Quantification of Plasmodium knowlesi versus Plasmodium falciparum in the rhesus liver: implications for malaria vaccine studies in rhesus models

Malaria control efforts would benefit enormously from development of a highly effective and easily deployable malaria vaccine that achieves durable sterile protection against Plasmodium falciparum malaria. The Plasmodium sporozoite and liver stages are considered strategic targets for intervention, and a number of strategies designed to block these stages have been developed including protein subunit vaccines, viral-vectored vaccines, and live-attenuated vaccines such as radiation-attenuated sporozoites and genetically- or chemically-attenuated parasite vaccines [1].

Despite numerous protective vaccine candidates being characterized in rodent malaria models, translating such findings to humans has been difficult and many clinical trials have shown underwhelming efficacy [2‐4], suggesting that rodent models may not faithfully predict vaccine efficacy in humans. To date, purified, cryopreserved, live-attenuated P. falciparum sporozoite (PfSPZ) vaccines have provided the best vaccine efficacy in humans, achieving up to 100% sterile protection against controlled human malaria infections in clinical trials [5‐7]. However, although there has been significant progress in this area [8], to achieve the goals of durable, high level vaccine efficacy at low cost, there is still a need to develop new malaria vaccine strategies and translationally-relevant models for their evaluation.

Rhesus macaques are valuable pre-clinical models that allow for assessment of malaria vaccine safety, immunogenicity and protective efficacy. Rhesus are considered excellent models of the human immune system [9], and the availability of immunological tools, together with the ability to study the liver and other tissues inaccessible in humans, makes them ideally suited for translationally-relevant vaccine research [10]. Both Plasmodium knowlesi and P. falciparum have been used in rhesus for malaria vaccine testing. Plasmodium knowlesi parasites naturally infect non-human primates and cause zoonotic malaria infections in humans [11]. The P. knowlesi/rhesus model can be used to evaluate malaria vaccine immunogenicity and efficacy, and because P. knowlesi is highly virulent in rhesus, provides a stringent model for assessing vaccine strategies prior to human trials [11, 12]. The P. falciparum/rhesus model is an alternative platform that allows direct evaluation of vaccines targeting this parasite species. However, since rhesus do not support P. falciparum blood stage infections [13], this platform is typically used to assess vaccine immunogenicity as animals are not amenable to challenge with a blood stage efficacy endpoint [14, 15].

For live-attenuated sporozoite vaccines against P. falciparum, which almost certainly rely on successful sporozoite invasion in the liver for antigen delivery [15], the use of rhesus for vaccine assessment critically hinges on their susceptibility to sporozoite infection. Fortunately, despite being unable to support the P. falciparum blood stage, previous data have demonstrated that rhesus have some susceptibility to P. falciparum sporozoite infection in the liver. Data from two independent studies have shown that P. falciparum sporozoites can invade and develop in rhesus primary hepatocytes in vitro, albeit with less efficiency than P. falciparum in human hepatocytes or P. knowlesi in rhesus hepatocytes [15, 16]. These findings validated the use of rhesus for testing the immunogenicity of live-attenuated PfSPZ vaccines, and enabled pivotal non-human primate studies [15] that guided subsequent clinical trials [5, 17]. Complementary studies in the P. knowlesi/rhesus challenge model contributed further insight into the immunogenicity of live-attenuated sporozoite vaccines [18] and allowed for more definitive identification of correlates of protection in the liver [19]. Rhesus have thus been key for elucidating mechanisms of vaccine-induced immunity that are expected to translate to humans. However, integrating data from these two platforms would undoubtedly be aided by a greater understanding of rhesus susceptibility to P. falciparum versus P. knowlesi sporozoites in vivo.

Recent data are beginning to shed light on this question. In a prior study, rhesus were inoculated intravenously with a high dose (6.5 × 10⁶) of freshly-dissected wild-type P. falciparum sporozoites and parasite liver burden was quantified 3 or 6 days later using ultra-sensitive 18S rRNA RT-PCR [20]. Parasite 18S rRNA was detected in the rhesus liver at both time points, with RNA copy number increasing 100-fold between days 3 and 6 [20]. This demonstrated measurable P. falciparum liver burden could be supported in rhesus, and provided the first quantification of rhesus susceptibility to P. falciparum sporozoites in vivo. These data also suggested the possibility of quantifying P. falciparum 18S rRNA in the rhesus liver to directly measure efficacy of vaccines targeting sporozoites and liver stages. Here, this study aims to extend upon this previous work by directly comparing the susceptibility of rhesus to wild-type P. falciparum versus P. knowlesi sporozoites in vivo, and theoretically evaluating the potential to use measurements of P. falciparum 18S rRNA in the liver as a vaccine efficacy endpoint.

