Age-Related Differences in Naturally Acquired T Cell Memory to Plasmodium falciparum Merozoite Surface Protein 1

Naturally acquired immunity to Plasmodium falciparum malaria in malaria holoendemic areas is characterized by the gradual, age-related development of protection against high-density parasitemia and clinical malaria. Animal studies, and less commonly, observations of humans with malaria, suggest that T-cell responses are important in the development and maintenance of this immunity, which is mediated primarily by antibodies that slow repeated cycles of merozoites through erythrocytes. To advance our rather limited knowledge on human T-cell immunity to blood stage malaria infection, we evaluated CD4 and CD8 T-cell effector memory subset responses to the 42 kDa C-terminal fragment of Merozoite Surface Protein 1 (MSP142), a malaria vaccine candidate, by 49 healthy 0.5 to ≥18 year old residents of a holoendemic area in western Kenya. The proportion of individuals with peripheral blood mononuclear cell MSP142 driven IFN-γ ELISPOT responses increased from 20% (2/20) among 0.5–1 year old children to 90% (9/10) of adults ≥18 years (P = 0.01); parallel increases in the magnitude of IFN-γ responses were observed across all age groups (0.5, 1, 2, 5 and ≥18 years, P = 0.001). Less than 1% of total CD4 and CD8 T-cells from both children and adults produced IFN-γ in response to MSP142. However, adults had higher proportions of MSP142 driven IFN-γ secreting CD4 and CD8 effector memory (CD45RA− CD62L−) T-cells than children (CD4: 50.9% vs. 28.8%, P = 0.036; CD8: 52.1% vs. 18.3%, respectively P = 0.009). In contrast, MSP142 driven IFN-γ secreting naïve-like, transitional (CD45RA+ CD62L+) CD4 and CD8 cells were the predominant T-cell subset among children with significantly fewer of these cells in adults (CD4: 34.9% vs. 5.1%, P = 0.002; CD8: 47.0% vs. 20.5%, respectively, P = 0.030). These data support the concept that meaningful age-related differences exist in the quality of T-cell immunity to malaria antigens such as MSP1.


Introduction
Malaria is a global health problem that affects primarily infants and children less than 5 years old [1,2], whereas older children and adults in most endemic regions develop naturally acquired immunity that protects against high-density parasitemia and malaria morbidity [3,4]. It is clear that antibodies directed against blood stage parasites are critical to this protection since passive transfer of Ig isolated from sera of immune African adults to children with acute malaria rapidly reduces the level of asexual parasitemia and severity of malaria illness [5]. Unlike the clear-cut importance of antibodies in mediating naturally acquired immunity, the role of T-cell memory in the development and maintenance of this immunity is poorly understood. Studies of rodent malaria models and limited observations of malaria-naïve humans who became resistant to blood stage P. falciparum challenge after repeated inoculation and drug cure with a small number (,300) of infected erythrocytes suggest that T-cells and IFN-c responses, even in the absence of antibodies, confers a degree of protective immunity [6,7]; however it is unclear whether residual anti-malarial drugs may have contributed to the protection seen [8]. In addition, due to the technical challenges of conducting more elaborate T-cell studies, limited information is available on human memory T-cells particularly in response to defined blood-stage malaria antigens. Greater understanding of how malaria specific T-cell memory subsets contribute to immunity in malaria endemic populations is important to the design and testing of blood stage malaria vaccines as well as understanding how decreasing malaria exposure due to vector control in Africa and elsewhere may affect age-related susceptibility to malaria infection and clinical illness.
Merozoite Surface Protein 1 (MSP1) is one of the most abundant antigenic proteins expressed by asexual parasites of all malaria species. In the case of P. falciparum, MSP1 is a 200 kDa glycoprotein expressed and sequentially processed to yield a 42 kDa (MSP1 42 ) fragment, which is essential to the initial low affinity attachment of the merozoite to the erythrocyte surface [9]. T-cell epitopes recognized by humans with P. falciparum infection are contained within MSP1 33 sub-fragment that is shed from MSP1 42 before erythrocyte invasion [10,11]. Although the mechanisms by which CD4 T-cells contribute to protective immunity are not well understood, it is likely that this occurs through cytokines that provide help to antigen specific B-cells, e.g. Ig isotype and IgG subclass switching and/or by direct cellular communication with macrophages, CD8 T-cells and B-cells [12,13,14,15]. MSP1-driven IFN-c responses have been observed in T cell receptor transgenic mice that resolved P. chabaudi chabaudi by generating T-cell responses to MSP1 33 , which augment antibody responses to MSP1 19 [16] and through induction of IL-4 [17]. Vaccination of rhesus monkeys with recombinant MSP1 42 [18,19] and human vaccine trials with MSP1 [20], MSP1 19 [21], and MSP1 42 [20,21,22] lend further support to the role of T-cells in protective immunity. In essence, depletion of IFN-c and CD4 T-cells abrogates protective immunity in mice immunized with MSP1 [23].
