by Peter Dumoulin A dissertation submitted to Johns Hopkins University in conformity with the requirements for the degree of Doctor of Philosophy

Transcription

1 STUDIES ON EXOERYTHROCYTIC DEVELOPMENT OF PLASMODIUM FALCIPARUM IN VITRO: DETECTION AND ISOLATION OF PLASMODIUM LIVER STAGES AND ANALYSIS OF CIRCUMSPOROZOITE PROTEIN ANTIGEN PROCESSING by Peter Dumoulin A dissertation submitted to Johns Hopkins University in conformity with the requirements for the degree of Doctor of Philosophy Baltimore, Maryland May, 2015

2 Abstract Malaria, the disease caused by Plasmodium parasites, remains a major global health burden despite efforts to eradicate the parasite. Infection is initiated when a mosquito deposits sporozoites into the dermis that then actively invade hepatocytes. Sporozoites and the resulting liver stage of Plasmodium falciparum infection are leading targets for generation of a protective vaccine. The circumsporozoite protein is the major surface protein on sporozoites and is the target of vaccines aimed at preventing blood stage infection. During the process of invasion, sporozoites cleave the N-terminus of CSP and consequently two forms of CSP are potentially exposed to the immune system. We utilized ectopic expression of two natural form of CSP to quantify their abilities to serve as targets for MHC class I-restricted CD8 + T cells. We determined that presentation of both forms of CSP on MHC class I depends on proteasomal activity, however, presentation of full length CSP is more efficient and is conferred by the presence of N-terminal lysine residues absent in the cleaved form of CSP. To evaluate the presence of these two forms during in vitro infection, we developed methods to allow for detection and isolation of developing live P. falciparum liver stages by flow cytometry. Using this technique we compared the susceptibility of five immortalized human hepatocyte cell lines and primary hepatocyte cultures from three donors to infection by P. falciparum sporozoites. We showed that exoerythrocytic forms can be detected and isolated from in vitro infected cultures of the HC-04 cell line and primary human hepatocytes. We confirmed the presence of developing parasites in sorted live human hepatocytes and characterized their morphology by fluorescence microscopy. Furthermore, we present an experimental evidence, that our ii

3 methods can be applicable for the analysis of human host factors limiting development of P. falciparum EEFs. iii

4 Main goals 1. Assess the ability of two forms of Plasmodium falciparum CSP, to serve as targets for MHC class I-restricted CD8 + T cell responses Specific Aims: A. To develop an in vitro system for monitoring the MHC class I processing and presentation of peptide epitopes derived from the P. falciparum CSP antigen B. To assess the ability of targets expressing one of the CSP forms to induce specific CD8 + T cell activation in a MHC class I-restricted and non-restricted manner C. To examine the role of major cellular proteolytic machineries in the processing of two forms of CSP D. To investigate the role of the ubiquitin system in the MHC class I-restricted processing of peptide epitopes from two forms of the CSP antigen 2. To develop an experimental system for detection and isolation of Plasmodium falciparum EEFs in vitro Specific Aims: A. To develop an in vitro hepatocellular model permissive for propagation of P. falciparum EEFs B. To develop flow cytometry-based detection and specific isolation of P. falciparum EEFs C. To compare dynamics of P. falciparum EEF development in vitro using different experimental hepatocellular systems D. To validate the applicability of our experimental model for the analysis of human host factors limiting development of P. falciparum EEFs. iv

5 Table of Contents Abstract... ii Main goals... iv Table of Figures... viii Chapter 1 : General Introduction... 1 History of elucidating the Plasmodium life cycle... 5 From skin to merosome... 9 Targeting exoerythrocytic forms: whole organism vaccines Mechanisms of immune protection Biology of CSP Techniques to study the Plasmodium liver stage Concluding remarks Chapter 2 : Processing and presentation of two forms of CSP on MHC class I Introduction Presentation of peptides on MHC class I Targeting Plasmodium: role of the MHC class I antigen processing and presentation pathway Two forms of CSP as targets of the CD8 + T cell immune response Results Ectopic expression of CSP in hepatocytes Identification of CSP in traversed and infected hepatocytes during in vitro infection.. 41 Two forms of CSP differentially activate CD8 + T cells Targets expressing one of two forms of CSP are differentially lysed by CD8 + T cells Two forms of CSP do not alter immune phenotypes of targets or overall susceptibility to death stimuli Half-life of two forms of CSP and dependence of proteasomal degradation on MHC class I presentation CSP-Full forms polyubiquitin chains more efficiently than CSP-Short Individual lysine residues in the N-terminus of CSP are required for efficient MHC class I-restricted antigen presentation Discussion Chapter 3 : Specific detection and isolation of P. falciparum EEFs in vitro v

6 Introduction Results Immortalized human hepatocyte cell lines are permissive for traversal by P. falciparum sporozoites Human hepatocyte cell lines exhibit differential ability to support development of P. falciparum EEFs P. falciparum-infected HC-04 hepatocytes can be specifically isolated Primary human hepatocytes support P. falciparum 3D7HT-GFP EEFs in vitro Discussion Chapter 4 : Flow cytometry based assays of P. falciparum liver stages in vitro: from sporozoite infectivity to EEF development Introduction Results CD81 is required for P. falciparum infection of primary human hepatocytes, but is not essential for the infection of HC Humanized CSP-specific antibody 2A10 inhibits traversal and reduces the number of P. falciparum EEFs in human hepatocytes in vitro Discussion Chapter 5 : Future Directions Chapter 6 : Materials and Methods Human hepatocyte culture Collagen coating and hepatocyte plating densities Infection with P. falciparum sporozoites Detection of hepatocyte traversal Detection of EEFs by flow cytometry Quantification of parasite development by RT- PCR Immunofluorescence of EEFs in HC-04 and primary human hepatocytes CD81 detection and surface neutralization Evaluation of the efficacy of mouse serum containing humanized anti-csp mab 2A10 on parasite motility Generation and maintenance of CTLs Construction of CSP containing mammalian expression plasmids Transfection of hepatocytes and purification by MACS Detection of CSP by western blot Detection of ectopically expressed CSP vi

7 T-cell activation assay: IFNγ release Cytotoxicity lactate dehydrogenase assay Chemical inhibitor treatments Immunoprecipitation of HA-ubiquitin complexes Peptide synthesis and target pulsing Statistical analysis References Curriculum Vitae vii

8 Table of Figures Figure 1.1 The lifecycle of the human Plasmodium parasites Figure 1.2 First microscopic identification of the pre-erythrocytic stage of P. falciparum Figure 1.3 Cartoon representation of the sporozoite interacting with the liver sinusoid Figure 1.4 Cartoon representation of merosomes budding and release into the blood stream Figure 1.5 Visualization of the CSP reaction Figure 1.6 Schematic of CSP and the RTS,S vaccine Figure 1.7 Estimated relative population sizes (biomass) between stages of the P. falciparum lifecycle Figure 1.8 Schematic of the structure of CSP from P. knowlesi based on nucleotide sequence Figure 1.9 Regions of sequence identity in CSP between P. knowlesi and P. falciparum sequences Figure 2.1 Schematic representation basic antigen presentation on MHC class I Figure 2.2 Schematic of the structure of HLA-A Figure 2.3 Schematic representation of CSP during infection Figure 2.4 Ectopic and natural expression of the two forms of P. falciparum CSP Figure 2.5 Identification, isolation and expansion of HLA-A2 restricted viral peptide specific CD8 + T cells Figure 2.6 Verification of CTL specificity Figure 2.7 Identification and purification of transfected HC Figure 2.8 MHC Class I restricted recognition of cells expressing CSP chimeras by peptide specific CD8+ T-cells viii

9 Figure 2.9 Specific lysis of targets expressing CSP-Full or CSP-Short by CD8+ T-cells Figure 2.10 Susceptibility of targets expressing two forms of CSP to T cell effector functions Figure 2.11 Degradation of ectopically expressed CSP in HC-04 cells Figure 2.12 Identification of ubiquitination patterns of two forms of CSP Figure 2.13 Role of the N-terminus on presentation of CSP and the contributions of individual lysines Figure 3.1 Specific detection of traversal by P. falciparum sporozoites in vitro Figure 3.2 Human hepatocyte cell lines can be traversed by P. falciparum sporozoites Figure 3.3 Identification of P. falciparum parasites by flow cytometry Figure 3.4 Specific detection of infection by P. falciparum sporozoites in HC-04 cultures in vitro (48 hours postinfection) Figure 3.5 Quantification of parasite-encoded 18S rrna in cultures of human hepatocytes infected with P. falciparum Figure 3.6 Visualization of parasites in HC-04 cells infected with P. falciparum Figure 3.7 Primary hepatocytes have a high degree of autofluorescence and require adjustment to voltages Figure 3.8 Comparison of detection of P. falciparum infection between primary human hepatocytes and HC Figure 3.9 Specific detection of infection by P. falciparum sporozoites in primary hepatocytes in vitro (96 hours postinfection) Figure 3.10 Parasite morphology and size in cryopreserved human hepatocytes from three donors Figure 3.11 Parasite detection and persistence over time Figure 4.1 Expression of surface CD81 by human hepatocyte cell lines and primary hepatocyte donors ix

10 Figure 4.2 Influence of CD81 blocking by mab 1D6 on P. falciparum hepatocyte infection Figure 4.3 Influence of a humanized anti-csp mab 2A10 on P. falciparum traversal and invasion x

11 Chapter 1 : General Introduction 1

12 Malaria is a mosquito-borne infectious disease caused by protozoan parasites of the genus Plasmodium. The parasite is initially deposited by an infected mosquito into the dermis of the intermediate host [1,2] (Figure 1.1A) where it actively enters the blood stream and develops as a clinically silent stage in the liver [3,4] (Figure 1.1B). Mature liver stage parasites eventually seed the blood and initiate the erythrocytic cycles associated with clinical malaria [5,6] (Figure 1.1C,D). A portion of asexually reproducing parasites in the blood will lead to male and female gametocyte production, which can initiate infection of a new mosquito if taken up in a blood meal (Figure 1.1E). B A C D E Figure 1.1 The lifecycle of the human Plasmodium parasites. (A) Infection of the human host begins when sporozoites are deposited in the dermis. (B) Sporozoites actively reach hepatocytes and invade. (C) Once infectious merozoites are formed the host cell is ruptured and merozoites are released. (D) Parasites invade red blood 2

13 cells. (E) A subset of parasites will commit to formation of gametocytes capable of infecting a mosquito if ingested with a blood meal. Adapted from [7] Five species of Plasmodium are capable of causing disease in humans: Plasmodium falciparum, Plasmodium vivax, Plasmodium ovale, Plasmodium malariae, and Plasmodium knowlesi [8,9]. Of these species, infection with P. falciparum results in the most complications and deaths, primarily in sub-saharan Africa [10]. An estimated 207 million new cases of malaria occurred in 2012, resulting in an estimated 627,000 deaths [10]. The current morbidity and mortality of malaria are concentrated in impoverished countries [11], particularly in tropical and subtropical regions. Examples of GDP growth rate increases after local eradication suggest that malaria is a cause of poverty and not just a confounding characteristic of malaria endemic countries [11]. Undoubtedly, resourcepoor countries also have difficulty in sustaining control measures, resulting in a poverty trap [12]. Successful elimination has historically occurred in relatively temperate geographic regions with less intense transmission and a higher degree of seasonality [13]. Consequently, current efforts of eradication or elimination are now focused on the most intransigent bastions of malaria. A vaccine capable of slowing transmission has the potential to aid in these efforts. Pathogenesis of P. falciparum can manifest as metabolic acidosis, severe anemia, respiratory distress, or coma and with complications during pregnancy [14 17]. Persistent infection and disease progression are often the result of strategies employed by P. falciparum to avoid clearance by the immune system [18] or mechanical destruction by the spleen [19]. A disproportionate number of polymorphisms in parasite invasion genes [20 23] allow the parasite to persist at a population level. In the case of individual subunit 3

14 vaccination against AMA1, a significant allele-specific efficacy has been observed [24] but was ineffective at targeting most circulating parasite strains. Considerable effort has been aimed at replicating the naturally acquired immunity to blood stage parasites and at generating a vaccine capable of halting blood stage growth completely. However, these approaches are largely confounded by the same evolutionary escape mechanisms that persist at the population level of natural parasite transmission. In addition to allelic variation between parasites, individual parasites have the ability to change expression of important genes within a single host. Cytoadherence is a tactic employed by the parasite to sequester in the blood vessels and tissues to escape clearance. This process is achieved through expression of PfEMP1 by the parasite at the RBC surface to mediate adherence [25]. The parasite encodes a family of these genes that are antigenically distinct, and which are expressed one at a time [26]. The ability of the parasite to sequester has been implicated in serious clinical outcomes such a cerebral malaria and complications in pregnant women [27,28]. As the host generates an adaptive immune response to one variant of PfEMP1, the fitness of parasites expressing that gene decreases and a minor subset of parasites expressing an alternative allele gains the selective advantage within the host [29]. In this case, many of the same principles of selection and diversity that apply at the population level are present in an individual host. This antagonistic relationship is not surprising considering the extensive period of co-evolution that has existed between Plasmodium and its intermediate hosts. Based on comparisons of Plasmodium gene sequences collected from wild apes in Africa, P. falciparum originally entered the human population as a gorilla parasite [30]. The timing and number of cross-species transmission events remain controversial [31,32] 4

15 and largely based on assumptions of mutation rate in P. falciparum. Even without knowing an exact timeframe, it is apparent that Plasmodium parasites have generated considerable selective pressure on the human genome [33]. These protective variants are often loss of function mutations that affect the red blood cell and include hemoglobinopathies, G6PD deficiency and ovalocytosis. History of elucidating the Plasmodium life cycle Prior to the dissection of the etiological details of malaria, many theories existed to explain the presence of the cyclical fevers originally described by Hippocrates [34]. Even before usage of the term malaria, the general disease theory of antiquity focused on the concept of bad-air (later miasma) as contagion combined with the notion that individual susceptibility is mediated by a balancing of the four humors, as first proposed by Galen (reviewed in [35]). The characteristic fevers observed in malaria patients were first attributed to an imbalance in one of the humors, either yellow or black bile. These theories relied on the concept of spontaneous generation of toxic air or organisms from non-related matter. The underpinning of this theory, in which living things are generated from the nonliving, were left unchallenged for centuries. In 1668 Francesco Redi experimentally tested and refuted the theory of spontaneous generation by determining that the formation of maggots on meat required contact from flies [36]. Soon after, Antonie van Leeuwenhoek directly observed single celled organisms by using his newly developed single lens microscope [37]. Louis Pasteur was instrumental in providing formal experiments describing microbial causes of disease and suggesting that common diseases have the potential to be explained by germ theory [38]. 5

16 These conceptual and technological advancements in microbiology set the stage as scientists began to search for a microbial determinant of malaria. The first observations of a parasite associated with malaria pigment in a patient s blood were made by the French military doctor Alphonse Laveran [39]. His samples were unstained, which allowed him to observe motility or what we now know is the process of exflagellation of male gametocytes. In 1897 William George MacCallum described an additional non-motile parasitic form in avian blood and postulated that this was due to sexual dimorphism and observed temperature dependent ookinete formation [40]. However, this observation still did not explain the mechanism of parasite transmission or the reason behind geographic risk factors. Considerable attention was paid to environmental sampling as others suggested the possibility that mosquitoes vectored the parasite. Ronald Ross worked under the premise that mosquitoes could cultivate the malaria parasite, while noting that: The study is a difficult one, as there is no a priori indication of what the derived parasite will be like precisely nor in what particular species of insect the experiment will be successful [41]. After two years of attempts, he was successful at identifying pigmented parasites larger than typical mosquito cells only in mosquitoes fed on patients whose blood contained crescents [41]. It was also noted that these cells were larger when observed later after blood meal, suggesting parasite development within the mosquito. However, before he was able to establish transmission back to humans he was posted to Calcutta, forcing him to utilize an avian model, Plasmodium relictum, due to the paucity of human malaria in the region [42]. Using this model he was able to observe sporozoites in the mosquito salivary gland and transmission to a new host [43]. Ross correctly assumed that a similar process 6

