Drug-cured experimental Trypanosoma cruzi infections confer long-lasting and cross-strain protection

Background The long term and complex nature of Chagas disease in humans has restricted studies on vaccine feasibility. Animal models also have limitations due to technical difficulties in monitoring the extremely low parasite burden that is characteristic of chronic stage infections. Advances in imaging technology offer alternative approaches that circumvent these problems. Here, we describe the use of highly sensitive whole body in vivo imaging to assess the efficacy of recombinant viral vector vaccines and benznidazole-cured infections to protect mice from challenge with Trypanosoma cruzi. Methodology/Principal findings Mice were infected with T. cruzi strains modified to express a red-shifted luciferase reporter. Using bioluminescence imaging, we assessed the degree of immunity to re-infection conferred after benznidazole-cure. Those infected for 14 days or more, prior to the onset of benznidazole treatment, were highly protected from challenge with both homologous and heterologous strains. There was a >99% reduction in parasite burden, with parasites frequently undetectable after homologous challenge. This level of protection was considerably greater than that achieved with recombinant vaccines. It was also independent of the route of infection or size of the challenge inoculum, and was long-lasting, with no significant diminution in immunity after almost a year. When the primary infection was benznidazole-treated after 4 days (before completion of the first cycle of intracellular infection), the degree of protection was much reduced, an outcome associated with a minimal T. cruzi-specific IFN-γ+ T cell response. Conclusions/Significance Our findings suggest that a protective Chagas disease vaccine must have the ability to eliminate parasites before they reach organs/tissues, such as the GI tract, where once established, they become largely refractory to the induced immune response.

Introduction Chagas disease is caused by the insect-transmitted protozoan Trypanosoma cruzi and is the most important parasitic infection in the Americas. More than 5 million people are infected with this obligate intracellular parasite [1], resulting in a financial burden estimated at $7 billion annually [2]. In humans, the disease is characterised by an acute stage that occurs 2-8 weeks post-infection, during which bloodstream parasites are often detectable. Symptoms during this period are normally mild, although lethal outcomes can occur in 5% of diagnosed cases. The parasite numbers are then controlled by a vigorous adaptive immune response. However, sterile immunity is not achieved and infected individuals transition to a chronic stage, which in most cases, appears to be life-long [3]. Around 30-40% of those infected eventually develop chronic disease pathology, a process that can take decades to become symptomatic. Cardiomyopathy is the most common clinical manifestation [4,5], although 10-15% of people can develop digestive tract megasyndromes, sometimes in addition to cardiac disease.
Attempts to control Chagas disease have been challenging. For example, although public health measures have been successful in reducing disease transmission in several regions of South America, there is a vast zoonotic reservoir that complicates disease eradication by this route [6][7][8]. The only drugs currently available to treat the infection, the nitroheterocycles benznidazole and nifurtimox, have limited efficacy and cause toxic side effects that can impact on patient compliance [9,10]. There have been no new treatments for almost 50 years, but progress in discovering new chemotherapeutic agents is now being accelerated by a range of drug development consortia encompassing both the academic and commercial sectors [11]. For many years, vaccine development against Chagas disease has been inhibited by concerns that autoimmunity could play a role in disease pathogenesis [12,13]. Although not excluded as a contributory factor, the current consensus is that the risk has been overstated, and that the continued presence of the parasite is required to drive disease pathology [14,15].
The host response to T. cruzi infection involves a complex combination of both innate and adaptive immune mechanisms [16,17]. The innate system is key to controlling parasite proliferation and dissemination during the initial stages of infection [18], with important roles for both Toll-like receptor (TLR)-mediated inflammatory responses [19] and TLR-independent processes [20,21]. As the acute phase progresses, the development of an antigen-specific immune response, in which CD8 + IFN-γ + T cells are the key effectors [16,22], is the critical step in controlling the infection. In both humans and mice, the major targets of this cellular response are a small set of immunodominant epitopes within specific members of the transsialidase super-family of surface antigens [23]. The observation that the pattern of this recognition displays strain variation has been interpreted as indicative that immune evasion could be operating at a population level. The adaptive response reduces the parasite burden by >99%, with the infection becoming highly focal, and in BALB/c mice at least, confined predominantly to the large intestine, stomach, and to a lesser extent, the gut mesentery tissue and sites in the skin [24,25]. The reason why the immune system is not able to eradicate the infection is unresolved. It does not appear to involve exhaustion of the CD8 + IFN-γ + T cell response, which continues to suppress, but not eliminate, the parasite burden throughout the long chronic stage [26]. These findings have questioned the feasibility of developing an effective anti-T. cruzi vaccine.
Experimental vaccination of animal models against T. cruzi infection has a long history [27], although there have been few instances in which unequivocal sterile protection has been reported. Approaches have included the use of attenuated parasites [28,29], immunisation with cell fractions [30], purified or recombinant proteins [31,32] and the use of DNA vaccines. In the latter case, viral backbones based on vaccinia [33], yellow fever [34] and adenovirus [35] have been used to facilitate expression of a range of parasite antigens such as trans-sialidase, Tc24, and the amastigote surface protein-2 (ASP-2) (also a member of the trans-sialidase super-family). Reported outcomes include protection from lethal infection, reduction in the acute stage parasite burden, induction of a favourable cytokine profile, and reduction in disease pathology. However, detailed analysis of vaccine efficacy has been limited by an inability to accurately monitor parasite levels during the chronic stage and technical difficulties in assessing the effect of the immune response on tissue distribution following challenge infections.
Recently, in vivo imaging approaches have been exploited to provide new insights into infection dynamics during experimental chronic Chagas disease [24,25,36]. These studies have revealed the pantropic nature of acute stage infections, and shown that during the chronic stage, the adaptive immune response restricts parasites to small infection foci, predominantly within the GI tract. Other tissues and organs, including the heart and skeletal muscle, are infected sporadically, the extent of which is influenced by host:parasite genetics and immune status. Additional factors such as nutrition, environmental stimuli, age and co-infections could also play a role in this complex chronic infection profile [37]. The survival of the small parasite foci within apparently tolerant sites is crucial for long-term infection, although the immunological context of these reservoirs is unknown. Another contributor to the long-term nature of T. cruzi infections could be the phenomenon of parasite dormancy; individual intracellular amastigotes can enter an apparently quiescent state in which they cease to replicate and exhibit reduced drug sensitivity [38]. Neither the mechanisms involved, nor the potential implications for immune evasion have yet been established.
Highly sensitive bioluminescence imaging involves the use of T. cruzi strains that have been modified to express a red-shifted luciferase reporter [39]. The system allows the real-time monitoring of parasite burden in experimental mice during chronic stage infections. There is a robust correlation between parasite numbers and whole animal bioluminescence, with a limit of detection close to 100 parasites [24]. Here, we describe the use of this imaging technology to assess the extent of protection in benznidazole-cured mice following re-challenge with homologous and heterologous strains. Our findings have important implications for vaccine strategies.

