTR gene is involved in the emergence of resistance to BZ and could be used as a reliable marker of resistance in patients treated for T. cruzi infection.

The authors declare that they have no competing interests.

Conceived and designed the experiments: AMMJ, GJF, OTC. Performed the experiments: AMMJ, GJF, LP. Analyzed the data: AMMJ, GJF, LP. Wrote the paper: AMMJ, LP, OTC. All authors read and approved the final manuscript.