17 of mosquito incubation and transmission occurred in human malaria parasite transmission but was unable to conclusively demonstrate the life cycle of such parasites. This formal proof was provided in 1899 by Bignami and Grassi in Italy, which conclusively demonstrated transmission from the mosquito to human [44]. This new understanding of the lifecycle, from human to mosquito back to human, provided an important framework to begin public health measures aimed at vector control. However, a cryptic or negative phase was repeatedly observed, in which parasites were not observable until about 10 days after mosquito bite. It was noted during this phase that blood was not infectious and that quinine was ineffective [45]. Grassi himself suggested that the morphological difference between the sporozoite nucleus and early stage blood trophozoites might be indicative of a developmental stage prior to blood stage parasitemia [46]. Yet the spurious observation that Plasmodium vivax sporozoites were able to penetrate and invade red blood cells persisted as evidence that sporozoites could progress directly to erythrocytic development [47,48]. The theory of direct progression to blood stage failed to explain the details of the observed cryptic phase and after several decades the theory was challenged by experiments with avian parasites showing development of sporozoites outside of erythrocytes [49,50]. None of these studies, however, directly implicated the site of development of primate trophic Plasmodium species. The study of Hepatocystis kochi in Kenya and Uganda by P.C.C. Garnham led to the discovery of developmental stages in the liver parenchyma [51]. Since H. kochi lacks an erythrocytic cycle and only invades red blood cells for gametocyte formation, Garnham rightly postulated that the liver may also serve as a developmental site for Plasmodium. Soon after, using Plasmodium cynomolgi infected Anopheles maculipennis fed on rhesus 7

18 macaques, Garnham observed plasmodial masses undergoing schizogony in the liver [52]. Similar forms were found in liver biopsies from human subjects experimentally exposed to enormous doses (770 mosquito bites) of P. falciparum and P. vivax infected mosquitoes [45,53] (Figure 1.2). These works represent the completion of the basic understanding of the Plasmodium life cycle begun by Laveran almost 70 years prior. Figure 1.2 First microscopic identification of the pre-erythrocytic stage of P. falciparum. Biopsy of a liver from a volunteer after 770 mosquito bites from P. falciparum infected mosquitoes. The above image (x 700) is attributed to parasites that have been developing in the liver for 116 hours. Since multiple mosquito feeds were done on different days and only one biopsy was performed the age of the parasite in the liver can only be assumed by size. Adapted from [45] Fig. 2 8

19 From skin to merosome Understanding the life cycle of Plasmodium was the first step in designing interventions aimed at interrupting parasite transmission. The concept that malaria transmission was caused by mosquitoes was quickly followed by the proof that individuals could be protected from malaria by protecting them from mosquito bites [42,46]. In a similar fashion, the discovery of the liver stage as an obligate replicative cycle prior to symptomatic blood stage infection offered the possibility of generating immunity that could eliminate the parasite prior to clinical episodes. Since the definitive demonstration in the 1940s that the cryptic phase is indeed an obligate liver stage of development, a considerable amount of work has elucidated the processes that allow for sporozoite targeting to the liver and progression to the blood stages. P. falciparum has a limited number of hosts, including humans and some primates [54 56], and consequently study of the liver stage has been undertaken largely with the use of rodent trophic parasites originally isolated from Thamnomys and Grammomys rodents in Africa [57,58]. The journey to the liver begins when sporozoites are injected into the host dermis [1,2] and gradually move from the bite site through a process of non-productive cell invasion termed traversal [59,60], a process first observed as sporozoite penetration of macrophages [61]. Once sporozoites enter the circulation they attach through CSP region II to the heparin sulfate proteoglycans (HSPGs) present in the extracellular matrix of the liver [62,63] (Figure 1.3A). Sporozoites detect these sulfation levels and convert from a migratory mode to an active mode for invasion [64]. The presence of HSPGs are not obligatory for invasion as shown during in vitro infections [65] but rather are important under flow conditions [66] similar to what the parasite experiences in vivo. This suggests 9

20 that HSPGs serve as a ligand for sporozoites to concentrate near hepatocytes after reaching the circulation in vivo and become activated for productive invasion. Identification of the parasite genetic determinants of traversal [67,68] has confirmed the necessity of traversal in reaching the liver in vivo where direct access is blocked by the sinusoid endothelium and Kupffer cells [69]. Depletion of sentinel Kupffer cells allows entry of traversal-deficient sporozoites to the liver when injected intravenously [67,70]. Traversal competent parasites have multiple routes available to traverse that can include but are not restricted to direct movement through Kupffer cells [71,72] (Figure 1.3B,C). Once sporozoites reach the liver parenchyma they continue to traverse hepatocytes [71] but induce a switch in motility to allow for productive invasion allowing for proper parasitophorous vacuole (PV) formation in hepatocytes [59,60,73] (Figure 1.3D). The particular hepatocyte ligands that mediate productive parasite invasion are not completely characterized. 10

21 A B C D Figure 1.3 Cartoon representation of the sporozoite interacting with the liver sinusoid. (A) Sporozoites recognized sulfation levels in the liver sinusoid and become activated for infection. (B,C) Sporozoites cross the sinusoid layer through cell traversal activity that does not require Kupffer cell invasion (D) Sporozoites actively invade hepatocytes. Adapted from [5] Fig 2. One of the host surface molecules that has been shown to be important for P. falciparum sporozoite invasion into hepatocytes is CD81 [74,75]. Even though CD81 is present at the hepatocyte surface, it is not liver specific and is commonly found on the 11

22 surface of lymphocytes [76]. Prior to the discovery of the importance of CD81 for hepatocyte invasion by P. falciparum, the extracellular loop had been shown to be important for the binding of hepatitis C virus E2 and hepatocyte infection [77]. Initial experiments found that CD81-deficient mice are resistant to P. yoelii exoerythrocytic form (EEF) formation [74]. In the same report, blocking of CD81 was shown to inhibit P. falciparum EEF formation in a dose-dependent manner. Subsequent work utilizing ectopically-expressed CD81 in the CD81 negative cell line HepG2 established that in the case of P. yoelii EEF formation, CD81 was sufficient to drive infection [75]. However, expression of CD81 was necessary but not sufficient to allow for P. falciparum EEF formation. These data suggest that additional receptors and/or mechanisms exist that are necessary for P. falciparum sporozoites to initiate productive infection in hepatocytes. Interestingly, the type I scavenger receptor (SR-BI) has been shown to be important for both invasion and development of Plasmodium sporozoites [78] in vitro but was found to be unnecessary in vivo [79]. After sporozoites productively invade a host hepatocyte and establish a parasitophorous vacuole (PV), they take up residence adjacent to the host nucleus [80] and recruit host organelles to the PV. The developmental time required for Plasmodium EEFs to reach maturity varies widely between species. For instance, the rodent parasites P. berghei and P. yoelii complete full development in less than three days post-infection [81,82]. In contrast, P. falciparum requires a minimum of six to seven days to reach maturity [83] and at maturity the parasites are considerably larger [84,85]. From the mature EEF of the parasite, bags of hepatic merozoites termed merosomes bud into the liver sinusoid and avoid phagocytosis by inhibiting the presentation of phosphatidylserine 12

23 residues at the cell surface [6] (Figure 1.4). In the case of P. yoelii, the merosomes reach the lung where individual merozoites are released and begin the erythrocytic cycle [86]. Figure 1.4 Cartoon representation of merosomes budding and release into the blood stream. Merosomes bud to gain access to red blood cells to initiate an erythrocytic infection. Targeting exoerythrocytic forms: whole organism vaccines Vaccines protecting against infectious diseases have been one of the most potent and cost effective public health measures. Smallpox is the only human infectious disease to be successfully eradicated, and is also the first disease to be safely vaccinated against [87]. The notion that smallpox-infected or cowpox-exposed individuals remained resistant to smallpox infection was well established before the 1750s [88] but it was not until 1796 that Edward Jenner performed the first definitive vaccination. Arguably more importantly, he also communicated his findings to the scientific community at large [89]. Eradication efforts begun in the 1960s by the World Health Organization (WHO) were aided by the absence of an animal reservoir, a stable vaccine, and a protective immune response that lasted for five years after a single immunization [90]. The characteristics of both the pathogen and vaccine were optimal in the case of smallpox eradication. When D.A. Henderson, the head of the successful smallpox eradication campaign, was asked about prospects for additional human pathogen eradication campaigns he said: 13

24 No. Not at this time given presently available relevant technologies, our understanding of the epidemiology of the infectious diseases, and pilot programmes of national disease elimination programmes. [91] This sentiment highlights the need for continued research and development of novel vaccine strategies and public health solutions for diseases such as malaria. The generation of a vaccine capable of blocking either Plasmodium sporozoite invasion of hepatocytes or development of EEFs has been a focus of research for decades. A successful formulation targeting the liver stage has the potential to protect the individual from symptomatic blood stage infection while simultaneously eliminating transmission. The first evidence that a sterilizing immune response was capable of halting the liver stage was generated with whole parasite vaccination, using first irradiated P. berghei sporozoites [92] and subsequently using the bites of infected, irradiated whole mosquitoes [93]. This concept of attenuated whole-organism vaccination was shown to be effective in protecting humans using hundreds of irradiated, infected mosquitoes [94]. This study established the possibility of generating a sterilizing immune response and demonstrated that immune sera reacted to sporozoites (circumsporozoite precipitation test) (Figure 1.5). This anticircumsporozoite reaction was correlated with protection from infection, suggesting an important role for antibodies in whole-organism vaccination with sporozoites [95]. 14

25 Figure 1.5 Visualization of the CSP reaction. Sporozoites treated with normal serum (top) and immune serum (bottom). Adapted from [96] Fig 2. The stage specificity of this response was established by demonstrating that individuals protected from the bite of non-irradiated, infected mosquitoes were still completely susceptible when directly transfused with blood stage parasites [94]. Strain transcendent immunity was evident following vaccination with attenuated P. falciparum sporozoites (Burma) and subsequent protection following challenge with strains from Panama, Malaya and the Philippines [97]. However, this same cross-strain immunity did not protect against the Chesson strain of P. vivax, suggesting a species-specific immunity [97]. These seminal studies established the possibility of sterilizing immunity, even when natural exposure is unable to generate this response [98]. The protective immunity in these studies persisted for up to six months. Ideally, protective immunity would persist through entire transmission and dry seasons and considerable work has been done to extend the immune response following whole-organism vaccinations [99]. Since the administration of 15

26 heat-killed sporozoites did not elicit a protective response, further efforts have been made using metabolically active attenuated parasites [100]. A major consideration since the demonstration of sterilizing protection using attenuated sporozoites for vaccination has been the production hurdles and economic feasibility of large scale formulations [101]. However, various whole-organism vaccines have been successfully developed for veterinary purposes [102]. Development of these veterinary vaccines have benefited from the ability to test formulations directly in their cognate host and in some cases due to a need for only short lived immunity [102]. An alternative to irradiation, attenuation through elimination of liver stage specific genes resulting in immature parasite arrest in the liver, is also being developed [103]. This method has several theoretical advantages compared to irradiation which include exposure of the host to late stage parasite antigens and generation of a homogenously attenuated parasite. Parasites that progress further into the liver stage before arresting have been shown to produce a greater immune response [104]. This observation has directed the discovery of parasite genes important for late stage development in the liver [105]. However, the generation of an immune response against late liver stage antigens is not boosted by additional attenuated sporozoite vaccinations. The inability to further boost immunity against these antigens is due to a dominant immune response against early stage parasites inhibiting EEF development or invasion (as measured by parasite 18S rrna), including a strong response to CSP [106]. It is critical to note for whole-organism vaccine development that genetic attenuation must be complete because breakthrough parasites have the potential to cause disease [107]. Candidate parasite genes are typically identified in rodent species and often 16

27 do not translate into completely attenuated human parasites resulting in the necessity of multiple gene deletions and substantial safety testing prior to human vaccine administration [105,108]. Mechanisms of immune protection The protective mechanisms of the immune response to irradiated sporozoites have been studied to identify correlates of protection. This information has been used to inform subunit vaccine design with the goal of generating a sterilizing immune response comparable to the gold standard of irradiated sporozoites. Initial observations on the ability of sera from mice immunized with irradiated sporozoites to specifically react with the surface of sporozoites [109] and the observation of the CSP reaction correlation with protection in humans [95] suggested that antibodies are important for a sterilizing immune response. Additional studies found that passive transfer of antibodies could protect against P. berghei infection and that monoclonal antibodies against a 44,000 (Pb44) molecular weight protein could abolish infectivity of P. berghei sporozoites in vitro [110,111]. The protein identified in these studies (Pb44), now referred to as the circumsporozoite protein (CSP), is the major surface protein of the sporozoite [112,113] and is shed during sporozoite motility. Inhibition of P. falciparum sporozoite invasion into hepatocytes by anti-csp antibodies was confirmed in vitro [114] suggesting that the observations originally made with rodent Plasmodium species may be applicable to human trophic parasites. However, antibody responses to CSP occur in areas of natural high transmission and do not correlate with protection from development of parasitemia, both symptomatic and asymptomatic [115]. This observation supports the notion that the immune response generated in endemic areas to the pre-erythrocytic stages does not protect against clinical 17

28 infection (blood stage parasitemia) [98]. Several scenarios are possible to explain the presence of serum anti-csp antibodies without measurable protection: 1) the antibodies generated are of a different specificity or avidity (possibly enhancing or neutral for infection) than those generated with irradiated sporozoites, or 2) the antibodies are maintained in the serum at levels below what is required for protection. Even though the ability of anti-sporozoite antibodies to react and inhibit development of EEFs is well characterized, it was first shown in 1977 that mice deficient in B-cell generation could still be protected by irradiated P. berghei vaccination without detectable anti-csp serum reactions [116]. This observation became the start of defining the cell-mediated immune response to pre-erythrocytic stages of Plasmodium. It was then shown, in a similar manner to transfer of protective serum, that cell-mediated protection can also be adoptively transferred [117]. A recombinant Pb CSP vaccine that generated antibody titers higher than those achieved by irradiated sporozoite vaccination, had lower efficacy than whole-organism vaccination [118]. By comparison, the irradiated sporozoite vaccination induced a stronger cell-mediated immune response than the recombinant vaccine. This work highlighted the ability of irradiated sporozoite vaccination to generate both humoral and cell-mediated immune responses, but also demonstrates the importance of the cellular immune response with regards to sterilizing protection. Selective neutralization or depletion of components of the cellular immune response aimed to determine the mechanism of this protection. Studies found that in vivo depletion of CD8 + T cells but not CD4 + T cells ablated the protection from irradiated sporozoites [119,120]. Combinations of T cells and sera in adoptive transfer acted synergistically, supporting the 18

29 notion that irradiated sporozoites induce a complex immune response with a strong cellular response important for sterilizing immunity [119]. Experiments aimed at defining the specificity of the cytotoxic immune response determined that an epitope from P. berghei CSP ( ) could provide protection through passive transfer of T cells in the context of MHC class I molecule H-2Kd [121]. Yet, generation of an anti-csp cytotoxic immune response that provides sterilizing protection requires a higher number of memory CD8 + T cells than other pathogens [122]. The genetic background in murine models is important when determining the details of protection [123]. For instance, to generate a sterilizing immune response to P. yoelii, BALB/c mice require CD8 + T cells, IFNγ, IL-12 and inos whereas C57BL/6 additionally required CD4 + T cells [124]. Host genetics both inside and outside the MHC class I locus determine the ability to elicit sterilizing immunity from irradiated sporozoite vaccination [125]. As noted above, CSP is a major target of the immune response to pre-erythrocytic stages of Plasmodium and has subsequently been of considerable focus for human vaccine design (Figure 1.6). 19

