Differential Immune Response Associated to Malaria Outcome Is Detectable in Peripheral Blood following Plasmodium yoelii Infection in Mice

Malaria infection in humans elicits a wide range of immune responses that can be detected in peripheral blood, but we lack detailed long-term follow-up data on the primary and subsequent infections that lead to naturally acquired immunity. Studies on antimalarial immune responses in mice have been based on models yielding homogenous infection profiles. Here, we present a mouse model in which a heterogeneous course of Plasmodium yoelii lethal malaria infection is produced in a non-congenic ICR strain to allow comparison among different immunological and clinical outcomes. Three different disease courses were observed ranging from a fatal outcome, either early or late, to a self-resolved infection that conferred long-term immunity against re-infection. Qualitative and quantitative changes produced in leukocyte subpopulations and cytokine profiles detected in peripheral blood during the first week of infection revealed that monocytes, dendritic cells and immature B cells were the main cell subsets present in highly-parasitized mice dying in the first week after infection. Besides, CD4+CD25high T cells expanded at an earlier time point in early deceased mice than in surviving mice and expressed higher levels of intracellular Foxp3 protein. In contrast, survivors showed a limited increase of cytokines release and stable circulating innate cells. From the second week of infection, mice that would die or survive showed similar immune profiles, although CD4+CD25high T cells number increased earlier in mice with the worst prognosis. In surviving mice the expansion of activated circulating T cell and switched-class B cells with a long-term protective humoral response from the second infection week is remarkable. Our results demonstrate that the follow-up studies of immunological blood parameters during a malaria infection can offer information about the course of the pathological process and the immune response.


Introduction
The pathophysiological mechanisms that lead to a given outcome in malaria patients are thought to be influenced by epidemiological and immunological factors [1] along with the mechanisms of immune evasion of the parasite [2]. Natural acquired immunity against Plasmodium falciparum is incomplete, non sterilizing and can be progressively acquired only after years of repeated infection in adults, but generally not in pregnant women or young children, and does not persist over long periods of time [1].
In the immune response to malaria, innate mechanisms are able to limit parasite density [3], but antibodies (Abs) and T cells are required to completely eliminate blood-stage parasites. APCs are particularly important to activate T CD4 cells which fight against the parasite by producing inflammatory cytokines which activate other cells such as macrophages and helping B cell activation to produce Abs [4]. These Abs have a protective role in malaria [5] and act by blocking merozoite invasion [6,7,8], by inhibiting cytoadherence of mature parasite-infected RBCs (iRBCs) [9], by binding to effector cells to trigger parasite-killing effector responses, such as opsonisation and phagocytosis of merozoite or iRBCs [10,11] or the mechanism known as Ab-dependent cellular inhibition of intracellular parasites [12,13].
Peripheral blood (PB) sampling has so far been the main provider of information on human immune responses against malaria since it is the only readily accessible source of leukocytes. However, white blood cells (WBC) may not reflect the global response to malaria since the activated cells during the infections may appear in secondary lymphoid organs. Hence, a better understanding of measurable immune system cells and proteins in PB could help identify malaria clinical states in humans. Although studies in animal models have provided useful information on the mechanisms involved in developing protective immunity to malaria, most rodent malaria studies have examined lymphoid organs rather than circulating PB cells because of the large quantity of cells available in these organs. This determines that the extrapolation of experimental data to the human response to infection is not straightforward. A wide variety of host-parasite models have addressed malaria immunity since any single rodent model replicates all the features of human malaria [14]. Despite high genetic variability in human populations, most bioassays in mice have used combinations of Plasmodium species and inbred mouse strains, which explains the homogeneous outcomes obtained.
By convention, P. yoelii yoelii 17XL (PyL) is considered a uniformly lethal parasite strain when used to infect the mouse strains most commonly used, including BALB/c, C57BL/6, Swiss and CBA [15]. Consequently, to date little evidence has been compiled on natural resistance to PyL parasites, only DBA/2 strain survives PyL infection after developing only moderate parasitemia [16,17]. Previous results from our laboratory show spontaneous recovery from lethal PyL infection of around 20% of the mice from the non-congenic ICR strain [18]. In the present study, we aim to formally characterize this new malaria model and identify potential immune response profiles associated to the different infection courses and final outcome. After a first PyL challenge, 20% of outbred ICR mice naturally developed a protective humoral response that confers long-term immunity against homologue re-infections. Besides, repeated individualized cytometric analysis of WBC revealed that cell mobilization and phenotypes vary in mice showing different infection severities and outcomes. Collectively our data revealed dramatic WBC changes that take place during malaria infection and show, for the first time, a heterogeneous course with different blood immune responses to the disease in ICR mice.

Ethics statement
All animal care and experimental procedures carried out at the Universidad Complutense de Madrid complied with Spanish (R.D. 32/2007) and European Union legislation (2010/63/CE) and were approved by the Animal Experimentation Committee of this institution. The experiments here described involving animals are reported following the ARRIVE guidelines [19]. The number of animals was calculated using Statgraphics Centurion 16.1.18 software (Statpoint Technologies, Inc, Warrenton, VA, USA) to provide about 80% of statistical power with 95% confidence level, and always following the 3Rs principles.

