| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Discipline of Biochemistry, School of Molecular and Microbial Biosciences, The University of Sydney, Sydney NSW 2006, Australia;
* Department of Pathology, The University of Sydney, Sydney NSW 2006, Australia; and
Department of Anatomy and Histology, The University of Sydney, Sydney NSW 2006, Australia
1Correspondence: School of Molecular and Microbial Biosciences, The University of Sydney, Sydney, NSW, 2006, Australia. E-mail c.rae{at}mmb.usyd.edu.au
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, interferon-
, and lymphotoxin. These changes are consistent with a state of cytopathic hypoxia. By contrast, in P. berghei K173 infection the brain showed increased metabolic rate, with no deleterious effect on bioenergetics. This was accompanied by mild up-regulation of expression of metabolic enzymes. These changes are consistent with benign hypermetabolism whose cause remains a subject of speculation.—Rae, C., McQuillan, J. A., Parekh, S. B., Bubb, W. A., Weiser, S., Balcar, V. J., Hansen, A., Ball, H., Hunt, N. H. Brain gene expression, metabolism, and bioenergetics: interrelationships in murine models of cerebral and noncerebral malaria.
Key Words: Plasmodium berghei • cytokines • hypoxia • glycolysis • brain metabolism
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
; 2) the "cytokine" hypothesis, which proposes that effector molecules produced by the immune system during its response against the malaria parasite adversely affect CNS function (2)
. It has been suggested that both mechanisms are involved, with the presence of intravascular pRBCs in the brain serving to focus the production of immune system effector molecules, with adverse consequences for CNS function (3)
.
A variety of different technical approaches to the study of CM has been taken. These include clinical studies of CM patients; postmortem studies of brain and retina tissues from CM victims; in vitro model systems of pRBC adhesion to various substrates or cells; animal models, particularly in the mouse; genetic studies of susceptibility/resistance loci in humans or mice. The most commonly used mouse model of CM is Plasmodium berghei ANKA in susceptible mouse strains, including CBA and C57BL/6 (4
5
6)
. Although not identical to human CM in every respect, the histopathological and immunopathological similarities are striking (7
, 8)
. A number of key concepts about the pathogenesis of human CM have arisen from studies in the mouse model (7)
.
Studies of biochemical changes in human and murine CM have revealed common features. For example, the kynurenine pathway of tryptophan metabolism is active in both murine (9)
and human (10
, 11)
CM. It was demonstrated that cerebrospinal fluid (CSF) levels of lactate are elevated in CM in Thai adults (12
, 13)
and Malawian children (14)
, which has been interpreted as providing support for the sequestration hypothesis (1)
. Brain lactate levels are increased in murine CM concurrently with clinical illness, but not in malaria infection without CNS involvement (15)
.
The mechanisms responsible for the increases in CNS lactate in human and mouse CM have not been elucidated. Hypoxia is a potential cause, but not the only possibility. The demonstration that brain lactate changes occur in murine CM allows investigation of the possible mechanisms involved. In the current study we have examined brain biochemistry in cerebral and noncerebral malaria infections in mice. We have examined the hypoxia and cytokine theories by correlating metabolic outcomes with changes in gene expression patterns.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Murine model
All experiments adhered to Australian NHMRC guidelines for animal research and were approved by the University of Sydney Animal Ethics Committee. Female CBA/T6 mice (6–8 weeks old) weighing 20 to 25 g were obtained from Blackburn Animal House, University of Sydney. The malaria parasites used were Plasmodium berghei, either P. berghei ANKA (PbA, from Dr. G. Grau, Université de la Méditerranée, Marseilles, France) or P. berghei K173 (PbK, from Dr. Ian Clark, Australian National University, Canberra, Australia).
Mice were inoculated by intraperitoneal injection of 106 pRBCs that had been obtained from the blood of infected animals and resuspended in 100 µL of phosphate-buffered saline (PBS), pH 7.4. Controls were uninfected CBA mice injected with 100 µL of PBS. Mice injected with PbA developed cerebral complications and became terminally ill between days 6 and 7 postinoculation (p.i.) (6)
. Mice inoculated with PbK became moribund between days 14 and 22 p.i. but without cerebral involvement at any stage (5)
. Since the rate of progression of parasite growth in the two infections differs slightly, two sets of PbK-inoculated mice were used as malaria-infected controls for PbA mice. These were 1) mice whose length of infection was matched to those of PbA mice (PbK(Dm)) and 2) mice whose parasitemia levels were matched to those of PbA mice (PbK(Pm)).