30 A B Figure 1.6 Schematic of CSP and the RTS,S vaccine. (A) Representation of the CSP molecule with an N-terminal signal sequence, conserved regions RI and TSR along with a central repeat region. (B) Portion of P. falciparum CSP fused to the Hepatitis B surface antigen with immunogenic potions indicated. Adapted from [126] Fig 2. Transgenic mice that express P. yoelii CSP in the germ line to force tolerance demonstrated that CSP is the immunodominant antigen in irradiated sporozoite vaccination [127]. However, tolerized mice can be protected when hyper-immunized, showing that even though an anti-csp response is sufficient to drive sterilizing immunity, it is not necessarily required, or that repeated immunization overcame CSP tolerance [128]. This mouse model is important in establishing the presence of non-csp CD8 + dependent antigens and suggests that antigens processed and presented by Kupffer cells or traversed cells could control infection through bystander activation [127]. This mechanism would be in contrast to a direct (or proximal) recognition and killing of P. berghei EEFs by CD8 + T cells 20

31 [129,130], a possible explanation for the discrepancies in species-specific protection of CSP-tolerized mice. The presence of persisting parasites in the liver post-irradiated sporozoite vaccination has been proposed to be a potential mechanism driving sterilizing protection. This process is absent in subunit vaccination and may help explain differences observed in clinical trials [131,132]. This phenomenon was initially identified in rats immunized with P. berghei sporozoites that had detectable parasite DNA up to six months post-vaccination [133]. Administration of primaquine was able to clear persisting parasites and subsequently abrogated the time of protection post-immunization, suggesting that persisting parasites maintain or enhance the immune response [133]. Later work provided evidence that persisting parasites are capable of being presented to the immune system and enhance a prolonged CD8 + T cell response important for protection [134]. In addition, primaquine treatment of mice immunized with genetically attenuated parasites (Pb uis3(-)) lost CD8 + mediated sterilizing immunity, implying that irradiated sporozoite vaccination and genetically attenuated parasite vaccination both rely on parasite persistence [135]. The choice of inbred mice as a model for Plasmodium infection has been, and will continue to be, paramount in understanding basic parasite biology and the critical components of the host immune response. However, the natural reservoir for P. berghei is the tree rat Grammomys surdaster, which requires a lower infectious dose of sporozoites for progression to a blood stage infection and which is considerably more difficult to protect with irradiated sporozoite vaccination than mice [136]. This is expected when considered in the context of extensive host-parasite co-evolution and mirrors the difficulty of whole-organism vaccination in humans [137]. These differences could be attributable to 21

32 factors such as parasites driving hepatocyte resistance to apoptosis [138] or antigenic mimicry resulting in T cell tolerance and immune evasion [139]. Consequently, mice represent an invaluable resource to study Plasmodium, but their use with a non-native parasite may not always provide conclusions relevant to all Plasmodium species or to relevant hosts. In this sense, the difficulty in protecting cognate parasite-host combinations would speak to the difficulty in generating a broadly sterilizing vaccine. Yet, the liver stage of Plasmodium has remained an attractive target for vaccination because parasite biomass is not equal throughout the parasite lifecycle and has a considerable bottleneck during establishment of a liver stage infection. These points in the lifecycle have been proposed to be ideal targets of intervention [140] when considering strategies of elimination and eradication (Figure 1.7). However, the converse argument would suggest that any intervention against these stages requires some degree of perfection since one surviving parasite can continue transmission. At a population scale, even if sterilizing protection is only achieved in a subset of individuals it has the potential to alter transmission. 22

33 B A Figure 1.7 Estimated relative population sizes (biomass) between stages of the P. falciparum lifecycle. (A) Low numbers of Oocysts in the mosquito and (B) injection of few sporozoites represent bottlenecks in the lifecycle. Adapted from [140] Fig 3. Biology of CSP Since the identification of CSP as the major target of the humoral response to irradiated sporozoite vaccination in 1980 [110,111], considerable work has aimed to define its structure, conservation, and importance to parasite biology. Initially the amount of cdna obtainable was a limiting factor in characterizing the sequence of CSP. In 1983, using P. knowlesi sporozoites, a cdna library was constructed containing CSP [141], and the first CSP sequence was reported [142]. This report identified the central tandem repeat region (12 amino acids, 12 times) and demonstrated competitive binding of synthetic repeat peptides with native CSP [142] (Figure 1.8). This central repeat region has since been 23

34 shown to be characteristic of CSP from all Plasmodium species, but varies in the composition and total number of repeats [143]. Figure 1.8 Schematic of the structure of CSP from P. knowlesi based on nucleotide sequence. Areas of charge and tandem repeats were discerned from the cdna sequence of P. knowlesi CSP. Adapted from [96] Fig 3. A protocol to digest intragenic regions of P. falciparum genomic DNA facilitated the generation of libraries to screen for P. falciparum CSP using asexual cultures not expressing CSP mrna [144]. In 1984 [145], the CSP sequence (412aa) was obtained from P. falciparum and several architectural observations were made. This original sequence had 41 tandem repeats, Asn-Ala-Asn-Pro (37 times) and Asn-Val-Asp-Pro (4 times) composing the central repeats region. At the N-terminus the first 16 amino acids were predicted to be a parasite signal sequence and the remaining amino acids up to the repeat region carried an overall positive charge. The C-terminal region to the repeats was predicted to be hydrophobic, and it was postulated that it may constitute an anchor region [146]. Since the P. knowlesi CSP sequence had previously been published, several observations about conserved regions were made. Two regions of homology were noted, region I directly N-terminal to the repeats and region II C-terminal to the repeats (Figure 1.9). The high degree of conservation of region II was suggested to be due to a role in liver invasion by binding to a putative receptor; this region also bears resemblance to the 24

35 adhesive protein thrombospondin and is identified as a type I thrombospondin repeat (TSR) [139]. Experimentally, the TSR has been shown to be important for HSPG binding [147] and mediates initial parasite attachment in vivo [66]. Region I has been demonstrated to be a cleavage site [148] for a parasite protease that removes the N-terminus [149]. This cleavage process is triggered by high levels of HSPGs [64] and unmasks the C-terminal TSR [148]. The N-terminal cleavage event is not required for parasite traversal but is important for productive invasion of hepatocytes [148]. Figure 1.9 Regions of sequence identity in CSP between P. knowlesi and P. falciparum sequences. Region I and Region II (TSR) were identified as being conserved between P. knowlesi and P. falciparum. This observation implied a potential functional role of these regions. Adapted from [145] Fig 6. Paradoxically the repeat regions of CSP appear immunodominant (Pf (NANP3)) in the humoral immune response [150] but the repeat sequences are conserved within each species of Plasmodium [143]. Conservation of the repeats could be due to a functional constraint on parasite biology, but may also suggest that immune pressure is not exerted on this portion of CSP. Antibodies are protective in vitro [150] and in vivo [110,111] but the inability of anti-csp antibodies to protect in settings of natural transmission [98,115] 25

36 suggests that naturally acquired immunity is insufficient to drive selective pressure of the repeats. In contrast, the T cell responses from persons living in malaria-endemic areas were greater against known polymorphic regions of CSP [151,152], implying that certain T cell epitopes may be under immune pressure. However, these polymorphisms in the C terminus appear stable over time in a given region and do not vary between age groups or correlate with hazard of infection [23]. The role of CSP in attachment and invasion in the liver [62,63,65,66, ] has been well characterized. In addition, several reports have suggested that CSP is involved in modulation of the host immune response through inhibition of host cell protein synthesis [153], modulation of Kupffer cell activation [154,155], or inhibition of NFκB nuclear import [156]. Techniques to study the Plasmodium liver stage Since direct access to Plasmodium-infected patient material (especially hepatocytes) is difficult and often cost prohibitive, the development of both in vitro and in vivo models of liver stage infection are paramount in basic research. Study of the erythrocytic stage of P. falciparum has benefited from two achievements. First, the demonstration that asexual cultures could be grown ex vivo using human red blood cells, serum, and a low oxygen environment [157] allowed for generation of considerable parasite material without the need for an infected primate host, such as Aotus. Second, a technique utilizing electroporation was shown to be capable of introducing foreign DNA into the parasite [158]. These methods provided the ability to grow continuous cultures of asexual parasites and also to manipulate their expression. 26

37 Study of the pre-erythrocytic stages of Plasmodium have depended heavily on the establishment of parasite lines capable of infecting laboratory rodents [57] and the development of in vitro methods for cultivating EEFs [159]. Rodent parasites can also be transfected in a similar manner to P. falciparum [160,161]. Using this technique fluorescent parasites have been generated to constitutively express GFP [162]. These lines allow for live parasite identification and the utilization of FACS to purify and profile EEFs [163,164] and their host cells. Plasmodium yoelii parasites expressing luciferase have provided a noninvasive measure of parasite liver development in vivo [165]. The use of these rodent parasites in vitro has benefited from a broad host cell line tropism that allows for complete EEF development in vitro [159]. P. falciparum EEFs can be obtained in primary human hepatocytes [55] but have a limited ability to infect many cells lines that are susceptible to rodent parasites [159]. Primary hepatocytes have long been the only method in which to routinely generate P. falciparum EEFs [55,166] and techniques have been developed to optimize this procedure [167]. Identification of P. falciparum EEFs in vitro has relied on simple staining techniques. Considerable effort has been made to identify cell lines that could replace the need for primary human hepatocytes for in vitro infections and thereby reduce costs and to better standardize experiments across labs. To date only two cell lines have been reported to allow for complete development of P. falciparum EEFs in vitro: the cell line HHS-102 [168] reported extremely low levels of infection and the cell line HC-04 [169] reported more substantial levels of infection of both P. vivax and P. falciparum. Recent development of a P. falciparum strain constitutively expressing GFP [170] and the availability of susceptible cell lines suggest that techniques 27

38 pioneered using rodent parasite lines may also be capable of routinely isolating and identifying P. falciparum EEFs in vitro. An alternative approach to the growth of P. falciparum EEFs in vitro has been the development of mice with a liver partially reconstituted with human hepatocytes [85]. Generation of these liver chimeric mice is prohibitively expensive for routine use and the technique used to reconstitute the liver completely ablates the immune system, making study of an in vivo immune response impossible [171]. Similar to rodent lines expressing luciferase, a P. falciparum line expressing a GFP-luciferase transgene has been developed [172] but only luciferase is detectable in hepatocytes. Therefore, the development of new techniques that can be applied to both in vitro and in vivo models of P. falciparum EEF development will aid in answering basic questions about host-pathogen interaction and will potentially serve as methods to validate interventional approaches such as vaccine development. Concluding remarks The transmission of Plasmodium parasites has continued throughout written human history. Only relatively recently has a microbial cause of malaria been determined. Yet, considerable strides have been made to parse the determinants of Plasmodium persistence and the correlates of the host response. This understanding takes place in the context of millennia of co-evolution, resulting in an evasive parasite that has left a distinct imprint on the human genome. The scientific community continues to investigate unknown facets of the host-pathogen interaction and to design potentially protective measures. An expanding understanding of parasite biology will continue to rely on the available techniques and the relevance of our models. 28

39 Chapter 2 : Processing and presentation of two forms of CSP on MHC class I 29

40 Introduction Presentation of peptides on MHC class I The importance of the thymus in the immune response was initially characterized by thymectomy. These experiments showed profound, systemic reductions in lymphocytes and impaired cell-mediated immunity [173]. The difference in this phenotype compared to removal of other lymphoid organs designated the thymus as a central lymphoid organ. Subsequent work focused on defining the cell populations present in the thymus and their roles in the immune response. Using allogeneic stimulation, it was shown that a subset of the thymus-derived cells had specific lytic activity [174] and that other cells could provide helper activity to the immune response [175]. The commitment step of T cells to follow these different functional subsets was shown to occur before stimulation and could be delineated by surface markers from unlinked genetic loci [176]. These experiments using allogeneic stimulation determined the ability of T cells to specifically lyse or to provide help to non-thymus immune cells during the immune response. However, classification of effector function subsets did not determine the stimuli responsible for an immune response in a syngeneic system. In the context of lymphocytic choriomeningitis (LCM) infection it was found that adoptively transferred T cells could home to lymphoid tissues in both syngeneic or allogeneic hosts, but could only continue to proliferate in a syngeneic host (specifically the H-2 gene complex) [177,178]. Yet, the restriction of lysis to genetically matched targets did not explain the mechanism of recognition of viral antigen and could not be explained by activation from free virus. Rather, the viral infection resulted in what was originally 30

41 termed altered self and suggested the importance of viral modifications to a component of the H-2 complex [178]. These findings set the original framework wherein specific lysis by T cells requires recognition of pathogens (peptides) in the context of self (MHC class I). Target cells are recognized during the effector phase of a CD8 + T cell immune response through TCR recognition of peptides presented in the context of MHC class I. For antigen presentation to occur, proteins are degraded and the resulting peptides are loaded onto MHC class I, which is trafficked to the cell surface [179] (Figure 2.1). Figure 2.1 Schematic representation basic antigen presentation on MHC class I. 31

42 Degradation of antigens and translocation of peptides for presentation of peptides in the context of MHC class I at the cell surface. Adapted from [179] Fig. 1. CD8 + T cell recognition of peptide complexed with MHC class I at the cell surface triggers the activation of CD8 + T cells with cytolytic potential [179]. Direct cytotoxic T lymphocyte (CTL)-mediated lysis can be achieved through two main mechanisms. First, mice deficient in the pore-forming protein perforin were shown to be deficient in lysis of both virally infected and allogeneic targets [180]. Alternatively, death receptor mediated cytotoxicity, including members of the tumor necrosis factor (TNF) family such as Fas/FasL [181,182] and TRAIL [183,184] can induce apoptosis of a target cell through a perforin-independent process. In addition to direct cytotoxicity of target cells, CTLs have the potential to control infection through non-cytolytic mechanisms [185] that include activation-dependent cytokine and chemokine production [186]. The MHC molecule is a cell surface glycoprotein with two structural motifs. The distal portion to the cell membrane is composed of eight antiparallel β-strands with two α- helix domains above [187,188]. The pocket between helices forms a groove that is filled by 8-10 amino acid peptides (Figure 2.2). 32

43 Figure 2.2 Schematic of the structure of HLA-A2. Top down view of the binding pocket for peptides in the groove of MHC class I. β-strands form the bottom of the groove with two α-helix domains shown. Adapted from [187] Fig. 2. The binding of peptides in this groove provides the necessary stability for the MHC class I heterodimer to be trafficked from the endoplasmic reticulum (ER) to the cell surface and prevents the translocation of empty MHC molecules to the cell surface [189,190]. Unlike the presentation of peptide on MHC class II, the presentation of peptides on MHC class I is generally unaltered by the inhibition of endosomes or lysosomes [191]. This finding implicated cytoplasmic or nuclear generation of peptides for subsequent presentation. Using inhibitors of the proteasome it was shown that a majority of peptides loaded onto 33