Animals and parasites
Sixty outbread Hsd:ICR (CD-1) and forty inbred BALB/ cAnNHsd pathogen-free female mice, aged 7 weeks were purchased from Harlan Laboratories (Italy) and housed at random in airy racks containing LignocelH three-fourths bedding (Rettenmaier & Sohne, Rosenberg, Germany) and kept under constant standard conditions of light (12:12 h light:dark cycles), temperature (22-24uC) and humidity (around 50%) at the Animal House of the Universidad Complutense de Madrid. All mice were fed a commercial diet (2018 Teklad Global 18% Protein Rodent Diet, Harlan Laboratories) ad libitum. The rodent malaria parasite P. yoelii yoelii 17XL (PyL) was kindly provided by Dr Virgilio Do Rosario (Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa) and stored in liquid nitrogen after serial blood passages in mice.

Experimental infection
Hsd:ICR(CD-1) mice were infected intraperitoneal (i.p.), using a 30 G one-half needle under an approximately 10-15u angle -in the lower quadrant of the abdomen off midline, with 2610 7 PyL-iRBCs obtained from donor PyL-infected mice. Parasitemia was monitored sequentially in each mouse by performing Wright's eosin methylene blue solution-stained thin tail blood smears. Animals that exhibited a loss of weight more than 15% or respiratory distress or other signs of severe disease were sacrificed by cervical dislocation earlier than the final infection endpoint in order to minimize the animal suffering. Total parasite clearance in cured mice was also confirmed by PCR analysis and i.p. subinoculation of 50 ml of blood from the mice into naïve BALB/c recipients. Mice that recovered from 1 st infection were reinfected on days 60 and 420 post-first infection (pi) following the same infection protocol. RBCs were counted sequentially in each mouse using a hemocytometer and anemia was defined as a RBC number below normal for age (p,0.05). Age-matched uninfected mice were used as controls. Three independent experiments are shown (each n = 20).

Cell preparations
Single-cell suspensions were prepared sequentially from PB, taken from each mouse, for flow cytometry analysis. Around 40 ml of blood were collected from tail of each mouse (without any sign of anxiety or pain during the following 2 hours after the intervention) in PBS containing 0.1 M EDTA and, after RBC lysis with ACK Lysing Buffer (Gibco), WBCs from each animal were individually stained with different mixes of fluorescent Abs. Viable cell counts were always made by Trypan Blue exclusion using a hemocytometer. Total bleeding was always #100 ml/ mouse/wk to minimize the biological effects of blood loss [20].

P. yoelii protein extraction from infected whole blood
PyL protein lysates were extracted from the whole blood of infected Hsd:ICR(CD-1) mice, which were anesthetized previously by isoflurane inhalation showing .50% parasitemia. Whole blood was collected in tubes containing 0.1 M EDTA and kept at 280uC until protein extraction. The extraction protocol began with erythrocyte lysis using 10 vol of saponin 0.1% (w/v) in PBS. After twice washing in cold PBS, the pellet was treated with 2-3 vol of extraction buffer (50 mM Tris-HCl, pH 8.0; 50 mM NaCl; 0.5% Mega 10; CHAPS 3%) containing a protease inhibitor cocktail (Roche) and subjected to four freeze-thaw cycles. Finally, lysates were centrifuged and PyL total protein samples stored at 220uC until use. Protein concentration was determined by the Bradford protein assay (Bio-Rad).
PCR-quantification of parasite DNA in blood P. yoelii DNA was extracted from peripheral iRBCs using the NuncPrep TM Chemistry Isolation of DNA from Whole Blood protocol of the ABI PRISMH 6100 Nucleic Acid Prepstation (Applied Biosystems) according to the manufacturer's instructions. Oligonucleotide primers and probes for the P. yoelii yoelii 18S ribosomal gene subunit (GenBank Accession No. U44379) were taken from [21]. Parasite DNA quantification was assessed employing the 59 fluorogenic nuclease assay (TaqMan) using a FAMTM dye-labeled specific probe. The primers/probe used were (59-39 sequences): Forward, CTTGGCTCCGCCTCGATA; Reverse, TCAAAGTAACGAGAGCCCAATG; Probe, CTGGCCCTTTGAGAGCCCACTGATT. PCR reactions were done in triplicate. Amplification, data acquisition and data analysis were carried out using the ABI 7700 Prism Sequence Detector system (Applied Biosystems).

Western blotting
10 mg of PyL total protein extract were fractionated on 10% SDS-PAGE (Bio-Rad), transferred to nitrocellulose membranes and blocked with 5% non-fat skimmed milk in PBS. Membranes were incubated overnight at 4uC with the different collected sera at 1:5000 dilutions and then with the secondary HRP-labeled antimouse IgG (Amersham Bioscience) at a 1:5000 dilution.

