Inferior T cell immunogenicity of a Plasmodium berghei model liver stage antigen expressed throughout pre‐erythrocytic maturation

Sporozoite antigens are the basis of a number of malaria vaccines being tested, but the contribution of antigens expressed during subsequent liver stage development to pre‐erythrocytic stage immunity is poorly understood. We previously showed that, following immunisation with radiation attenuated sporozoites (RAS), a model epitope embedded in a sporozoite surface protein elicited robust CD8+ T cell responses, whilst the same epitope in a liver stage antigen induced inferior responses. Since RAS arrest early in their development in host hepatocytes, we hypothesised that extending parasite maturation in the liver could considerably improve the epitope‐specific CD8+ T cell response. Here, we employed a late liver stage arrested parasite model, azithromycin prophylaxis alongside live sporozoites, to increase expression of the model epitope until full liver stage maturation. Strikingly, this alternative immunisation strategy, which has been shown to elicit superior protection, failed to improve the resulting epitope‐specific CD8+ T cell responses. Our findings support the notion that liver stage antigens are poorly immunogenic and provide additional caution about prioritising antigens for vaccine development based solely on immunogenicity.


| INTRODUC TI ON
The malaria pre-erythrocytic stages are comprised of the extracellular sporozoites, which are inoculated by infected mosquitoes to the mammalian host, followed by the intracellular exo-erythrocytic forms (EEF; also known as liver stages), resulting from the development and maturation of sporozoites within a parasitophorous vacuole (PV) in hepatocytes. 1 Whilst there is an abundance of investigations delineating Plasmodium sporozoite antigens, the immune responses they induce, and their potential for use in malaria pre-erythrocytic vaccines, [2][3][4][5][6][7][8][9] little is known about antigens solely expressed in EEFs.
Vaccination-induced protection against pre-erythrocytic stages was first shown to be feasible in animal models and humans using whole sporozoite vaccine approaches, particularly with radiation attenuation sporozoites (RAS), [10][11][12] which are considered the benchmark for anti-malarial vaccines. Sterile protection induced by RAS has been shown to be mediated primarily by CD8 + T cells. 13,14 Efficient recall of CD8 + T cell responses following presentation of parasite antigens on hepatocytes is crucial due to the short duration that the parasites are in the liver (~48-52 h for P. berghei 15 ). Despite the high level of protection achieved, P. berghei RAS do not develop into large, mature hepatic schizonts but are arrested prior to onset of replication and growth approximately 24 h post-immunisation. 16,17 This early hepatic arrest raises questions as to the magnitude of contributions of EEF-specific over sporozoite antigens in immunity.
Enhanced protection has been reported following the administration of anti-malarial chemoprophylaxis and live sporozoites in both animal models and in humans. [18][19][20][21] Similar results have been reported after immunisations with late arresting genetically arrested sporozoites in preclinical studies in mice. 22,23 Given that this alternative whole sporozoite vaccine approach ensures full EEF development, the results suggest that the longer the parasites are allowed to develop and mature before arrest, the greater the protection induced by the immunisation strategy. This outcome has been initially interpreted as having more antigens expressed and an increase in antigen biomass during extended parasite development, eliciting a broader range of EEF-specific immune responses needed for protection. 22,23 However, immunogenic proteins expressed in the late EEF, which may increase CD8 + T cell responses, are currently not well defined.
The circumsporozoite protein (CSP) is the major surface coat antigen of sporozoites and upregulated in sporozoites 4 (UIS4) is a protein mainly associated with the parasitophorous vacuole membrane (PVM) surrounding the EEF. 24,25 Upon P. berghei sporozoite infections of H-2K b restricted C57BL/6 mice, no immunodominant epitopes in either CSP or UIS4 were identified. 26 Thus, in the absence of known EEF epitopes allowing for the quantification of specific T cell responses following sporozoite immunisation, we and others have used a surrogate marker, that is the upregulation of CD11a, in combination with the downregulation of CD8α, as a durable and 'accurate' phenotyping method for infection or vaccine induced antigen-experienced T cells. 27,28 To understand the immunogenicity of EEF-specific antigens, as proxies for detecting CD8 + T cell epitopes in sporozoite and EEF antigens, we previously generated P. berghei transgenic parasites that express the SIINFEKL epitope from ovalbumin as part of either CSP or UIS4, CSP SIINFEKL and UIS4 SIINFEKL , respectively. 29 Following RAS immunisation, a striking difference between the larger SIINFEKL-specific CD8 + T cell response elicited by CSP SIINFEKL and the smaller response induced by UIS4 SIINFEKL was found. This divergence could be due to the abrupt cessation of or limited UIS4 expression following early arrest of RAS.
In this report, we tested the hypothesis that prolonged PVM protein expression increases CD8 + T cell responses against EEF vacuolar membrane proteins. Using CSP SIINFEKL and UIS4 SIINFEKL , we compared the resulting CD8 + T cell responses, both in the spleen and in the liver, following RAS immunisation or infection with live parasites under azithromycin (AZ/Spz) prophylaxis, which is an apicoplast-targeting, delayed death-inducing anti-malarial drug allowing for the complete maturation of EEFs in the liver. 30 2 | MATERIAL S AND ME THODS