Evaluation of effector memory T-cell subsets in malaria exposed human populations has been constrained by the complexity of the assays involved in the identification of low frequency antigenspecific T-cell subsets, the limited number of peripheral blood lymphocytes that can be obtained during field studies (particularly from infants and children), and the inability to access primary lymphoid organs. However with recent technologic advances, human T-cell memory subsets can be defined by multi-parameter flow cytometry using a panel of functional and phenotypic markers [24]. To this end, CD4 and CD8 T-cell central memory (T CM ), effector memory (T EM ), terminally differentiated RA re-expressing effector memory (T EMRA ) and naïve (T N ) cell subsets can be characterized according to expression of CD45RA, CCR7, and CD62. T CM are CD45RA 2 CCR7 + CD62L + ; T EM are CD45RA 2 CCR7 2 CD62L 2 ; T EMRA are CD45RA + CCR7 2 CD62L 2 ; and T N are CD45RA + CCR7 + CD62L + [25]. One recent study used a similar approach to characterize T-cell memory subsets specific to P. falciparum MSP1 42 from malarianaïve adult volunteers who participated in a phase I vaccine trial. This study reported that memory CD4 + CD45R0 + CD154 + cells were elicited after vaccination [22]. Further, multifunctional cytokine secreting T-cell subsets specific to P. falciparum Apical Membrane Antigen 1 (AMA1) have been defined in malaria-naïve individuals vaccinated with this antigenic protein [26]. However, to our knowledge, there are no published data defining MSP1specific memory T-cell subsets in populations naturally exposed to P. falciparum. In the present study, we characterized the frequency and quality of naturally acquired IFN-c producing MSP1 42 specific CD4 and CD8 T-cell effector memory subsets in children and adults residing in a malaria holoendemic area of western Kenya.

Results
Malaria infection status, white blood cell counts and T lymphocyte subsets of study participants Table 1 describes the age-stratified median malaria parasite density, complete blood count (i.e. white blood cells, (WBC), total lymphocytes, monocytes, granulocytes, red blood cells (RBC), and platelets), CD4 and CD8 cell absolute counts and frequencies of the study participants. The upper range of asexual parasitemia among these healthy malaria asymptomatic individuals was highest among participants who were 1 to 5 years old compared to infants with a median age of 0.5 years, presumably due to protective maternally acquired antibodies during infancy and the development of malaria-specific immunity later in life. None of the adults were parasitemic by blood smear. Median WBC, lymphocyte, and monocyte counts determined by Coulter Counter were significantly lower among the adults than 0.5 to 5 year olds (P,0.001). However, the relative proportion of CD4 and CD8 cells was not different across the various age groups (P = 0.310). These latter data are similar to observations of non-African children showing that the absolute number, but not ratio of CD4 and CD8 cells, change normally between birth and approximately 5 years [27,28].
PBMC IFN-c responses to MSP1 42 increase after age 5 years Due to the limited blood volume obtainable from young children, we conducted time-course experiments comparing 24, 48, 60 and 72 hour incubation periods (data not shown). These condition-optimization experiments demonstrated that a 60 hour incubation period resulted in measureable IFN-c recall responses to MSP-1 42 without resulting in appreciable cell death or inducing high background in vitro, and therefore was selected as the optimal incubation period for ELISPOT assays used in this study. The proportion of IFN-c ELISPOT responders to MSP1 42 did not significantly change from age 0.5 to 5 years, but increased from 20% for infants with median age 0.5 years to 90% for adults $18 years (P = 0.01, Table 2). However, significant increases with age were observed for the magnitude of IFN-c responses, with a median of 5 to 155 SFU/10 6 PBMCs across age groups (P = 0.001, Fig. 1). PBMCs from all 49 individuals examined generated IFN-c in response to SEB super antigen (data not shown), confirming that all age groups were immune competent by this criterion.
Ex vivo resting CD4 and CD8 T cell effector memory subsets differ by age The proportions of various CD4 T-cell memory subsets were determined for freshly isolated PBMCs prior to in vitro exposure to MSP1 42 . Representative dot plots for PBMCs from an adult and a 1 year old child are shown in Fig. 2A and 2B. Overall, adults (n = 10) had a higher frequency of CD4 T CM (CD45RA 2 CD62L + ) and T EM (CD45RA 2 CD62L 2 ) compared to children #5 years (n = 39, P,0.001; Fig. 3A and 3C). In contrast, children had a higher frequency of T N (CD45RA + CD62L + ) than adults (P,0.001, Fig. 3B); whereas the proportion of CD4 T EMRA (CD45RA + CD62L 2 ) was ,1%, and similar across age groups (P = 0.719, Fig. 3D).