44 MHC class I molecules were generated through proteasomal-dependent degradation [192] and ubiquitination of target proteins [193]. However, alternative pathways for the generation of peptides for presentation on MHC class I have been suggested, including a role for autophagy in antigen processing [194]. These peptides have been shown to be transported from the cytoplasm to the ER for loading onto MHC class I molecules by the transporter of antigen presentation (TAP) protein [195]. Each MHC class I allele has the potential to bind a myriad of peptides, but the exact repertoire is dependent on the individual characteristics of the binding pocket that accommodates peptide loading [195]. Consequently, on a population scale, the MHC locus represents the most polymorphic portion of the human genome and also has some of the highest recombination potential [196]. In fact, HLA-B is the most polymorphic gene identified in the human genome with a calculated 86 single nucleotide polymorphisms (SNPs) per kilobase and a recorded 1,605 alleles with 1,268 unique peptide sequences [197,198]. This diversity is attributed to the Red Queen hypothesis in which organisms must continually evolve to avoid extinction [199]. However, many pathogens, most notably viruses, have evolved capabilities to impair antigen presentation on MHC class I, presumably to prevent host cell lysis by CTLs. These methods include inhibition of antigen degradation, transportation, and MHC class I complex trafficking [200,201]. Targeting Plasmodium: role of the MHC class I antigen processing and presentation pathway Several facets of vaccine design have focused on blocking sporozoite entry into the liver and/or eliminating EEFs before they reach maturity to prevent progression to blood 34

45 stage. The circumsporozoite protein (CSP) is the major protein on the surface of Plasmodium sporozoites and contains both B and T cell epitopes [132]. Rodent models of malaria, particularly in the BALB/c host, have demonstrated the importance of a CD8 + T cell mediated, CSP-specific response in protective immunity [121,202]. In rodent models CSP is the major antigen involved in protection from whole organism-based vaccines and is a leading candidate for generation of a protective vaccine for use in humans [127,203]. The only stage of the Plasmodium life cycle that involves the infection of MHC class I expressing cells is the liver stage. In mouse models of vaccination with irradiated sporozoites, the gold standard of inducing a sterilizing immune response to the Plasmodium liver stage, is dependent on the generation of CD8 + T cells [119,120]. CSP is the immunodominant antigen in this model [127]. CD8 + T cell clones specific to C-terminal peptides of CSP are sufficient to provide liver stage specific immunity and protection in both P. berghei [121] and P. yoelii [204]. These protective CD8 + T cells are initially primed in the draining lymph node [205] and are maintained as a resident memory population in the liver [206]. In mouse models utilizing irradiated sporozoite immunization, sterilizing protection that is dependent on CD8 + T cells was not mediated by perforin/granzyme or Fas mechanisms of killing [124]. Rather, in these mouse models IFN-γ produced by activated CD8 + T cells was a correlate of protection, suggesting the potential for control of EEFs without direct recognition. However, in a P. yoelii model, IFN-γ deficient CS specific CD8 + T cells were able to eliminate liver stage parasites similarly to wild-type [207]. This study was in contrast to data obtained using IFN-γ deficient mice and IFN-γ neutralization experiments, suggesting a potential non-cd8 + source of IFN-γ or potential systemic defects in IFN-γ deficient mice. 35

46 During natural transmission, CD8 + T cells are primed against liver stage antigens [152,208] and antibodies are generated that target the sporozoite [209]. Yet, a sterilizing effector immune response is not achieved in naturally exposed individuals but rather they gain clinical protection from disease caused by the erythrocytic stage [98]. Possible reasons for the absence of a naturally generated protective liver stage immune response include parasite allelic diversity [23,210], parasite resistance to immune effector mechanisms [138] or failure to maintain a protective memory T cell population [211]. However, the HLA-B53 allele has been implicated in protection from severe malaria by recognition of liver stage antigen-1 (LSA-1) [ ]. This HLA-B allele is enriched in West African populations, possibly due to selection for a particularly protective immune response. Immune protection of naturally exposed individuals through a specific MHC class I allele directed against a liver stage antigen [212] may seem paradoxical since sterilizing liver stage immunity fails to develop [98]. Measurement of protection of HLA types used severe malaria as a metric of protection [213] as assessed by passive surveillance and not protection from development of blood stage parasitemia. Yet, a lack of natural sterilizing immunity was found using a prospective study [98] that did not rely on clinical metrics and measured time to blood stage infection in each individual. Importantly, this study did not utilize a drug treatment baseline to avoid recrudescent infections [215] nor skewing of the immune response [216,217]. Therefore, a natural immune response targeting the liver stage may not provide sterilizing protection, but could, through an undefined mechanism, influence the severity of a resulting blood stage infection and influence clinical disease. This could help explain the protective nature of a MHC class I allele in the absence of an observed sterilizing immunity. 36

47 The inability of immunity to protect against Plasmodium infection in naturally exposed individuals [98] even in the presence of CD8 + T cells specific to CSP [152] suggests that the parasite may have evolved to inhibit presentation of CSP on MHC class I. In fact, CSP has been shown to be capable of altering host cell protein synthesis [153] and promoting development through interference of NFκB signaling [156]. The generation of a protective CD8 + T cell mediated response to CSP requires a high level of memory CD8 + T cells [122] in animal models. This requirement may reflect a parasite driven resistance to apoptosis [138] or an ability of the parasite to limit the amount and/or scope of peptides capable of being presented on MHC class I. Alternatively, the observation that immunodominant T cell epitopes of CSP are also regions of a relatively higher number of polymorphisms [151] suggest that the parasites may be capable of escaping natural T cell immune pressure. In contrast to blood stage immunity [210], clear, CSP allele-specific protection has not been established in humans [23]. This may be due to the differences in CSP derived peptides presented by various HLA types [152], extensive HLA diversity in Africa [218] or inability of naturally exposed individuals to be protected by low percentage anti-csp CD8 + T cell response. In this case, the interaction between parasite antigen alleles and individual HLA would ultimately determine the ability of CSP to be processed and presented in a single individual. Design of a vaccine capable of completely halting the liver stage must overcome these hurdles inherent in the natural immune response. Various adjuvant formulations are capable of enhancing the magnitude of the immune response but do not alter the breadth of the immune response that may be importance in a setting of heterologous challenge. Additionally, identify mechanisms the parasite may employ to misdirect or avoid a 37

48 protective immune response is essential in informing an intelligent vaccine formulation. The RTS,S vaccine is designed to target the sporozoite and liver stage of P. falciparum and is currently seeking licensure. The formulation is composed of a portion of CSP fused to the hepatitis B surface antigen (HBsAg) in combination with an adjuvant formulation (Figure 1.6). Active detection has observed a reduction in time to measurable parasitemia [131,219] and clinical data show a reduction in severe disease [131, ]. However, protection is not sterilizing and differs between age groups. The roles of antibodies and CD4 + T cell in protection have been modeled [222] and suggest that infections may be the result of individual sporozoites productively infecting the liver. Antibody titers against CSP correlate with protection in non-endemic human volunteers [223,224]. In these studies a level of 20µg/ml anti-csp antibody was considered a general threshold necessary for protection. Re-formulations and adjuvant designs are aimed at boosting the immune response to RTS,S with the goal of inducing sterilizing immunity. Additionally, the current CSP- HBsAg fusion protein does not include the N-terminus of CSP and it is unclear if inclusion of a full length CSP could influence a protective immune response or could enhance the limited cytolytic response observed in prior studies [225,226]. Two forms of CSP as targets of the CD8 + T cell immune response Prior to invasion, a parasite protease cleaves the N-terminus of CSP at the conserved region I, adjacent to the repeat region [149]. Consequently, CSP may exist in two forms in the host (Figure 2.3). The majority of work involving cleavage of CSP has 38

49 focused on the necessity of cleavage during invasion and subsequent importance to parasite biology. It is unclear what forms of CSP are present in host hepatocytes. Nothing is known about the individual contributions of these forms to antigen presentation and target cytotoxicity has not been addressed. Understanding both the nature of CSP found in host cells and the ability of these forms to be processed and presented is important for understanding any parasite mechanisms to avoid the immune response. Figure 2.3 Schematic representation of CSP during infection. Proposed conformational changes to CSP that occur from the mosquito to the eventual infection of hepatocytes (N-terminus in green). Adapted from [148] Fig 8. Here we describe differences in T cell activation, target cytotoxicity, and ubiquitination between two forms of CSP using an ectopic expression system. We identify which form of CSP predominates in cells traversed and infected by the Plasmodium parasite. These data suggest that inclusion of N-terminal lysine residues of CSP are important in antigen presentation and their removal may represent a parasite mechanism to lessen antigen processing and presentation. 39

50 Results Ectopic expression of CSP in hepatocytes We constructed mammalian expression plasmids capable of expressing either the full-length circumsporozoite protein (CSP-Full) or the cleaved N-terminal form (CSP- Short). Due to the difficulty in obtaining and identifying lymphocytes specific to CSP from naturally exposed individuals we designed a chimeric CSP sequence that contained well characterized HLA-A2 epitopes. To evaluate presentation of CSP via MHC class I and the resulting CD8 + T cell activation we inserted known MHC class I viral epitopes from either influenza A matrix protein 1 or Epstein-Barr virus protein BMLF1 in the C-terminus (Figure 2.4A). The amino acid substitutions used to alter the putative epitopes in CSP were not predicted to be appreciably different than the native epitope [227]. A HLA-A2 restricted response to the native epitope, GLIMVLSFL is primed and recall responses can be identified [152]. We utilized a model hepatocyte cell line, HC-04 that is reported to have the ability to support P. falciparum EEF development in vitro [169]. Using our CSP expression constructs we determined that CSP can be expressed in HC-04 following transfection, and that CSP-Full and CSP-Short can be differentiated by their electrophoretic mobility (Figure 2.4B). Both CSP-Full and CSP-Short contain the central repeat region (NANP) that is recognized by the monoclonal antibody 2A10 [150]. To determine if the N-terminus influenced the cellular localization of CSP in hepatocytes we transfected cells with either CSP-Full or CSP-Short and stained the central repeat region (IFA). We observed that both CSP-Full and CSP-Short are predominantly cytoplasmic and have similar localization 40

51 patterns in hepatocytes, when expressed ectopically (Figure 2.4C). In contrast to previous reports we do not observe localization of CSP to the host nucleus [156]. However, in vivo CSP can be delivered or expressed in hepatocytes in a variety of ways, resulting in differential localizations over time including a diffuse cytoplasmic pattern (data not shown). Therefore, the localization of ectopically expressed CSP is representative of at least a portion of CSP found in vivo and has the potential to be classically processed for presentation on MHC class I. Identification of CSP in traversed and infected hepatocytes during in vitro infection Plasmodium sporozoites have been observed to interact with hepatocytes prior to productive invasion and parasitophorous vacuole formation [72]. This process of traversal, which also occurs in the skin, involves breaching of the host cell plasma membrane, intracellular movement, and egress [228]. Since CSP is shed during parasite motility [229] we aimed to identify which form of CSP is present in traversed cells and has the potential to activate CD8 + T cells. We sorted traversed hepatocytes from P. falciparum-infected cultures and identified the presence of both forms of CSP (Figure 2.4D). We primarily detect the presence of the cleaved form of CSP in traversed cells as compared to our ectopically expressed variants of CSP. To identify the form of CSP present in infected cells, we sorted cultures infected with a GFP-expressing P. falciparum parasite 72 hours postinfection. Infected hepatocytes were found to contain CSP-Short (Figure 2.4E). Consequently, the short form of CSP is present and predominates in both traversed and infected cells. Even though CSP transcript 41

52 has been identified in EEFs [230], it is unclear if CSP is translated in the EEF and cleaved or the presence of cleaved CSP in EEFs originated in the sporozoite. Next, we asked if the two forms of CSP have a differential ability to be processed and presented on MHC class I and presumably have different potential contributions to the effector phase of a CD8 + T cell immune response. Figure 2.4 Ectopic and natural expression of the two forms of P. falciparum CSP. (A) Schematic representation of the mammalian expression construct pvitro2-neo-mcs encoding CSP and a marker of transfection, the murine CD8 alpha chain. C-terminal chimeric sequences are indicated. (B) Western blot detection of CSP-Full (F) and CSP- 42

53 Short (S) at 12, 24 and 48 hours post-transfection. (C) Immunofluorescence detection of CSP in HC-04 cells ectopically expressing either CSP-Full or CSP-Short. (D) Western blot detection of CSP in HC-04 cells traversed with P. falciparum sporozoites compared to the electrophoretic mobility of ectopically expressed CSP-Full and CSP-Short. Traversed (dextran+) and non-traversed (dextran-) cells were separated by flow cytometry based cell sorting. (E) Western blot detection of CSP in HC-04 cells infected with P. falciparum 3D7- HTGFP compared to the electrophoretic mobility of ectopically expressed CSP-Full and CSP-Short. Infected (GFP+) HC-04 were isolated 72 hours postinfection by flow cytometry based cell sorting. Two forms of CSP differentially activate CD8 + T cells The conservative substitution of the native C-terminal sequence of CSP (GLIMVLSFL) with two well characterized HLA-A2 restricted viral peptides allows the use of CD8 + T cells that are readily identifiable with existing reagents. The wild-type CSP epitopes and viral epitopes have similar predicted binding strengths to HLA-A2 [227]. To generate CD8 + T cell effectors and establish lines of monospecific, polyclonal CD8 + T cells we first identified HLA-A2 positive donor PBMCs. These donor lymphocytes positive for HLA-A2 were stained with pentamer specific to either GLCTLVAML or GILGFVFTL in the context of HLA-A2 (Figure 2.5). These cultures were co-cultured with irradiated target hepatocytes pulsed with synthetic peptide (GLCTLVAML or GILGFVFTL) and expanded on allogenic feeders. After expansion, cultures were separated based on CD8 positivity and pentamer positivity by fluorescence-activated cell sorting (FACS) (Figure 2.5). 43

54 Figure 2.5 Identification, isolation and expansion of HLA-A2 restricted viral peptide specific CD8 + T cells. Lymphocytes of HLA A2-positive blood donors were screened for the presence of CD8 + HLA A2-restricted CTLs specific to GILGFVFTL (peptide sequence from influenza A, 44

55 matrix protein 1) and GLCTLVAML (peptide sequence from Epstein-Barr virus, BMLF1) peptide epitopes. Three lymphocyte cultures contained either GLC and/or GIL-specific CD8 + CTLs identified by pentamer staining (panel screening ). Lymphocyte cultures were stimulated with relevant synthetic peptide and expanded on allogenic feeders (panel expansion ). Peptide-specific CD8 + CTLs were sorted using specific pentamers and further expanded on allogenic feeders for expansion (panel post-sorting and expansion ). Specific activation of sorted and expanded CD8 + T cell cultures was demonstrated by using targets pulsed with either GLCTLVAML or GILGFVFTL synthetic peptides and subsequent measurement of intracellular IFNγ production by flow cytometry (Figure 2.6). 45

56 Figure 2.6 Verification of CTL specificity. HC-04 cells pulsed with either synthetic GLCTLVAML (EBV, BMLF1) or GILGFVFTL (Inf A, M1) peptide were used as stimulators for activation of sorted and expanded lymphocyte cultures. Intracellular IFNγ was measured after 8 hours of co-incubation using an anti-ifnγ antibody followed by flow cytometry. 46

57 Our CSP expression constructs also include an independent transcript encoding the murine CD8 alpha chain (mcd8a) (Figure 2.4A). This molecule is trafficked to the cell surface and can be used to identify transfected hepatocytes (Figure 2.7A). Live, transfected cells can be purified using magnetic-activated cell sorting (MACS) based on mcd8a surface expression (Figure 2.7B). Figure 2.7 Identification and purification of transfected HC-04. (A) Staining of transfected HC-04 with anti-mcd8a or isotype control antibody 24 hours after transfection. (B) Purification of transfected cells 24 hours post-transfection by antimcd8a beads. MACS column flow-through (solid line) and purified fraction (dashed line) are shown. 47