| Ethics and animal experimentation
Animal experiments performed in the United Kingdom were con- NMRI, CD-1 and C57BL/6 laboratory mouse strains were bred in house at LSHTM or purchased from Charles River Laboratories (Margate). NMRI and CD-1 were used for cycling of parasites between vertebrate and mosquito hosts. Female C57BL/6 mice were used for immunology experiments at age 6-8 weeks.

| P. berghei ANKA immunisation
P. berghei WT, CSP SIINFEKL and UIS4 SIINFEKL (strain ANKA; clone c15cy1) parasites 29 were maintained by continuous cycling between murine hosts (NMRI or CD-1) and Anopheles stephensi mosquitoes, as previously described. 29 Mice were immunised with 10,000 freshlyisolated sporozoites intravenously in the lateral tail vein. Sporozoites were either γ-irradiated at 1.2 × 10 4 cGy or administered under prophylactic AZ cover. Azithromycin (Pfizer) was administered at a dose of 240 mg/kg intraperitoneally on the same day as parasite inoculation and one day after. 30

| Infection of Huh7 hepatoma cells with P. berghei sporozoites
In vitro infections were performed in Huh7 cells that were grown in

| Quantification of SIINFEKL-specific CD8 + T cell responses
Spleens and livers perfused with 5 ml PBS were harvested from immunised and naive mice and filtered by passing the organs through 70 µm cell strainers (Corning). Liver-infiltrating lymphocytes were isolated following published protocols using a Percoll gradient. 33 Following red blood cell lysis and resuspension in complete media, cells were diluted in Trypan Blue (ThermoFisher Scientific) and counted by microscopy using a Neubauer 'Improved' haemocytom-
Statistics were calculated using one-way ANOVA with Tukey's multiple comparison test. Normality was calculated using the Shapiro-Wilk test.

