Metabolic Fingerprints of Serum, Brain, and Liver Are Distinct for


Metabolic Fingerprints of Serum, Brain, and Liver Are Distinct for...

9 downloads 95 Views 2MB Size

Article pubs.acs.org/jpr

Metabolic Fingerprints of Serum, Brain, and Liver Are Distinct for Mice with Cerebral and Noncerebral Malaria: A 1H NMR Spectroscopy-Based Metabonomic Study Soumita Ghosh,† Arjun Sengupta,† Shobhona Sharma,*,‡ and Haripalsingh M. Sonawat*,† †

Department of Chemical Sciences and ‡Department of Biological Sciences, Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai 400005, India S Supporting Information *

ABSTRACT: Cerebral malaria (CM) is a life-threatening disease in humans caused by Plasmodium falciparum, leading to high mortality. Plasmodium berghei ANKA (PbA) infection in C57Bl/6 mice induces pathologic symptoms similar to that in human CM. However, experimental CM incidence in mice is variable, and there are no known metabolic correlates/ fingerprints for the animals that develop CM. Here, we have used 1H NMR-based metabonomics to investigate the metabolic changes in the mice with CM with respect to the mice that have noncerebral malaria (NCM) of the same batchmates with identical genetic backgrounds and infected simultaneously. The metabolic profile of the infected mice (both CM and NCM) was separately compared with the metabolite profile of uninfected control mice of same genetic background. The objective of this study was to search for metabolic changes/fingerprints of CM and identify the pathways that might be differentially altered in mice that succumbed to CM. The results show that brain, liver, and sera exhibit unique metabolic fingerprints for CM over NCM mice. Some of the major fingerprints are increased level of triglycerides, VLDL-cholesterol in sera of CM mice, and decreased levels of glutamine in the sera concomitant with increased levels of glutamine in the brain of the mice with CM. Moreover, glycerophosphocholine is decreased in both the brain and the liver of animals with CM, and myo-inositol and histamine are increased in the liver of CM mice. The metabolic fingerprints in brain, sera, and liver of mice with CM point toward perturbation in the ammonia detoxification pathway and perturbation in lipid and choline metabolism in CM specifically. The study helps us to understand the severity of CM over NCM and in unrevealing the specific metabolic pathways that are compromised in CM. KEYWORDS: cerebral malaria, NMR, metabonomics, serum, brain, liver, OPLS-DA



INTRODUCTION

Experimental infection in a murine model is the most convenient and traditional alternative to a human malaria study of the pathogenesis of CM. The most widely used animal model of CM is PbA, used in a susceptible mouse strain such as C57BL/6 mice.8 Although this model does not reflect all of the complications of human CM, it can induce some of the neurological and pathological symptoms that mimic human CM.9 The pathological characteristics of human CM that this model represents are breakdown of the blood−brain barrier, sequestration of blood cells, increased expression of adhesion molecules, pro-inflammatory cytokines, and the activation of

Malaria is an infectious disease transmitted through the bites of infected female Anopheles mosquitoes, leading to a large number of deaths across the world.1 One of the major causes of fatality is cerebral malaria (CM),2 and it is caused by the human malarial parasite Plasmodium falciparum.3 The clinical course of CM is characterized by seizures, encephalopathy, and loss of consciousness.4 Most of the symptoms of CM have been attributed to a specific interaction between parasitized RBCs (pRBC) and endothelium, which reduces blood flow and induces hypoxia.5 However, the coexistence of pRBC sequestration with CM is controversial.6,7 Therefore, it is important to study molecular mechanisms in CM patients, which are distinct from non-CM patients. © 2012 American Chemical Society

Received: June 22, 2012 Published: July 27, 2012 4992

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article

microglia.10,11 The histopathological changes involve loss of vascular cell integrity, tissue edema, and congestion of microvessels.4 In the experimental model too, as in humans, not all of the mice infected with the parasite succumb to CM but exhibit variations. The mice that do not succumb to CM eventually die of high parasitemia and are grouped as noncerebral malaria (NCM). In several studies, specifically, the metabolic profiles of brains of CM and NCM animals are compared by using different genetic strains of mice as well as parasites, for example, NCM models such as Balb/C with PbA versus the CM model CBA mice with PbA12 or CBA with Plasmodium berghei K173 versus CBA with PbA.13 Moreover, it is known that the 1H nuclear magnetic resonance (NMR) urine profiles of C57BL/6 and Balb/C are different,14 which indicate that genetic differences affect metabolic constitution. Here, we have deliberately avoided using mice of different genetic backgrounds as the NCM model or different parasite strains for comparison. In our set up, approximately 50% of infected mice (C57BL/6 infected with PbA) succumbed to CM, allowing us to compare CM and NCM metabolic profiles within animals of the same genetic background and with the same batch and within the same experiment. The host response to parasite infection is rarely confined to one particular tissue, and usually, host homeostasis is achieved at the cost of a cascade of events in different tissues.15 It is axiomatic that liver, kidneys, and lungs are important components that help the organism maintain homeostasis. Plasma/serum composition reflects the status of this homeostasis and is also an excellent reporter of disturbances caused by environmental stressors. The liver was selected since it is an organ that is involved in toxic control. We have, therefore, studied the changes in liver along with sera to understand the metabolic networks/pathways that are altered due to malarial parasite infection. We anticipated that certain serum metabolites would be identified as markers of the disease. Investigation of the brain would, in a direct manner, help us understand the metabolic alterations that occur in this tissue specifically for CM and hence would indicate the severity of the disease. We have used 1H NMR-based metabolomics16 for our investigations. The metabolite profiles were generated by this systems biological strategy. 1H NMR is an unbiased method that gives the overall metabolic profile of the biofluids or the tissue extracts. Metabolomics is a hypothesis-free study that helps in nontargetable quantification of all assayable metabolites of the biofluid or the tissue under study and aims toward identification of biomarker(s) of the diseased condition.17 The technique has been used to study various host−parasite and host−symbiotic interactions.18,19 It is apparent from various studies that transgenomic interactions are better studied in vivo, and metabolomics is a potential tool to study such systems. We have compared the metabolic profiles of the sera, brain, and liver of mice with NCM vs CM and each group with respect to the uninfected controls to develop a mechanistic approach of the disease specifically in CM. We have studied the multiorgan perturbation caused by the disease to get a global response of mice with CM and NCM; therefore, we believe that we have an insight into metabolic alterations that occur along with (or responsible for) the disease progression. In this study, we find novel metabolic correlates in the sera, liver, and brain of mice with CM, while keeping NCM mice of the same genetic background as the control.

We report here the specific signature metabolites of CM. The results from our study point toward some of the metabolic pathways that appear to be compromised during CM and help us in understanding the various complications that arise specifically in CM with respect to NCM animals. Pathways such as ammonia detoxification and lipid/lipoproteins and glucose metabolism, the latter involving myo-inositol formation, are perturbed in CM.



EXPERIMENTAL PROCEDURES

Animal Handling

The animals were treated in accordance with the guidelines set forth by the Institutional Animal Ethics Committee of TIFR. Three separate experiments were performed in a batch of 20, 25, and 24 animals. Female C57BL/6 mice, 6−8 weeks old and weighing 20−25 g, were used for the study. They were maintained in a 12 h day and night cycle and had free access to water and standard food pellets. The temperature was maintained at 22 ± 2 °C. Experiments were repeated to get a statistically significant number and to see whether the markers from the one experiment predict CM in a separate set. Inoculation of Mice, Disease Progression, and Assessment