PBMCs were stained in parallel for CD8 T-cell memory subsets as described above. This approach was necessary because a 4color flow cytometer was the only instrument available to our laboratory in Kenya at the time the studies were performed. Figure 2C and 2D show representative histograms of the distribution of freshly isolated CD8 T-cell subsets from an adult and a 1-year old child. The age-related distribution of CD8 T-cell memory subsets was similar to that of CD4 cells, with T CM (Fig. 4A) and T EM (Fig. 4C) frequencies being higher among adults compared to children (P = 0.019 and 0.016, respectively) and the frequency of CD8 T N (Fig. 4B) being higher in children compared to adults (P = 0.002). However, unlike the case for CD4 T EMRA cells that showed no difference in frequency by age, CD8 T EMRA cells (Fig. 4D) were readily detectable in both age groups, and were significantly higher among adults than children (24.4% and 12.2%, respectively, P = 0.006). Table 1. Parasitemia, MSP1-specific IgG antibody responses and age-related hematologic indices.   42 -specific CD4 and CD8 T-cells were detected after in vitro stimulation followed by intracellular staining for IFN-c and flow cytometric analysis, as described above. The median frequency (%) of MSP1 42 driven CD4 and CD8 cells that stained positive for intracellular IFN-c was less than 1% (ranging from nil to ,3.8%), and did not differ significantly according to age ( Fig. 5A and 5B). In order to determine the phenotype of the T-cell producing IFN-c in response to MSP1 42 stimulation, IFN-c positive CD4 and CD8 T-cells were identified by flow cytometry and then back-gated into the memory T-cell subsets defined by CD45RA and CD62L expression. This analysis revealed that both adults and children had MSP1-specific IFN-c producing CD4 T CM (Fig. 6A) with median frequencies of 45.0% and 24.5%, respectively (P = 0.345). These differences were not significant due to the wide range of responses within each age group (from nil to ,75% of all IFN-c positive CD4 cells within the gate). However, the frequency of IFN-c + CD4 T EM was significantly greater among adults than children (50.9% vs. 28.8%, Fig. 6C, P = 0.036). In contrast, the median frequency of MSP1 42 driven IFN-c + CD4 T N (Fig. 6B) was higher in children compared to adults (34.9% and 5.1%, respectively, P = 0.002). The median frequency of MSP1 42 -specific IFN-c + CD4 T EMRA (Fig. 6D) was not significantly different among children and adults, 7.2% and 0.4%, respectively.
Even though classical endogenous MHC class I-restricted antigen processing and presentation may be considered unlikely for parasites that reside in erythrocytes, CD8 T-cells specific to blood stage antigens have been described in mouse malaria models [29,30]. Therefore, we also performed studies to detect MSP1 42specific IFN-c expressing CD8 T-cells and their effector/memory phenotypes. Of the CD8 T-cells that produced IFN-c, the frequencies of T CM (Fig. 7A) and T EMRA (Fig. 7D) were similar among children and adults (P = 0.707 and 0.855, respectively). In contrast, the median frequency of T EM cells (Fig. 7C) was significantly higher among adults compared to children (52.1% and 18.3%, respectively, P = 0.009) while CD8 T N (Fig. 7B) were significantly more frequent among children than adults (47.0% and 20.5%. respectively, P = 0.030). The median frequency of MSP1 42 -specific IFN-c + CD8 T-cell responses mirrored the overall age-dependent shift in total CD8 T-cell effector memory subsets independent of antigen specificity (as shown in Fig. 4).

The proportion of CD4 and CD8 T-cell subsets differs by age
In order to visualize the relative contributions of CD4 and CD8 T-cells subsets to the resting pool of T cells as well as to MSP1 42specific IFN-c responses present in children #5 years of age and adults, proportions of each cell type were compared. Figure 8 reveals that children have predominantly (68.7%) CD4 naïve-like, transitional CD4 T-cells (T N ) cells compared to adults who have more (48.6%) CD4 T CM cells in their resting pool. Even though children produce the same level of IFN-c in response to stimulation with rMSP1 42 as shown in Figure 5, the CD4 T cell subset responsible for this function in order of rank for children is T N (34.9%).T EM (28.8%).T CM (24.5%).T EMRA (7.2%), in contrast to IFN-c producing CD4 T cell subsets in adults which rank from highest to lowest median frequency: T EM (50.9%).T CM (45.0%).T N (5.1%).T EMRA (0.4%). Figure 9 demonstrates similar age-dependent differences in CD8 T-cell subset proportions. Children have predominantly (62.8%) CD8 T N cells compared to adults who have significantly fewer (35.6%) CD8 T N cells (P-value = 0.002). Again, children appeared able to produce similar levels of IFN-c as adults in response to stimulation with rMSP1 42 as shown in Figure 5, but the CD8 T-cell subset contributing to this function differs in order of rank for children with median frequencies for T N (47.0%).T EM (20.0%).T EMRA (18.3%).T CM (6.9%), in contrast to IFN-c producing CD8 T-cell subsets in adults which rank T EM (52.1%).T EMRA (22.2%).T N (20.5%).T CM (5.8%).