Generation of recombinant ASP-2/TS vaccines
The fusion gene encoding the ASP-2 and TS peptides (Fig 1A) was generated by linking sequences corresponding to the mouse Ig kappa chain signal peptide, ASP-2 amino acids 1-694 (GenBank accession no. U77951) and TS amino acids 1-624 (GenBank accession no. L38457). The furin 2A splice site linker was inserted between the trypanosome sequences to ensure the subsequent generation of two separate peptides from a single open reading frame [40]. The ASP-2/TS fusion gene was cloned into the ChAdOx1 [41] and MVA [42,43] viralvectored vaccine platforms, and confirmed by sequencing prior to use in protection studies.

Assessment of recombinant vaccine immunogenicity
Vaccines were prepared in PBS and administered intramuscularly into the left and right quadriceps of mice. The ChAdOx1 vaccine was administered at 1x10 8 infectious units per dose. With MVA:ASP-2/TS, each dose was equivalent to 1x10 6 plaque forming units. ELISpots were carried out using either peripheral blood mononuclear cells (PBMCs) or splenocytes [44]. Briefly, MAIP ELISpot plates (Millipore) were coated at 4˚C overnight with anti-mouse IFN-γ mAb AN-18 (Mabtech), at 250 ng per well, and then blocked for 1 h with complete DMEM medium (plus 10% foetal calf serum). Whole blood was sampled by venesection of the tail vein and PBMCs were isolated using histopaque 1083 (Sigma), and plated at 5x10 5 cells per well in complete DMEM medium as above, with 20-mer specific peptides overlapping by 10 amino acids (10 μg ml -1 ) (Pepscan Presto). Splenocytes from naïve mice were plated at 2.5x10 5 per well. After 16 h incubation, cells were discarded and plates washed with PBS. 50 μl of biotinylated anti-mouse IFN-γ mAb RA-6A2 (1:1000 in PBS) was then added to each well and incubated for 2 h. After another washing step, streptavidin peroxidase (Sigma) was added and incubated at 37˚C for 1 h. The plates were washed and developed with TMB substrate solution (Mabtech). When spots were visible, the reaction was stopped by washing the plate with water. Spots were analysed using an ELISpot reader, and the number of spot-forming cells/10 6 PBMCs producing IFN-γ was calculated.