In the three batches, the number of uninfected mice was kept as 5, 5, and 6, respectively. The rest of the mice were inoculated with 107 RBCs infected with PbA. The parasites were maintained in Swiss female mice of 6 weeks. The donor mice were sacrificed when the parasitemia was around 20%, and the blood was collected in ACD (acid, citrate, and dextrose) diluted by physiological saline to 107 iRBCs per 100 μL. The rectal temperature of all of the mice was measured daily using a digital thermometer. Parasitemia of the infected mice was determined daily. This was done by counting the number of infected cells per 1000 RBCs in a thin blood smear stained with Giemsa.20 The mice were monitored daily for symptoms of the disease after day 5 postinfection. Mice were considered to have CM if they had a rectal temperature 34 °C. Thus, categorization of animals as having CM or NCM was unambiguous. Some of the mice died during the night, and those were excluded from our study. A total of eight mice died during the night (six mice before day 8 and two mice at day 8). On day 8, the conditions of some of the CM mice (a total of 10 mice in three experiments together) were critical so they were sacrificed, while we sacrificed all other CM and the NCM mice (a total of 35 mice) on day 9 postinfection because the neurological symptoms in CM were clearly visible in addition to the reduced rectal temperature. The uninfected control mice were also sacrificed on day 9 (mean temperature, 37.2 °C). Sample Preparation for 1H NMR of Sera

Blood (approximately 650 μL) was collected by retro-orbital bleeding just before dissection. Here, we deliberately avoided collection of blood by postmortem heart puncture because it resulted in the mixing of blood with the pleural effusion.34 The blood samples were incubated at 37 °C for 10 min and centrifuged for 10 min at 13100g at 4 °C. The supernatant was collected, frozen immediately in liquid N2, and stored at −80 °C. For NMR experiments, 300 μL of sample was mixed with the same volume of buffer (0.075 M Na2HPO4·7H2O, 4% 4993

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article 1

NaN3, and 0.02% DSS, pH 7.4). The buffer recipe was provided by Bruker Biospin, Metabonomic unit. The pH of the samples was checked before experiments.

H NMR of Extracts of Brain and Liver

1

H NMR spectra were acquired for all of the extract samples of brain and liver on AVANCE 700 MHz Bruker spectrometer. The experiments were recorded at 300 K. The pulse sequence adopted for the experiment was 1D nuclear Overhauser effect spectroscopy presat (NOESYPR1D). The pulse sequence is of the form RD-90-t1-90-tm-90-ACQ. The RD of 4 s was used for all of the experiments. The spectral width was kept at 20.06 ppm, and the mixing time (tm) used was 150 μs. The number of transients used was 32, and the time domain data points were 32000. In all of the experiments, the 90° pulse lengths were determined individually. The baseline and the phase correction were done with the help of AU program provided by Bruker Biospin.

Preparation of Tissue Extract

After the animals were sacrificed, the organs were quickly excised out, snap frozen in liquid N2, and stored at −80 °C until further processing. Each whole brain and whole liver was thawed and homogenized in ice-cold pH 7.4 buffer containing 0.075 M Na 2 HPO 4 ·7H 2 O and 4% NaN 3 in a glass homogenizer. To the homogenate, perchloric acid (5 mL/g of the tissue) was added. It was followed by three freeze−thaw cycles and sonication on ice. It was kept on ice for 10 min and centrifuged at 13000g for 10 min at 4 °C. The supernatant was neutralized to pH 7.4 by 2 M K2CO3 and recentrifuged. The supernatants were dried in vacuum concentrator and stored at −20 °C until further use for NMR experiments. The dried mass was reconstituted in D2O containing 0.01% DSS for NMR spectra acquisition.

Data Preprocessing

The 1H NMR spectra obtained after phase and baseline corrections were subjected to statistical processing. The spectral region of 0.5−4.14 ppm was bucketed into a frequency window of 0.02 ppm (0.04 ppm for extracts of brain and liver). The region corresponding to water (4.15−5.5 ppm) was excluded during binning to avoid possible artifacts due to presaturation of water. Likewise, the regions (3.53−3.59 and 1.12−1.15 ppm) arising from ether were excluded from analysis of the spectrum from serum. Furthermore, the aromatic region, which had a poor signal-to-noise ratio, was also excluded and was analyzed separately. The bins were integrated, and the resulting integrals were normalized to the working region (0.50−4.14 ppm) of the spectrum to correct for intersample variation in dilution. The binning and the normalizations were done using AMIX 2.0 (Bruker). The matrix so obtained was imported to SIMCA P+ 12.0 (Umetrics AB, Sweden) for further multivariate data analysis. In addition, specific peaks in the aromatic regions were integrated and processed using univariate analysis.

1

H NMR of Sera

1

H NMR spectra at 310 K were acquired on AVANCE 700 MHz Bruker spectrometer with triple resonance probe using D2O (10%) as the field-frequency lock. The samples were placed in a 5 mm NMR tube. The pulse sequence used for the experiment was of the form -RD-(τ-180-τ-)n-90°-ACQ. A relaxation delay (RD) of 4 s was used between consecutive pulses. Carr−Prucell−Meiboom−Gill (CPMG) spin echo block (τ-180-τ-)n with a loop count (n) of 128 was used to attenuate the broad signals from the macromolecules. The echo time for the experiment (τ) was 300 μs. In these experiments, 32 transients were collected into 88640 data points using a spectral width of 20.06 ppm. Ninety degree pulse lengths were determined for all of the samples individually. FIDs were subjected to exponential multiplication, leading to an additional line broadening of 1 Hz, a Gaussian multiplication of 0.01 Hz, and a sine-squared bell apodization function prior to Fourier transformation. The resulting spectra were phased and baseline corrected by the AU program provided by Bruker Biospin. The assignments of metabolites were based on 2D (1H−1H) correlation spectroscopy (COSY), 2D (1H−1H) total correlation spectroscopy (TOCSY), and 2D J-resolved spectroscopy experiments. The assignment was further aided using Human Metabolome Databases HMDB.23 The assignments were also cross-checked from the literature.24 For COSY and TOCSY experiments, 64 transients per increment and 256 increments were collected in the indirect dimension. A SINE function with 2048 and 1024 digital points was used for processing. Exponential multiplications of 0.20 and 0.30 Hz in the direct and indirect dimension were used, respectively. TOCSY was processed using the SINE function. Exponential multiplications of 0.20 and 0.30 Hz in the direct and indirect dimension were used, respectively. In addition, a Gaussian multiplication of 0.1 Hz in the indirect dimension was applied for the TOCSY. Twodimensional J-resolved spectra were also recorded to aid the assignments of the metabolites. The pulse sequence is of the form -RD-90-t1-180-t1-ACQ. The RD was 2 s, and the number of scans was 1 for all experiments. The numbers of data points in direct and indirect dimensions were 16k and 256, respectively. The spectral widths in direct and indirect dimensions were 16.63 ppm and 0.15 Hz ppm, respectively. After the Fourier transformation, baseline correction was performed followed by tilt (45°) of the spectrum.

Multivariate Data AnalysisPCA and Orthogonal Partial Least Square Discriminant Analysis (OPLS-DA)

The 2D data matrix was subjected to multivariate statistical analyses. The first step was principal component analysis (PCA),25 which is an unsupervised method. PCA was done to see if there is a trend in the data set and to find out if there are any outliers present. This was followed by OPLS-DA,26 which is a supervised method and gives segregation between two classes along the predictive component. The two parameters, namely, the R2 (cum) and Q2 (cum), judge the model. While R2 (cum) explains the total variation, Q2 is a cross-validation parameter that indicates the predictability of the model. In each cross-validated round, 1/7 of the data were taken out. The scores were visualized as 2D scores plot. The differences in the relative concentration of metabolites in the sample were interpreted by “S” plot and the variable importance plot (VIP). The S plot gives the loading as well as the correlation values of the chemical shifts of metabolites related to the class segregation denoted by p[1] and p(corr), respectively. The VIP values from the VIP plot reflect the importance of the chemical shifts both with respect to class segregation and the loading (p[1]). From these two plots, a “V” plot was generated by plotting the p(corr) versus VIP. Both the S plot and the V plot were considered during choosing the relevant chemical shift of the metabolites. A threshold value >1.0 was used for the VIP, for the selection of the chemical shift of the bins for further investigation because these bins are relevant to class segregation,27 while a threshold value of 0.7 for p(corr) was used as cut off since p(corr) 34 °C and did not exhibit neurological symptoms were considered to be NCM.21,22 The grouping of the mice as CM and NCM was done on the basis of daily measurement of rectal temperature34 and symptoms for the disease. In a separate experiment, the infected animals were observed and were allowed to die naturally. The resultant survival curve is shown in Figure 1. It is

error in coefficient plots of the corresponding OPLS-DA model. To further investigate the metabolites that are specifically affected in one of the infected classes (CM/ NCM) with respect to the uninfected controls, a shared and unique structures (SUS) plot was generated. The SUS plot was generated by plotting the p(corr) values of two models, namely, CM versus CTRL and NCM versus CTRL, against each other. Because the p(corr) values are normalized (ranging from +1 to −1), the chemical shifts in the SUS plots with a very high positive or negative value in the x coordinate and a very low positive or negative value in the y coordinate in the SUS plot or vice versa are the markers that change differentially in one group over the other. Univariate Data Analysis

The aromatic peaks were separately quantified and judged by the nonparametric Kruskal−Walis test (SigmaPlot 11.0). The significance of the level of metabolites was checked against each other. The level was considered to be significant if p < 0.05. Integration of NMR Peaks

In the 1H NMR spectra, the peaks representing the potential biomarkers that were obtained from OPLS-DA were integrated and normalized to the total spectral intensity using TOPSPIN 2.1. The significance was checked by nonparametric Kruskal− Walis test of SigmaPlot 11.0.