Discussion
Our study demonstrated profound qualitative differences in Tcell memory responses to an essential merozoite invasion ligand and antigen protein, MSP1 42 , that correlate with age-related naturally acquired malaria immunity in residents of a holoendemic area of western Kenya. Although our study is limited by the small samples size, results show that the overall frequency of antigenspecific CD4 and CD8 T-cells that produced IFN-c in response to MSP1 42 were low and did not appear to differ according to parasitemia or ages ranging from 0.5 to 5 years and $18 years but there was a shift in hierarchy among various T-cell subsets responsive to MSP1 42 such that the T EM subset was the dominant cell type in adults in contrast to children who had more phenotypically naïve-like, T N cells. These shifts were observed not only for IFN-c producing CD4 cells, which might be anticipated based on the presumed importance of CD4 helper T-cells in generating and maintaining antibody mediated responses by B-cells, but also for CD8 T-cells, which are thought to be important in generating immunity to pre-erythrocytic rather than blood stage malaria antigens [31,32]. Categorizing samples based on the presence of parasitemia or parasite density did not appear to influence IFN-c responses T-cell subset dominance, however most of our study participants were aparasitemic and asymptomatic. Furthermore, the observed patterns of CD4 T-cell responses did not correlate with antibody responses to MSP1 42 but there was weak correlation between CD8 responses and IgG responses to MSP1 42 . The potential importance for both CD4 and CD8 T-cells in determining resistance to blood stage parasitemia has been suggested by experimental infection of mice with P. chabaudi [33]. Absence of parasitemia in adults and differential parasite density in children suggest that different levels of immunity exist within this population and could be affected by parasite, human and external factors such as intensity of repeated malaria exposure. Longitudinal cohort studies looking at how age, repeated parasite exposure and clinical episodes influence the development of T-cell immunity to multiple antigens are underway. In addition, malaria-specific T-cell responses may differ in those residing in hypoendemic areas who are exposed to malaria antigens intermittently and with less intensity. The overall magnitude of MSP1 42 driven IFN-c responses observed for CD4 and CD8 T-cells was generally weaker than that reported for recent studies of regulatory T cell responses in Gambian children with severe and mild malaria [34,35] and malaria naïve volunteers challenged experimentally a single time with P. falciparum infected erythrocytes [36]. One likely explanation for this difference is that that the current study used a single, defined malaria antigen to stimulate T-cell recall responses, whereas the above mentioned studies and other reports used schizont extracts or infected erythrocytes as stimuli. Schizont extracts and infected erythrocytes contain not only MSP1 (presumably the non-processed pre-protein as well as various processed fragments) but also multiple other merozoite antigens and parasite moieties that may stimulate innate as well adaptive responses by T-cells and IFN-c secreting NK cells [34,37]. IFN-c production by non-T cell types were not ascertained within this study but would explain the differences in frequencies observed between adults and children by ELISPOT (Fig. 1) that were not reflected in the T-cell specific ICS results (Fig. 5) that did not appear to differ by age. The frequencies reported here are consistent with IFN-c secreting peripheral blood and cord blood mononuclear cells in our previously published studies using recombinant MSP1 33 , MSP1 42 and peptide epitopes contained within MSP1 33 to study T-cell IFN-c recall responses in adults as well as newborns in Kenya [38,39]. These earlier studies also established that the 3D7 strain of malaria is among the most common alleles of the T-cell epitope containing MSP1 33 fragment circulating in western Kenya [38].