Murine infections and bioluminescence imaging
Animal work was performed under UK Home Office project licence (PPL 70/8207) and approved by the LSHTM Animal Welfare and Ethical Review Board. Procedures were in accordance with the UK Animals (Scientific Procedures) Act 1986 (ASPA). BALB/c mice were purchased from Charles River (UK), and CB17 SCID mice were bred in-house. Animals were maintained under specific pathogen-free conditions in individually ventilated cages. They experienced a 12 h light/dark cycle, with access to food and water ad libitum. SCID mice were infected with 1x10 4 bioluminescent bloodstream trypomastigotes (BTs) in 0.2 ml PBS via intraperitoneal (i.p.) injection [24,25,36]. BALB/c female mice, aged 8-10 weeks, were infected i.p with 1x10 3 BTs derived from SCID mouse blood. At experimental end-points, mice were sacrificed by exsanguination under terminal anaesthesia.
For in vivo imaging, mice were injected with 150 mg kg -1 d-luciferin i.p., then anaesthetized using 2.5% (v/v) gaseous isoflurane. They were placed in an IVIS Lumina II system (Caliper Life Science) 5-10 min after d-luciferin administration and images acquired using Living-Image 4.3. Exposure times varied from 30 s to 5 min, depending on signal intensity. After imaging, mice were revived and returned to cages. For ex vivo imaging, mice were injected with d-luciferin, and sacrificed by exsanguination under terminal anaesthesia 5 min later. They were then perfused via the heart with 10 ml 0.3 mg ml -1 d-luciferin in PBS. Organs and tissues were removed and transferred to a Petri dish in a standardized arrangement, soaked in 0.3 mg ml -1 d-luciferin in PBS, and imaged using maximum detection settings (5 min exposure, large binning). The remaining animal parts and carcass were checked for residual bioluminescent foci, also using maximum detection settings [24,25]. To estimate parasite burden in live mice, regions of interest were drawn using LivingImage v.4.3 to quantify bioluminescence as total flux (photons/second), summed from dorsal and ventral images. The detection threshold for in vivo imaging was determined using uninfected mice.

Drug treatment and immunosuppression
Benznidazole was synthesized by Epichem Pty Ltd., Australia, and prepared at 10 mg ml −1 in an aqueous suspension vehicle containing 5% (v/v) DMSO, 0.5% (w/v) hydroxypropyl methylcellulose, 0.5% (v/v) benzyl alcohol and 0.4% (v/v) Tween 80. It was administered by oral gavage. To detect any residual infection following treatment, mice were immunosuppressed with cyclophosphamide monohydrate (Sigma) in D-PBS (200 mg kg −1 ), administered by i.p. injection every 4 days, for 3 doses. Two weeks after the end of immunosuppression, mice that were bioluminescence negative by both in vivo and ex vivo imaging were designated as cured.