ELISAs
Total IgM, IgG and PyL-specific IgG Abs were quantified by Ab isotype-specific ELISA. Total IgM and IgG were quantified using anti-mouse IgM or IgG as the capture Ab (Bethyl Laboratories) while PyL specific IgG isotypes were quantified using 0.5 mg/well of PyL total protein lysates as coating antigen (Ag) prepared in carbonate-bicarbonate buffered solution (Sigma). Coating Ags were incubated for 2 h at RT and subsequently overnight at 4uC. From this step onwards, the manufacturer's protocol was followed (Bethyl Lab.). Briefly, plates were blocked with 1% BSA in Trisbuffered saline solution and duplicate diluted serum samples were added for 1 h at RT (1:5000 for IgM; 1:50000 for IgG; 1:40-1:2000 for PyL-specific IgGs). Total IgM and IgG Abs were detected with HRP-labeled goat anti-mouse IgM or IgG at a 1:75000 dilution and IgG subclasses were determined with 1:45000 diluted HRP-labeled goat anti-mouse IgG1, IgG2a, IgG2b or IgG3 Abs. The enzyme reaction was developed using 3,39,5,59tetramethyl benzidine (TBM) as the enzyme substrate (Thermo Scientific). Samples were read at 450 nm in a Varian Cary 50 Bio spectrophotometer (Agilent Technologies). Sera from uninfected mice were used as negative controls. Purified myelomaderived mouse IgG, IgG1, IgG2a, IgG2b, IgG3 and IgM (Bethyl Lab.) were used to generate a logistic four-parameter sigmoidal standard. The lower limit of positivity (cut-off) was determined by the mean of negative healthy controls + 2 SD.

Adoptive transfer experiments
In groups of 4-5 animals, 6-7 wk-old female BALB/c naïve mice were injected i.v. with 150-200 mg of total IgG from pooled sera obtained from late deceased mice on days 8-11 pi which showed 1.57610 6 60.36 ng/ml of total IgGs in serum; or from surviving or uninfected mice on day 70 pi (10 days after the 2 nd infection) that had 3.98610 6 60.20 ng/ml and 1.70610 6 6 0.72 ng/ml of total IgGs in serum, respectively. Mice injected with PBS were used as infection controls. After 2 h of passive transfer of serum, mice were challenged with 2610 7 PyL iRBCs. One mouse in each group was left uninfected as a healthy control. Two independent experiments were performed. Animals that exhibited a loss of weight more than 15% or respiratory distress or other signs of severe disease were sacrificed earlier than final infection endpoint in order to minimize the animal suffering.

Cytokine antibody arrays
To determine cytokines and chemokines in the mouse sera, the Mouse Cytokine Ab Array II kit (AAM-CYT-2-2, RayBio) was used according to the manufacturer's protocol. Pooled serum samples from each group of mice (of days 3 or 7 pi) were applied to the membranes and, after incubation with the detection Ab, membranes were developed with streptavidin-HRP followed by a chemiluminescence reagent (Thermo Scientific). Membranes were then exposed to X-ray film. Pixel densities were calculated for each spot of the array using Quantity One software (Bio-Rad Laboratories) and mean values for duplicate spots were compared. Arbitrary cut-off expression values higher than 1.5-fold were set to consider changes between groups or differences within a group.

Statistical analysis
To assist sorting of the infected animals into the 3 experimental groups (early deceased, late deceased and surviving), linear regression analysis was used to determine the slope of the parasitemia increase during the first 15 days of the infection. Differences in multiple groups were analyzed using one-way analysis of variance (ANOVA) or Kruskal-Wallis followed, respectively, by the Student's t-test or the Mann-Whitney test to compare significant differences between individual groups in Prism 5 software (GraphPad Software Inc.). Significance was set at p, 0.05. Data are shown as means 6 SEM.