58 Using purified effectors (CD8 + T cells) and targets (mcd8a + HC-04 cells) we determined that targets expressing CSP-Short were less efficient at activating CD8 + T cells than targets expressing CSP-Full in a HLA-A2 restricted peptide-specific manner (Figure 2.8). CD8 + T cells were only activated by their cognate chimeric CSP and the observation of differential activation was seen in three of the four separate T-cell cultures. Figure 2.8 MHC Class I restricted recognition of cells expressing CSP chimeras by peptide specific CD8+ T-cells. 48

59 HC-04 cells transfected with DNA constructs coding for different CSP chimeras were sorted by MACS and used as stimulators for viral peptide-specific CTLs at 1:1 ratio. Flowcytometry-based detection of intracellular IFNγ was measured after 8 hours of coincubation. Even though the T-cell culture GIL displayed antigen specificity (Figure 2.6), it responded minimally to transfected targets. Targets expressing one of two forms of CSP are differentially lysed by CD8 + T cells After determining that hepatocytes expressing CSP-Full and CSP-Short differentially activate CD8 + T cells, we tested the ability of these targets to be lysed by CD8 + T cells. We detected cytotoxicity by measuring the presence of an enzymatically active cytosolic protein, lactate dehydrogenase (LDH) using a commercially available kit, following addition of effector T cells. Similarly to experiments measuring T cell activation, targets expressing CSP-Short were less susceptible to lysis by antigen specific CD8 + T cells (Figure 2.9A). We hypothesized that the observed difference in cytotoxicity was due to different molar amounts of processed and presented peptide at the hepatocyte surface. Therefore, we attempted to indirectly measure the amount of peptide presented on the hepatocyte surface and quantify the difference we see between CSP-Full and CSP-Short We generated a cytotoxicity curve by titrating synthetic peptide on hepatocytes and fit a line using a one site specific binding model (y=bmax*x/(kd+x)) (Figure 2.9B). Simultaneously we repeated the cytotoxicity assay using MACS-purified chimeric CSP expressing targets. Using the cytotoxicity measurements of CSP expressing targets and the titration curve, we determined indirect estimations of presented peptide on the target 49

60 surface (Figure 2.9B). The relative molar values suggest that 2.5 fold more peptide is presented on the cell surface in CSP-Full than CSP-Short expressing hepatocytes. However, these measurements assume that the inherent susceptibility of hepatocytes to cytotoxicity is the same between CSP-Full and CSP-Short. Figure 2.9 Specific lysis of targets expressing CSP-Full or CSP-Short by CD8+ T- cells. (A) Release of LDH from MACS sorted HC-04 targets expressing CSP-Full-GLC or CSP- Short-GLC chimeras upon co-incubation with GLC CD8 + T cells used as effectors. Mean±SD of triplicates are shown. Significance was determined by two-way ANOVA, post-hoc comparisons shown, **p<0.01. (B) Titration of GLCTLVAML synthetic peptide on HC-04 targets and resulting cytotoxicity by GLC CD8+ T-cells. A best fit line was determined using a one site specific binding model. Concurrent cytotoxicity of transfected HC-04 is shown with indirect quantification of presented peptide. 50

61 Two forms of CSP do not alter immune phenotypes of targets or overall susceptibility to death stimuli In addition to peptides presented on MHC class I, resistance or susceptibility to cytotoxicity can be influenced by the target cell surface phenotype or by an intrinsic resistance to apoptosis or apoptotic stimuli. To determine if the differential susceptibility to cytotoxicity is in fact due to differences in presented peptide, we surface stained for the presence of CD54, an adhesion molecule important for cell-cell contact, and total MHC class I (Figure 2.10A) on cells expressing the two forms of CSP. We did not observe a significant difference in the intensity of staining for the adhesion molecule CD54 or MHC class I complexes at the surface of hepatocytes expressing CSP-Full or CSP-Short. Different responses to engagement of the death receptors Fas or TRAIL Receptors have the potential to influence target cytotoxicity. We engaged these death receptors with recombinant TRAIL or anti-fas antibodies to evaluate their individual contributions to cytotoxicity and found no difference in cytotoxicity between CSP-Full and CSP-Short (Figure 2.10B). Our initial measurement of cytotoxicity were made using CD8 + T cells. Consequently, we evaluated if the presence of CSP alters overall target susceptibility to CD8 + mediated cytotoxicity by using constructs that express wild-type CSP and therefore do not contain the chimeric viral epitope recognized by our CD8 + T cells. By pulsing these wild-type CSP expressing targets with synthetic peptide we activated T cells independently of peptides derived from CSP. Using this methodology we determined that targets expressing CSP-Full and CSP-Short are lysed with equal efficiencies when loaded with exogenous peptide (Figure 2.10C). These data support the conclusion that targets expressing either form of CSP maintain their susceptibility to apoptosis but differentially 51

62 process and present epitopes on MHC class I at the plasma membrane. Therefore, the difference in killing between targets expressing CSP-Full or CSP-Short may be due to different abilities of the two forms of CSP to generate peptides for presentation on MHC class I. Figure 2.10 Susceptibility of targets expressing two forms of CSP to T cell effector functions. (A) Expression of CD54 and MHC Class I on the surface of HC-04 cells transfected with plasmids encoding CSP-Full or CSP-Short. (B) Susceptibility of transfected HC-04 cells to cytotoxicity mediated by SuperKiller TRAIL or CH11. (C) Cytotoxicity by GLC CD8 + T cells of HC-04 cells expressing CSP-Full-WT or CSP-Short-WT pulsed with or without synthetic GLCTLVAML peptide. Mean±SD of triplicates are shown. Significance was determined by two-way ANOVA, post-hoc comparisons shown, ***p<

63 Half-life of two forms of CSP and dependence of proteasomal degradation on MHC class I presentation Prior to presentation of peptides on MHC class I, proteins are degraded by proteolytic machinery into typically 8-10 amino acid long peptides [231]. We blocked global translation with cyclohexamide and monitored persistence of CSP-Full and CSP- Short over time. We observed no difference in the rate of protein degradation over time between the two forms of CSP (Figure 2.11A) suggesting that they have a similar stability in hepatocytes. Next we sought to determine which pathways are responsible for CSP degradation and necessary for antigen presentation. One of the major pathways leading to generation of peptides on MHC class I is through the proteasome. Blocking proteasomal degradation of proteins using the inhibitor lactacystin significantly reduced the ability of hepatocytes to present both forms of chimeric CSP and to be lysed by CD8 + T cells (Figure 2.11B). Since proteasomal degradation of ectopically expressed CSP is required for presentation on MHC class I we speculated that access to the proteasome through formation of ubiquitin chains may influence the amount of peptides generated for presentation. 53

64 Figure 2.11 Degradation of ectopically expressed CSP in HC-04 cells. (A) Western blot detection and quantification of CSP band intensity following cyclohexamide treatment beginning at 24 hours post-transfection. (B) CTL-mediated lysis of HC-04 targets expressing CSP-Full-GLC or CSP-Short-GLC chimeras with or without prior exposure to lactacystin. Mean±SD of triplicates is shown. Significance was determined by two-way ANOVA, post-hoc comparisons shown, ***p< CSP-Full forms polyubiquitin chains more efficiently than CSP-Short The formation and composition of ubiquitin chains on a substrate has the potential target the substrate to one of many independent proteolytic machineries in the cell [232]. Since presentation of CSP on MHC class I is proteasome dependent, degradation of CSP- Ub complexes by alternative proteolytic pathways may alter the amount of peptide available for the proteasome and subsequent presentation. We used available plasmids [233] that express ubiquitin (Ub) tagged with hemagglutinin (HA) to identify and compare ubiquitination of both forms of CSP. Hepatocytes were transfected with CSP expression 54

65 constructs and Ub-HA constructs at the same time. Lysates were immunoprecipitated for HA and blotted for CSP. We first treated these transfected cultures with inhibitors of the proteasome (lactacystin), lysosome (chloroquine), and autophagy (spautin) to determine the major proteolytic systems that contribute to the degradation of ubiquitinated CSP. Accumulation of CSP-ubiquitin linkages was most prominent during inhibition of the proteasome, suggesting that ubiquitinated CSP-Full and CSP-Short proteins are predominantly degraded by the proteasome (Figure 2.12A). However, we observed substantially more ubiquitin chains formed on CSP-Full than CSP-Short. The greater amount of ubiquitin chains formed on CSP-Full correlates with the observed greater susceptibility to cytotoxicity and dependence on proteasomal degradation for presentation. Since ubiquitin has the potential to form chains using seven lysines we considered the possibility that the types of chains formed on CSP-Full and CSP-Short may also differ. Classically, lysine-48 linked chains are targeted for proteasomal degradation and lysine-63 linkages traffic to the lysosome [234,235]. Using ubiquitin expression plasmids that restrict linkages to only lysine-48 (K48), lysine-63 (K63), or a plasmid that is devoid of lysine (K- KO) we evaluated the contribution of individual chain linkages to Ub-CSP complexes. The two forms of CSP differed in their ubiquitination by wild-type ubiquitin but no appreciable difference was seen in their linkages restricted to K48 or K63 (Figure 2.12B). However, we noted that CSP-Full was able to form linkages independent of ubiquitin lysines (Figure 2.12B). The ability of ubiquitin to form lysine independent, linear chains has been reported but the contribution of these linkages to degradation and antigen presentation have not been well established [236]. 55

66 Figure 2.12 Identification of ubiquitination patterns of two forms of CSP. (A) Transfection of HC-04 with plasmids containing HA-tagged ubiquitin and either CSP- Full or CSP-Short. Cells were lysed and supernatant was immunoprecipitated for HA. (B) Transfection of HC-04 with wild type HA-tagged ubiquitin or mutated ubiquitin to restrict lysine chain formation along with either CSP-Full or CSP-Short. Cells were lysed and supernatant was immunoprecipitated for HA. Individual lysine residues in the N-terminus of CSP are required for efficient MHC class I-restricted antigen presentation We constructed two additional N-terminal truncation mutants of CSP to evaluate the specific contribution of N-terminal amino acids to presentation on MHC class I and cytotoxicity. N-terminal truncations beginning at amino acids 59 and 68 of CSP resulted in a significant reduction in cytotoxicity when compared to CSP-Full (Figure 2.13A). The ability of the truncation mutants to serve as targets supports the notion that N-terminal 56

67 residues present in CSP-Full allow for efficient processing and presentation of C-terminal epitopes on MHC class I. This attribute of CSP would be in contrast to a system in which the N-terminal amino acids specifically found only in CSP-Full are the direct determinants of protein degradation [237]. However, it is uncertain if presentation is altered by the overall N-terminal structure or is dependent on specific N-terminal lysines for ubiquitin chain formation. Three lysine mutants were generated from CSP-Full, by substituting lysines to alanines. Mutation of lysine 58 to alanine significantly reduced cytotoxicity (Figure 2.13B) and in parallel reduced ubiquitination (Figure 2.13C). Mutations at positions 66 and 67 of lysine to alanine did not significantly alter CTL mediated cytotoxicity (Figure 2.13B). However, substitution of both lysine 66 and 67 simultaneously reduced CTL mediated cytotoxicity and the concurrent amount of ubiquitinated CSP (Figure 2.13B,C). These data support the hypothesis that individual lysines present in the N-terminus of CSP contribute to antigen presentation of C-terminal epitopes and that cleavage of the N-terminus of CSP by sporozoites may alter the immunogenicity of CSP. In our system, presentation of ectopically expressed CSP is dependent on peptide generation by the proteasome and can be regulated by altering the ability of ubiquitin chain formation, specifically at the N-terminus of CSP. 57

68 Figure 2.13 Role of the N-terminus on presentation of CSP and the contributions of individual lysines. (A) Cytotoxicity by GLC CD8 + T cells of HC-04 targets expressing N-terminal deletions of CSP. (B) Cytotoxicity by GLC CD8 + T cells of HC-04 targets expressing CSP with individual lysine to alanine mutations. Significance was determined by one-way ANOVA, post-hoc Tukey s HSD comparisons shown, **p<0.01 ***p< (C) Transfection of HC-04 with HA-tagged ubiquitin and CSP with individual lysine to alanine mutations. Cells were lysed and supernatant was immunoprecipitated for HA. 58

69 Discussion In rodent models of malaria, the circumsporozoite protein is the major target of the immune response in whole organism vaccination and can be protective in single antigen vaccination [238]. Currently, the only human vaccine against malaria in phase III trials targets CSP [221]. However, current forms of the vaccine do not provide complete, sterilizing protection [239]. Sterilizing protection is more difficult to achieve when native rodent hosts are infected with their natural Plasmodium pathogen [136]. This supports the idea that host-pathogen co-evolution alters the efficacy of the host immune response to favor transmission of the parasite. Prior to productive invasion of the host hepatocyte, CSP must be cleaved by a parasite protease [148,149]. This cleavage event produces two forms of CSP that have potentially different abilities for presentation, independent of their functional roles in sporozoite invasion. Differences in presentation may be due to alternations in the N- terminal amino acids, overall protein structure or changes in the ability of the protein to be degraded [237]. As the sporozoite moves, CSP is shed from the surface and therefore has the potential to be processed and presented in cells traversed but not infected by the parasite. Additionally, CSP has the potential to be cross-presented by dendritic cells [240]. It is conceivable that shed CSP can activate CD8 + T cells and aid in elimination of EEFs in a bystander fashion. To identify which form of CSP predominates during the process of sporozoite traversal prior to EEF formation, we sorted hepatocytes based on dextran retention six hours after infection. Both dextran positive and dextran negative populations from the same infected culture contained CSP but the amount of CSP in traversed cells was considerably 59

70 greater, suggesting that the act of traversal leads to deposition of CSP in hepatocytes. The presence of CSP in dextran negative cells is possibly due to extracellular, attached sporozoites, the presence of a proportion of traversed cells that did not take up dextran or from infected cells. In traversed cells we detect both forms of CSP, however, the cleaved form predominates (Figure 2.4D). These data support the idea that any CD8 + T cell activation by traversed cells is predominantly due to the immunogenicity of the cleaved form of CSP. Previous imaging studies in mice show that CD8 + T cells in close proximity to EEFs can eliminate parasites, presumably though direct recognition of peptides presented by infected cells [241]. We sorted live hepatocytes three days after infection with a GFP-expressing P. falciparum line [170] and determined that developing EEFs contain the cleaved form of CSP (Figure 2.4E). Transcriptome analysis determined that CSP transcript is detectable in EEFs [230]. The detection of only the cleaved form of the protein could be due to CSP being both translated and cleaved in the EEF or because CSP found in the EEF predominantly originated in the sporozoite. In either scenario the cleaved form of CSP predominates in EEFs and is potentially the antigen processed and presented if direct recognition of EEFs by CD8 + T cells does occur. To assess the immunogenicity of these two forms, we generated a CSP expression system that allows for expression of either CSP-Full or CSP-Short and contains a chimeric C-terminal sequence capable of being presented in the context of HLA-A2. This HLA allele is present across Africa [218] and has been used to identify the presence of naturally occurring reactivity to CSP epitopes [152]. Using this expression system in a HLA-A2 positive hepatocyte line, HC-04 and purified CD8 + T cells specific to viral epitopes (Figure 2.6), we determined that T cells are differentially activated by these two forms of CSP 60