Monitoring ASP-2/TS vaccine efficacy using highly sensitive bioluminescence imaging
The T. cruzi amastigote surface protein 2 (ASP-2) and trypomastigote cell surface protein trans-sialidase (TS) have shown promise as vaccine candidates against T. cruzi infections [45][46][47]. To further assess their efficacy, we used two replication-deficient recombinant vaccine platforms, a chimpanzee adenovirus (ChAdOx1) and a Modified Vaccinia Ankara virus (MVA), expressing ASP-2 and TS peptides from a single open reading frame ( Fig 1A) (Materials and Methods) [40]. We used a homologous prime-boost vaccination strategy in which BALB/c mice received intramuscular injections administered one week apart ( Fig 1B) and confirmed immunogenicity of the vaccine delivery system using an ex vivo IFN-γ + ELISpot. Three weeks after receiving either a prime only, or a prime-booster vaccination, splenocytes were plated on antibody coated ELISpot plates. When these cells were then stimulated with a peptide pool representing the entire ASP-2/TS sequence, there was a pronounced increase in the number of peptide specific IFN-γ + splenocytes, particularly in those mice that had received the booster (Fig 1C).
To test protective efficacy, mice were vaccinated using the homologous prime-boost strategy outlined above. Three weeks after the MVA booster, they were challenged by i.p. injection with 10 3 bioluminescent T. cruzi blood trypomastigotes (strain CL Brener) (Fig 2A). The resulting infection was monitored by in vivo imaging [36] (Materials and Methods). In BALB/ c mice, parasites rapidly disseminate and proliferate, with the infection reaching a peak after approximately 2 weeks. Thereafter, a vigorous adaptive immune response reduces the parasite burden by >2 orders of magnitude, and the infection transitions to the life-long chronic stage [24]. No differences were observed in the bioluminescence-inferred parasite burden between vaccinated and control mice at the earliest time-point assessed (day 7, post-infection) (Fig 2B  and 2C). However, by day 15, the peak of the acute stage, the ASP-2/TS vaccinated mice displayed a 77% reduction in the bioluminescence-inferred parasite burden. This protective effect was maintained until day 21. By day 28, against a background of immune-mediated reduction in the parasite burden, there were no significant differences between the vaccinated and control groups. From that point onwards, the parasite burden remained similar between the groups (Fig 2B and 2C). Following termination of the experiment (day 95), ex vivo imaging of internal organs and tissues revealed that the profile of infection in the vaccinated cohort was typical of the chronic stage. The colon and/or stomach were the major tissues persistently parasitized, with infections in other organs being sporadic (S1 Fig). There were no apparent differences in the tissue-specific parasite burden between vaccinated and control mice ( Fig 2D). Therefore, although vaccination with the ASP-2/TS constructs can reduce parasite burden during the acute stage, bioluminescence imaging has revealed that it does not impact on the long-term burden of chronic infections.

Drug-cured T. cruzi infection confers significant protection against challenge with a homologous strain
To place the recombinant vaccine results into context, we sought to establish the extent to which drug-cured infections could enhance the capacity of the murine immune response to protect against challenge. We hypothesised that this would induce a high level of protection against which to compare the efficacy of vaccine candidates. BALB/c mice were first inoculated i.p. with bioluminescent parasites (CL Brener strain) (n = 12). At three different points postinfection (4, 14 and 36 days) (Fig 3A), we initiated treatment with benznidazole (100 mg kg -1 ), once daily, for 20 consecutive days. This dosing regimen was shown to be curative (S2 Fig), in line with previous results [48,49]. The plasma concentration of benznidazole falls below the in vitro EC 50 value in approximately 12 h [49]. 21 days after the cessation of treatment, the cured mice were re-infected i.p. and monitored regularly by in vivo imaging (Fig 3B and 3C). 70 days after challenge, mice found to be bioluminescence-negative were immunosuppressed to facilitate the outgrowth and dissemination of any residual parasites, then assessed further by ex vivo imaging two weeks later (Fig 3D) (Materials and Methods). We have previously shown that this methodology is considerably more reliable than PCR-based approaches at detecting surviving parasites [48].
When curative treatment was initiated at 36 days post-infection, none of the mice exhibited a distinct acute stage infection peak following challenge (Fig 3B and 3C). There was >99% reduction in the inferred parasite burden in all cases, compared to primary infection control mice. At the experimental end-point, in 6 out of 12 mice, parasites were undetectable (Table 1, Fig 3D, S3 Fig). In cases where the primary infection was allowed to proceed for 14 days prior to the initiation of curative benznidazole treatment, the infection profile following challenge was very similar to the cohort where treatment was initiated 36 days post-infection. However, a greater number of foci were detectable during the period corresponding to the acute stage of primary infections (Fig 3B), with full protection achieved in 3 out of 12 mice. In contrast to both of the above, when curative treatment began 4 days after the primary infection, there was a clear acute stage peak in the bioluminescence profile following challenge (Fig 3C). The kinetic profile mirrored that in control infected mice, although the maximum parasite burden was 85% lower, and only a single mouse was judged to be protected (Table 1, S3 Fig). Therefore, although the immune response induced by a short course infection is able to impact on the burden of re-infection, the effect is significantly limited compared with what is achievable when the primary infection is allowed to progress fully into the acute stage, prior to treatment.
To assess whether this protective effect was dependent on the route of inoculation, we repeated the 36 days challenge experiment using the subcutaneous (s.c.) route. T. cruzi