Figure 1. Representative survival graph of C57BL/6 female mice infected with P. berghei ANKA. The red dots represent the NCM, while the black dots represent the survival curve for the CM.

Analysis of Cholesterol

The total cholesterol concentration in the serum was estimated using the cholesterol oxidase (CHOD-PAP) method. The total triglyceride (TG) was determined by using the colorimetric (Griffing George Ltd., England) method as described earlier.28 The VLDL-C was calculated by the Friedewald formula, that is, total TGs/5.29

apparent that the survival was lower for mice with CM than those with NCM. By days 8−9 postinfection, the neurological symptoms were clear for CM, and the mice were sacrificed to collect various tissues and blood serum. In this study, metabolic profiles of sera, brain, and liver of the mice with CM were compared with the mice with NCM and with that of uninfected controls. The 1H NMR spectra of sera, brain, and liver extracts of the uninfected C57BL/6 mice are presented in Figure 2.

Two-Dimensional Spearman Correlation Matrices

The chemical shifts that significantly changed across groups as obtained from the OPLS-DA and SUS plots were further analyzed separately using Topspin 2.1. In brain extract, the discrete and nonoverlapping peaks of individual metabolites were integrated and normalized with respect to total spectral intensity. Thus, taurine (3.43, triplet), GABA (2.98, triplet), NAA (2.0, singlet), myo-inositol (3.60), glutamine (2.44), glutamate (2.35), BCCA (0.92−1.04), aspartate (2.80, 2.82 dd), choline (3.20 singlet), PC (3.21, singlet), glycerophosphocholine (GPC) (3.22, singlet), alanine (1.48, doublet), and lactate (1.32, doublet) resonances were integrated. Three different 2D Spearman correlation matrices, with a p value of each correlation, were calculated by an in-house script on Matlab7.0 platform. The differential correlations would throw light on the possible biochemical correlations in mice with CM, NCM, and controls, respectively. The correlation between the metabolites is represented by the colored boxes. The dark redcolored box represents a high positive correlation, while the dark blue-colored box represents the highest negative correlation.



Metabolic Markers in Serum for the CM Mice

The OPLS-DA scores of 1H NMR spectra of sera of three groups of animals, namely, CM, NCM, and uninfected controls, are segregated from each other and are clustered separately (Figure 1 in the Supporting Information). The OPLS-DA models were prepared with sera of all CM, NCM, and controls of three different experiments considered altogether. To investigate the metabolic markers of CM, a separate model is generated with only CM and NCM (Figure 3A). The Q2 value of the scores plot for the metabolic profiles of sera for the CM and NCM is 0.78. The metabolites that contribute to the distinct clustering of sera of the two classes are lipids/ lipoproteins, glutamine, and lactate (Figure 3B and Table 1). While lipids/lipoproteins are increased in concentration in the sera of CM, the metabolites that are decreased in mice with CM in comparison to NCM are glutamine and lactate. The corresponding p(corr) and VIP are provided in Table 1. To validate this observation, we separately plotted OPLS-DA scores and loadings of CM versus NCM of one experiment (Figure 1b,c in the Supporting Information) and validated it by a second experiment; that is, the scores of the observation of second experiment were predicted on the basis of previous one (Figure 1d in the Supporting Information). This analysis was performed by importing the data set of the second experiment (each 0.02 ppm bin of NMR spectra normalized to whole spectra for CM and NCM) into SIMCA and was assigned as the “prediction set” for the prepared OPLS-DA model of the first experiment. The predicted scores of the second experiment

RESULTS

Transition to CM in Mice

C57BL/6 mice infected with PbA is a well-known animal model of experimental CM.30,31 The fraction of animals transiting into CM is variable.31−33 In our experiments on a murine model of the disease, about 40−60% of the C57BL/6 mice infected with PbA develop CM.34 The mice that exhibited 4995

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article

Figure 2. Typical 700 MHz CPMG 1H NMR spectra of (A) serum, (B) perchloric extract of brain, and (C) perchloric acid extract of liver of control mice. Peaks: 1, lipids [1a, −CH3; 1b, −(CH2)n−; 1c, −CH2CH2CO; 1d, −CH2CC; 1e, −CH2CO; and 1f, C−(CH2)−C]; 2, leucine; 3, isoleucine; 4, valine; 5, lactate; 6, alanine; 7, acetate; 8, glutamine; 9, glutamate; 10, methionine; 11, pyruvate; 12, citrate; 13, lysine; 14, creatine; 15, choline; 16, phosphorylcholine; 17, GPC; 18, glucose; 19, GABA; 20, NAA; 21, succinate; 22, aspartate; 23, myo-inositol; 24, scylloinositol; 25, taurine; 26, glycine; 27, O-phosphoethanolamine; 28, glutathione; 29, ethanolamine; 30, betaine; and 31, unidentified.

(tPS vs toPS) were plotted and color coded according to their original class (red for CM and black for NCM). In this plot, we see that CM and NCM of the second experiment form their respective distinct cluster and in same manner as that in first

experiment (negative OPLS-DA scores for NCM and positive ones for the CM), which points to the fact that the OPLS-DA model built for the first experiment correctly predicts CM and NCM in the second experiment. This analysis was followed by 4996

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article

Figure 3. 1H NMR-based OPLS-DA scores and corresponding “V” plot of serum of C57BL/6 female mice infected with P. berghei ANKA. (A) Scores of NCM and CM, (B) V plot of panel A, (C) scores of CM and control, (D) V plot of panel C, (E) scores of NCM and control, and (F) V plot of panel E. Symbols: red, CM; black, NCM; and blue, control. The ellipse in the scores is a 95% Hotelling T2 ellipse. Each data point in the scores represents a serum sample, and those in the “V” plot represent a NMR bin identified by the chemical shift.

corresponding p(corr) and VIP of the metabolites are provided in Table 1. To further corroborate this fact, a SUS plot was created with the p(corr) values of the two models, namely, CM versus control and NCM versus control (Figure 1g in the Supporting Information), which clearly shows that lipids/ lipoproteins are specifically up-regulated, while glutamine is specifically down-regulated in CM. In addition, the sera levels of certain aromatic amino acids were distinct across the three classes. For example, there is an increase and increasing trend of histidine and phenylalanine, respectively, in the sera of mice with CM along with a decrease of tyrosine with respect to NCM (Figure 4). Among all of the metabolites perturbed between CM and NCM, serum lipids, glutamine, and lactate were in the aliphatic region of the spectra, hence with considerable spectral intensity across all of the samples. Therefore, they were further analyzed for their candidature as markers of CM. To achieve this, a receiver operating characteristic plot (ROC) of those metabolites was plotted (Figure 5). The area under the curve (AUC) values are 0.96, 0.87, 0.92, and 0.89 for lipids (1.27 and 0.87 ppm), glutamine (2.45 ppm), and lactate (4.11 ppm), respectively. From the ROC plot, it is deduced that the lipid/ lipoprotein moiety (CH2)n, resonating at 1.27 ppm, is the best candidate in the serum as a marker for CM.