71 (Figure 2.8). Three of the four T cell lines were capable of being activated by CSP expressing hepatocytes only if the relevant peptide epitope was present in the C-terminus of CSP (Figure 2.8). Of the T cell lines that were specifically activated, all three had higher percentages of IFNγ producing CD8 + T cells when activated by hepatocytes expressing CSP-Full rather than CSP-Short. This difference was observed using two epitopes and three different T cell lines, supporting the assertion that the absence of the N-terminus of CSP hinders activation of T cells specific to a C-terminal peptide. Next, we determined whether the difference in T cell activation of CSP-Full and CSP-Short resulted in differential target cytotoxicity. Using release of lactate dehydrogenase as a metric for cytotoxicity we found that CSP-Full-expressing cells were killed at a higher percentage than those expressing CSP-Short (Figure 2.9A). This trend is consistent with the difference in T cell activation. One possibility is that different amounts of peptides are generated from CSP-Full and CSP-Short and that this leads to different numbers of specific peptide-mhc complexes reaching the hepatocyte surface. After considerable effort we were unable to obtain enough peptide from the cell surface to quantify our specific epitope by mass spectrometry. Instead we used an indirect measure of peptide by titrating synthetic peptide on the cell surface and generating a killing curve as a function of peptide (Figure 2.9B). Using this curve and simultaneous measurements of cytotoxicity of CSP expressing targets we estimated that over 2.5 fold more peptide is presented at the cell surface from CSP-Full than CSP-Short (Figure 2.9B). However, we were unable to formally exclude the potential contribution that different forms of CSP might have on the inherent ability of hepatocytes to activate or to be killed by CD8 + T cells. CSP has been shown to alter host cell processes to promote EEF 61

72 development [153,156], but it is unclear if CSP alone can modulate hepatocyte sensitivity to cytotoxicity. We measured the levels of MHC class I and the adhesion molecule CD54 in CSP-transfected cells and found no difference in their surface staining phenotypes (Figure 2.10A). To test the susceptibility of CSP expressing targets to cytotoxicity independent of antigen processing, we triggered death by crosslinking of TRAIL receptors or FAS. We did not observe any difference in the susceptibilities of CSP expressing targets to TRAIL- or FAS-induced lysis (Figure 2.10B). Since we did not observe any difference in surface phenotype or susceptibility to death receptor mediated cytotoxicity, we assessed if two forms of wild-type CSP could alter susceptibility to CD8 + mediated cytotoxicity when targets were pulsed with exogenous peptide. In this case we did not observe any difference in cytotoxicity (Figure 2.10C), supporting the conclusion that the two forms of CSP do not differentially alter the intrinsic ability of targets to be killed by T cells. These data are in line with previous observations [242]. These data suggest that the differences in T cell activation and cytotoxicity by CSP-Full or CSP-Short are due to differences in the processing and presentation of epitopes in CSP. To better understand this process we needed to determine what pathways are involved in CSP degradation leading to presentation on MHC class I. First, by halting nascent translation of CSP and following CSP over time we determined that CSP-Full and CSP-Short do not have appreciable differences in their halflives when ectopically expressed (Figure 2.11A). Since both forms of ectopically expressed CSP have similar stabilities we aimed to determine what proteolytic systems degrade these proteins and have the ability to contribute to antigen presentation. Degradation of proteins by the proteasome is the major source of peptides loaded onto MHC class I. Using 62

73 lactacystin to inhibit degradation of proteins by the proteasome we determined that both CSP-Full and CSP-Short presentation on MHC class I are completely dependent on the proteasome. If both forms of CSP require access to the proteasome for generation of peptides to be presented on MHC class I, we hypothesized that access to proteasomal degradation may influence the ultimate generation of antigenic peptides. Proteins are targeted to the proteasome after being covalently tagged with ubiquitin, which forms multisubunit chains at lysine residues. To specifically identify ubiquitin chains on CSP we expressed a HA-tagged ubiquitin along with CSP, immunoprecipitated for HA, and blotted for CSP. We first asked which proteolytic pathways are responsible for degrading ubiquitinated CSP by chemically inhibiting their proteolytic activity. We observed accumulation of ubiquitinated CSP only with inhibition of the proteasome, suggesting that the lysosomal and autophagic pathways are not influencing access of ubiquitinated cytosolic CSP to the proteasome. However, upon inhibition of the proteasome we observed more ubiquitination of CSP-Full than CSP-Short (Figure 2.12A). This difference suggests that the two forms of CSP form ubiquitin linkages at different efficiencies and is consistent with CSP-Full presenting more peptide on MHC class I than CSP-Short. Ubiquitin has the potential to form lysine chains in a variety of linkages [232]. The types of linkages formed have the potential to direct protein degradation through different pathways. For instance, linkages at lysine 48 (K48) direct proteins for proteasomal degradation and linkages at lysine 63 (K63) direct proteins to the lysosome [234,235]. To determine if CSP-Full and CSP-Short differ in the types of ubiquitin chain linkages they form we transfected with HA-tagged ubiquitin that is restricted to K48, K63, or lysine independent chain formation. We found that the amount of K48 and K63 chains was similar 63

74 between CSP-Full and CSP-Short. However, CSP-Full was able to form more lysine independent ubiquitin linkages than CSP-Short (Figure 2.12B). To our knowledge the contribution of lysine independent ubiquitin linkages to antigen presentation has not been previously investigated in any system. Since CSP-Full forms more ubiquitin linkages and is more susceptible to cytotoxicity than CSP-Short we aimed to determine what characteristics of the N-terminus of CSP contribute to presentation of CSP-Full. We generated two intermediate N-terminal truncation mutants that were shorter than CSP-Full but longer than CSP-Short. We measured the ability of these mutants to be killed by CD8 + T cells and determined that both mutants were killed less than CSP-Full (Figure 2.13A). These data support the notion that amino acid residues in the N-terminus influence presentation of CSP. Both these mutants contain the region I putative cleavage site but have alterations in their presentation. Since the difference in cytotoxicity correlates with differences in ubiquitination, we hypothesized that specific lysine residues in the N-terminus may be important for ubiquitin chain formation, subsequent proteasomal degradation, and antigen presentation. To test this hypothesis we mutated three lysine residues in the N-terminus of CSP to alanines. Of the single point mutation substitutions we found that K58 replacement significantly reduced cytotoxicity (Figure 2.13B). We confirmed that lysine at position 58 had reduced ubiquitin chain formation (Figure 2.13C), suggesting that the observed reduction in cytotoxicity by CD8 + T cells is due to the loss of ubiquitin chain formation at lysine 58 in the N-terminus. Even though individual mutations of lysines at position 66 and 67 did not significantly alter ubiquitination and antigen presentation, simultaneous substitution significantly decreased both ubiquitination and CTL mediated cytotoxicity (Figure 64

75 2.13B,C). These data support the hypothesis that specific lysines in the N-terminus influence CSP presentation through formation of ubiquitin chains, but to different extents depending on the specific position in the N-terminus. The N-terminal cleavage of CSP has been well documented to be important for the establishment of EEFs in the liver. Yet, the contributions these two forms have to activation of the immune response has not been previously addressed. Using an ectopic expression system we are able to compare differences in the processing and presentation of two naturally occurring forms of CSP. We found CSP-Full activates CD8 + T cells more than CSP-Short and can be killed to a greater degree. These data led us to examine how this mechanistically occurs, whether by alterations in presentation or modification of host cell biology. We found no evidence that CSP alters the overall susceptibility of hepatocytes to be killed; rather we found that ubiquitination of specific lysines in the N-terminus drives degradation by the proteasome and subsequent presentation of C-terminal peptides. We found that CSP-Short predominates in traversed and infected hepatocytes. It is known that T cell responses against the C-terminus of CSP can be primed and recalled [152] but do not appear protective in naturally exposed individuals [98]. It is possible that in addition to a functional role of CSP cleavage, the parasite benefits by abrogating the ability of CSP to be processed and to activate the effector arm of the immune response. This is an important consideration because in animals models of vaccination peptides from the C-terminus of CSP are sufficient to protect mice [121,204] and therefore any modifications to their presentation may influence the protective nature of an effector immune response. 65

76 Current formulations of the RTS,S vaccine based on CSP do not include the N- terminal portion of CSP. Even though we have not specifically tested the ability of CSP- Full or CSP-Short to prime an immune response, it is possible that processing and presentation may be enhanced through inclusion of the N-terminus and contribute to priming the immune response. 66

77 Chapter 3 : Specific detection and isolation of P. falciparum EEFs in vitro Portions of this chapter have been adapter form Flow cytometry based detection and isolation of Plasmodium falciparum liver stages in vitro (under review) 67

78 Introduction Direct access to in vivo infected human hepatocytes is untenable due to ethical and logistic constraints. Biopsy of a heavily infected human liver was crucial for the original identification of the P. falciparum liver stage [45] but is unreasonable for routine experimentation. Consequently, studies of the liver stage of Plasmodium infection have relied mainly on the use of rodent parasites both in vivo and in vitro [159,243]. The rodent parasites Plasmodium berghei and Plasmodium yoelii complete full development in less than three days after mosquito bite and can fully develop into infectious merozoites in human hepatocellular carcinoma cell lines in vitro [81,82]. However, the human parasite P. falciparum requires at least 144 hours for full EEF development in the liver and has a limited ability to infect human hepatocellular carcinoma cell lines [159]. Thus, the majority of scientific work concerning the biology of the Plasmodium liver stage has been accomplished using the tools of rodent parasites. Multiple experimental models utilizing primary human hepatocytes for P. falciparum EEF development have been reported. Infection of primary hepatocytes in vitro by P. falciparum was first described almost thirty years ago [55]. Recent work using micropatterned primary hepatocytes surrounded by stromal cells has allowed for both complete development of P. falciparum EEFs and possibly generation of P. vivax hypnozoites in vitro [167]. The inability of primary hepatocytes to be continually propagated in culture adds to the cost of experimentation, as does the introduction of variability between donors and even individual cell preparations. The determinants of a robust in vitro P. falciparum liver stage infection have not been clearly defined. The use of 68

79 a standardized, immortalized hepatocyte cell line with slow growth kinetics and susceptibility to P. falciparum EEF formation would be ideal. The first mouse models relying on the engraftment of human hepatocytes into immune-compromised animals capable of generating mature EEFs were reported more than two decades ago [244] and were further used to obtain isolated infected cells from fixed frozen liver tissues through micro-dissection [245]. Complete development of P. falciparum liver stages and liver-to-blood transmission were later demonstrated in vivo in immune-compromised and fumarylacetoacetate hydrolase-deficient animals backcrossed with NOD mice [85]. Recently, SCID mice with chimeric human livers were used to show the protective effect of parasite antigen-specific human monoclonal antibodies derived from RTS,S vaccine recipients [246]. These mice lack a functional immune system and therefore can only be used to test individual, reconstituted components of immune protection. The in vivo and in vitro methods described above demonstrated the generation of P. falciparum merozoites capable of infecting red blood cells. However, the technical complexity and the high associated costs restrict the widespread use of these methodologies for routine studies on P. falciparum liver stages. Additionally, these methods rely on immunofluorescence or quantification of total parasite biomass and are unable to isolate live, individual P.falciparum EEFs. Therefore, a technically reproducible and costeffective experimental system for in vitro monitoring and purification of P. falciparum EEFs is still needed. Mouse models of the liver stage using rodent parasites suggest a role for both CD8 + T cells and sporozoite antigen-specific antibodies in sterilizing immunity [132]. However, 69

80 understanding the contributions of the humoral and cell-mediated immune responses directed against P. falciparum EEFs during the natural course of infection [83,247] or induced upon vaccination [248,249] requires a robust in vitro system. Two modes of interaction between sporozoites and host hepatocytes are currently described in vitro [60,250] and in vivo [59,73]: (i) breaching of the host cell plasma membrane followed by intracellular movement and subsequent exit, referred to as traversal, and (ii) productive invasion and parasitophorous vacuole (PV) formation within hepatocytes. The influence of traversed cells on infection, the subsequent immune response, and parasite biology are largely unknown. Thus, an optimal experimental in vitro system recapitulating the liver stage of P. falciparum should allow for specific identification and isolation of traversed from non-traversed cells and infected from uninfected cells. Utilizing in vitro models of Plasmodium infection, non-traversed and uninfected cell populations are similarly exposed to a plethora of biological factors from the salivary glands of infected mosquitoes and serve as the most accurate control population to study the immunology and developmental biology of P. falciparum liver stage infection in vitro. However, in vivo, sporozoites are deposited into the host dermis and the direct impact of salivary gland factors on hepatocytes is therefore diminished [1]. Here we report a flow cytometry-based in vitro system to monitor P. falciparum liver stages that permits detection and isolation of P. falciparum EEFs. A previously reported method for detection of P. falciparum infection in vitro utilized an anti-csp stain [251] that was unable to differentiate between cells traversed by sporozoites and productively infected cells. Additionally, the use of an intracellular antigen to identify parasites required fixation and thus the inability to purify live samples. 70

81 Isolation of individual EEFs creates the potential to study both the metabolic and/or transcriptional activity of both the parasite and the host hepatocyte. Expression profile libraries of the liver stage have been generated using rodent parasites [252] but comparable libraries have not been created for the human parasite P. falciparum. 71

82 Results Immortalized human hepatocyte cell lines are permissive for traversal by P. falciparum sporozoites Traversal of hepatocytes by Plasmodium sporozoites has been previously documented in vitro [60,250] and in vivo [59,73]. The in vitro traversed cell population can be identified by flow cytometry by its ability to retain fluorescent dextran during migration of sporozoites through the host cell [250]. Detection of traversal was dependent on the number of sporozoites added (Figure 3.1) and could be reversed by pre-incubation of sporozoites with the actin inhibitor cytochalasin D (Figure 3.1B,C). This demonstrates the specificity of detection of traversal and the absence of any factors other than sporozoites in quantification of traversal in vitro. 72

83 73

84 Figure 3.1 Specific detection of traversal by P. falciparum sporozoites in vitro. Traversal was assessed by flow cytometry as an ability of cells to uptake and retain fluorescent high molecular weight dextran during sporozoite migration. (A) Gating strategy for detection of dextran retention. Cells gated on forward and side scatter. Dextran percentages determine by signal in FL-2 (B) Traversal is inhibited by incubation of sporozoites with the mycotoxin cytochalasin D (C) Traversal is dependent on the initial sporozoite to hepatocyte ratio and in all cases inhibited by cytochalasin D. Performed in biological triplicate, Mean±SD shown. Ratio and cytochalasin D p< by two-way ANOVA, post-hoc Bonferroni adjusted comparisons shown ***p< In search of cell lines that allow for efficient detection of traversal and invasion of P. falciparum sporozoites, we turned to immortalized cell lines derived from human hepatocytes that included THLE-2 and THLE-3 [253], HepG2, HepG2-CD81 [75] and HC- 04. HC-04 is the only cell line previously reported to support progressive development of P. falciparum liver stages [167,169]. However, the process of traversal occurs in several tissues and is likely not specific to any one cell type. We found that all five cell lines tested had a detectable population of hepatocytes traversed by P. falciparum 3D7HT-GFP sporozoites (Figure 3.2A). HC-04, HepG2 and HepG2-CD81 had comparable percentages of cells retaining fluorescent dextran, which was about 2-3 fold higher than in cultures of THLE-2 and THLE-3 cell lines (Figure 3.2B). We concluded that each of the tested cell lines can be utilized to study traversal by P. falciparum sporozoites in vitro. 74

85 75

86 Figure 3.2 Human hepatocyte cell lines can be traversed by P. falciparum sporozoites. (A) Representative plots for each cell line are shown. Numbers indicate percentage of dextran-positive cells. (B) Percentage of traversed cells normalized to HC-04 obtained from three sporozoite preparations each in triplicate. Mean±SD shown, ***p<0.001, *p<0.05. Human hepatocyte cell lines exhibit differential ability to support development of P. falciparum EEFs Next, to assess the ability of immortalized hepatocyte cell lines to support the development of P. falciparum 3D7HT-GFP parasites, we used a novel, quantitative, flow cytometry-based analysis of infected cultures. We utilized relative FL1 (530/30) to FL2 (585/42) fluorescence ratios (non-overlapping) to detect the weak GFP signal of the parasite in relation to background autofluorescence of the host cell while simultaneously excluding dead cells using propidium iodide. To validate our ability to specifically detect developing parasites we quantified parasite 18S expression using real-time PCR and visualized P. falciparum EEFs by immunofluorescence microscopy following flow cytometry-based sorting of infected cells. Here we show that GFP + events could be detected not only in HC-04, but also in HepG2 and HepG2-CD81, which had been previously described as non-permissive for the development of P. falciparum EEFs (Figure 3.3A) [159]. However, the relative frequencies (percentages) of GFP + PI - events were 2.5-fold higher (Figure 3.3B) and the total number of GFP + events per well were 8-fold higher (Figure 3.3C) in HC-04 as compared to HepG2 and HepG2-CD81 cells. Both THLE-2 and THLE-3 had significantly fewer total GFP + events than the other cell lines tested (Figure 3.3C). In contrast, the fluorescence intensity 76