PLOS NEGLECTED TROPICAL DISEASES
transmission normally occurs when parasite-infected faeces from the insect vector are rubbed into the wound produced by blood feeding; therefore, s.c. inoculation probably reflects more closely how the majority of human infections occur. The mice were inoculated s.c. with 10 3 bloodstream trypomastigotes for both the primary and challenge infections, but the protocols and treatment timelines were otherwise identical to those used previously (Fig 3A, 36 day infection). In control mice, the bioluminescence profile of s.c. infections, and the resulting organ-specific tropism during the chronic stage was similar to that in i.p. infections (Fig 4A,  4B and 4C), as shown previously [24,25]. When the challenge cohort was assessed by in vivo imaging (Fig 4A), none of the mice displayed an acute stage peak, and bioluminescence was at, or close to background levels. At the experimental end-point, all of the mice were immunosuppressed and then subjected to post-mortem ex vivo organ imaging to test for foci of infection below the limit of in vivo detection. 4 out of 5 were found to be bioluminescence negative in all analyses and were designated as protected (Table 1).

PLOS NEGLECTED TROPICAL DISEASES
It has been reported that the capacity of a cured mouse to resist re-infection is dependent on the size of the challenge inoculum [50]. To investigate this using the in vivo imaging system, mice where curative treatment was initiated 36 days post infection were challenged i.p. with 6x10 4 CL Brener trypomastigotes, 60 times the number used previously. The outcome was similar. None of the mice displayed an acute stage parasite burden profile, and only a single mouse (out of 6) was designated non-protected (Fig 4D, 4E and 4F). Therefore, the level of protection conferred by a cured infection is similar when a higher challenge inoculum is used.
We next sought to assess the extent of the immunological memory conferred by a primary drug-cured infection. BALB/c mice were infected i.p. with CL Brener trypomastigotes, and curative benznidazole treatment was initiated after 36 days. On this occasion, however, the mice were not re-infected until 338 days after the cessation of treatment. Following homologous challenge, the level of protection was comparable to that achieved in previous experiments when mice were re-infected 3 weeks after the last dose of benznidazole. They were able to prevent the onset of parasite proliferation during the acute stage, and parasites were nondetectable in 4 out of 6 mice (Fig 4G, 4H and 4I).

Assessing circulating IFN-γ + T cells in mice after primary infection and challenge
We sought to determine if the duration of the primary infection, prior to commencement of curative drug treatment, impacted on the extent of the T. cruzi-specific immune response to challenge with homologous parasites. Blood was collected, 2 days prior to re-infection and at regular intervals thereafter, from mice that had been infected for 4, 14 and 36 days (Fig 5A). PBMCs were isolated, re-stimulated with an ASP-2 and TS peptide pool, and the IFN-γ + cell frequencies measured by ELISpot (Materials and Methods). As expected for mice infected with T. cruzi [51], the control group receiving their first parasite exposure showed a delayed peptide-specific response, with the frequency of IFN-γ + cells on day 10 not significantly different to pre-infection levels. There were substantial increases by days 25 and 40 (Fig 5B), co-incident with the period when the parasite burden had been controlled and was undergoing major reduction (Fig 3B and 3C).

PLOS NEGLECTED TROPICAL DISEASES
With benznidazole-cured mice, the pre-challenge levels of circulating T. cruzi peptide-specific IFN-γ + T cells varied, depending on the duration of the preliminary infection ( Fig 5B). Initially, mice that had been infected for 36 days prior to cure displayed higher levels than the control cohort (day -2). However, these levels did not increase significantly following re-infection, although in the case of the day 14 group, there was a slight trend in this direction. In the control group, the robust adaptive response, which controlled the infection, was associated with levels of circulating IFN-γ + T cells that were significantly higher than in any of the benznidazole-cured mice by 40 days post-challenge (Fig 5B). Mice that had been infected for only 4 days prior to drug-cure, displayed IFN-γ + T cell kinetics that were initially more similar to naïve mice receiving their preliminary infection, with very low levels before and 10 days postchallenge, followed by a major increase by day 25. The levels decreased thereafter, in contrast to the control group, where they continued at higher levels until day 40.
We also examined circulating peptide-specific IFN-γ + T cells in mice that had been reinfected 338 days after drug cure (Fig 6A). Prior to re-challenge, the levels were similar to