The study was designed to directly compare the susceptibility of the rhesus liver to wild-type P. falciparum and P. knowlesi sporozoites in vivo (Fig. 1). Blood was drawn from animals at baseline for CRP and CBC to evaluate their health, and animals were inoculated at day 0 by DVI of either P. falciparum and P. knowlesi sporozoites according to their treatment group. Two animals per group were inoculated with sporozoites as follows: Group A—1 × 10⁶ PfSPZ, Group B—5 × 10⁶ PfSPZ, Group C—0.5 × 10⁶ PkSPZ. Groups A and B were designed to assess the dose-dependence of P. falciparum 18S rRNA liver burden following sporozoite inoculation, and to identify if either dose would theoretically allow for measurement of parasite 18S rRNA in the liver as a potential vaccine efficacy endpoint. Group C was designed to be a comparator, using the highest feasible dose of P. knowlesi sporozoites based on the PkSPZ vialling amount. At exactly 5 days post-sporozoite inoculation, blood was collected for 18S rRNA RT-PCR analysis to assess if parasites had egressed from the liver. Immediately thereafter, animals were euthanized, and livers collected, divided into four lobes, then processed and analyzed to determine parasite burden in each lobe using a slightly modified version of a well-established 18S rRNA RT-PCR biomarker assay [20, 23].

The results for parasite burden in the liver are expressed as log₁₀ copies 18S rRNA per g of tissue (Fig. 2a and Table 1). Three of the four animals inoculated with PfSPZ had measurable parasite 18S rRNA in the liver at 5 days post-infection, indicating P. falciparum liver burdens following sporozoite administration were somewhat unreliable. The animal with undetectable parasite 18S rRNA had received a dose of 1 × 10⁶ PfSPZ. The other animal that received this dose had detectable parasite 18S rRNA in every liver lobe, with a mean of 5.57 log₁₀ copies per g of liver. Both animals that received the higher dose of 5 × 10⁶ PfSPZ had detectable parasite 18S rRNA in the liver, suggesting some impact of sporozoite dose on assay outcome. One animal in this group had 18S rRNA signal in two of four liver lobes, with a mean of 5.32 log₁₀ copies per g of liver in the infected lobes, while the other animal had parasite 18S rRNA in every lobe with a mean of 6.27 log₁₀ copies per g of liver. Both animals that received PkSPZ had much higher liver parasite 18S rRNA concentrations that were consistent across all four lobes, with means of 9.52 and 9.72 log₁₀ copies per g of liver. Parasite 18S rRNA copy number was thus 3–5 log higher in animals that received PkSPZ versus PfSPZ, despite the PkSPZ dose being 2–10 fold lower than in the PfSPZ groups.

The magnitude of parasite 18S rRNA in the liver did not appear to correlate with baseline CRP levels or animal body weight (Table 1). While the small sample size of this study precluded any formal correlation analysis, qualitative assessment of the data suggested that these factors were not likely confounding variables. Specifically, for CRP levels, although animals had a range of baseline CRP levels, there was no apparent correlation between this and assay outcome. For body weight, although the smallest animal had the highest 18S rRNA signal following PkSPZ inoculation, the largest animal had the highest 18S rRNA signal following PfSPZ inoculation, giving no clear relationship between body weight and signal. Similarly, for CBC measures, all animals had blood cell counts that were within normal limits based on their age and experience of veterinary staff at the primate center. By contrast, preliminary observations suggest there may be a potential correlation between sex and liver parasite burden following PfSPZ infection. This was evident as both males receiving PfSPZ had detectable 18S rRNA in every liver lobe, but females in the equivalent treatment groups either had no detectable 18S rRNA or signal in only some liver lobes.

The results for parasite burden in the blood are expressed as log₁₀ copies 18S rRNA per mL of whole blood (Fig. 2b). Regardless of the treatment group, no animal was found to have detectable parasite 18S rRNA in the blood at 5 days post sporozoite inoculation. This was consistent with the known duration of the P. falciparum liver stage in humans and the P. knowlesi liver stage in rhesus [21, 22], and the previous data showing no P. falciparum 18S rRNA was detectable in the blood of rhesus 6 days post-sporozoite inoculation [20]. This demonstrated that despite the high sporozoite doses given, no detectable parasites had egressed from the liver into the blood at the time of euthanasia and liver collection. Thus, the entirety of the parasite 18S rRNA biomarker signal in the liver as measured above was determined to have originated from parasites in the liver tissue.