another set of multivariate analysis in which a separate OPLSDA model was built for the second experiment, and the scores and the loadings of this model were subsequently plotted (Figure 1d in the Supporting Information). In this score plot, it is further evident that the CM and the NCM are clustered separately and similarly to experiment 1. Moreover, the chemical shift bins that contributed to the segregation of CM and NCM in the first experiment are identical to those of the second set (Figure 1e,f in the Supporting Information). The results further corroborate the consistency of the marker molecules for CM and NCM. Separate OPLS-DA models of CM versus control and NCM versus control were generated to elucidate the changes in the infected animals from common baseline (i.e., uninfected controls). The scores plot of the OPLS-DA models created for CM versus controls showed distinct segregation (Figure 3C), the Q2 of this model being 0.89. The V plot (Figure 3D) for the corresponding OPLS-DA model indicates that lipids/ lipoproteins are up-regulated in CM. The other metabolite, which is increased in concentration in mice afflicted with CM with respect to uninfected controls, is lactate. The metabolite, which is decreased in concentration in CM mice, is glucose. The p(corr) and VIP of the resonances of corresponding metabolites are provided in Table 1. The scores plot generated for NCM and control and its corresponding V plot are given in (Figure 3E,F). It is seen that the scores of the two groups in this model are clearly segregated. The Q2 of this model is 0.81. The metabolites that contribute toward discrimination between the NCM and the uninfected controls are lactate and glucose. While lactate is high in NCM, glucose levels are low. The

Lipid Profiles of Serum of Control, NCM, and CM

It is observed that total TGs, VLDL-cholesterol, and the total cholesterol concentrations are enhanced in the sera of the mice with CM as compared to that of mice with NCM (Table 2). TG and VLDL-cholesterol levels are significantly higher in CM 4997

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article

Table 1. OPLS-DA Correlation p(corr) and VIP of the Metabolites Toward Class Segregation (Classes Referred in the Panel of Models) model CM vs NCM

metabolites lipids CH3, (CH2)n, CH2CH2CO, CH2CC, CH2CO lactate glutamine glucose

CM vs control

lipids CH3, (CH2)n, CH2CH2CO, CH2CC, CH2CO lactate glucose

NCM vs control

lactate glucose

CM vs NCM

glutamine GPC succinate myo-inositol glutamine GPC PC, choline succinate PC, choline GABA BCAA O-phosphoethanolamine

CM vs control

NCM vs control

CM vs NCM

CM vs control

NCM vs control

taurine, GPC taurine, glucose histamine myo-inositol unidentified glucose BCAA glutamate, glutamine lactate myo-inositol glucose glutamate/glutamine taurine, GPC BCAA ethanolamine alanine

p(corr)

VIP

0.75, 0.84, 0.81, 0.76, 0.88, 0.83 −0.73 −0.75, −0.77 −0.50, −0.44, −0.49, −0.50, −0.47, −0.51 0.77, 0.83, 0.79, 0.75, 0.72, 0.73, 0.82, 0.77. 0.71, 0.64 −0.95, −0.92, −0.90, −0.94, −0.94, −0.94 0.94, 0.91 −0.95, −0.91, −0.90, −0.96, −0.96, −0.95

2.63, 4.64, 4.07, 1.56, 1.59,1.62 2.32 0.95, 1.37 1.10, 0.56, 1.14, 0.90, 0.91, 1.05 1.91, 2.45, 3.14, 2.65, 0.98, 0.97, 1.16, 1.20 3.00, 1.26 2.44, 2.29, 1.61, 2.13, 2.30, 2.46 2.49, 4.96 2.41, 1.31, 2.16, 2.15, 2.33, 2.48

−0.88 0.72 −0.84 0.75 0.93 −0.75 −0.75 0.85 0.79 0.75 0.81 −0.78

3.58 2.96 1.87 2.17 2.73 2.69 2.82 1.14 3.2 2.14 1.85 1.42

0.90, 0.78 −0.91 −0.89 −0.88 0.80, 0.67, 0.70, 0.74 −0.78 −0.77, −0.69 −0.69 −0.64 0.83, 0.66, 0.69, 0.66 −0.79, −0.76 −0.78 −0.70 −0.77 −0.71

3.32, 3.47 2.62 2.74 2.78, 1.68 1.49, 1.22 1.19 2.63, 1.69, 1.47 1.39 1.21 1.58

resonances (ppm) serum 0.87, 1.27, 1.29, 1.57, 2.23 4.11 2.45, 3.77 3.83, 3.85, 3.89, 3.49, 3.45 0.87, 1.25, 1.27, 1.29, 2.01, 2.03, 2.21 1.33, 4.11 3.83, 3.89, 3.85, 3.49, 3.45 4.11, 1.33 3.83, 3.85, 3.89, 3.49, 3.45 brain 2.44 3.24 2.40 3.68 2.44 3.24 3.20 2.4 3.2 2.28 0.92 4.0 liver 3.24, 3.4 3.32 4.04 4.0 3.72, 3.84, 3.64, 3.48 0.96 2.15, 2.04 4.12 4.04 3.72, 3.84, 3.64, 3.48 2.12, 2.04 3.24 0.96 3.12 1.48

2.01,

3.47, 1.57,

3.47,

3.47,

3.20

2.65, 2.22, 1.83 1.16

2.52, 2.33, 1.48 1.24

Figure 4. Box plot of the normalized intensity of amino acids in the serum of CM, NCM, and control female C57BL/6 mice. (A) Phenylalanine, (B) tyrosine, and (C) histidine. Significant difference (p < 0.05) between the levels in CM vs control (*) and CM vs NCM (#) by Kruskal−Walis test. The horizontal line in the boxes denotes the median.

sera in comparison to controls as well as NCM (p < 0.05). The TG concentration in CM is nearly 2-fold with respect to that of

controls, while TG is not significantly different in NCM with respect to controls. In addition to this, total VLDL-cholesterol 4998

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article

Figure 5. ROC plot (sensitivity vs 1 − specificity) of lipids, glutamine, and lactate for sera from C57BL/6 female mice with CM and NCM. (A) Lipids (1.27 ppm), (B) lipids (0.87 ppm), (C) glutamine, and (D) lactate.

Table 2. Quantitation of Total Cholesterol, TGs, and Free Fatty Acids in Sera of CM, NCM, and Control Female C57BL/6 Mice mg/dL

a

category

total cholesterol

TGs

VLDL-cholesterol

total free fatty acids

control NCM CM

117.0 ± 4.8 62.5 ± 5.5a 89.0 ± 8.1

111.0 ± 5.7 138.0 ± 8.1 198.0 ± 10.6a

31.0 ± 1.1 27.6 ± 1.6 39.6 ± 2.1a

296.0 ± 2.8 289.5 ± 3.1 290.5 ± 2.5

Significant difference (p < 0.05) between CM/NCM with respect to controls using Kruskal−Walis one-way test of ANOVA.

Figure 6. Box plot of the normalized intensity levels of the following metabolites in the brain of CM, NCM, and control C57BL/6 female mice. (A) Glutamine, (B) GPC, (C) succinate, and (D) phenylalanine. Significant difference (p < 0.05) in *CM vs control and #CM vs NCM.

is also increased in CM sera with respect to controls, while in NCM, the change is not significant. The total cholesterol in

NCM is significantly lower than that of controls, while the change in the cholesterol level is not significant in the case of 4999

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article

Figure 7. Box plot of the normalized intensity of the following metabolites in the liver of CM, NCM, and control C57BL/6 female mice. (A) myoInositol, (B) glucose, and (C) GPC. * and # denote a significant difference (p < 0.05) in CM vs control and CM vs NCM, respectively.

pathophysiological changes during CM, more insights into the process can be gained by investigating the relationship among them. To achieve this, we created the Spearman correlation matrix using the normalized integrated values of the NMR peak from the metabolites that are perturbed in CM and NCM with respect to control animals. We observed a significantly distinct correlation pattern of three different classes. The correlation maps created with different metabolites of brain in the three categories of animals, namely, CM, NCM, and control, clearly show distinct features (Figure 3a−c in the Supporting Information). It is clearly seen that glutamine has a very high Spearman correlation with taurine with p < 0.05, which is absent both in controls and in NCM mice. In addition, GPC has a negative correlation with PC and choline in the case of CM mice (p < 0.05), while the corresponding p value of the correlation is not significant either in NCM or in controls. The other significant correlation is a positive correlation of GABA with NAA (p = 0.057) and glutamate with NAA (p < 0.05) in CM brain, while insignificant in NCM and control.