Protection conferred by a benznidazole-cured infection is not dependent on the route of infection, size of the challenge inoculum or the time-period until re-infection. (A) BALB/c mice, infected by the subcutaneous (s.c.) route with 10 3 bioluminescent trypomastigotes (CL Brener strain)
, were subjected to curative benznidazole treatment 36 days post-infection. 21 days after the end of treatment, they were re-infected (s.c.). Control mice were also infected by the s.c. route. (B) Total body bioluminescence of drug-cured mice following s.c. re-infection (means ± SD). (C) Ex vivo bioluminescence imaging of organs and carcass from a control and the re-infected mouse that was found to be non-protected after immunosuppression (Materials and Methods). (D) BALB/c mice infected i.p. with CL Brener trypomastigotes, were subjected to benznidazole treatment 36 days post-infection. 21 days after the end of treatment, they were re-infected (i.p.) with 6x10 4 trypomastigotes. (E) Total body bioluminescence of drug-cured mice following re-infection. (F) Ex vivo bioluminescence imaging of organs and carcass of a control and the re-infected mouse that was found to be non-protected after immunosuppression. (G) As above, BALB/c mice were infected i.p. and subjected to curative benznidazole treatment. 338 days after the end of treatment, they were re-infected (i.p.) with 10 3 trypomastigotes. (H) Total body bioluminescence of drug-cured mice following re-infection. The increased mean bioluminescence of the re-infected mice towards the end of the monitoring period was due to an intense bioluminescence focus in one of the two non-protected animals. (I) Ex vivo bioluminescence imaging of organs and carcass of a control and a re-infected mouse that was found to be non-protected after immunosuppression. In all cases, the original cohort size was n = 6. During the course of the s.c. infection experiment (A-C), one mouse failed to recover from anaesthesia, and was excluded from the analysis. https://doi.org/10.1371/journal.pntd.0007717.g004

PLOS NEGLECTED TROPICAL DISEASES
Monitoring immunity against T. cruzi infections by in vivo imaging those in non-infected control mice. However, by 10 days post-challenge, there had been a 5-fold increase, in contrast to the delayed peptide-specific response typical of a T. cruzi infection ( Fig 6B). This difference was not maintained, and on days 25 and 40 after challenge, the level of IFN-γ + T cells was not significantly different from control mice.

The capacity of drug-cured infections to confer a cross-strain protective response
T. cruzi displays significant genetic diversity, with the natural population subdivided into 6 lineages known as discrete typing units (DTUs), each of which has the ability to infect humans [52]. We therefore investigated if the capacity to confer homologous protection is a general feature of T. cruzi infections by performing an analogous experiment using JR strain parasites from the genetically distant TcI lineage. In BALB/c mice, infections with this strain are slightly slower to reach the peak of the acute stage, but the bioluminescence profile is otherwise similar to that of the CL Brener strain (DTU VI lineage) [25]. Mice were benznidazole-treated 36 days into a JR infection, and re-infected with the same strain 20 days after the end of treatment ( Fig  7A). As before, we observed that the cured infection conferred significant protection. None of the mice exhibited a distinct acute stage peak, with the majority remaining close to bioluminescence background levels (Fig 7B and 7C). However, only a single mouse (out of 6) exhibited full protection as assessed by ex vivo imaging following immunosuppression (Materials and Methods) (Fig 7D).

PLOS NEGLECTED TROPICAL DISEASES
To assess the scope for vaccine-induced species-wide immunity, we next investigated the effectiveness of protection conferred against a heterologous challenge using the strains CL Brener (TcVI) and JR (TcI). Following the strategy outlined above, BALB/c mice were infected i.p. with 10 3 bloodstream CL Brener or JR trypomastigotes, benznidazole-treated, and then challenged with the heterologous strain 20 days after the end of the curative therapy. In both experiments, we observed a strong protective response (>99%) (Fig 7E-7J), with no distinct acute stage peak and a reduced number of bioluminescent foci in the period corresponding to the transition to the chronic stage. However, in both cases, all mice exposed to cross-strain challenge displayed small but clear parasite foci following re-infection. None were bioluminescence-negative when examined by ex vivo imaging, with each displaying the type of GI tract infections characteristic of the chronic stage (Fig 7G and 7J). Therefore, although infection with a heterologous strain can have a major impact on the subsequent parasite burden, it did not lead to parasite elimination in any of the mice examined.