To determine the ex vivo frequency of phenotypically distinct Tcell populations, we compared freshly isolated CD4 and CD8 cell effector-memory subsets defined by CD45RA and CD62L [25]. As expected, there was a significantly greater proportion of T CM and T EM CD4 cell subsets among adults compared to children, whereas most CD4 cells were T N in children relative to adults. This is consistent with other studies that compared the proportion of T cell subsets in older children and adults and observed a similar maturation pattern associated with age [40,41]. There were very few terminally differentiated (T EMRA ) CD4 cells present in these healthy individuals regardless of age. Similarly, the proportion of CD8 T CM and T EM cells increased with age and formed the majority of the CD8 T-cell compartment in adults as compared to more naïve-like (T N ) CD8 T-cells found in young children. In contrast to CD4 T EMRA that represented less than 1% of the total CD4 cell pool (Fig. 3), the frequency of T EMRA within the CD8 cell pool increased from ,15% in children 5 years and younger to over 20% in adults (Fig. 4). The observed age-related increase in the population of mature memory T-cell phenotype could be explained by the fact that as individuals age, they are exposed to a myriad of microbial antigens hence expanding the memory pool to these pathogens. When comparing the proportion of ex vivo T-cell subsets to the proportion of MSP1-specific IFN-c producing T-cell subsets, in general the IFN-c producing T-cell subsets reflect the proportion of ex vivo T-cell subsets present in each age group. For example, children have more T N cells compared to adults who have relatively more T CM and T EM -and not surprisingly this distribution of T-cell subsets is reflected by the MSP-1 specific IFN-c producing T-cell subsets. The exception to this 'proportional-distribution by age' is the CD4 T EMRA cells, which do not differ between adults and children ex vivo but are over represented as a source of IFN-c production in response to MSP-1 in the children but not the adults. The picture is slightly different for the CD8 T-cell subsets, where the proportion of ex vivo T CM and T EMRA are higher for adults compared to children, but the MSP-1 specific T CM and T EMRA making IFN-c appear the same across age groups, with the dominant cell type making IFN-c in the children being the T N cells in contrast to the T EM for the adults. Future studies will reveal if this particular CD8 T-cell subset is associated with fewer high-density infections over time and/or fewer episodes of clinical malaria. To date, no studies have compared the relative contributions from CD4 and CD8 T-cell effector subsets (i.e., T EM versus T EMRA ) in protection from asexual parasitemia or clinical infections.
The necessity of T-cell maturation and differentiation for the acquisition of malaria immunity is further supported by changes in Figure 3. Proportion of resting CD4 T cell subsets across age groups. The frequency of each T cell subset was compared between adults ($18 years old) and children (#5 years old) for (A) Central memory, T CM : CD4 + CD45RA 2 CD62L + ; (B) naïve T cells, T N : CD4 + CD45RA + CD62L + ; (C) effector memory, T EM : CD4 + CD45RA 2 CD62L 2 ; and (D) RA-expressing effector memory, T EMRA : CD4 + CD45RA + CD62L 2 . The median frequency of CD4 + T CM and T EM was significantly lower for children compared to adults (P,0.001, Mann Whitney test). Whereas the median frequency of CD4 + T N was significantly higher for children compared with adults (P,0.001). The median frequency of CD4 + T EMRA was similar for the two age groups (P = 0.719). doi:10.1371/journal.pone.0024852.g003 the quality of IFN-c recall responses to MSP1 42 stimulation across age groups as presented here. IFN-c has been associated with resolution of Plasmodium infection in humans and animal models [42,43], and synergizes with other cytokines and nitric oxide to eliminate parasites [44]. An age-related increase in malariaspecific IFN-c recall responses has been observed in several other studies of individuals from malaria endemic areas [45]. Children in Africa and Papua New Guinea were more efficient producers of IFN-c to sporozoite or merozoite antigens, suggesting an association between antigen-specific IFN-c production and reduced pathology [42]. Further, fewer CD4 IFN-c producing cells in Gabonese children with acute malaria were associated with hyperparasitemia [46] and higher prevalence and magnitude of IFN-c were observed at the end of rainy season (transmission season) than dry season in a seasonally endemic area of Gambia [47]. Taken together, these studies suggest that IFN-c is a  biomarker of recent exposure and/or clinical immunity to malaria. It is known that long term cultures may skew antigen specific responses, but our observation that children consistently secreted IFN-c from naïve T-cells while adults had mature memory T-cells as the primary source of the cytokines argues against selective survival of a specific subset by the culture condition. The differential secretion of IFN-c by CD4 and CD8 T-cells in adults and children (Figs. 6 and 7) has implications for the sustainability of this essential cytokine response.
Lack of sterile immunity for those residing in malaria endemic areas and waning of semi-protective immunity observed in residents of malaria endemic regions who travel to non-malarial zones [48] have suggested that long lasting immunologic memory to P. falciparum is not possible under natural exposure conditions. Cytokine recall responses could therefore be attributed to immunologic 'boosting' by repeat infections that may be derived from short-lived T-cells [49] or possibly other IFN-c producing cells such as NK cells or monocytes. Nevertheless, adults returning to endemic areas appear to re-acquire clinical immunity more rapidly than children [50,51] suggesting that some form of longterm memory to malaria was established. An alternate explanation is that a mature immune system acquires antigen-specific effector memory and/or central memory T-cells more efficiently than an immature immune system. An infant's ability to develop immunologic memory may thus be impaired due to age-inherent differences in antigen recognition, presentation and dendritic cell function [52,53].