87 of the GFP + populations, a probable indication of parasite development, was comparable in all cell lines tested but had a large within sample variability (Figure 3.3D). Asynchronous development of rodent EEFs has been previously reported [254] and the broad range of the GFP signal intensity detected may be due to asynchronous development of P. falciparum parasites. Figure 3.3 Identification of P. falciparum parasites by flow cytometry. Infected cells were identified by flow cytometry at 96 hours postinfection as GFP-positive events in PI-negative (viable) cell populations. Uninfected cultures propagated in parallel with infected cultures were used to define specificity of GFP-positive events. A similar number of total events were acquired from infected and uninfected cultures of each cell lines. (A) Data from one representative culture is shown for each cell line. Event number corresponding to the GFP-positive gate (middle panels) and the geometric mean of events 77

88 in the GFP-positive gate is indicated (right panels). (B) Percentage of GFP-positive PInegative events. Mean+SD shown, ***p<.001. The percentage of GFP positive events in HC-04 cultures at 48 hours postinfection is dependent on the initial sporozoite to hepatocyte ratio and is inhibited if sporozoite motility is abrogated by cytochalasin D prior to infection (Figure 3.4). Even though our sporozoite preparations involve purification from a density gradient and significantly removes contamination from mosquito material, it is important to establish the specificity of our detection method. Figure 3.4 Specific detection of infection by P. falciparum sporozoites in HC-04 cultures in vitro (48 hours postinfection). Infection was assessed by flow cytometry. Detection of GFP events was dependent on sporozoite to hepatocyte ratio and inhibited by pretreatment of sporozoites with 78

89 cytochalasin D. Performed in biological triplicate, Mean±SD shown. Ratio and cytochalasin D p< by two-way ANOVA, post-hoc Bonferroni adjusted comparisons shown ***p<0.001, *p<0.05. Quantification of P. falciparum 18S rrna at two time points postinfection supported our flow cytometry findings (Figure 3.5). The ratio of parasite 18S copy number relative to a hepatocyte housekeeping gene (GAPDH arbitrary units) was 2-fold higher in HC-04 than in HepG2 and HepG2-CD81, and about 30-fold higher than in THLE-2 and THLE-3 (Figure 3.5A). In HC-04, a 3-fold increase in the Pf18S rrna copy numbers per well was observed between 48 and 96 hours postinfection (Figure 3.5B). However, a decrease in the 18S:GAPDH ratio was observed in the same cultures from 48 to 96 hours postinfection (Figure 3.5A). This can be explained by extensive proliferation of uninfected HC-04 cells (Figure 3.5C) resulting in the relative drop in percentage of infected hepatocytes. 79

90 80

91 Figure 3.5 Quantification of parasite-encoded 18S rrna in cultures of human hepatocytes infected with P. falciparum. Real-time PCR analysis of 18S rrna expression was done at 48 and 96 hours postinfection. Each culture (n=3) was analyzed in technical triplicate. (A) 18S copy number normalized using CT values from GAPDH (B) 18S copy number per culture. (C) Growth curve of uninfected hepatocyte cell lines (n=3 per time point). Mean±SD shown for all graphs, ***p< P. falciparum-infected HC-04 hepatocytes can be specifically isolated Infected hepatocytes from mice have been sorted ex vivo using P. yoelii parasites expressing GFP [164]. We have previously demonstrated that HC-04 cells infected with P. berghei ANKA-GFP can be isolated by flow cytometry-based cell sorting based on GFP expression in vitro [242]. Utilizing a similar approach we performed infection of HC-04 with P. falciparum 3D7HT-GFP sporozoites and sought to verify that GFP + HC-04 cells contain developing parasites. Hepatocytes were sorted based on a GFP positivity and cytospun, followed by immunofluorescence analysis of GFP + cell populations. This process is particularly harsh and may disrupt native host and parasite morphology, but it is necessary due to the low number of GFP + cells and the rapid outgrowth of uninfected HC- 04 during the 96 hours pi (Figure 3.5C). As shown in Figure 3.6, GFP + cells isolated at 96 hours pi contained P. falciparum parasites. We used three fluorescent markers to identify the presence of EEFs: parasiteencoded GFP, DAPI to stain both host and parasite DNA, and a monoclonal antibody against PfHsp70. We observed three patterns in sorted GFP + cells (Figure 3.6): (a, b) a GFP-positive, Hsp70 positive parasite located perinuclear and morphologically similar to parasites in primary hepatocytes [55,167] (c, d) a GFP-positive Hsp70-positive parasite 81

92 adjacent to a dividing or fragmented host nucleus and (e, f) a GFP-positive Hsp70-positive cell with indistinguishable host DNA that was either devoid of a host nucleus or that had a host nucleus containing the parasite. Punctate DNA staining within the parasites is indicative of development. We do not exclusively observe P. falciparum localized in the nucleus of HC-04 cells as has been previously reported for HepG2 [75], however, we cannot formally exclude that this may occur in a subset of GFP + HC-04 cells. 82

93 Figure 3.6 Visualization of parasites in HC-04 cells infected with P. falciparum. 83

94 GFP-positive PI-negative events were isolated by flow cytometry-based cell sorting 96 hours postinfection and cytospun. Representative micrographs of EEFs show DNA staining with DAPI, GFP expression and immunofluorescent detection of PfHsp70 by anti- Hsp70 monoclonal antibody (Blue = DAPI, Green = GFP, Red = PfHsp70. Scale bars indicate 10 μm). Primary human hepatocytes support P. falciparum 3D7HT- GFP EEFs in vitro We next asked if P. falciparum infection of HC-04 is comparable to that of primary human hepatocytes and if our flow cytometry based method can identify parasites. Higher levels of autofluorescence observed in primary hepatocyte cultures (Figure 3.7) required adjustments of flow cytometer voltages prior to acquisition of primary hepatocytes versus HC-04 cells. 84

95 Figure 3.7 Primary hepatocytes have a high degree of autofluorescence and require adjustment to voltages. HC-04 and primary hepatocytes (donor 4051) have different levels of background autofluorescence in live cells (PI negative). Acquisition of HC-04 required use of the high voltage settings listed and primary hepatocytes were acquired using the low voltage settings. Infected hepatocyte cultures from three donors had viable infected (GFP + PI - ) events detectable by flow cytometry at 96 hours pi (Figure 3.8A). Direct comparisons of GFP 85

96 mean fluorescence intensity between HC-04 and primary hepatocyte cultures are not feasible or relevant due to the necessary voltage adjustment. Unlike HC-04, GFP + events detected in primary hepatocyte cultures were within the range of background autofluorescence in the FL1 fluorescence channel (530/30) (Figure 3.8A). This highlights the necessity of using a ratio of closely related fluorescent emission channels to distinguish GFP + events from the autofluorescence of primary hepatocytes. The relative frequencies of GFP + events detected in primary hepatocytes at 96 hours pi were 20-fold higher (Figure 3.8A-B) than HC-04. However, the numbers of GFP + PI - cells in each individual culture (Figure 3.8C) did not significantly differ between HC-04 and two of the primary hepatocyte cultures (donors 4055A and 4059). Infected hepatocyte cultures from donor 4051 supported 3-fold more GFP + cells than HC-04 and primary cells from the two other donors. The drastic difference in percentages of infected cells found in HC-04 versus primary hepatocyte cultures at 96 hours pi (Figure 3.8A,B) was due to extensive proliferation of uninfected HC-04 cells (Figure 3.5C). To confirm that GFP + primary hepatocytes contained parasites we sorted and stained cells with an Hsp70-specific antibody (Figure 3.8D) in a similar fashion to GFP + HC

97 Figure 3.8 Comparison of detection of P. falciparum infection between primary human hepatocytes and HC-04. Infected cells were identified by flow cytometry at 96 hours postinfection as GFP-positive events in PI-negative (viable) cell populations (n=3). Uninfected cultures were used to define the positive gates. (A) Representative gating for detection of GFP events using three human hepatocyte donors and HC-04. Number of GFP-positive PI-negative events indicated (middle panels). Geometric mean of GFP-positive PI-negative events indicated (right panels). (B) Comparisons of the percentage of GFP-positive events and (C) total number of GFP events obtained per well. Mean±SD shown. (D) Representative examples of GFP-positive PI-negative events isolated by flow cytometry-based cell sorting 96 hours postinfection and cytospun (Scale bars indicate 10 μm). 87

98 Similar to HC-04, the percentage of GFP positive events is specific to the initial sporozoite to hepatocyte ratio and is inhibited by treatment of sporozoites with cytochalasin D (Figure 3.9). Figure 3.9 Specific detection of infection by P. falciparum sporozoites in primary hepatocytes in vitro (96 hours postinfection). Infection is dependent on the initial sporozoite to hepatocyte ratio and inhibited by cytochalasin D. Performed in biological triplicate, Mean±SD shown. *p<0.05, **p<0.01. The lack of proliferation of primary hepatocytes allowed for visualization of unsorted EEFs maintained on coverslips throughout infection (Figure 3.10A). This technique has been used by others to visualize EEFs and allows us to demonstrate the presence of EEFs in our lab by a well-established protocol. We defined parasites by Hsp70 staining (red) and observed no significant differences in parasite size between hepatocytes 88

99 from different donors at 96 hours pi (Figure 3.10B, C). Similar to previous studies [55,167], Pf3D7-HTGFP parasites were either adjacent to or touching the host nucleus (Figure 3.10D). A distance of >10μm was observed between some parasites and the closest host nucleus (Figure 3.10E). However, unlike GFP+ HC-04 cells, the high background of green autofluorescence, as seen by flow cytometry, obscured the parasite GFP signal. 89

100 Figure 3.10 Parasite morphology and size in cryopreserved human hepatocytes from three donors. Cultures were fixed and stained directly on coverslips 96 hours postinfection. (A) Representative micrographs of EEFs observed from each donor. DAPI (blue) represents host and Plasmodium nuclei and PfHsp70 (red) denotes EEF. Arrows indicate EEF in phase contrast micrographs (Scale bar = 10 μm). EEF size was measured by (B) surface area and (C) maximum diameter. (D) Proportion of EEFs observed to be directly adjacent to host 90

101 nucleus. (E) Of EEFs distant from host nucleus, the distance to the closest nucleus is indicated. Mean shown (horizontal line) for individual dot plots, ***p<0.001, **p<0.01. Dynamics of P. falciparum EEF detection by flow cytometry differ in HC-04 and primary hepatocytes Our prior experiments were performed at 96 pi to allow for detection of developing parasites. However, the overall ability to detect GFP + parasites by flow cytometry is a function of their brightness over background and attrition over time. To determine the progression of PfEEF development as well as the sensitivity of flow cytometry-based detection of Pf3D7HT-GFP EEFs at different time points post infection, we monitored two parameters: the number of viable infected (GFP + PI - ) cells and the intensity of GFP fluorescence in these cells as a marker of parasite development (Figure 3.11). The maximal numbers of parasites developing in HC-04 cells were detectable starting at 48 hours pi followed by a steady decrease in the number of GFP + events per culture over time (Figure 3.11A). In contrast, at 48 hours pi detection of parasites in primary hepatocytes was partially obscured by background autofluorescence, even when using the ratio of closely related fluorescence channels. Beginning at 72 hours pi a higher number of parasite-infected cells were detectable in primary hepatocyte cultures by flow cytometry (Figure 3.11B). The relative parasite survival rate between hours pi was 29% for HC-04 and 65% in primary hepatocytes (Figure 3.11). The parasite-specific GFP fluorescence intensity increased steadily in both HC-04 (Figure 3.11A) and primary hepatocytes (Figure 3.11B) and reached its maximum at 144 hours pi, suggesting development of the remaining persisting parasites. Since HC-04 and primary hepatocytes required adjustments to the voltages prior to acquisition (Figure 3.7) direct comparison of 91

102 the geometric MFI (Figure 3.11 y-axis) is not valid. Based on these data, for all subsequent experiments we chose to monitor Pf3D7HT-GFP development at 48 hours pi in HC-04 and at 96 hours pi in primary hepatocytes, when the maximal numbers of parasite-infected cells were observed in their respective cell cultures. Figure 3.11 Parasite detection and persistence over time. Cultures were infected and collected for flow cytometry starting at 48 hours postinfection in duplicate. Plots are shown for the number of GFP-positive events acquired at each time point and the geometric MFI (acquired at voltages specific to the cell type) of these events for (A) HC-04 and (B) primary donor 4051 (shading indicates time when a portion of the GFP population is obscured). Mean±SD shown. 92

103 Discussion The liver stage of P. falciparum infection, a leading target for generation of protective vaccines, still remains the least characterized part of the parasite lifecycle. Since the demonstration of P. falciparum EEF development in primary human hepatocytes, a number of alternative methodologies relying on microscopy have been identified to study P. falciparum liver stages in vitro [159,167]. The generation of fluorescent rodent parasite strains along with a wide range of host cell tropisms has allowed for routine identification and sorting of rodent Plasmodium EEFs [164], though such an approach has not been described for P. falciparum. Using a recently developed strain of P. falciparum expressing GFP [170] we show that live P. falciparum-infected hepatocytes are identifiable by flow cytometry and, therefore, can be quantitatively assessed and isolated. This approach allows for sensitive and rapid screening of hepatocyte cell lines and primary human hepatocytes from different donors for their ability to support PfEEF development in vitro. Prior to EEF formation in the liver, Plasmodium sporozoites traverse through several host cell types including hepatocytes [59,60,67,72]. We found that in vitro, the amount of cells retaining dextran after parasite traversal differed depending on the hepatocyte cell line used (Figure 3.2). These differences in detection of traversed cells could be a function of inherent susceptibilities to traversal by sporozoites or differences in the cells to retain dextran. No correlation was observed between the numbers of traversed cells detected 6 hours pi and the frequencies of infected cells detected 96 hours pi (Figure 3.3). These data suggest that in the cell lines tested, the process of invasion and development is distinct from traversal. CSP is shed from the sporozoite surface and is found in both cells 93

104 permissive or non-permissive to EEF formation [255]. We recently found that flow cytometry-based detection of CSP in P. falciparum-infected hepatocyte cultures does not represent an accurate measure of hepatocyte infection, but rather reflects the retention of CSP shed during parasite traversal (Trop S. et al., submitted). Therefore, we used parasiteencoded GFP as a marker for specific identification of developing parasites by flow cytometry. Using a ratio of two fluorescent channels, a FL1 (530/30) for detection of parasite GFP and FL2 (585/42) for detection of propidium iodide (PI) positive cells, we compared five cell lines (Figure 3.3) and three primary hepatocyte cultures (Figure 3.8) for their ability to support PfEEF formation and development. Intrinsic autofluorescence from cells during flow cytometry has the potential to generate false positives (bright events classified as positive) and false negative (dim events obscured in the background). To avoid improper classification of artifacts as true positive or true negatives, techniques using multiple fluorescent channels have been utilized [256]. Previous attempts at identifying fluorescent parasites by flow cytometry may have been obscured due to the presence of false negatives [172] resulting from high autofluorescence. The use of two channels allows for discrimination between the dead/pi-positive (FL1<FL2), autofluorescent (FL1=FL2) and GFP-positive (FL1>FL2) cells in infected hepatocyte cultures. We are not able to quantify directly the individual contributions of autofluorescence to each channel but can adjust the voltage characteristics to allow for detection of autofluorescence as shown previously [256]. This setup allows for the exclusion of autofluorescent false positive while detecting weak GFP signals that would be considered false negative if only a single parameter was used. 94