CM. The total serum fatty acids in CM and NCM are not significantly different with respect to that of control. Metabolic Fingerprint of Brain

Typical 1H NMR spectra of the extracts of brain of control mouse are shown in Figure 2B. The resonances are assigned with the help of COSY and TOCSY and a database (www. hmdb.ca). Here, one batch of animal experiment was used for the analysis. The CM, NCM, and uninfected controls are clustered distinctly. This is as evident from the OPLS-DA scores plot of perchloric acid extract of the whole brain (Figure 2a in the Supporting Information). To investigate the metabolic response in the brain of mice with CM, a separate model was built between CM and NCM (Figure 2b in the Supporting Information). The Q2 of the model is 0.59. However, to know the actual trend of metabolites in CM and NCM, both of the infected groups were compared to that of uninfected controls (Figure 2c,d in the Supporting Information). The Q2 of the two OPLS-DA models (CM vs control) and (NCM vs control) are 0.92 and 0.84, respectively. The difference between the metabolic profiles of the brain in CM with respect to NCM is due to the significantly elevated level of glutamine and lower concentration of the GPC in brain of mice with CM (Figure 2e in the Supporting Information). The p(corr) and the VIP of the significant spectral peaks are listed in Table 1. The difference in metabolite profile of brain of CM and the uninfected control are due to lower levels of choline, GPC, phosphocholine (PC), and higher levels of glutamine in mice with CM (Figure 2f in the Supporting Information). The distinction in the brain metabolites of NCM and uninfected controls are due to the lower levels of PC, choline, and γ-amino butyric acid (GABA) in NCM (Figure 2g in the Supporting Information). These results were further confirmed using a SUS plot (Figure 2h in the Supporting Information). It is evident from the SUS plot that myo-inositol is specifically decreased in CM brain [p(corr) of −0.6 in CM vs ctrl and p(corr) of 0.09 in NCM vs ctrl] and succinate is specifically increased in CM [p(corr) of 0.85 in CM vs ctrl and p(corr) of −0.13 in NCM vs ctrl]. Furthermore, the O-phosphoethanolamine level is enhanced, and branched chain amino acids (BCAA) are decreased in NCM. Moreover, the relative levels of glutamine, GPC, succinate, and phenylalanine (Figure 6) and O-phosphorylethanolamine, BCAA, myoinositol, and GABA in (Figure 2i in the Supporting Information) for CM, NCM, and uninfected controls further corroborate that these metabolites have altered levels in CM with respect to NCM and controls. In the earlier section, we discussed several metabolites that were found to be specifically perturbed in the brain of the CM mice with respect to NCM and control mice. Although the change in their levels are important in terms of understanding

Metabolic Fingerprint of Liver

The 1H NMR spectrum of liver from an uninfected mouse is shown in Figure 2C. Here, one batch of animal experiment was used for the analysis. The OPLS-DA scores of 1H NMR spectra of liver extract for CM, NCM, and control show distinct clustering of the three groups of mice (Figure 4a in the Supporting Information). The Q2 for the OPLS-DA model between the liver extract of CM and NCM is 0.72 (Figure 4b in the Supporting Information). A separate model of the infected mice (CM/NCM) with that of the uninfected group is also prepared (Figure 4c,d in the Supporting Information). The Q2 of OPLS-DA model prepared with the liver of CM and control is 0.85, whereas that of NCM and control is 0.56. The segregation between CM and NCM is due to taurine, GPC, myo-inositol, and histamine. While taurine and GPC are downregulated, the levels of histamine and myo-inositol are increased in the liver of mice with CM with respect to that of NCM (Figure 4e in the Supporting Information). Furthermore, from the analysis of the OPLS-DA model created for CM and controls, it is seen that the level of glucose is lower in CM in comparison to control liver (Figure 4f in the Supporting Information). In addition, metabolites such as histamine, myoinositol, glutamine, glutamate, and lactate are up-regulated in the liver of mice with CM (Figure 4f in the Supporting Information) with respect to the controls. The metabolites that are responsible for the segregation of liver of mice with NCM from control are glucose, glutamine, glutamate, taurine, alanine, and ethanolamine (Figure 4g in the Supporting Information). While the glucose level is decreased, glutamine, glutamate, taurine, and ethanolamine are elevated in the liver of mice with 5000

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article

NCM (Figure 4g in the Supporting Information). However, lactate is poorly correlated toward the segregation of NCM and control. From these results, it can be said that down-regulation of glucose and up-regulation of glutamine and glutamate in the liver are due to malaria in general, regardless of CM or NCM. The SUS plot (Figure 4h in the Supporting Information) shows that histamine and myo-inositol are up-regulated in the liver of CM in contrast to NCM and that lactate is up-regulated in CM liver in contrast to NCM liver when compared to that of uninfected controls. In addition, ethanolamine and taurine are up-regulated in NCM liver. From the plot of relative levels of metabolites in liver (Figure 7 and Figure 4i in the Supporting Information), it is clearly seen that there is an increase of myoinositol and a decrease of GPC and glucose in the liver of CM. Moreover, lactate is enhanced in the liver of CM, while ethanolamine and taurine are elevated in NCM liver.

Glutamine accumulation is known to adversely affect the brain function by astrocyte swelling.52 As seen in the correlation matrix, glutamine has a high positive correlation with taurine (Figure 3A), which is not present either in NCM or in controls. Hyperammonemia induces the transfer of taurine in brain capillary.53 Thus, an indication of hyperammonia is evident in the brain of CM. GPC also contributed toward segregation between CM and NCM. GPC is a breakdown product of phosphatidylcholine, which is a major constituent of cell membrane and is internalized by P. falciparum.54 An uptake of phosphocholine lipids by the parasite would modify the turnover of these lipids, which would result in decreased formation of GPC in the brain of CM mice. Furthermore, there is a negative correlation between GPC between both PC and choline (Figure 3 in the Supporting Information), which indicates a perturbed metabolism of phosphatidylcholine pathway. In addition to this, we see that that there are certain features that are specific to CM. For example, the GPC of brain showed a significant positive correlation with the level of Nacetylaspartate (NAA) in the brain. Such a feature was not observed in the NCM or control mice. This suggests that in addition to perturbation in the phosphatidylcholine pathway, CM mice have an associated complication in terms of operation of the NAA pathway. Therefore, it is likely that the functions of NAA in the brain are affected. NAA is known to act as an important neuronal osmolyte. It is known to remove excess water by acting as a molecular water pump.55 Perturbation in neuronal osmosis and brain edema were implicated in experimental CM earlier.56 Our data suggest that this might have an important connection to the alteration in the phospholipid metabolism during the cerebral stage. Moreover, NAA is known to be important in the synthesis of neurotransmitters like N-acetylaspartylglutamate (NAAG). Our observation that glutamate and NAA in the brain bear a significant positive correlation in the brain of CM mice suggests that the pathways dependent on and/or participated in by NAAG may be perturbed during the cerebral stage of the disease. An increase in succinate in the brain of CM is an indication of reduced activity of succinate dehydrogenase under hypothermia.57 This, indeed, is the case with CM mice as stated in the Experimental Procedures. Although not statistically significant, we observe a decreasing trend of myo-inositol in CM brain (Figure 2i in the Supporting Information). myo-Inositol is converted to glucuronic acid in the liver, which is used as a substrate for detoxification.58 This is corroborated by the fact that in the case of CM mice, the liver myo-inositol is enhanced with both respect to NCM and to control mice. Another perspective of this finding is a simultaneous increase of glutamine and reduction in myoinositol in brain, which might be associated with liver encephalopathy.59 Hepatic encephalopathy is a result of failure of the liver like chirrosis. In acute cases, liver encephalopathy leads to coma, brain edema, and death. Furthermore, a decrease in the concentration of myo-inositol is associated with a compensatory role of increased osmolarity created by increased glutamine concentration. This is turn might lead to astrocyte swelling.59 Moreover, it is also evident from our study that Ophosphorylethanolamine, a precursor of phospholipids metabolism, is specifically up-regulated in NCM, which might play an important role in the pathogenesis in NCM.