Primary P. yoelii 17XL infection leads to three malaria infection profiles in ICR mice
The i.p. infection with 2610 7 PyL in ICR outbred mice resulted into three different infection profiles according to their parasitemia and survival kinetics (Table 1). A diagram showing the experimental design is provided in Fig. 1.
A 20% of mice spontaneously resolved the infection and were designated as surviving mice (S). S mice showed a slow increase in parasitemia with a peak of 59% and resolved the infection by day 22 pi (Fig. 2A). The infection was lethal before day 15 pi in the remaining 80% of the animals. Among mice that would die after the infection, two different profiles were observed: Early deceased mice (ED) showed rapid-onset fulminating parasitemia with a peak of 83% and died before day 8 pi ( Fig. 2A), being significantly associated the day of their death and the slope of the parasitemia increase (p,0.05; R = 0.82). Different from this behavior, the rest of mice which died after infection were designated as late deceased mice (LD) and underwent a slow increase in parasitemia, similar to that of S mice, which peaked at 54%, but followed fatal outcome around day 11 pi. The slope of parasitemia growth was significantly different between ED mice and LD or S mice (both p,0.0001), but no differences were found between LD and S mice (Fig. 2B). In LD mice the time of death was not associated with the slope of the parasitemia (p.0.05; R 2 = 20,05). Total clearance of parasites in S mice after infection was confirmed by microscopy examination of blood smears, PCR and sub-inoculation of blood in naïve BALB/c mice. These three infection profiles remained unchanged when the infective dose was 20-fold lower (1610 6 PyL-iRBCs) (data not shown). To ascertain whether the intrinsic properties of the parasites might contribute to the development of the heterogeneous infection profile observed in ICR, 2610 7 PyL iRBCs were collected from ICR mice with high-level parasitemia (35%) on day 3 pi and from mice with low-level parasitemia (those with ,0.5% on day 3 pi and 35% on day 6 pi) on day 6 to inoculate into 5 BALB/c inbred mice. The course of infection in both BALB/c groups were similar and followed the typical profile since they died on day 6.062.3 and day 6.662.0, respectively (data not shown).
After their recovery, S mice were reinfected twice on days 60 and 420 pi using the same challenge PyL doses. 100% of these animals survived both reinfections and none of them exhibited parasites in PB ( Fig. 2A). All mice showed 7.760.3610 9 RBCs/ml of blood at the beginning of experiment, but anemia was detected in all groups of animals (Fig. 2C). In ED mice, RBC loss was significantly evident from day 3 pi until death (p = 0.01) whereas in LD mice, the drop in RBCs started on day 9 pi (p = 0.02). In S mice, RBC counts fell from days 6 (p = 0.03) to 14 pi (p = 0.04), but thereafter recovered and counts were comparable to initial levels. In both LD and S mice, infection induced an increase in WBC (Fig. 2C). In S mice, WBC increased 3.6-fold by day 21 pi (p = 0.03), but by the start of the 2 nd infection counts returned to baseline.

Circulating monocytes and DCs show a marked increase in ED mice
Both dendritic cells (DCs) [22] and macrophages [23] have been shown to have a protective effect in malaria infection. However, the acute phase of lethal PyL may impair DC function [24]. Changes produced in PB activated monocytes (Mac-3 + MHC II + ) and DCs (CD11c + MHC II + ) during PyL infection and possible links to different outcomes were assessed (Fig. 3A). Initially, mice showed 10.6962.57610 4 macrophages and 3.8060.30610 5 DCs per ml of blood which accounted for a 0.5760.16% and a 2.0560.18% of WBCs, respectively. ED mice showed most changes in PB innate immune cells. Thus, 10-fold and 5-fold increases were observed in the frequencies of monocytes and DCs respectively on day 6 pi (Fig. 3B, C; p,0.01), when parasitemia was at its maximum level ( Fig. 2A). Total cell numbers showed similar kinetics, with an 18-fold increase in monocytes and 10-fold increase in DCs detected on day 6 pi in ED mice (both p,0.01).  Remarkably, S mice showed the earliest monocyte expansion on day 3 pi (p = 0.01) although this increase was reduced relative to the expansion in ED mice on day 6 pi.