| RE SULTS AND D ISCUSS I ON
We first compared the EEF development in vitro of RAS versus sporozoites with AZ ( Figure 1). Irrespective of the method used, sporozoites retained their invasive capacities. Forty eight h after infection, as expected, morphological analysis revealed normal EEF differentiation by both untreated sporozoites and sporozoites cultured with AZ; large, mature hepatic schizonts, which underwent multiple nuclear divisions and surrounded by remodelled PV, as shown by UIS4 staining, were observed. 30 In contrast, RAS gave rise to arrested EEFs, which had reduced growth and showed absence of nuclear divisions. 17 Despite this developmental arrest, these small, round intracellular parasites expressed HSP70 in their cytoplasm and are surrounded by a UIS4 containing remodelled PV, comparable to untreated parasites at earlier stages of intracellular development 29 ( Figure 1A). The striking differences in parasite maturation in the liver rather suggest a higher amount of EEF-specific antigens upon AZ treatment. We quantified UIS4 protein levels by measuring the UIS4 fluorescence intensity over the area of individual EEFs. We show that UIS4 protein levels were comparable in AZ treated and untreated control EEFs, but significantly lower in RAS ( Figure 1B).
For the subsequent experiments, we compared the CD8+ T cell responses induced following RAS immunisation or sporozoites with AZ, utilising the PbCSP SIINFEKL and PbUIS4 SIINFEKL parasites we previously generated. 29 We then measured the in vivo magnitude of antigen-experienced cells after parasite immunisation. Immunisation with both RAS or sporozoites under AZ cover, produced quantifiable CD11a hi CD8 + (CD8α lo ) T cells, around 8% in the spleen (Figure 2A,C,E) and 30% in the liver ( Figure 2B,D,F), 2 weeks after immunisation as compared to 5% and 20%, respectively, in naïve mice. These findings are consistent with previous work showing that both RAS and AZ attenuation induce comparable high levels of antigen-experienced CD8 + T cells in peripheral blood following immunisation. 21 Next, we compared epitope-specific CD8 + T cell responses by ex vivo stimulation with SIINFEKL peptide (Figure 3). Consistent with our previous work, 29 following RAS immunisation, significantly higher proportions and numbers of SIINFEKL-specific IFNγ-producing CD8 + T cells were induced with PbCSP SIINFEKL as compared to PbUIS4 SIINFEKL (Figure 3A-D). Notably, extension of antigen availability and/or increased antigen biomass, due to prolonged UIS4 expression permitted by AZ administration, did not improve the proportions and numbers of IFNγ producing CD8 + T cells, as compared to RAS immunisation. Together, these results clearly demonstrate that extending EEF development, resulting in elevated vacuolar membrane antigen expression, does not amplify IFNγ producing CD8 + T cell responses.
Whole-parasite immunisation strategies that allow the parasite to complete EEF development in the liver ensure immunisation against many antigens expressed in the pre-erythrocytic stages. These span from those expressed by the sporozoite to those expressed very late in the EEF prior to merozoite release into the blood, as well as those antigens that are expressed throughout the pre-erythrocytic stages. Given that CD8 + T cells are crucial for protection against pre-erythrocytic stages, these strategies were originally thought to augment the pool of immunisation-induced CD8 + T cells that are specifically targeted against EEF antigens. 18 However, we have demonstrated poor CD8 + T cell responses to an epitope contained within UIS4, a PVM protein expressed constitutively in the liver following sporozoite invasion of a hepatocyte. 29 We show that AZ prophylaxis would have allowed UIS4 to increase in size for the full 48-52 h of EEF development, in contrast to ~24 h when RAS were used. Despite these striking differences in parasite development, adaptive immune responses to sporozoite immunisation remained unaltered as evidenced by the comparable numbers of antigen-experienced CD8 + T cells, based on the quantification of CD11a proxy marker co-expression. These results are consistent with our previous findings that responses to UIS4 could not be enhanced by increasing the dose of RAS used for immunisation either. 29 We hypothesised that extending parasite maturation in the liver might improve the epitope-specific CD8 + T cell response. It is noteworthy that in order to obtain direct evidence for prolonged antigen exposure in sporozoites attenuated by AZ cover versus RAS immunisations, elution of MHC-bound peptides from infected hepatocytes over time could be performed. Thus far, elution of MHC-bound epitopes from dendritic cells that had been co-cultured with asexual blood stages has been achieved. 34 Establishing this approach for pre-erythrocytic antigens and in a time-course experiment will be considerably more challenging. Our data showing that AZ prophylaxis failed to improve the resulting epitope-specific CD8 + T cell responses, lending further support to the notion that liver stage antigens are poorly immunogenic.
AZ has been shown to specifically impede the biogenesis and inheritance of the apicoplast in malaria liver stages; EEFs continue to develop, but blood stage infection is not established. 30 In vitro work has indicated that AZ treatment of P. berghei -infected hepatoma cells allows for the detachment of merozoite-containing infected liver cells, and these merosomes fail to initiate blood stage patency. 30 The immunogenicity of these merosomes is of interest, particularly because mice immunised with sporozoites under AZ cover are susceptible to blood-stage challenge, demonstrating that protective immunity offered by this form of parasite attenuation is primarily against the pre-erythrocytic stages and that CD8 + T cells are the prime effector mechanisms. 21 In contrast, immunisation with sporozoites under chloroquine (CQ) cover also leads to full EEF development and successful initial blood infection. Accordingly, recent studies revealed the generation of cross-stage protection involving both pre-erythrocytic and blood stages. 35,36 Understanding the extent to which blood stage antigens are exposed to the immune system following immunisation with whole sporozoites is important to the identification of vaccination strategies that can combine T cell responses against the pre-erythrocytic stages and antibodies against sporozoites and blood stages.
This study provides a confirmation that EEF antigens are poorly immunogenic. Yet, CD8 + T cells must recognise peptides directly processed and presented by parasitised hepatocytes to employ their protective functions. Taken together with our recent findings that EEF antigens are nonetheless excellent targets of vaccine-induced CD8 + T cells, 29 our results challenge the use of immunogenicity in prioritising antigens for the design and evaluation of next-generation pre-erythrocytic vaccines. Standard immunological assays, endeavoured at discovering highly immunogenic antigens, may fail to identify those candidates with the ability to evoke superior levels of protective immunity. An in-depth characterisation of the complex biology of pre-erythrocytic stages, the immune responses they generate or not, coupled with a strategic identification of vaccine targets, should drive progress towards a highly efficacious malaria vaccine.

ACK N OWLED G EM ENTS
We would like to thank Jason Mooney for his critical reading of the manuscript. Kai Matuschewski was supported by the Max Planck