DISCUSSION In this study, we report for the first time metabolic fingerprints of serum, brain, and liver of CM, NCM, and control mice of the same genetic background. Lactate and glucose in serum and liver contribute significantly toward the segregation between NCM/CM and control groups analyzed as individual models. Lactate is relatively higher in the infected groups and glucose in the control. This result is consistent with the previous studies.35,36 A decrease in glucose is the result of enhanced glycolysis in both CM and NCM, which, in turn, would lead to hypoglycemic conditions reported in malaria patients.37 The low glucose concentration in the serum of CM and NCM could be attributed to the enhanced consumption of glucose by pRBCs.37 The malarial parasite lacks a TCA cycle38 and so meet their energy requirements primarily through anaerobic glycolysis. This would lead to the accumulation of lactate. Lactic acidosis is a known phenomenon and is a predicator of fatality in patients with malaria. These changes can be attributed to the alterations brought about by malaria. When the serum profiles of NCM are compared with those of CM, the markers for the latter are a high concentration of lipids/lipoproteins and a high concentration of TGs and VLDLcholesterol and a low concentration of glutamine and lactate. Although malarial parasite infection is known to alter the lipid parameters,39−42 the impact of CM on serum lipid profile is poorly understood. Our study clearly demonstrates that the lipid concentration increases in serum during CM in contrast to NCM as well as controls. High levels of TGs in host serum could result from a parasite-induced lipolysis for fulfilling its requirements of building blocks for lipids. The excess of the TGs might then be taken up by the liver to synthesize VLDL.43 A low glutamine concentration in CM is suggested to be associated with sepsis during malaria.44 Furthermore, the sera in mice with CM exhibit a lower level of tyrosine and a high concentration of histidine and phenylalanine. Phenylalanine is known to increase in CM and in severe malaria and is likely to be associated with the disruption of the activity of phenylalanine hydroxylase enzyme.45 Histidine up-regulation has been attributed to hepatic failure46 and/or kidney pathology.47 An increased level of glutamine in the brain of CM mice could mean an enhanced synthesis and/or lower clearance rate of the amino acid. Brain is essentially devoid of urea cycle,48,49 and the major route for the elimination of ammonia in brain is through the formation of glutamine. The amino acid has been reported as a marker of subarachnoid hemorrhage and has been correlated with vasopasm and traumatic brain injury.50,51 5001

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article

In addition to this, our observations indicate an up-regulation of phenylalanine in the brain of mice with CM. High levels of serum phenylalanine could lead to elevated levels of phenylalanine in the brain, which, in turn, has been reported to be neurotoxic.60 It has also been suggested that an altered concentration of tetrahydrobiopterin in brain could lead to hyperphenylalanine.45 From our study, it seems that high levels of glutamine, succinate, and phenylalanine concomitant with low levels of GPC are the perturbants in the brain of mice with CM. Liver from the CM, NCM, and control animals exhibits different metabolite profiles. In comparison to NCM, the liver from mice with CM is characterized by relatively lower levels of taurine, GPC, and glucose and an up-regulation of histamine and myo-inositol. Taurine is known to play a positive role in ammonia detoxification.61 It is possible that a relative increase of taurine in the liver of NCM gives the ability to protect mice from cerebral outcome. In this context, it is important to note that mice with CM had enhanced ammonia toxicity in the brain as compared to NCM, denoted by increased glutamine levels. Furthermore, an increase in myo-inositol in CM could be sourced from glucose.62 myo-Inositol is converted to glucuronic acid, which is used as a detoxification substrate. The increase of histamine in CM liver might be significant since it increases vascular permeability and acute inflammation,63 and its increase in liver might play an important role in disease progression in CM. Histamine is normally inactivated by liver. However, in mice with CM, an elevated level of histamine indicates a disruption of this important liver function.64 The increase in the levels of ethanolamine in liver and Ophosphoethanolamine in the brain of NCM mice is again an indication of lesion/s in specific pathway/s and need further investigations. GPC is seen to be perturbed in brain and liver; hence, its pathway specifically seems to be compromised in mice with CM. In addition, it is also important to note that glucose is the prime source of myo-inositol and is downregulated serum and liver, whereas there is an increase in myoinositol in liver and a decrease of the same in the brain of animals with CM. Therefore, a study of glucose and myoinositol metabolism in CM would shed light on the role of the relevant pathways in disease progression to cerebral complications. The ammonia detoxification pathway seems to be specifically perturbed in CM. Enhanced levels of glutamine and down-regulation of myo-inositol along with a positive correlation of taurine with glutamine in the brain of CM clearly points toward ammonia toxicity. Further lipids/lipoproteins and TGs metabolism are also perturbed specifically in the case of CM. The metabolic perturbations in CM over NCM are listed in Table 3 and schematically depicted in Figure 8.

Table 3. Altered Metabolite Profile in C57BL/6 Female Mice with CM (as Compared to NCM) CM tissue/body fluid

high

sera

lipids/lipoproteins (↑) TGs (↑) VLDL-cholesterol (↑) phenylalanine (↑) histidine (↑) glutamine (↑) phenylalanine (↑) succinate (↑) histamine (↑) myo-inositol (↑) lactate (↑a)

brain

liver

low lactate (↓) glutamine (↓) tyrosine (↓)

myo-inositol (↓a) GPC (↓) glucose (↓a) GPC (↓a)

a Decreasing/increasing trend wrt NCM but not significant by Kruskal−Walis test.

Figure 8. Overall metabolic changes in serum, brain, and liver of C57BL/6 female mice infected with P. berghei ANKA. The metabolites in red font are perturbed in mice with CM as compared to NCM.



CONCLUSION We have elucidated the metabolic changes that appear to be specifically correlated to CM in serum, brain, and liver. The fingerprints of the parasite-infected mice are distinctly different from that of control mice. The specific clustering of CM and NCM mice shows distinct metabolic features. Because the mice were inbred and the same litters were used for the study under identical feed and environment, the metabolic changes are unlikely to be due to genetic background and are likely to be due to specific infection pathology. A combinatorial metabolic profile with an increase or decrease in the levels of certain metabolites is observed in the animals of the three groups. Overall, the concomitant increase of glutamine, succinate, and

decrease of GPC in the brain along with an increase of lipids (TGs and VLDL-cholesterol) and a decrease of glutamine in serum appear to be some of the indicators of murine CM. Other characteristics of CM are an increase of myo-inositol and histamine and a decrease of GPC and glucose in the liver. This study also indicates toward the possible perturbation of ammonia detoxification pathway in the brain of CM animals. Moreover, because lipids/lipoproteins and TGs are specifically up-regulated in CM sera, a compromise of its metabolism could be present in CM. The study thus suggests further investigations into lipid metabolism (specifically TGs), 5002