Mean parasitemia for PbK(Dm) mice [day 6; 4.0 (1.5) % (SD)] was significantly (P=0.0001) less than that for PbK(Pm) (day 9; 13.4 (3.8) %) mice and significantly (P=0.0005) less than that for PbA mice (9.8 (3.8) %). PbA and PbK(Pm) parasitemias were not significantly different (P>0.05, Mann-Whitney U test).
Cerebral metabolism of D-[1-13C]glucose
Cerebral metabolism is highly sensitive to changes in brain pharmacology and activity, producing changes that are often unique to their cause (16
17
18
19
20)
. Therefore, we investigated the cerebral metabolism of D-[1-13C]glucose and studied the resultant isotopomer distribution in order to gain insight into the mechanism by which metabolism is disturbed in malaria.
On days 6–7 in the case of PbA mice and on days 6–7 (PbK(Dm)) or day 14 (PbK(Pm)) in the case of PbK mice, mice (N=5 for each group, including age-matched controls) were injected in the tail vein with 50 µL D-[1-13C]glucose (40% w/v, 1 g/kg body weight). After 20 min, mice were killed by cervical dislocation and the head was snap frozen in liquid nitrogen. Brains were removed while frozen, extracted in ice-cold perchloric acid (6% w/v), and centrifuged (4500 rpm) at 4°C. The pellet was retained for protein estimation (15)
and the supernatant was neutralized to pH 7.2 with NaOH. After removal of KClO4 by centrifugation, lyophilized samples were stored at -20°C until required for NMR analysis.
Samples were resuspended in 0.65 mL D2O containing 2 mM sodium [13C]formate as an internal intensity and chemical shift reference (
171.8). 13C{1H-Decoupled} spectra (14,000–18,000 transients, duty cycle 4 s, 83,300 data points) were acquired at 9.4 T, on a Bruker DRX-400 WB spectrometer using a 5 mm dual 1H/13C probe. Fully relaxed 1H and 1H{13C-decoupled} spectra (duty cycle 30 s, WURST-40 (21)
with a 112-step phase cycle (22)
, decoupling during acquisition) were obtained at 600.13 MHz on a Bruker DRX-600 spectrometer with a broadband inverse xyz gradient probe. Assignments were aided as described previously (18)
.
After zero filling to 128k, 13C{1H-decoupled} spectra were transformed using 3 Hz exponential line-broadening and peak areas were determined by integration using standard Bruker software (XWINNMR, Version 2.6). Peak areas were adjusted for nuclear Overhauser effect, saturation and natural abundance effects, and quantified by reference to the area of the internal standard resonance of [13C]formate. Metabolite pool sizes (lactate, alanine, GABA, glutamate, glutamine, and aspartate) were determined by integration of resonances in fully relaxed 600 MHz 1H{13C-decoupled} spectra using [13C]formate as the internal intensity reference. 13C data are expressed as absolute concentrations (net flux, after adjusting for natural abundance contributions) or as fractional enrichments (% of total metabolite pool that has been 13C labeled). Statistical significance was tested using the Mann-Whitney U test (=0.05) and data are expressed as means (SD).
31P Magnetic resonance spectroscopy
Brain bioenergetics are highly responsive to cerebral insults, including hypoxia/ischemia (23)
. Accordingly, bioenergetics were measured noninvasively by 31P magnetic resonance spectroscopy.
Mice with CM (PbA; N=9) were studied on day 6–7 postinoculation (p.i.). Noncerebral malaria (PbK) mice were studied on day 6 p.i. (PbK(Dm); N=12) and on day 9 p.i. (PbK(Pm); N=9). Noninfected control mice were age-matched (N=12). Mice were anesthetized lightly with pentobarbitone (30 mg/kg, i.p.) and placed in the animal bed of a Bruker microimaging probe. A 1 cm surface coil was positioned directly on top of the head and the entire probe was inserted into a 400 MHz wide-bore magnet interfaced with a Bruker DRX-AVANCE spectrometer. Total scanning time, including tuning and shimming, was 15–20 min. Animal temperature was maintained at 37°C by feeding 37°C water through the external gradient coil in which the animal handling system was located. After spectral acquisition, the mouse was removed from the magnet and decapitated such that the head fell into liquid nitrogen. The frozen brains were extracted for 1H NMR analysis as above.