S mice show the enhanced mobilization of CD8 and CD4 T cells
It is widely accepted that CD4 T cells are essential to control blood-stage malaria infection and that CD8 T cells play a role during the liver-stage of the parasite cycle [25]. However, the contribution of the latter T cells to blood-stage infection remains unclear. At day 0, mice had 4.5360.12610 6 CD4 + cells and 1.9860.06610 6 CD8 + cells per ml of blood which constituted a 29.6060.68% and a 13.1060.46% of total leukocytes, respectively. In S mice, CD4 + and CD8 + T cells showed similar kinetics in blood (Fig. 4A, B). Percentages of circulating CD4 and CD8 T cells decreased from day 9 pi onwards during the 1 st infection (p, 0.05, except for CD8 T cells on day 14 pi). In contrast, total numbers of CD4 T cells were elevated on day 21 pi (p = 0.01) and numbers of CD8 T cells increased on days 14 and 21 pi (p = 0.03). During the 2 nd infection in S mice, initial CD4 T cells levels were recovered, but CD8 T cells remained in lower number and proportion than in uninfected mice. ED mice showed a reduction in the number (p,0.01) and proportion (p,0.01) of CD8 T cells and unchanged CD4 T cell levels, whereas LD mice showed a similar trend to that observed in S mice.
During primary infection elevated levels of activated lymphocytes are detected in the peripheral blood of surviving mice CD44 is a ligand for hyaluronic acid, which is up-regulated in activated/memory cells mediating rolling and adhesion during the traffic of activated lymphocytes to target sites of immunity [26]. While standard number and frequency of activated leukocytes (CD44 + ) were 2.8960.19610 6 /ml of blood and 18.8061.10% of total WBCs, these values markedly augmented in S mice, from day 9 to day 21 pi (all p = 0.03) (Fig. 4C). To differentiate peripheral T cells, which can be naïve or previously activated Ag-experienced memory cells, we examined their expression of the CD44 receptor. In S mice, the T CD4 + population showed a higher frequency of CD44 + cells on day 21 pi (p = 0.01) and higher CD44 + cell numbers on days 14 (p = 0.03) and 21 pi (p = 0.04). T CD8 + CD44 + cells showed similar increases on day 21 pi (p = 0.03 both frequency and number) (Fig. 4D).
Circulating CD4 + CD25 + T cell expansion and Foxp3 expression vary with outcome Expression of the IL-2Ra chain, CD25, is a widely used, but not exclusive marker for T regulatory cells (Treg) [27] and the suppressor activity of CD4 + CD25 + cells is well documented [28,29] (Fig. 5A). The frequencies and numbers of CD4 + CD25 high cells were early increased in mice with fatal malaria (Fig. 5B), while in healthy mice numbers remained at levels of 1.9060.10610 5 cells/ml in blood that accounted for a 1.2160.07% of total WBCs. In S mice, only an increase in cell numbers was detected at the end of the 1 st infection (p = 0.03) once the parasitemia had been controlled, and these returned to the normal range after the 2 nd challenge.
Next, we analyzed the intracellular expression of Foxp3 during the first infection to detect natural Tregs, defined as CD4 + CD25 high Foxp3 + cells [29] (Fig.5C) and reported to regulate the immune response in malaria [30]. On day 3 pi. the frequency of circulating Tregs increased in ED mice and decreased in LD and S mice (Fig. 5C). Cell levels remained unchanged along the remaining length of the first infection (data not shown).
Transitional, mature and switched-class B cell kinetics differ according to infection severity Mice lacking B cells reveal the importance of B cells in malaria as they are unable to clear P. yoelii [31] and P. chabaudi infections  [32]. To explore the maturation of B cells during blood-stage malaria infection according to severity, the expression of surface IgM and IgD was determined (Fig. 6A). When immature B cells leave the bone marrow, they develop into transitional 1 (T1) (IgM hi IgD low ) B stage cells, which via the bloodstream reach the spleen. They then become transitional 2 (T2) (IgM hi IgD hi ) cells and progress to mature cells (IgM low IgD int ), which re-circulate to find an Ag or remain in the follicular zone of the spleen or progress to marginal zone B cells [33,34]. After Ag-induced differentiation, B cells commonly switch from expressing IgM to IgG Abs, generating IgM 2 IgD 2 cells that include short or long-lived plasma cells and memory B cells [35]. In blood, cells gated as IgM hi IgD low also include B-1 cells [36]. In our study, CD11b + CD5 2 B-1b cells (B-1a cells, which are CD5 + , were absent from blood) were detected at very low percentages (always ,4.5% of IgM hi IgD low cell levels) while CD11b 2 CD5 2 T1 B cells comprised the majority of the IgM hi IgD low cell gate (always . 70% of IgM hi IgD low cell levels) (Fig. 6B). All subtypes, except T2 cells, were detected in the PB of all mice. T1 B cells increased their frequency in all infected mice in the first week of infection (p, 0.05), in contrast to mature and class switched B cell proportions that mainly decreased (p,0.01) (Fig. 6C). Healthy mice showed 4.6760.23610 6 of mature B cells and 8.0060.60610 5 of T1 B cells per ml of blood, which accounted respectively for 30.7961.38% and 5.3860.40% of total B cells, but the infection produced changes in those values. Although elevated T1 B cell numbers were detected in blood from day 3 in both ED (p,0.01) and S mice (p = 0.03), the increase was highest in mice with the worst prognosis. Total mature cells presented minor changes; first a decline in all infected mice on day 3 pi and then an increase in LD and S mice on day 6 pi (p = 0.03). Isotype-switched B cell numbers showed a small increase in ED (p = 0.03) and LD mice (p,0.01) on day 6 pi, but in S mice at the end of infection, the increase produced was 8-fold (p,0.01).
The B220 + MHC II + subset and proportion of B220 low cells MHC class II is constitutively expressed by mature B cells at modest levels, but is dramatically overexpressed on activated B cells and lost after differentiation to plasma cells. Thus, the ability of B cells to up-regulate MHC II expression following activation is likely to be critical for their ability to function as APCs. We examined activated B cells (B220 + MHC II + ) in blood a decrease was observed in their frequency in all infected mice, but only significantly so in S mice on days 14 (p = 0.01) and 21 pi (p,0.01) (Fig. 7A). Healthy mice showed 4.1560.17610 6 B220 + MHC II + cells during all the experiment that was 29.1661.45% of total WBCs. The absolute B220 + MHC II + cell numbers in S mice, even when a gross increase in WBC was observed, represented half the initial cell numbers on day 14 (p = 0.02) and day 21 pi (p = 0.04). Normal levels of active B cells in PB were recovered at the beginning of the 2 nd infection (data not shown).
We could distinguish two populations of B220 + MHC II + cells according to the B220 levels: B220 high MHC II + (B high ) and B220 low MHC II + (B low ) as shown in the dot plot in Fig. 7B. Descriptions in the literature already exist of the distribution of B220 high/low populations in mice suffering non-lethal P. yoelii 17XNL (PyNL) infection [37], amyloidosis [38] or mammary tumor virus infection [39] and in healthy neonatal or old mice [40] and ovariectomized mice [41]. Our healthy mice constantly showed about 25% B high cells in PB whereas the B low subset only represented around 1.5% of the total WBC count (data not shown); the normal proportion of B high /B low cells being around 9:1 among total activated B cells (Fig. 7B). However, malaria infection promoted a rise in B low cells and drop in B high cells in all mice from day 3 or 6 pi.