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article

Pukritayakamee, S.; Nagachinta, B.; et al. An immunohistochemical study of the pathology of fatal malaria. Evidence for widespread endothelial activation and a potential role for intercellular adhesion molecule-1 in cerebral sequestration. Am. J. Pathol. 1994, 145, 1057− 1069. (11) Medana, I. M.; Hunt, N. H.; Chan-Ling, T. Early activation of microglia in the pathogenesis of fatal murine cerebral malaria. Glia 1997, 19, 91−103. (12) Penet, M. F.; Kober, F.; Confort-Gouny, S.; Le Fur, Y.; Dalmasso, C.; Coltel, N.; Liprandi, A.; Gulian, J. M.; Grau, G. E.; Cozzone, P. J.; Viola, A. Magnetic resonance spectroscopy reveals an impaired brain metabolic profile in mice resistant to cerebral malaria infected with Plasmodium berghei ANKA. J. Biol. Chem. 2007, 282, 14505−14514. (13) Sanni, L. A.; Rae, C.; Maitland, A.; Stocker, R.; Hunt, N. H. Is ischemia involved in the pathogenesis of murine cerebral malaria? Am. J. Pathol. 2001, 159, 1105−1112. (14) Cloarec, O.; Dumas, M. E.; Craig, A.; Barton, R. H.; Trygg, J.; Hudson, J.; Blancher, C.; Gauguier, D.; Lindon, J. C.; Holmes, E.; Nicholson, J. Statistical total correlation spectroscopy: an exploratory approach for latent biomarker identification from metabolic 1H NMR data sets. Anal. Chem. 2005, 77, 1282−1289. (15) Lederberg, J. Infectious history. Science 2000, 288, 287−293. (16) Nicholson, J. K.; Lindon, J. C.; Holmes, E. 'Metabonomics': Understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica 1999, 29, 1181−1189. (17) Zhang, G. F.; Sadhukhan, S.; Tochtrop, G. P.; Brunengraber, H. Metabolomics, pathway regulation, and pathway discovery. J. Biol. Chem. 2011, 286, 23631−23635. (18) Saric, J.; Li, J. V.; Wang, Y.; Keiser, J.; Bundy, J. G.; Holmes, E.; Utzinger, J. Metabolic profiling of an Echinostoma caproni infection in the mouse for biomarker discovery. PLoS Neglected Trop. Dis. 2008, 2, e254. (19) Nicholson, J. K.; Holmes, E.; Wilson, I. D. Gut microorganisms, mammalian metabolism and personalized health care. Nat. Rev. Microbiol. 2005, 3, 431−438. (20) Rajeshwari, K.; Patel, K.; Nambeesan, S.; Mehta, M.; Sehgal, A.; Chakraborty, T.; Sharma, S. The P domain of the P0 protein of Plasmodium falciparum protects against challenge with malaria parasites. Infect. Immun. 2004, 72, 5515−5521. (21) Amani, V.; Boubou, M. I.; Pied, S.; Marussig, M.; Walliker, D.; Mazier, D.; Renia, L. Cloned lines of Plasmodium berghei ANKA differ in their abilities to induce experimental cerebral malaria. Infect. Immun. 1998, 66, 4093−4099. (22) Franke-Fayard, B.; Janse, C. J.; Cunha-Rodrigues, M.; Ramesar, J.; Buscher, P.; Que, I.; Lowik, C.; Voshol, P. J.; den Boer, M. A.; van Duinen, S. G.; Febbraio, M.; Mota, M. M.; Waters, A. P. Murine malaria parasite sequestration: CD36 is the major receptor, but cerebral pathology is unlinked to sequestration. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 11468−11473. (23) www.hmdb.ca. (24) Nicholson, J. K.; Foxall, P. J.; Spraul, M.; Farrant, R. D.; Lindon, J. C. 750 MHz 1H and 1H-13C NMR spectroscopy of human blood plasma. Anal. Chem. 1995, 67, 793−811. (25) Wold, S.; Esbensen, K.; Geladi, P. Chemom. Intell. 2 1987, 2 (1− 3), 37−52. (26) Trygg, J.; Wold, S. J. Chemom. 2002, 16 (3), 119−128. (27) Umetrics AB. Multi- and Megavariate Data Analysis. Part 1: Basic Principles and Applications, 2nd ed.; Umetrics AB: Sweden, 2006; p 390. (28) Faucher, J. F.; Ngou-Milama, E.; Missinou, M. A.; Ngomo, R.; Kombila, M.; Kremsner, P. G. The impact of malaria on common lipid parameters. Parasitol. Res. 2002, 88, 1040−1043. (29) Friedewald, W. T.; Levy, R. I.; Fredrickson, D. S. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin. Chem. 1972, 18, 499−502.

glucose−myo-inositol pathway and also the ammonia detoxification pathway in CM over NCM.



ASSOCIATED CONTENT

S Supporting Information *

Supplemental figures 1−4. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Tel: +91-22-22782625. Fax: +91-22-22804610. E-mail: [email protected] (S.S.). Tel: +91-22-22782394. Fax: +91-2222804860. E-mail: [email protected] (H.M.S.). Author Contributions

S.G. performed the animal and NMR experiments, statistical analyses and interpretation of the data, and drafted the manuscript. A.S. participated in animal experiments and in drafting the manuscript. H.M.S. and S.S. conceived the study and were involved in interpreting the data and drafting of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank the staff of the National Facility for High Field NMR at TIFR for their help and support. The help of Dr. Sheetal and Dr. Sachin and other personnel of the Experimental Animals Facility at TIFR is gratefully acknowledged. A.S. thanks the CSIR, Govt of India, for the SPM fellowship. We acknowledge Debanjan Bhowmik for standardizing the extraction procedure for liver experiments.



ABBREVIATIONS CM, cerebral malaria; NCM, noncerebral malaria; NMR, nuclear magnetic resonance; OPLS-DA, orthogonal partial least square discriminant analysis; GPC, glycerophosphocholine; CPMG, Carr−Prucell−Meiboom−Gill



REFERENCES

(1) http://www.rbm.who.int/keyfacts.html. (2) http://www.malariasite.com/malaria/Complications2.htm. (3) Schofield, L.; Grau, G. E. Immunological processes in malaria pathogenesis. Nat. Rev. Immunol. 2005, 5, 722−735. (4) Bagot, S.; Boubou, M. I.; Campino, S.; Behrschmidt, C.; Gorgette, O.; Guenet, J. L.; Penha-Goncalves, C.; Mazier, D.; Pied, S.; Cazenave, P. A. Susceptibility to experimental cerebral malaria induced by Plasmodium berghei ANKA in inbred mouse strains recently derived from wild stock. Infect. Immun. 2002, 70, 2049−2056. (5) Turner, G. Cerebral malaria. Brain Pathol. 1997, 7, 569−582. (6) Berendt, A. R.; Ferguson, D. J.; Gardner, J.; Turner, G.; Rowe, A.; McCormick, C.; Roberts, D.; Craig, A.; Pinches, R.; Elford, B. C.; et al. Molecular mechanisms of sequestration in malaria. Parasitology 1994, 108 (Suppl.), S19−S28. (7) Grau, G. E.; de Kossodo, S. Cerebral malaria: mediators, mechanical obstruction or more? Parasitol. Today 1994, 10, 408−409. (8) Collette, A.; Bagot, S.; Ferrandiz, M. E.; Cazenave, P. A.; Six, A.; Pied, S. A profound alteration of blood TCRB repertoire allows prediction of cerebral malaria. J. Immunol. 2004, 173, 4568−4575. (9) Hunt, N. H.; Grau, G. E. Cytokines: Accelerators and brakes in the pathogenesis of cerebral malaria. Trends Immunol. 2003, 24 (9), 491−499. (10) Turner, G. D.; Morrison, H.; Jones, M.; Davis, T. M.; Looareesuwan, S.; Buley, I. D.; Gatter, K. C.; Newbold, C. I.; 5003