31P magnetic resonance spectra (400 acquisitions, repetition rate 2 s) were acquired at 310 K. The pulse-width was calibrated to give maximum signal at this duty cycle.
Spectra were processed using jMRUI (version 1.0) (24)
. Quantification of the reconstructed signals was performed in the time domain. AMARES (25)
was used to fit the resonances of phosphomonoesters (PME, derived from phosphoethanolamine and phosphocholine), inorganic phosphate (Pi), phosphocreatine (PCr), and the , , and ß peaks of ATP (ATP). The area of the ß peak of ATP was used for quantification, as it is the only resonance from ATP that is not coresonant with resonances from ADP and NAD+/NADH. Intracellular pH was also determined (26)
. Results are expressed as peak ratios and relative to total signal intensity.
Statistical analysis of NMR-derived data was carried out initially using ANOVA. Where statistically significant, differences were indicated and between-group comparisons were conducted using a conservative nonparametric comparison (Mann-Whitney U test; =0.05). Correlations were tested using Spearman rank correlation.
RNA extraction
Mice were killed by cervical dislocation on designated days p.i. with malaria parasites. RNA was extracted from brain tissue with addition of 1 mL of TRI-reagent and 0.5 mL of zirconia beads (Biospec Products Inc. Bartlesville, OK, USA) and vigorous homogenization for 20 s in a Savant Fastprep (Holbrook, NY, USA). Chloroform (200 µL) was added and this mixture spun at 10,000 g for 20 min (4°C). To the aqueous layer, 500 µL of isopropanol was introduced. After standing at room temperature for 5 min, the sample underwent another identical centrifugation step. The pellet was washed in ice cold 70% (v/v) ethanol (RNase free; Amresco, Solon, OH, USA), air dried, and resuspended in water.
cDNA synthesis and real-time PCR
cDNA was synthesized using the Invitrogen Moloney murine leukemia virus (MMLV)-RT 2 step reverse transcriptase kit. Approximately 2 µg of RNA was added to a reaction containing 1x RT buffer, 0.5 mM dNTP, 0.1 µg of random primers, 0.1 mM dithiothreitol, 20 U RNase inhibitor, and 100 U of MMLV-RT. The samples were incubated for 1 h at 37°C, then subjected to an inactivation step at 92°C for 5 min. The cDNA was diluted with water and used as a template for real-time polymerase chain reactions.
The PCR reaction contained 1x SYBR green reaction mix, 1x ROX dye, 2 µM of each primer and 8 µL of template in a 20 µL reaction volume. Amplification was carried out on an ABI 7700 with incubation times of 50°C/2 min, 95°C/10 min, followed by 40–60 cycles of 95°C/15 s and 60°C/1 min. Specificity of the amplification was checked by melting curve analysis and the mRNA content was measured relative to that of hypoxanthine phosphoribosyltransferase (HPRT) RNA. Primer sequences are shown in Table 1
.
|
Statistical analysis of gene expression data was carried out using the Kruskal-Wallis test to check for variance. Where statistically significant, differences were indicated, probabilities were calculated post hoc using Mann-Whitney U test.
Uptake of D-[3H]aspartate and 2-deoxy-D-[1-14C]glucose by prisms of cerebral cortex in vitro
D-[3H]Aspartate uptake is used as a marker for study of the rates of uptake of L-glutamate, which has been shown to be sensitive to stimulation by cytokines (27)
. Uptake of 2DOG was investigated to determine whether glucose uptake rates, which may be altered by increased glucose transporter densities, were altered and hence contributing to differences in glucose metabolism.
Uptake of radiolabeled compounds was studied using described methodology (28)
. Animals were killed by cervical dislocation and the cerebral cortices were rapidly excised. Minislices (prisms) of the tissue were prepared using a McIlwain tissue chopper (Mickle Co., Gomshall, UK) and suspended in buffer at 100 mg/mL.