Historically demonstration of immunologic memory to malaria has been supported primarily by observations from serologic studies [54,55,56]. However, there is a growing consensus that antibody levels may be a more accurate measurement of cumulative malaria exposure history as opposed to the sole means by which protection is conferred [57]. Though it is well accepted that memory T-cells are an important component for protection against a variety of infectious pathogens [50,51], there is limited information on human memory T-cells specific to malaria [45,49]. The nature of memory . The phenotype of IFN-c expressing CD4 + T cells in adults were predominantly T CM and T EM compared to the children who had IFN-c expressing CD4 + T cells generated from all four subsets including those phenotypically T EMRA and T N . Significant differences (Mann Whitney test) in median frequencies were observed for CD4 + T EM subsets, which were more prevalent (P-value = 0.036) and CD4 + T N , which were less prevalent (P = 0.002) in adults compared to children. Differences observed in the frequencies of CD4 + T CM and T EMRA between age groups did not reach significance due to small sample size. doi:10.1371/journal.pone.0024852.g006 T-cell populations elicited in malaria-naïve North Americans vaccinated with the 3D7 and FVO strains of MSP1 42 demonstrated induction of antigen-specific memory CD4 T-cells [22], however the contribution of CD8 T-cell memory subsets was not evaluated. Studies on individuals residing in P. vivax malaria endemic areas have shown an increased proportion of memory CD8 T-cells subsets [52], supporting the premise that T-cell maturity is as important as antibodies in the development of protection against malaria. Although we cannot conclusively determine the origin of the blood stage specific CD8 T-cells within this study, it is clear that CD8 T-cells play a crucial function by secreting IFN-c, a cytokine that has been shown to be important in control of malaria parasites. It has been demonstrated in mouse models of malaria that CD8 Tcells are able to 'see' malaria blood stage antigens through the process of cross-presentation [58], thus it is possible this mechanism is also important in human immunity to malaria.
Advances in polychromatic flow cytometry have demonstrated the heterogeneity of T-cell immunity and characterized 'atypical' antigen-specific memory T-cell populations [12]. Thus, our unexpected observation that IFN-c was generated from pheno-typically naïve-like T cells (T N ) in response to MSP1 42 from children, but not adults, could be an indication that malaria is driving an effector function from a population of atypical ''transitional'' T-cells. Since these cells have not undergone full maturation, there is high likelihood of effector T N attaining a state of anergy and, thus leading to failure to contain infection. It is not known whether antigen experienced T N differentiate into T EM or T CM , become anergic or undergo apoptosis. Given that infants born to P. falciparum infected mothers may experience malaria attacks earlier than those whose mothers do not have malaria during pregnancy [55], it will be important to incorporate into longitudinal cohort studies the impact of fetal antigenic sensitization on malaria-specific T-cell differentiation during early childhood.

Ethics Statement
Ethical approvals were obtained from the University Hospitals of Cleveland Institutional Review Board for Human Studies at . The predominant IFN-c expressing CD8 + T cell in adults displayed a T EM phenotype and this was significantly different than children (P = 0.009) who, similar to CD4 + T cells, had IFN-c expression in response to MSP1 42 from all four CD8 + T cell subsets with a predominance of phenotypically naïve secreting T cells (P = 0.030) compared to adults. IFN-c responses observed for CD8 + T CM and T EMRA were of similarly low frequencies across age groups. doi:10.1371/journal.pone.0024852.g007 Case Western Reserve University and the Ethical Review Committee at the Kenya Medical Research Institute. Informed, written consent was obtained from each adult study participant and from the parent or guardian of minor study participants prior to enrollment in this study.

Study population and study area
This was an age-structured study involving 49 participants grouped according to median age of 0.5, 1, 2, 5 and $18 years. All participants were healthy lifelong residents of a malaria holoendemic region of Nyanza Province in western Kenya. The study  was conducted in the region near the Chulaimbo Rural Health Training Center managed by the Kenya Ministry of Health. Insecticide treated bed nets were not widely distributed in the area at the time the study was conducted (February and March 2007). Historically, the entomological inoculation rate in this area has been estimated at 0.65 to 0.79 infectious bites per person per day [54]. None of the study participants had a history of clinical malaria or taken anti-malarial drugs within the previous 4 weeks. Signed informed consent was obtained from adults, defined as residents who were $18 years old, and from the parent or guardian of participants who were 0.5 to 5 years old.