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004

Journal of Proteome Research

Article

(30) Mackey, L. J.; Hochmann, A.; June, C. H.; Contreras, C. E.; Lambert, P. H. Immunopathological aspects of Plasmodium berghei infection in five strains of mice. II. Immunopathology of cerebral and other tissue lesions during the infection. Clin. Exp. Immunol. 1980, 42, 412−420. (31) Rest, J. R. Cerebral malaria in inbred mice. I. A new model and its pathology. Trans. R. Soc. Trop. Med. Hyg. 1982, 76, 410−415. (32) Neill, A. L.; Hunt, N. H. Pathology of Fatal and Resolving Plasmodium-Berghei Cerebral Malaria in Mice. Parasitology 1992, 105, 165−175. (33) Grau, G. E.; Piguet, P. F.; Engers, H. D.; Louis, J. A.; Vassalli, P.; Lambert, P. H. L3T4+ T lymphocytes play a major role in the pathogenesis of murine cerebral malaria. J. Immunol. 1986, 137, 2348− 2354. (34) Ghosh, S.; Sengupta, A.; Sharma, S.; Sonawat, H. M. Multivariate modelling with 1H NMR of pleural effusion in murine cerebral malaria. Malar. J. 2011, 10, 330. (35) Krishna, S.; Waller, D. W.; ter Kuile, F.; Kwiatkowski, D.; Crawley, J.; Craddock, C. F.; Nosten, F.; Chapman, D.; Brewster, D.; Holloway, P. A.; et al. Lactic acidosis and hypoglycaemia in children with severe malaria: pathophysiological and prognostic significance. Trans. R. Soc. Trop. Med. Hyg. 1994, 88, 67−73. (36) Li, J. V.; Wang, Y.; Saric, J.; Nicholson, J. K.; Dirnhofer, S.; Singer, B. H.; Tanner, M.; Wittlin, S.; Holmes, E.; Utzinger, J. Global metabolic responses of NMRI mice to an experimental Plasmodium berghei infection. J. Proteome Res. 2008, 7, 3948−3956. (37) Agbenyega, T.; Angus, B. J.; Bedu-Addo, G.; Baffoe-Bonnie, B.; Guyton, T.; Stacpoole, P. W.; Krishna, S. Glucose and lactate kinetics in children with severe malaria. J. Clin. Endocrinol. Metab. 2000, 85, 1569−1576. (38) Lang-Unnasch, N.; Murphy, A. D. Metabolic changes of the malaria parasite during the transition from the human to the mosquito host. Annu. Rev. Microbiol. 1998, 52, 561−590. (39) Adekunle, A. S.; A., O. C.; Egbewale, B. E. Serum status of selected biochemical parameters in malaria: An animal model. Biomed. Res. 2007, 18, 109−113. (40) Mohanty, S.; Mishra, S. K.; Das, B. S.; Satpathy, S. K.; Mohanty, D.; Patnaik, J. K.; Bose, T. K. Altered plasma lipid pattern in falciparum malaria. Ann. Trop. Med. Parasitol. 1992, 86, 601−606. (41) Krishna, A. P.; Chandrika; Kumar, S.; Acharya, M.; Patil, S. L. Variation in common lipid parameters in Malraiainfected patients. Indian J. Physiol. Pharmacol. 2009, 53, 271−274. (42) Nilsson-Ehle, I.; Nilsson-Ehle, P. Changes in plasma lipoproteins in acute malaria. J. Intern. Med. 1990, 227, 151−155. (43) Maurois, P.; Charet, P.; Fournet, B.; Fruchart, J. C. Metabolism of lipoproteins in rodent malaria, relationship between lipolysis, steatosis and increased biosynthesis of V.L.D.L. Ann. Parasitol. Hum. Comp. 1981, 56, 9−19. (44) Cowan, G.; Planche, T.; Agbenyega, T.; Bedu-Addo, G.; OwusuOfori, A.; Adebe-Appiah, J.; Agranoff, D.; Woodrow, C.; Castell, L.; Elford, B.; Krishna, S. Plasma glutamine levels and falciparum malaria. Trans. R. Soc. Trop. Med. Hyg. 1999, 93, 616−618. (45) Lopansri, B. K.; Anstey, N. M.; Stoddard, G. J.; Mwaikambo, E. D.; Boutlis, C. S.; Tjitra, E.; Maniboey, H.; Hobbs, M. R.; Levesque, M. C.; Weinberg, J. B.; Granger, D. L. Elevated plasma phenylalanine in severe malaria and implications for pathophysiology of neurological complications. Infect. Immun. 2006, 74, 3355−3359. (46) Sowunmi, A. Hepatomegaly in acute falciparum malaria in children. Trans. R. Soc. Trop. Med. Hyg. 1996, 90, 540−542. (47) Beisel, W. R. Herman Award Lecture, 1995: Infection-induced malnutrition–from cholera to cytokines. Am. J. Clin. Nutr. 1995, 62 (4), 813−819. (48) Shawcross, D.; Jalan, R. The pathophysiologic basis of hepatic encephalopathy: Central role for ammonia and inflammation. Cell. Mol. Life Sci. 2005, 62 (19−20), 2295−2304. (49) Cooper, A. J.; Plum, F. Biochemistry and physiology of brain ammonia. Physiol. Rev. 1987, 67 (2), 440−519. (50) Dunne, V. G.; Bhattachayya, S.; Besser, M.; Rae, C.; Griffin, J. L. Metabolites from cerebrospinal fluid in aneurysmal subarachnoid

haemorrhage correlate with vasospasm and clinical outcome: A pattern-recognition 1H NMR study. NMR Biomed. 2005, 18, 24−33. (51) Ashwal, S.; Holshouser, B.; Tong, K.; Serna, T.; Osterdock, R.; Gross, M.; Kido, D. Proton MR spectroscopy detected glutamate/ glutamine is increased in children with traumatic brain injury. J. Neurotrauma 2004, 21, 1539−1552. (52) Rama Rao, K. V.; Reddy, P. V.; Tong, X.; Norenberg, M. D. Brain edema in acute liver failure: inhibition by L-histidine. Am. J. Pathol. 2010, 176, 1400−1408. (53) Belanger, M.; Asashima, T.; Ohtsuki, S.; Yamaguchi, H.; Ito, S.; Terasaki, T. Hyperammonemia induces transport of taurine and creatine and suppresses claudin-12 gene expression in brain capillary endothelial cells in vitro. Neurochem. Int. 2007, 50, 95−101. (54) Simoes, A. P.; Moll, G. N.; Slotboom, A. J.; Roelofsen, B.; Op den Kamp, J. A. Selective internalization of choline-phospholipids in Plasmodium falciparum parasitized human erythrocytes. Biochim. Biophys. Acta 1991, 1063, 45−50. (55) Baslow, M. H. Evidence supporting a role for N-acetyl-Laspartate as a molecular water pump in myelinated neurons in the central nervous system. An analytical review. Neurochem. Int. 2002, 40 (4), 295−300. (56) Penet, M. F.; Viola, A.; Confort-Gouny, S.; Le Fur, Y.; Duhamel, G.; Kober, F.; Ibarrola, D.; Izquierdo, M.; Coltel, N.; Gharib, B.; Grau, G. E.; Cozzone, P. J. Imaging experimental cerebral malaria in vivo: Significant role of ischemic brain edema. J. Neurosci. 2005, 25, 7352− 7358. (57) Eliseev, V. A. Activity of succinate dehydrogenase of the rat brain and liver in hypothermia and after rewarming. Biull. Eksp. Biol. Med. 1969, 68, 43−44. (58) Ross, B. D. Biochemical considerations in 1H spectroscopy. Glutamate and glutamine; myo-inositol and related metabolites. NMR Biomed. 1991, 4, 59−63. (59) Cordoba, J.; Gottstein, J.; Blei, A. T. Glutamine, myo-inositol, and organic brain osmolytes after portocaval anastomosis in the rat: Implications for ammonia-induced brain edema. Hepatology 1996, 24, 919−923. (60) Kaufman, S. An evaluation of the possible neurotoxicity of metabolites of phenylalanine. J. Pediatr. 1989, 114, 895−900. (61) Delic, D.; Warskulat, U.; Borsch, E.; Al-Qahtani, S.; Al-Quraishi, S.; Haussinger, D.; Wunderlich, F. Loss of ability to self-heal malaria upon taurine transporter deletion. Infect. Immun. 2010, 78, 1642− 1649. (62) Majumder, A. L.; Chatterjee, A.; Ghosh Dastidar, K.; Majee, M. Diversification and evolution of L-myo-inositol 1-phosphate synthase. FEBS Lett. 2003, 553, 3−10. (63) Maegraith, B. G.; Onabanjo, A. O. The effects of histamine in malaria. Br. J. Pharmacol. 1970, 39, 755−764. (64) Eiseman, B.; Yeoh, K. S. Portal hypertension associated with idiopathic retroperitoneal fibrosis. Br. J. Surg. 1962, 50, 225−227.

5004

dx.doi.org/10.1021/pr300562m | J. Proteome Res. 2012, 11, 4992−5004