105 Among all cell lines tested, HC-04 [169] supported the highest absolute number of PfEEFs at 96 hours pi (Figure 3.3A,C). Surprisingly the HepG2 cell line, previously described as being non-permissive for P. falciparum sporozoite infection [159], had GFP + events detectable in the viable hepatocyte gate (Figure 3.3A). The SV40-immortalized hepatocyte lines THLE-2 and THLE-3 [253] had previously not been tested for their ability to support P. falciparum infection. Though these lines had a low proliferation rate, making them suitable to study the late stages of PfEEFs, both THLE-2 and THLE-3 showed a minimal ability to support Pf liver stages in vitro (Figure 3.3A,C). In addition to the numbers of GFP + events detectable in infected hepatocyte cultures, we compared the fluorescent intensity of these events as a measure of parasite development. The geometric mean fluorescence intensity (MFI) of these GFP + populations in each well were not significantly different (Figure 3.3D). To further verify the presence of developing P. falciparum EEFs we used quantitative real-time PCR for detection of Pf18S at two time points postinfection (Figure 3.5). Of the five cell lines tested, a time-dependent increase in Pf18S gene copy number was detected only in HC-04 cultures (Figure 3.5B), indicating that P. falciparum parasites can develop in these cells. It is unclear if the failure to increase Pf18s copy number in the other hepatoma cell lines is a function of parasites failing to develop or a decrease in the overall parasite number. Primary human hepatocytes remain the gold standard for generation of PfEEFs in vitro. However, recent work has questioned the susceptibility of all primary hepatocytes to P. falciparum infection, at least in vitro [167]. Using flow cytometry we demonstrated, that cryopreserved, commercially available human hepatocytes obtained from 3 different donors were permissive for infection with P. falciparum 3D7HT-GFP sporozoites. Though 95

106 efficient detection of parasite-specific GFP signal at the early stages of infection was hampered due to high autofluorescence in normal hepatocytes combined with the weak fluorescence of infected cells in which parasites have not yet multiplied [257,258]. Primary hepatocytes infected with Pf3D7HT-GFP were reproducibly detectable by flow cytometry between hours pi (Figure 3.11B). It needs to be seen if flow cytometry based detection can be applied to the ex vivo detection of the liver stage parasites in the recently developed humanized mouse models of P. falciparum infection [85]. This process could conceivable circumvent the need for mosquito dissection and sporozoite purification by directly feeding mosquitoes on humanized mice. It will be interesting to see if this could provide an even larger number of P. falciparum EEFs than in vitro infections. To address the dynamics of 3D7HT-GFP parasite detection and development in hepatocytes in vitro we used flow cytometry to monitor changes in parasite numbers and parasite-specific GFP fluorescence (MFI) as a reflection of parasite expansion within infected hepatocytes. The maximum numbers of GFP + events in HC-04 cultures were detected at 48 hours pi followed by a rapid decline (Figure 3.11A). In contrast, a larger proportion of infected cells persisted from 96 to 144 hours pi in primary hepatocytes (Figure 3.11B). A similar trend has been observed by others [167]. Importantly, an increase in GFP fluorescence over time was seen in infected cells that persisted in both HC-04 and primary hepatocyte cultures, indicative of parasite development. Taken together, these results show that P. falciparum parasites can develop in both HC-04 and primary hepatocytes. To visualize the presence of parasites in GFP + cells and to demonstrate the ability to selectively isolate live parasites from bulk infected cultures we sorted GFP + HC-04 cells 96

107 and primary hepatocytes (Figure 3.6). It does not appear that PfEEFs developing in HC-04 cells exclusively localize to the host nucleus, a phenomenon that has been previously observed in HepG2 [75]. Using GFP and PfHsp70 as parasite-specific markers, we identified three patterns of P. falciparum parasites in HC-04 cells at 96 hours pi. Parasites were found either adjacent to an intact host nucleus, surrounded by a fragmenting or dividing host nuclei or without an apparent host nucleus distinct from the parasite DNA (Figure 3.6). Additional studies are needed to describe phenotypes of parasites detected in HC-04 cells at the later time points of infection. In contrast to parasites developing in HC- 04, PfEEFs detected in sorted GFP + primary hepatocytes were mainly adjacent to the intact host nucleus easily distinguishable from the parasite nuclei (Figure 3.8D). In the context of low levels of infection, contrary to the rapidly dividing and vertically growing HC-04 cells, non-proliferating primary hepatocytes permit identification of EEFs by fluorescent microscopy without prior sorting (Figure 3.10). Unlike sorting and cytospin procedures, direct fixation of infected cells on coverslips better preserve both parasite and host hepatocyte morphology. This technique has been used by other groups and was used to compare our EEFs to previously published data. In all three primary infected hepatocyte cultures, PfHsp70-specific antibody identified developing parasites (Figure 3.10), but GFP fluorescence was not detectable in these cells by fluorescence microscopy (data not shown). As discussed above, the inability of GFP signal to be detected by fluorescence microscopy is likely due to high autofluorescence in normal hepatocytes [257]. At 96 hours pi PfEEFs developing in primary hepatocyte cultures were mainly adjacent to the host nucleus and comparable in size to parasites described in vivo [85] (Figure 3.10). 97

108 Previously, detection of PfEEFs by flow cytometry has only been possible using rodent Plasmodium species [159,164,242,250]. Although the use of rodent parasites and mouse models of Plasmodium infection will continue to be a cornerstone of malaria research, advancements in methods for studying PfEEFs in vitro are required to strengthen our understanding of basic parasite biology and to provide objective and quantifiable protocols to inform translational research. HC-04 and primary hepatocytes constitute complementary in vitro models to study PfEEF formation. Among five cell lines tested, HC-04 is able to support the greatest number of developing parasites. However, uninfected cells in HC-04 cultures rapidly proliferate, complicating parasite detection after hours pi. On the other hand, parasite traversal can be quantified in these cells early after infection and fluorescent parasites are clearly detectable by flow cytometry starting at 48 hours pi. Consequently, HC-04 is a suitable host cell to screen for the efficacy of interventions aimed at blocking the motility of sporozoites, such as anti-parasite antibodies, but is not fully suitable for study of late Pf liver stages. In contrast to HC-04 cells, primary hepatocytes do not proliferate in vitro, however, the PfEEFs can be clearly detected in these cells by flow cytometry only beginning at 72 hours pi. Though primary hepatocytes will continue to be the standard for study of PfEEFs, de novo generation or identification of additional human hepatocyte cell lines with low doubling times and limited autofluorescence still remains a priority. The development of a P. falciparum line expressing GFP through the parasite s lifecycle [170] was paramount in our ability to reproducibly detect and quantify PfEEFs by flow cytometry. In both HC-04 and primary hepatocytes detection of parasite GFP is specific, as shown by influence of sporozoite number on GFP percentages and inhibition by cytochalasin D. However, the 98

109 generation of a novel P. falciparum strain expressing a fluorophore brighter than GFP and/or use of a stronger liver stage promoter will facilitate flow cytometry-based detection and isolation of P. falciparum parasites in human hepatocytes even before 48 hours postinfection. 99

110 Chapter 4 : Flow cytometry based assays of P. falciparum liver stages in vitro: from sporozoite infectivity to EEF development Portions of this chapter have been adapter form Flow cytometry based detection and isolation of Plasmodium falciparum liver stages in vitro (under review) 100

111 Introduction Study of the blood stages of P. falciparum has been greatly facilitated by the development of continual in vitro blood cultures [157]. This method has contributed to a detailed understanding of merozoites invasion ligands [259], in vitro assays to screen for antibody-mediated growth inhibition [260], and high-throughput drug screening [261]. Prior to in vitro cultivation of the blood stages, human malaria parasites could only be propagated in susceptible primate hosts. In contrast, study of the liver stage of Plasmodium has benefited from the use of alternative Plasmodium species capable of infecting laboratory rodents [57,58]. These parasites have been instrumental in studies of parasite liver stage biology including global transcriptional profiling [252] and gene disruptions [161] to determine essential components of sporozoite infectivity [67,68,70] or EEF development [103]. In addition to routine infection of mice and rats, rodent Plasmodium species have a wide variety of susceptible cell lines in vitro [159], of both hepatocytic and non-hepatocytic origin. The specific determinants that allow for P. falciparum growth in one particular hepatocyte donor s cells over another are enigmatic [167]. Understandably, data obtained from rodent-trophic parasites do not always correlate with those from human parasites [105] and the results require confirmation in P. falciparum [107]. For instance, the human tetraspanin protein CD81 has been found to be necessary but not sufficient for invasion of P. falciparum sporozoites into hepatocytes [74] but is not required for rodent-trophic P. berghei EEF formation [75]. The necessity of CD81 in P. falciparum EEF formation has recently been confirmed in vivo [79]. Therefore, the choice of parasite and host is important when studying the nuances of the liver stage. 101

112 A variety of assays both in vitro and in vivo have been used to evaluate components of a protective immune response and the requirements for parasite development in the liver. In general these assays have aimed to measure inhibition of sporozoite invasion (ISI) or inhibition of liver stage development (ILSDA). Both of these assays have been limited by low invasion or infection rates and in the case of P. falciparum often rely on manual counting of parasites by an immunofluorescent assay (IFA). The general concept of the ISI assay is to measure interventions that target sporozoite motility, such as antibodies. Due to the wide variety of cultured cell lines that are susceptible to infection by rodent Plasmodium parasites [159], invasion assays have been used routinely to examine the inhibitory effect of antibodies against the sporozoite [262,263]. Similar assays were then developed for use with P. falciparum, which unfortunately used cell types (WI-38 and HepG2-A16) unsuitable for complete P. falciparum EEF development [264]. Consequently, the use of P. falciparum sporozoite inhibition in these assays is likely a measure of parasite traversal and not productive invasion rates. Both traversal and productive invasion are dependent on parasite motility, but the mechanics and dynamics greatly differ between the two processes. Therefore even though ISI assays can be performed with P. falciparum sporozoites using a non-permissive cell type, they lack the ability to identify any inhibition of initial productive invasion and subsequent PV formation in hepatocytes. Several methods have been developed to address this issue and to determine an assay more suitable for P. falciparum sporozoites and EEFs. Primary hepatocytes have been used as the gold standard for in vitro P. falciparum EEF evaluation [114]. These techniques originally relied on microscopic identification of parasites and recently, on 102

113 parasite RNA [166]. However, quantification of parasite development using 18S rrna is unable to differentiate between delays in parasite growth or abrogation of the total number of parasites in the liver. Alternatively, humanized mice have been used in tandem with the passive transfer of antibody to establish their inhibitory capacity [171,244,246]. Interestingly, purified IgG from chemoprophylaxis-immunized humans was inhibitory in an in vitro traversal assay and in an in vivo humanized mouse model, but measures of sporozoite gliding motility were inconclusive [171]. However, the use of humanized mice requires large amounts of purified monoclonal antibodies or pooled, purified IgG and still relies on the microscopic identification of parasites in liver sections or detection of luciferase [85,265]. Modification of rodent parasites to express specific antigens such as CSP from P. falciparum [266] or P. vivax [267] allows for the evaluation of responses directed toward candidate antigens. These systems provide a simplified and standardized method with which to evaluate immune responses in vivo but are limited in the scope of targets that can be studied. Ideally, any system that studies the inhibition of Plasmodium sporozoites or the inhibition of EEF development would have an unbiased method of detection with a high sensitivity in order to detect even low levels of hepatocyte infection. Our flow-cytometrybased detection protocol has the capability to detect parasite number (invasion, ISI) and parasite development (geomfi, ISLDA). However, an individual assay is not suitable for all experiments. The specific scientific question being asked will determine the strengths and weaknesses of each approach. 103

114 Results CD81 is required for P. falciparum infection of primary human hepatocytes, but is not essential for the infection of HC-04 It has been previously shown that functional antibody-based neutralization of CD81 on the surface of primary human hepatocytes abolishes infection by P. falciparum sporozoites in vitro, as measured by the numbers of P. falciparum EEFs detected by immunofluorescence [74]. Antibodies against CD81 were also capable of blocking P. falciparum sporozoite invasion in vivo, using liver chimeric mice [79]. To address the requirement of CD81 using our experimental model utilizing 3D7-HTGFP parasite infection of human hepatocytes in vitro, we first verified the expression of CD81 on the surface of hepatocyte cell lines and primary human hepatocytes. Primary hepatocytes from three donors as well as THLE-2 and -3 cells expressed surface CD81, whereas neither HC- 04 nor HepG2 had detectable expression by flow cytometry (Figure 4.1A). These data were further confirmed by RT-PCR (data not shown), ruling out our inability to detect low CD81 expression by flow cytometry. These observations in combination with the differential ability of hepatocyte cell lines to support development of P. falciparum EEFs (see Chapter 3) reveal no correlation between the expression of CD81 and the susceptibility of human hepatocyte cell lines to P. falciparum infection or to traversal in vitro. A lack of CD81 did not preclude infection of HC-04, and endogenous expression of CD81 on THLE-2 and -3 was not sufficient to allow for efficient infection. Accordingly, ectopic expression of CD81 on HepG2 cells did not result in more efficient P. falciparum infection. Similarly, transient 104

115 expression of CD81 in HC-04 cells did not significantly change the numbers and development of EEFs in these cultures (Figure 4.1C, D). Figure 4.1 Expression of surface CD81 by human hepatocyte cell lines and primary hepatocyte donors. Surface CD81 was stained using specific antibodies or an isotype control followed by flow cytometry for detection. (A) Representative plots for all cells are shown and geometric MFI is indicated for both isotype (black) and anti-cd81 staining (red). (B) Surface staining of a mock and transient transfection of HC-04. Transiently transfected HC-04 were infected and run on flow cytometry 96 hours postinfection; (C) GFP-positive number and (D) geometric MFI shown. Mean±SD shown. Using our experimental model, we recapitulated previously described conditions [74] for antibody-based neutralization of CD81 (Figure 4.2). We found that the CD81- blocking antibody (clone 1D6) had no effect on P. falciparum 3D7HT-GFP infection of CD81-negative HC-04, as assessed by percentages and the numbers of infected (GFP + PI - ) cells detected by flow cytometry at 48 hours pi (Figure 4.2A, B). In contrast, the same neutralizing antibody blocked P. falciparum infection in primary human hepatocytes 105

116 (Figure 4.2C, D). Moreover, 1D6 did not alter parasite numbers when added 6 hours pi, suggesting that CD81 is critical only during parasite invasion. These data are in agreement with the original observation by Silvie et al., demonstrating that CD81 blocking significantly reduces invasion of primary human hepatocytes by P. falciparum sporozoites [74]. Therefore, though invasion of HC-04 appears to be independent of CD81, we confirm that invasion of primary hepatocytes is CD81-dependent. Figure 4.2 Influence of CD81 blocking by mab 1D6 on P. falciparum hepatocyte infection. 1D6 or isotype were added to cultures at 10µg/ml prior to infection (-2 to 0 hrs), during invasion (0 to 6 hours) or after invasion (6 to 24 hours). Representative flow plots shown for (A) HC hours postinfection and (B) number and percentage of GFP-positive events in duplicate. (C) Flow plots for donor hours postinfection and (D) graphs indicated the number and percentage of GFP-positive events in duplicate. Mean±SD shown on all graphs, **p<0.01, *p<