Antibodies in serum during three consecutive PyL infections
Abs are crucial components of the protective immune response against malaria in human and animal models [5,44]. PyL infection modifies Ab production in LD and S animals (Fig. 8A). While serum IgM levels peaked in the first infection, IgG Ab production started to increase after the 2 nd wk of infection and peaked after the 2 nd challenge in S animals. Among the serum PyL-specific IgG isotypes, IgG2b were most abundant in S and LD mice (day 14 pi), though IgG2a and IgG1 also reached high levels in S mice (Fig. 8B). In the 1 st infection, LD and S mice differed mainly according to the presence of IgG2a in the latter on day 14 pi. The 2 nd challenge promoted a rapid 3-4.5-fold expansion in IgG2b, IgG2a and IgG1 and the appearance of IgG3 Abs, which were always the least abundant subclass during all infections. Third PyL challenge, induced again the Ab production. To assess specific IgG reactivity, we conducted a time-course IgG immunoblot analysis using total P. yoelii proteins (Fig. 8C). IgG Abs recognized a wide range of parasite Ag, the strongest signals appearing in the high To detect baseline signals, control immunoblots using RBC proteins showed that residual bands recognized by sera from infected mice from day 0 to 14 pi were comparable to those from healthy mice sera and therefore are credited to unspecific recognition of non-infected RBC proteins (Fig. 8C).

Serum from surviving mice partially protects BALB/c mice from PyL infection
To confirm the protective role of anti-PyL Abs in S mice, we performed passive transfer assays in BALB/c mice, which is a sensitive strain to PyL. Animals were inoculated with 200 mg of serum IgGs taken on day 74 pi from S mice or age-matched uninfected mice or on days 8-11 pi from LD mice and then infected 2 h later with 2610 7 PyL iRBCs. Transfers using PBS were performed as controls. No protection from infection was conferred by serum from LD mice, uninfected mice or PBS, but pooled sera from S mice was able to cure 40% of transferred naïve mice confirming the protective capacity of anti-PyL Abs elicited in spontaneously cured ICR mice (Table 2) Circulating cytokines in early deceased and surviving mice The balance between pro-and anti-inflammatory responses is essential to limit an immune-mediated disease [45]. In human malaria, evidence exists of a link among cytokine profiles in sera, disease severity and parasitemia [46,47]. We compared serum PyL iRBCs and classified as early deceased (ED), late deceased (LD) or surviving (S). Survivors were reinfected on day 60 pi. WBCs were isolated and (A) B220 + MHC II + cells percentages and absolute numbers were normalized to data obtained in uninfected mice (n = 5 per experiment). Standard numbers of B220 + MHC II + cells were 4.1560.17610 6 /ml blood. * p,0.05 with respect to uninfected mice. (B) Representative flow cytometry dot plots identifying B220 low MHC II + and B220 high MHC II + populations in infected mice. Graph shows the contribution of each subpopulation to the total number of B220 + MHC II + cells. As uninfected mice (n = 5) maintained in all time-points same proportions as day 0, for simplicity the graphic has not been shown. cytokine profiles during the 1 st wk of infection in S and ED mice by protein microarrays. S and ED mice showed different levels of cytokines examined on days 3 and 7 pi (Fig. 9). At 3 days pi, S mice secreted higher levels of cytokines such as hematopoietic IL-3, the Th2 cytokine IL-4, and the Th1 cytokines IFN-c and IL-2 than ED mice. Conversely at 7 days pi, ED mice showed higher levels of most of the markers, possibly as the result of their state of terminal decline and consequent physiological dysregulation. At this time point, S mice showed similar cytokine levels in serum to those observed on day 3 pi, though most were overall reduced. Greatest reductions from days 3 to 7 pi were produced in IL-4, TNF-a, IL-13, IL-2, IL-3, IFN-c and IL-17 while VEGF (vascular endothelial growth factor), eotaxin, IL-6 and sTNFR1 (soluble tumor necrosis factor receptor 1) levels were slightly higher.