In the case of D-[3H]aspartate uptake, the suspension was gently centrifuged (<250 g for 2 min), resuspended in fresh buffer, and the tissue was distributed (10 mg/10 mL) into 50 mL conical flasks that were immediately placed in a shaking water bath (25°C, 90 strokes/min). The suspension was preincubated for 15 min and D-[3H]aspartate added at 0.5–2.0 µCi/10 mL, depending on the substrate concentration used. After a further 7 min incubation, the tissue prisms were separated by vacuum filtration (20–25 p.s.i.) through Whatman No. 1 filters (2.5 cm diameter) and rapidly washed twice with 10 mL of fresh, room temperature buffer. Filters were placed into standard scintillation vials and extracted overnight in 1 mL distilled water. After addition of 7 mL Optifluor (Canberra Packard), radioactivity was determined by scintillation counting.
For 2DOG uptake, the method was modified as follows: the centrifugation step was eliminated; preincubation and incubation temperature was 37°C; the incubation time was 10 min; the concentration of radiolabel was 1 µCi/10 mL.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
A).
|
There were no significant differences between the brain metabolite pool sizes in PbK(Dm) and PbK(Pm) mice. Compared with uninfected mice, brain lactate, glutamate, GABA, and glutamine pool sizes were elevated in PbK(Dm) and PbK(Pm) infection (Fig. 1A
); aspartate was also elevated in PbK(Dm) mice compared with controls. Lactate, glutamate, GABA, and aspartate pools were significantly elevated in PbK(Dm) and PbK(Pm) mice compared with PbA mice (Fig. 1A
).
Typical 13C NMR spectra obtained from the extracted brains of mice are shown in Fig. 2
. Total free (unmetabolized) [1-13C]glucose was significantly higher (P=0.0003, ANOVA) in PbA mice than controls or PbK(Dm) or PbK(Pm) (Fig. 1B
). Net flux of 13C label was significantly increased in PbA mice (Fig. 1B
) into Glu C4, GABA C2, lactate C3, and Ala C3 (all P=0.009) compared with control. Net flux into lactate C3 was significantly higher than in PbK(Dm) (P=0.009) or PbK(Pm) (P=0.028).
|
There were no significant differences in net 13C fluxes between PbK(Dm) or PbK(Pm) mice (Fig. 1B
). Net flux of 13C label into Glu isotopomers (C2, C3, C4) was significantly greater in PbK(Dm) and PbK(Pm) mice than in controls (all P=0.009 and P=0.014, respectively) and significantly greater than in PbA mice (all P=0.009 and P=0.014, respectively). Net flux into GABA C2 was increased compared with control (PbK(Dm), P=0.009; PbK(Pm), P=0.014), as was flux into Gln C4 (P=0.009, PbK(Dm); P=0.014, PbK(Pm) and Asp isotopomers (C2 and C3; P=0.009, PbK(Dm); P=0.014, PbK(Pm).
The fractional enrichments of Glu C4, GABA C2, lactate C3, and Ala C3 were significantly increased in PbA compared with controls. Fractional enrichment of lactate C3 was significantly higher than in PbK(Dm) or PbK(Pm) mice (Fig. 1C
). In PbK mice, the fractional enrichment of Glu C2, GABA C2, Asp C2, and Asp C3 was increased compared with control mice. The fractional enrichment of Glu C2 and Asp C2 was also higher in PbK mice than PbA mice (Fig. 1C
).
31P MRS determination of brain bioenergetics
Typical 31P MRS spectra obtained from the brains of mice in vivo are shown in Fig. 3
. Brain bioenergetic status was significantly altered in mice with CM compared with uninfected controls (Fig. 4
A). The peak ratios of Pi/PCr and Pi/ß-ATP were significantly higher in PbA mice than controls (Fig. 4A
). Inspection of quantitative data suggested this was due more to increased Pi (P=0.0086) than to decreased PCr (P=0.055) or ATP (P=1). The peak area of PME increased in PbA mice compared with control (P=0.016).
|
|
By contrast, peak ratios of Pi/PCr and Pi/ß-ATP in PbK(Dm) and PbK(Pm) mouse brains were not significantly different from those in control, uninfected mouse brains. These peak ratios were also significantly lower than in PbA (both P=0.0002). The peak area of PME was increased compared with control in both day (P=0.044) and parasitemia-matched (P=0.0002) PbK mice. There were no significant differences in intracellular brain pH between any of the mice [control 6.99 (0.08), PbA 7.00 (0.09), PbK(Dm) 7.03 (0.05), PbK(Pm) 6.98 (0.09)].