Discussion
Although experimental T. cruzi vaccines have been widely shown to reduce the burden of infection in animal models [27][28][29][30][31][32][33][34][35], there is little unambiguous evidence for sterile protection. Despite this, there have been an increasing number of reports that vaccination could have therapeutic benefits in terms of decreased cardiac pathology [33,[53][54][55]. Therefore, the question as to whether the development of a Chagas disease vaccine might be a practical option for reducing the public health impact of this infection remains unanswered. Detailed assessment has been limited by difficulties in detecting the intermittent low-level parasitemia of chronic stage infections, and in identifying the tissue/organ location of persistent parasites. Here, we demonstrate that highly sensitive bioluminescent imaging can negate some of these issues, and provide novel insights into vaccine efficacy.

PLOS NEGLECTED TROPICAL DISEASES
Initially, we tested the protective properties of two viral vectors (MVA and ChAdOx1) that had been modified to express an ASP-2/TS fusion gene (Fig 1). The reduction in peak parasite burden (77%, Fig 2C) was in a range similar to that reported for other recombinant T. cruzi vaccines [56-58, as examples]. Interestingly, vaccination had no impact on the parasite burden once the infection had begun transition to the chronic phase, suggesting that if parasites can survive until this stage of the disease, they are less susceptible to clearance by vaccine-induced immunity. This has not been reported previously. At the experimental end-point, bioluminescence imaging allowed us to demonstrate that the remaining parasites were restricted predominantly to the GI tract, with other organs/tissues infected only sporadically (Fig 2D, S1 Fig). In BALB/c and other mice, the colon and/or stomach serve as a permissive niche, enabling parasites to persist in an otherwise hostile immune environment [24,25,37], although the mechanism(s) for this have yet to be elucidated. To determine if this might limit the utility of a Chagas disease vaccine, we therefore investigated the protective effect of benznidazole-cured infections, on the basis that these should provide optimal levels of immunity. Curative treatment has been associated with the development of a stable anti-T. cruzi CD8 + T cell population [22].