Blood collection and PBMC isolation
Approximately 2-5 ml and 8-10 ml venous blood samples were collected into heparin anti-coagulated tubes from 0.5-5 and $18 year olds, respectively. Total white blood cell counts (WBC) and differentials describing proportions of lymphocytes, monocytes and polymorphonuclear leukocytes were assessed using a Coulter Counter (Coulter AcT Diff 2, Beckman Coulter, Miami, FL). The absolute number of CD4 and CD8 cells per ml of whole blood was determined according to WBC and flow cytometry using anti-CD4 and anti-CD8 antibodies. Absolute numbers of lymphocytes per ml blood were obtained from the Coulter Counter and the percentages of CD3, CD4, and CD8 cells within the lymphocyte population were acquired from flow cytometry data. The percentages of CD3, CD4, and CD8 cells were matched to the absolute number of lymphocytes to calculate the absolute numbers of these T-cell subsets. PBMC's were separated from fresh whole blood by Ficoll-hypaque density gradient centrifugation and suspended in culture medium (RPMI 1640 (GIBCO, Invitrogen, Paisley, Scotland UK) supplemented with 10% heat inactivated human AB serum, 50 mg/ml gentamicin, 10 mM HEPES and 2 mM L-glutamine). PBMCs were used at a concentration of 5610 5 /well for IFN-c ELISPOT and T cell memory subset phenotyping.

Light microscopy to detect P. falciparum infection
Thick and thin smears were prepared from venous blood samples at the same time PBMCs were obtained. The slides were air dried, fixed in 100% methanol and stained with 5% Giemsa for enumeration of P. falciparum infected erythrocytes. A smear was deemed negative when microscopic inspection showed no parasites after counting fields that included at least 200 leukocytes. Density of parasitemia was expressed as the number of asexual P. falciparum/ml blood assuming a leukocyte count of 8,000/ml of whole blood.

Superantigens and malaria antigens
PBMCs were incubated with culture medium and PBS (blank control), Staphylococcal Enterotoxin B (SEB) at 2 mg/ml and recombinant MSP-1 42 (3D7 allele) at 5 mg/mL [56]. SEB served as the positive control. Recombinant MSP-1 42 was provided by Carole Long and Sanjay Singh (NIAID, NIH, Rockville MD). The construct contains T cell and B cell epitopes in the MSP1 33 and MSP1 19 fragments, respectively [57].

Cell surface staining for ex vivo T-cell memory phenotypes
Five hundred thousand PBMCs were suspended in 100 ml 0.5% BSA-PBS (wash buffer) and stained with the following panel of antibodies to characterize the memory T cell phenotype: CD3-APC, CD4-PerCP or CD8-PerCP, Cy7, CD45RA -FITC and CD62L-PE. All antibodies were purchased from BD Biosciences and used according to the manufacturer's instructions. Stained cells were incubated in the dark at room temperature for 30 minutes. Labeled cells were washed with 2 ml wash buffer and fixed with 500 ml 4% paraformaldehyde for 15 min at 4uC in the dark. At least 10,000 gated events were acquired using a FACSCalibur TM flow cytometer (Becton-Dickinson).

IFN-c enzyme linked immunospot (ELISPOT)
MultiScreen 96-well plates (Millipore, Bellirica MA) were coated overnight at 4uC with capture anti-human IFN-c antibody (Endogen M-700A, Worcester MA) at a final concentration of 5 mg/ml. The next day, the plates were washed 3 times with PBS and blocked with 100 ml 10% fetal calf serum for 2 hours at room temperature. Plates were then washed 3 times with PBS, and 5610 5 cells seeded per well. SEB and MSP1 42 were added in respective wells and cultured for 60 hours in a humidified incubator with 5% CO 2 at 37uC. The plates were subsequently washed with PBS and secondary anti-human IFN-c antibody (Endogen M-700B) added at a final concentration of 0.75 mg/ml culture medium followed by incubation for 90 minutes at 37uC. Plates were washed 3 times with PBS-Tween, horseradish peroxidase (HRP) conjugated streptavidin (DAKO P0397, Carpinteria CA) added at a 1:2000 dilution, and incubated for 2 hours at room temperature. Finally, the plates were washed 3 times with PBS and 1% 3-amino-9-ethyl-carbazole in 0.1 M acetate buffer (HRP substrate) was added to visualize spots. Plates were scanned and spot forming units (SFU) counted by Immunospot satellite analyzer (Cellular Technology, Cleveland OH). An individual was defined as an IFN-c responder to MSP1 42 if the frequency of SFUs per 10 6 PBMCs in the stimulated well was significantly greater than the well containing culture medium alone using Fisher's exact test. The range of IFN-c secreting cells in the unstimulated wells was 0-7 SFU/million PBMC.