Discussion
Here, we show that the serial analysis of blood immune factors in an animal model yielding different outcomes of infection can provide useful information on the course of the immune response and pathological processes. Our experiments reveal for the first time that the outbred ICR mouse strain mounts different  L immunobiological responses against primary PyL infection associated to the course of infection. Three different infection profiles were observed according to parasitemia and survival: a fulminating parasite growth that led to a fatal outcome before day 8 pi in 60% of animals (designated early deceased, or ED mice) or a more sustained parasitemia that was lethal before day 15 pi in 20% of animals (designated late deceased, or LD mice) yet self-resolved in a further 20% (designated surviving, or S mice). In S mice, parasitemia followed the typical non-lethal Plasmodium kinetics for rodents, peaking after 2 wk and clearing by the 3 rd -4 th wk.
Although mice surviving a malaria infection usually become immune to following infections, 5% of BALB/c naïve mice that recover from P. yoelii nigeriensis primary infection remain susceptible to reinfection a month later [48]. After parasite clearance, homologue reinfections in S mice revealed that these animals develop a protective long-standing immune response for 14 months.
In our model, PyL parasite contact promoted a late leukocytosis, consistent with the results of prior murine studies [49,50,51], but contrary to reports of acute malaria in humans in which leukocytopenia and lymphopenia are characteristic features [52,53]. This discrepancy and the drastic changes in mouse blood cell composition observed during blood-stage P. chabaudi [49,50] and PyL infections (present study) point to the importance of timing of blood sampling during a malaria infection. Some authors have even recently suggested that immunological studies based on WBC in mice may provide data comparable to the infection process in humans [14].
Our serial cytometric phenotyping of WBC served to identify different cell kinetics in ICR mice according to disease outcome. The first defense barrier against a 1 st wave of Plasmodium is the innate cell response that promotes the subsequent T cell-mediated response [3]. Macrophages play a critical role in the immune response to malaria due to their ability to phagocytose iRBCs in the absence of cytophilic or opsonizing Abs, to activate T cells through Ag presentation on MHC II and to release inflammatory cytokines. Particularly in PyL infections, the key role of macrophages has been clearly demonstrated [23]. The contributions of DCs include Ag uptake and stimulating T-helper cells [54]. In the present study, although ED mice experienced the highest increase in blood levels of both innate cell populations during the 1 st wk of infection, the innate response elicited was unable to control the high parasitemia (over 60%). The prompt increase observed here in circulating innate system cells in ED mice indicates that high numbers of circulating monocytes or DCs are not needed to control the infection, rather they could be markers of an inadequate innate response at early time points in acute malaria infection. In S and LD mice, minor changes in circulating innate cells during the first week of infection were observed. Hence, the general ''stress condition'' caused by severe infection in ED mice probably contributed to the dysregulation of immune responses, also reflected by the over production of the majority of cytokines measured. In human malaria, iRBCs can impair the IFN-cinduced class II MHC expression on monocytes [55] and hemozoin can inhibit maturation of human monocyte-derived DCs [56], mechanisms that would diminish the T cell responses [55]. However, our data indicate that the MHC II expression and the differentiation of monocytes into DCs expression during murine Plasmodium contact in vivo are not impaired, as described in other congenic mouse strains [57]. Hence, the low efficiency of monocytes in our ED mice could be the result of an insufficient level of IFN-c compared to levels in surviving mice, as detected in PB by microarrays at the start of infection. In agreement, higher IFN-c levels in splenocyte cultures have been reported in resistant DBA/2 than susceptible BALB/c mice during PyL infection [58].
Neither T nor B lymphocytes in general seem to be required to control the 1 st wave of P. yoelii infection [23]. However, the suppressive role of CD4 + CD25 + cells [28,29] make of this population an important player during bacterial [59], viral [60], helminth [61,62] or protozoan parasite infections including malarial [63] [60,61,62,63,64]. In the CD4 + CD25 + population are included effector CD4 T cells and Treg cells which play an important role through suppression of the Th1 response [30]. In our study, the increase of PB CD4 + CD25 + cells observed in mice with a fatal outcome occurred before the parasitemia peak, what could be impairing the development of effective protective immunity [30,65], whereas in surviving mice it was only observed when parasitemia almost cleared. In our mouse model and other mouse strains [66] expansion of CD4 + CD25 + cells correlates with disease outcome. CD4 + CD25 high Foxp3 + Treg cells followed different kinetics in PB according to parasitemia. Lower percentages where found in mouse groups with lower parasitemias (LD and S mice), while mice with high parasitemias and lethal outcome after a malaria infection was accompanied by an increase of Tregs at the beginning of infection. Thus, the suppressive activity that Foxp3 expression confers to naive T cells [67], might not be Figure 10. Summary of immune response in peripheral blood of ICR mice with different clinical courses of PyL infection. Left: ED mice on day 6 pi reached high parasitemia, accompanied by a large increase of CD4 + CD25 + cells, immature T1 B cells, innate response cells an overproduction of cytokines. Right: S mice on day 6 pi showed low parasitemia with a slight increase in the circulation of mature and immature B cells and a limited production of cytokines. On day 14 pi, S mice reached around 45% parasitemia and increased the number of activated B and CD4 + T cells in addition to high levels of circulating anti-parasite specific IgGs. Parasitemia clearance on day 21 pi was concurrent with the highest leukocytosis level, while increase in specific humoral response continued. In S mice the major increases of leukocyte populations in comparison to healthy mice were in activated CD4 + and CD8 + T cells, class-switched B cells and, to a lesser extent, T1 B cells and CD4 + CD25 + cells. doi:10.1371/journal.pone.0085664.g010 beneficial in the first days of malaria infection according to our results.