PLOS NEGLECTED TROPICAL DISEASES
Drug-cure was initiated after infection for 36 days (a time-point when the adaptive CD8 + T cell response is controlling the infection), 14 days (the peak of the acute stage), and 4 days (just short of one complete round of the intracellular replication cycle that leads to differentiation and parasite egress) (Fig 3). Mice infected for 36 days prior to benznidazole treatment were highly protected from challenge (Fig 3B), with complete absence of a typical acute stage peak. However, parasites were eventually detectable in half of the re-infected mice (Table 1). Therefore, although drug-cured infections can generate a highly effective immune response, that prevents a second acute phase, parasites that evade this initial encounter seem to be refractory to immune-mediated elimination and are able to persist long term. This outcome was not significantly influenced by the size of the challenge inoculum or the route of infection (Fig 4). Initiating treatment after 14 days was also able to prevent a detectable acute stage peak in bioluminescence when the mice were challenged, although there was a slight reduction in the number of mice where parasites were undetectable (Fig 3B, Table 1). In contrast, when drugcure was initiated after 4 days, a pronounced post-challenge acute stage peak could be observed, but even then, the parasite burden was 85% lower than in a naive infection. Therefore, 14 days exposure to an untreated infection is sufficient for the induction of a robust immune response, whereas with 4 days, the response appears to be less developed, although still sufficient to have a significant impact on the parasite burden. Studies on the protective potential of benznidazole-cured infections have been reported previously [59], where treatment was initiated 15-days post-infection. However, parasite recrudescence in these experiments was monitored after cyclophosphamide-mediated immunosuppression, making it difficult to compare the protective outcomes with those reported here.
A delayed onset of the CD8 + T cell response is a characteristic feature of Chagas disease [60], with the first round of intracellular infections passing largely undetected by the immune system. Upon invasion of mammalian cells, parasites rapidly escape from the phagolysosome, there is down-modulation of the host cell immunoproteasome [61], and minimal activation of the host-pattern recognition receptors. Induction of effective innate immunity requires a full cycle of parasite replication, host cell lysis, and the release of trypomastigotes into the extracellular milieu, a process that takes at least 4-5 days. This is followed by the production of pathogen-associated and damage-associated molecular patterns that promote innate immune responses, allowing parasitized cells to flag up their infected status by MHC class I antigen presentation. Full development of the CD8 + T cell response to T. cruzi infections takes around 3 weeks [62]. In mice where curative treatment was initiated 14 or 36 days post-infection, circulating parasite peptide-specific IFN-γ + T cells were readily detectable prior to challenge ( Fig  5B), and were associated with protection against the development of a second acute phase profile. Experimental challenge did not lead to a significant increase in the level of circulating IFN-γ + T cells. Although we did not study the distinct contributions of peptide-specific CD4/ CD8 T cell responses in the periphery, it is possible that the pre-existing effector population was able to contain the secondary challenge without further activation/expansion. Assays of peptide-specific T cell responses in the spleen were not feasible due to the requirement for end-point immunosuppression to promote outgrowth of the parasites. In agreement with previous studies, in some cases the parasites inoculated during the secondary challenge escaped from immunity conferred by the drug-cured infection, and were able to survive long-term.
In mice where curative treatment was initiated 4 days into the primary infection, the level of T. cruzi peptide-specific IFN-γ + T cells prior to re-infection was negligible, and the kinetics of the response induced over the first 25 days of the challenge infection was similar to that in the controls (Fig 5). Despite this, there was an 85% reduction in the parasite burden at the peak of the re-infection. Therefore, the induced partially protective effect in these mice is conferred either by an extremely low level of circulating parasite-specific IFN-γ + T cells, or by other factors that operate to moderate the infection. In mice challenged almost a year after curative treatment of the primary infection, the level of protection was similar to mice in which the gap between the end of treatment and challenge was only~20 days. However, unlike these mice, re-infection after almost a year was accompanied by induction of peptide-specific T cells, with kinetics that were more rapid than in the naive control cohort (Fig 6). This evidence for a memory response suggests that vaccine-mediated long-term protection against fulminant T. cruzi infection may be a feasible goal. Furthermore, if these results can be extrapolated to humans, it would imply that patients who have undergone curative drug treatment could have the added benefit of a high level of stable protection against re-infection.
We also investigated the extent to which benznidazole-cured infections could provide cross-strain protection. T. cruzi is highly diverse, with six major genetic lineages that display considerable geographic overlap. Taxonomy is further complicated by the widespread existence of hybrid strains [63]. Mice initially infected with the T. cruzi CL Brener (TcVI) were challenged with the JR strain (TcI), and vice-versa. Although, suppression of the parasite burden was similar to that in a homologous challenge (>99%), sterile protection was not achieved (Fig 7), with surviving parasites persisting at very low levels. This suggests that there is a potential for co-infection with a different lineage, but that the secondary infection would be unlikely to pass through an acute stage. It also highlights that there may be limits to the level of protection conferred on patients who have undergone curative treatment.
We have suggested a model for chronic Chagas disease [37] in which the gut (and perhaps other tissues, such as the skin or skeletal muscle) acts as an immunologically tolerant reservoir for T. cruzi persistence, with periodic trafficking to other sites, where the parasites are then destroyed rapidly by immune effector mechanisms. In the heart, this can lead to cumulative collateral damage that ultimately gives rise to cardiac pathology [64] We now further propose that when parasites establish infections in the GI tract, or other permissive sites, they become refractory to elimination by the vigorous adaptive responses induced by drug-cured infections. Thus, the effectiveness of a Chagas disease vaccine could depend on the efficiency with which the primed immune system prevents T. cruzi from reaching the relative safety of these sites of persistence, and its ability to maintain this response over time against a wide range of strains. The results presented here, and elsewhere [31-34, 45, 46, 53-58], highlight the possibility that current subunit/DNA vaccines may be unable to fulfil these requirements under the schedules used for administration, although their ability to prevent lethal outcomes [31-34, 45, 46, 54, 57] and provide therapeutic benefits [33,[53][54][55][56]58] merits further research. In addition, the long-term protection conferred by a drug-cured live infection suggests that the use of parasites, suitably attenuated by genetic modification, may be an approach worth exploring to identify an effective vaccine.