Previous studies with the P. falciparum/rhesus and P. knowlesi/rhesus platforms have demonstrated the value of these models for assessing live-attenuated sporozoite vaccines and elucidating mechanisms of vaccine-induced immunity relevant to humans [15, 17‐19]. Since these live-attenuated vaccines must rely on sporozoite invasion for antigen delivery to the liver, studies across these two platforms have only been possible because rhesus have demonstrated some susceptibility to P. falciparum sporozoites [15, 16]. However, despite evidence that rhesus are less susceptible to P. falciparum versus P. knowlesi sporozoite infection, their relative ability to infect and replicate in the rhesus liver has never been directly compared in vivo. Here, using an ultra-sensitive Plasmodium 18S rRNA biomarker assay, parasite rRNA copy number in the liver was found to be 3–5 orders of magnitude lower in animals administered PfSPZ versus PkSPZ at 5 days post-infection. While the sample size of this study was small, it was nonetheless sufficient to identify striking trends that have significant relevance for the ongoing use of rhesus in malaria vaccine studies. These findings highlight the fact that there are likely significant differences in host-parasite interactions between P. falciparum and P. knowlesi in the rhesus liver, which in turn are expected to impact antigen delivery and the immunogenicity of live-attenuated sporozoite vaccines.

The most notable finding of this study is that administration of wild-type cryopreserved P. falciparum sporozoites (PfSPZ) resulted in inconsistent, heterogeneous liver stage burdens in vivo. Given that the PfSPZ inoculum was 300 to 1500 times greater than required for 100% blood stage infection in humans [6] and 1000 times the inoculum required to achieve 100% blood stage infection with PkSPZ in rhesus (S. Chakravarty, unpublished), this suggests that P. falciparum sporozoites must face considerable barriers to infection in rhesus in vivo. These barriers could include sporozoite exit from the circulation, invasion of hepatocytes, proliferation within hepatocytes, or avoiding immune clearance mechanisms in the liver [24]. The impact of each of these events cannot be distinguished with the current study design, but it is possible to make several further observations. First, although P. yoelii sporozoite invasion and development in vivo can be affected by cryopreservation [25], the robust infections seen in the PkSPZ comparator group suggests that cryopreservation was not responsible for the striking difference seen in liver parasite burdens between animals infected with PfSPZ versus PkSPZ in this study. Indeed, although it remains formally possible that P. falciparum and P. knowlesi sporozoites may tolerate cryopreservation to different extents, the comparable infectivity of cryopreserved PfSPZ in humans and cryopreserved PkSPZ in rhesus described above argue against this as a significant confounding variable. Similarly, as P. falciparum sporozoites can invade rhesus hepatocytes in vitro [15, 16], this implies that sporozoite invasion is unlikely to be completely blocked in vivo. On the other hand, as P. falciparum liver stage schizont development is only supported at low frequencies in rhesus hepatocytes in vitro [16], this argues that liver stage development may be severely impaired in vivo. Given that avoidance of immune clearance mechanisms by liver stage parasites involves complex and species-specific manipulations of the host cell [26], it is also reasonable to speculate that any P. falciparum liver stages that develop in the rhesus liver will have poorer survival than simian-adapted parasite species.

The identification of a potential correlation between sex and liver parasite burden following infection with PfSPZ is of considerable interest. Although there were only two females in this study, the finding that one had no detectable parasite 18S rRNA, and the other had lower signal in fewer liver lobes than the male in the same group was noteworthy. Due to the small sample size of this study, it was not possible to exclude the alternate hypothesis that these findings were due to chance alone, or that despite rigorous procedures, the PfSPZ administration failed in the one animal that did not develop detectable liver stage parasite burden. However, as every effort was made to use age- and weight-matched animals, and to confirm correct administration of PfSPZ into the vein, additional studies in the future may be warranted to determine if this apparent sex-specific decrease in P. falciparum liver stage burden is reproducible. While comparatively little is known about sex-specific responses to primary malaria infection, there is an emerging body of literature about sex-specific responses to vaccination [27, 28]. As the animals used here were juveniles, it is unlikely that any sex-specific effects are due to the influence of adult sex hormones on the immune system as documented for live-attenuated sporozoite vaccines in mice [29]. However, other sex-specific immunological differences have been described in human infants and children [28], including in response to the RTS,S malaria vaccine, indicating that the sexes begin to diverge in relevant ways prior to puberty. Given that children are one of the primary target populations for PfSPZ vaccines in endemic areas, it may be beneficial to determine if juvenile rhesus recapitulate these sex-specific differences and responses to malaria vaccination. Further studies to investigate and compare the susceptibility of juvenile versus adult rhesus may also be important to pursue. These studies would enhance the value of rhesus for malaria vaccine testing, and are expected to be key topics for future investigation.