Data obtained from 600.13 MHz 1H NMR spectra of the extracted brains of these mice are shown in Fig. 4B
. There were no significant differences in metabolite levels between PbK(Dm) mice and controls. N-acetylaspartate, Glu, Asp, GABA, and myo-inositol were significantly decreased in PbA mice whereas Gln was increased.
Glutamate transport rates
There were no significant differences between D-[3H]aspartate uptake in brain slices from controls and PbA-infected mice at either lower (1.5 µM and 15 µM) or higher substrate concentrations (150 µM; P>0.05, t test). The observed rates of uptake were consistent with KM = 19.5 ± 3.0 µM and Vmax = 115 ± 3 pmol/mg/min (control, N=7) and KM = 15.8 ± 5.6 µM and Vmax = 94 ± 11 pmol/mg/min (PbA, N=9). The values indicated no significant difference between controls and malaria-infected tissue (P>0.05, t test.)
2-DOG uptake by mouse brain in vivo
Control mouse 2DOG uptake into brain [54.6 (7.5) cpm/mg tissue (SD)] did not differ significantly (ANOVA, P>0.05) from that of PbA (51.7 (5.8)), PbK (Dm) (54.0 (7.3)), or PbK (Pm) (61.1 (10.9).
RT-PCR
Levels of mRNA for genes related to metabolism relative to HPRT mRNA in PbA-infected mice during the course of malaria infection are shown in Fig. 5
. There were few significant increases despite the large changes seen in brain metabolism and bioenergetics. In contrast to PbK infected mice, in which levels of the glucose transporters GLUT1 increased 10-fold (Fig. 5)
, levels of GLUT1 in PbA infection increased only 2-fold on day 7. The largest change was observed in oxoglutarate dehydrogenase, where expression increased eightfold on day 7.
|
Levels of mRNA of genes related to metabolism, expressed relative to HPRT mRNA, in PbK mice are shown in Fig. 5
. Levels of mRNA for glutamine synthetase, alanine aminotransferase, and the astrocytic monocarboxylate transporter MCT2 increased from three- to fivefold on days 4–6 p.i. whereas expression of the glucose transporter GLUT1 and oxoglutarate dehydrogenase mRNAs continued to increase with time. Levels of mRNAs for the hypoxia-inducible transcription factor (HIF1), its target gene erythropoietin, and the erythropoietin receptor were not significantly altered.
Changes in expression of cytokine and cytokine receptor genes relative to HPRT RNA are shown in Table 2
. There were no significant changes compared with control in mRNA for any of the cytokines measured in PbK(Dm) mice. MRNA expression of interferon- and its receptor was significantly increased in both PbA-infected mice and PbK(Pm) mice, although its expression was 10-fold higher in PbA than PbK(Pm). Expression of mRNA for tumor necrosis factor- and its receptor was increased in PbA and PbK(Pm) mice although, again, expression was threefold higher in PbA than PbK(Pm). There was a modest increase in expression of lymphotoxin- and -ß mRNA in PbA-infected mice.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
13
14)
have been very influential in promoting the sequestration hypothesis concerning the pathogenesis of CM. Since these observations some 15 years ago, little additional biochemical insight into the mechanisms leading to the increases in CSF lactate has appeared from work in human CM. This is largely a consequence of the obvious ethical limitations. The demonstration that lactate and alanine levels were increased in the brain in the best-characterized model of murine CM (15)
opened new possibilities for investigating metabolic disturbances in this condition.
In the current work we have concentrated on changes in metabolism in the CNS. Systemic changes occur in both murine (5
, 29
, 30)
and human (31
, 32)
CM, and these may be due in part to elevated circulating levels of cytokines (2)
. However, increased CNS lactate is considered to be due to changes in metabolism within the brain and not to increased systemic production of lactate and uptake from the blood in human (13)
or mouse (15)
CM.