Long term culture and intracellular cytokine and surface staining for T cell memory subsets At the same time that IFN-c ELISPOT assays were started, parallel aliquots of PBMCs from all 49 donors were seeded at a concentration of 2.5610 5 cells/200 ml culture medium in duplicate wells of 96-well round bottom culture plates, washed 60 hours later and re-suspended in fresh culture medium supplemented with 20 U recombinant human IL-2. All 25 PBMC samples that had positive IFN-c ELISPOT responses (determined after 60 hours ex vivo incubation as described above) and those that had enough cells but lacked detectable IFN-c responses were further evaluated for T cell memory subsets in the long-term cultures supplemented with IL-2. On day 6, from the initial seeding, cells from duplicate wells were pooled, washed and incubated for 18 hours with fresh culture medium alone, SEB or MSP1 42 . During the last 6 hours of culture, brefeldin A was added to allow for intracellular accumulation of IFN-c. Cells were washed with 20 mM EDTA-PBS and transferred into 5 ml polystyrene tubes, washed with 0.5% BSA-PBS and labeled for detection of cell surface CD3, CD4, CD8, CD45RA, and CD62L at room temperature for 30 minutes. Fixing was done with 4% paraformaldehyde for 15 min at 4uC in the dark. Cells were then washed twice with HEPES plus 0.1% saponin (permeabilization buffer) and cell membranes permeabilized for 30 min at 4uC in the dark. 50 mL permeabilization buffer and anti-human IFN-c APC antibody were then added and incubation continued for 30 minutes. Finally, cells were washed twice with 500 ml permeabilization buffer, resuspended in 0.5 ml BSA-PBS and 5610 4 gated events per tube acquired for flow cytometric analysis. For IFN-c ICS, 50,000 gated lymphocyte events were acquired and any IFN-c spots from the unstimulated cells were subtracted from the MSP1 stimulated cells for each study participant to determine frequency of MSP1-specific IFN-c producing cells per age group as shown in Figure 5. The range of IFN-c positive cells found in the unstimulated CD4 and CD8 stained samples were 0.03-1.56% and 0.05-0.74%, respectively. For Figures 6 and 7 the proportion of MSP1-speicific IFN-c producing cells in each of the four T cell subsets (CD45 RA v CD62L) was then determined.
Enzyme linked immunosorbent assay (ELISA) for detection of total IgG antibodies IgG antibodies were measured by ELISA. Recombinant MSP142 protein was dissolved in 0.01 M phosphate-buffered saline (PBS) to a concentration of 0.1 ug/ml and added to immulon-4 plates (Dynex Technologies, Chantily, VA). After overnight incubation at 40C, washing and blocking in 5% non-fat powdered milk in PBS, duplicate 50 ml samples of serum diluted 1/100 in 5% powdered milk were added to wells, and incubation was continued for 2 hours at room temperature. After washing, 50 ml of alkaline phosphatase-conjugated goat anti-human IgG (Jackson ImmunoResearch, West Grove, PA), diluted 1/1000 in 5% powdered milk was added and removed after 1 hour. The plate was washed three times and substrate p-nitrophenyl phosphate was added in accordance with the manufacturer's instructions (Sigma-Aldrich, St. Louis, MO). The reaction was stopped with 3N NaoH and optical density (OD) was measured at 405 nm. An antibody response was considered positive if it was three standard deviations above the mean of the malaria-naïve negative controls [59].

Data analysis
Flow cytometric data were processed and analyzed using FlowJo software version 7.2 (Tree Star, San Carlos CA). Figure S1 is a representative histogram depicting the gating strategy used to capture MSP1 42 -specific IFN-c expressing CD4+ and CD8+ T cells. Figure S2 shows the gating strategy to determine the proportions of effector-memory CD4 and CD8 T cell subsets expressing CD45RA and CD62L. There was no minimum event number required for a positive response since the proportion of each T-cell subset back-gated from IFN-c producing cells (either CD4 or CD8, respectively) was reported for each individual (Figs. 6 and 7). The median frequency of each subset was used to demonstrate the phenotypic differences for the MSP1 IFN-c producing T populations in children compared to adults (Figs. 8  and 9). Fisher's exact test was used to compare the proportion of IFN-c responders across age groups and the difference between stimulated and unstimulated cells. Kruskal-Wallis test was used to compare the magnitude of IFN-c secretion by ELISPOT and frequencies of T cell subsets across the five age categories. All analyses were done using the Graphpad program (Graphpad Prism TM , La Jolla CA).