Regardless of the underlying reasons, the heterogeneity of P. falciparum burden in the rhesus liver suggests quantification of parasite 18S rRNA is unlikely to provide a useful efficacy measure for vaccines targeting sporozoites and liver stages in the P. falciparum/rhesus model. The potential of using the 18S rRNA biomarker assay for this purpose was suggested by a previous study, where rhesus macaques were intravenously administered 6.5 × 10⁶ freshly-purified wild-type P. falciparum sporozoites, and parasite 18S rRNA copy number was found to increase between animals sacrificed at days 3 and 6 [20]. This finding suggested that P. falciparum 18S rRNA signal could be measured in the rhesus liver over a dynamic range, and hence, that differences in copy number between naïve and vaccinated animals might be used to quantify vaccine efficacy as routinely done in rodent malaria models [30, 31]. Since large doses of freshly-dissected P. falciparum sporozoites can be difficult to obtain, lower doses of cryopreserved wild-type sporozoites were assessed here as these would be more ideally suited for vaccine studies. While P. falciparum 18S rRNA copy numbers were indeed lower in the current study consistent with the lower sporozoite dose, they were broadly concordant with the previous data, with the day 5 copy numbers measured here similar to or greater than the day 3 copy numbers in the previous study [20]. However, neither dose of PfSPZ resulted in consistent burdens, meaning that parasite 18S rRNA copy number measurements would be highly subject to where in the liver was sampled, or necessitate sampling of the entire liver as performed here. Unfortunately, this may be impractical for vaccine studies since protocols for simultaneous processing of liver samples for 18S rRNA and immunological assays are yet to be developed for rhesus, and costs and ethical concerns would preclude most researchers from performing immunogenicity and efficacy studies in parallel cohorts of rhesus. Taken together, this ultimately suggests that quantification of parasite 18S rRNA in the liver is unlikely to be a feasible way to assess vaccine efficacy in the P. falciparum/rhesus model.

Finally, this study has several important implications for research involving live-attenuated parasite vaccines in rhesus. Live-attenuated vaccines that rely on sporozoite invasion of the liver encompass a spectrum of vaccine concepts that include radiation-attenuated sporozoites, late-arresting genetically-attenuated parasites, and chemically-attenuated parasites [1], the latter including the PfSPZ-CVac chemoprophylaxis vaccine [6]. These vaccines differ in their methods of attenuation, the extent of parasite development in the liver, and the corresponding breadth of antigens delivered to the immune system [1]. Given the ability to meaningfully evaluate these vaccines is critically dependent on to what extent rhesus can support sporozoite invasion and liver stage development, the quantitative data presented here offer new insight into how the P. falciparum/rhesus and P. knowlesi/rhesus platforms can best be used to evaluate these vaccines. For radiation-attenuated sporozoites that arrest early in the liver, previous studies have already demonstrated that both the P. falciparum and P. knowlesi platforms can provide highly valuable immunogenicity data about these vaccines [15, 17, 18]. Given the relatively poor susceptibility of rhesus to PfSPZ revealed here, it may be necessary to re-interpret some of the data from past studies of P. falciparum attenuated sporozoite vaccines in rhesus to take this decreased susceptibility into consideration. However, this certainly does not impact the validity of using rhesus to assess antibody and T cell responses induced by P. falciparum attenuated sporozoite vaccines, but rather serves to refine understanding of the model and potentially improve how findings from these studies are extrapolated to humans. Equally, as there is evidence that P. falciparum sporozoites can invade rhesus hepatocytes in vitro [15, 16] and in vivo [20], this also suggests that both platforms can be used to study responses to pre-formed antigens induced by radiation-attenuated sporozoite vaccines [32], with the caveat that the decreased susceptibility of rhesus to P. falciparum sporozoites should again be taken into consideration. The decreased susceptibility of the rhesus liver to P. falciparum versus P. knowlesi sporozoites should also be considered for making comparisons across platforms, as it is expected to impact antigen load, antigen distribution across the liver, and the expression of liver stage-specific antigens. Conversely, for genetically- or chemically-attenuated parasites vaccines that arrest at later stages of development, the data presented here indicate that P. falciparum/rhesus model is unsuitable for evaluation of these vaccines, and suggests such vaccines should instead be assessed using alternative platforms.

This study provides the first direct quantitative comparison of the relative susceptibility of the rhesus liver to wild-type P. falciparum versus P. knowlesi sporozoites in vivo. This data will help guide pre-clinical studies of live-attenuated parasite vaccines in the P. falciparum/rhesus and P. knowlesi/rhesus models and enable better integration data from vaccine studies across platforms.