Cerebral metabolism of D-[1-13C]glucose
Flux of 13C label into brain metabolites from D-[1-13C]glucose follows well-characterized pathways (Fig. 6
). Analysis of labeling patterns in CM mouse brains showed changes consistent with the following: 1) decreased relative flux of label through the Krebs cycle (as evidenced by decreased flux of 13C into Glu C2 and C3, which would be labeled on the second turn of the Krebs cycle; Fig. 6
); 2) increased cycling of label into the glutamate/glutamine cycle (as shown by increased flux of 13C into Glu C4 and Ala C3 (Fig. 1B
) and increased fractional enrichment of Glu C4; Fig. 1C
), and hence increased glutamatergic activity (18)
; 3) increased flux through the GABA shunt as shown by increased flux into GABA C2 and Gln C4 (18)
; 4) increased flux of label into the hypoxia markers lactate C3 and Ala C3 (33)
; this is consistent with the greatly increased fractional enrichment of lactate C3 and Ala C3 (Fig. 1C
); 5) slower metabolic rate, indicated by increased residual C1 glucose resonances (Fig. 2)
.
|
As there was no significant difference in the uptake rates of 2DOG by mouse brain cortical tissue prisms or by whole mouse brains in vivo (34)
between control and malaria models, it is unlikely that any metabolic differences observed were due to different glucose transport efficacy as might be expected from the altered expression of GLUT 1 (Fig. 5)
. Although in normal brains local glucose use is coupled to glucose transporter expression (35)
, there are documented examples of disorders where this relationship is uncoupled (36)
.
Analysis of labeling patterns in noncerebral malaria mouse brains shows changes consistent with an increased metabolic rate. This can be seen from the even, relative increase in flux of 13C label into all iosotopomers measured (Fig. 1B
) and the lack of free (unmetabolized) glucose detected in the brains of these mice. Metabolite pool sizes in the brains of PbK mice given the large dose of [1-13C]glucose also uniformly increased (Fig. 1A
). This is in contrast to the lack of change in metabolite pool sizes seen in the brains of the PbK mice, which were not given a large bolus of glucose (i.e., those from which 31P MRS spectra were obtained (Fig. 4B
) and the lack of change in metabolite pool sizes in PbK mice reported by us previously (15)
. Administration of a large bolus of glucose, as occurred here, might be expected to translate to increased metabolite pool sizes given that the metabolic rate is 
twofold higher. The reason for the increased metabolic rate is not clear. Hypermetabolism is a commonly reported consequence of infection (37)
associated with fever and with the induction of cytokine expression. However, mice with cerebral (38)
and noncerebral malaria (S. B. Parekh, unpublished data) show reduced temperature, suggesting that fever is not causative of the increased metabolic rate seen in PbK infection. As little is known about the effect of cytokines on metabolism, it is not possible to state whether or not the hypermetabolism is related to altered cytokine levels, although this may be a fruitful area to investigate for a possible explanation.
31P Magnetic resonance spectroscopy
Brain bioenergetics showed distinctly different patterns in CM (PbA infection) as opposed to noncerebral malaria. Inorganic phosphate levels increased significantly in PbA infection compared with control and PbK-infected mice. Increased inorganic phosphate has been seen in conditions of hypoxia (39
, 40)
and is found in mitochondrial disorders (41)
.
The pattern of brain metabolite changes observed by 1H NMR spectroscopy in the brains of PbA mice from which 31P MRS data were obtained is consistent with an increase in the proportion of nonoxidative compared with oxidative, metabolism. This is indicated by increased lactate pools and decreased aspartate pools. Previous data also suggest decreased neuronal viability late in PbA infection, as indicated by decreased pool sizes of the neuronal viability marker N-acetylaspartate (42)
, levels of which decrease in response to inhibition of mitochondrial activity (43)
. There was a significant correlation between levels of the high-energy currency PCr and brain lactate (P=0.012. Rs=–0.46, Spearman Rank) and succinate levels (P=0.0034, Rs=–0.54), in agreement with this, as well as between levels of N-acetylaspartate and the bioenergetic ratios Pi/PCr (P=0.026, Rs=–0.41) and Pi/ß-ATP (P=0.029, Rs=–0.40).