In the first week of infection, when most of the mice succumbed to the infection, T1 B cells showed the greatest changes among all the B subpopulations. T1 B cells increased the frequency in all infected mice, in contrast to the rest of B cells, and their numbers were particularly enlarged in ED mice blood. This may suggest diminished competence of the immune response since peripheral immature B cells lead to B cell tolerance in the absence of T cell help [68,69,70].
A subset of B cells expressing low levels of B220 protein was observed in infected mice. Although it has been speculated that B220 low cells might be beneficial in malaria PyNL infections [37], similarly increased PB levels of these cells were observed in all our mouse groups and we characterized them mainly as immature B cells and not as B-1 B cells as previously described [37].
From the second week of infection, a late increase in activated T and class-switched B cells, together with the high production of specific Abs, conferred to S mice a very different immunological profile than in the first week. The role of CD8 T cells showed by transfer experiments in P. yoelii malaria models is still controversial [71,72,73]. Here, the increase in the total number of CD8 + cells with an activated phenotype detected here in S mice could point to a protective role or at least suggest they do not impair a proper immune response.
In S mice, the decline of B cells acting as APCs (B220 + MHCII + ) in blood paralleled to class-switched B cells increase in the PB could suggest the exit of B cells from the circulating population to lymphoid organs upon stimulation by parasite Ag, causing them to switch class to Ag-experienced and memory B cells that lose the expression of MHC II as no longer need to bind or present Ag [74]. The inability of erythrocytes to present Ags prevents iRBC destruction by a specific MHC-restricted T-cell response. Immunity to blood stage malaria parasites is thus primarily conferred by humoral immune responses in which the participation of IL-4 producing CD4 Th2 cells and B cells are of major importance [75,76]. In S mice, an efficient humoral immune response was mounted during 1 st infection and maintained for more than a year after the second infection. Microarray analysis showed that IL-4 increased in S mice serum before than in ED mice, what could induce an effective production of Abs [77]. In LD and S mice the parasitemia course as well as general changes in WBC populations and humoral response in the first weeks were quite similar, suggesting that the death of LD animals could be in part due to ineffective erythropoiesis, as described in both human and rodent malaria infection [78].
The development of immunological memory in the S animal group was patent since both reinfections produced a rapid protective specific Ab response. All the IgG isotypes examined, IgG1, IgG2a, IgG2b and IgG3, were detected in the sera of S mice, consistent with findings in PyNL-infected ICR mice [79]. In the first infection, the cytophilic IgG2a and IgG2b Abs [80], that would be induced by a Th1 response [81,82], predominated over non-cytophilic IgG1 and IgG3 Abs [80] that would respond to Th2 response [81,82]. However, this prevalence decreased in the following infections suggesting the establishment of a Th1 immune response during the first infection and a Th2 response in the reinfections. Studies in humans suggest that the presence of malaria-specific Abs may be dependent on the presence of chronic parasitemia [83], but the clearance of blood parasitemia in PyL-ICR mice after each infection was confirmed by microscopy, PCR and re-inoculating blood in naïve mice. The persistence of parasites in the spleen or other organs was, however, not investigated. Our immunoblots revealed that the repertoire of PyL Ag was recognized by the specific IgGs raised after each reinfection. This has also been observed in acquired immunity to human malaria and is likely to depend on the build-up of a wide range of antigenic specificities over a long period [84]. In our malaria model, parasite proteins exhibiting antigenicity spanned a wide MW range. Remarkably, circulating Abs against high MW PyL Ags were preferably maintained after several months without parasite re-exposure. In a recent proteomic study, we identified some of these PyL Ags with Abs from protected S mice, as a new strategy to develop multi-Ag-based vaccine therapies [85]. The goal of the serum-transfer experiments was to determine the protective capacity of the humoral response at the time reaching its maximum in a given group of infected mice. The passive transfer of hyperimmune sera has shown the concept that immunity to PyNL [44,86] and to lethal P. yoelii nigeriensis [48] is largely humoral, and this is now extended to include the PyL strain by our passive transfer results.
The results obtained in our rodent malaria model indicate rapid cell changes in the PB of ICR mice during blood-stage PyL malaria related to the severity of the infection and outcome (Fig. 10). During the first week of the infection, the immune response observed in the PB of highly parasitized mice (that had the worst prognosis), consisted of a rapid/higher increase in circulating CD4 + CD25 + T cells with higher expression of Foxp3, in T1 B cells and in activated innate cells in comparison with the rest of mice groups. In contrast, the immune response observed in the same week in S mice, showed a limited production of cytokines and mostly unchanged circulating innate cell kinetics. After the first week of infection S mice were characterized by the circulation of activated T and B cells, together with the generation of a longterm protective humoral response.
Here, we demonstrate that outbred mice strains that show different clinical outcomes are valuable animal models to distinguish between potentially effective and ineffective immune responses to malaria. Thus, once described this model, future experiments are necessary to study in detail the immunological mechanisms of malaria susceptibility and protection in ICR mice. Moreover, whether this effect could also take place in humans deserve to be investigated since can eventually allow to identify WBC markers of severity and outcome of malaria infection, particularly useful in clinical studies to assess new vaccine formulations.