By contrast, there were no significant changes in 31P MRS ratios in PbK(Pm) brains compared with control brains. Data from PbK(Dm) brains showed a dip in Pi (P=0.0003) and an increase in PCr (P=0.011) at day 6 p.i. that was not evident by day 9. The ratios of Pi/PCr (P=0.002) and Pi/ß-ATP (P=0.0035) were altered on day 6, but not different from control by day 9 p.i. There was no significant change in ATP, consistent with the tight control of ATP concentration in the brain (33)
. Higher levels of PCr and Pi/ATP, and lower levels of Pi are bioenergetically favorable and relate to improved brain performance in nondiseased human subjects (44)
.
Our findings of lack of change in bioenergetic status in PbK(Pm) mice are consistent with an earlier 31P MRS study in a different model system, Plasmodium yoelli, where severe anemia, induced by 70% parasitemia (much higher than in this study), produced no alterations in PCr, ATP or Pi. This was due to increased synthesis of 2,3-bisphosphoglycerate, resulting in a shift in the oxygen/hemoglobin dissociation curve (45)
. In P. berghei K173 infection, a similar compensatory effect may be occurring by day 14. This effect, by contrast, does not appear to occur in PbA infection despite comparable parasitemia levels at early stages of infection.
Time course of metabolic changes in cerebral malaria
We have examined metabolite pool sizes from the brains of mice with CM at a number of time points in their course of infection, both here and previously (15)
. We can now state that the increase in activity of the glutamate/glutamine cycle, seen early on day 6 of infection, occurs prior to the increase in lactate and alanine. This is supported by our observation in this work that, at parasitemia levels 10%, infected mouse brains showed evidence of increased glutamate/glutamine cycling (Fig. 1B
) without demonstrating increased lactate levels. PbA-infected mice with slightly higher parasitemia levels and high clinical scoring of behavioral changes, examined later in the time course of infection during the second half of day 6, showed both increased glutamate/glutamine cycling and augmented levels of lactate and alanine (15)
. This indicates that the increased glutamatergic activity does not occur in response to an hypoxic/ischemic insult, to which the levels of lactate and alanine in particular are very sensitive (33)
, but is due to some other metabolic perturbation. This is consistent with the model proposed by Lou and colleagues (46)
which shows hypoxia/ischemia occurring only after sequestration of blood cells in the microvasculature.
Brain and plasma lactate and plasma alanine levels have been shown to be predictive in human cerebral malaria (14
, 47)
both of likelihood of death and neurological outcome. It should be noted, however, that clinical measurements of neurological outcome are not necessarily proportional to the degree of cerebral "damage" (48)
.
Metabolic gene expression
Given the significant deleterious metabolic effects of PbA infection, it is remarkable that the resultant change in expression of genes directly involved in metabolism is so minor (Fig. 5)
. Expression of cytosolic and mitochondrial forms of aspartate aminotransferase was down-regulated, although by only twofold, and expression of oxoglutarate dehydrogenase was up-regulated. Both these changes would promote increased retention of carbon backbones by the Krebs cycle (i.e., carbon skeletons would be less likely to be lost as aspartate if aspartate amino transferase were less active and as glutamate if oxoglutarate dehydrogenase were more active), facilitating greater synthesis of ATP from available substrate. Given that metabolic rates have decreased (Fig. 1B
) and ATP production is down (Fig. 4A
) in PbA infection, this adjustment makes sense. However, there is no significant adjustment in expression of genes associated with the glutamate/glutamine cycle, such as glutamate dehydrogenase, glutamine synthetase or alanine aminotransferase, although glutamatergic activity is clearly increased.
This situation contrasts with PbK infection, where metabolic rates increase twofold. Net flux through the glutamate/glutamine is also increased, although proportional only to that through the Krebs cycle. Expression of glutamine synthetase and alanine aminotransferase is up-regulated and that of glutamate dehydrogenase is down-regulated. These changes may act to limit buildup of glutamate, which is known to be neurotoxic (49)
.
Cytokine gene expression
At least two possible mechanisms might underlie the changes in brain metabolism in PbA infection seen in this and earlier work (15)
: 1) hypoxia consequent upon microvascular obstruction, which has already been observed in this model of CM (50)
, albeit to a limited extent; 2) direct influences of cytokines on glycolysis and/or Krebs cycle activity.
Activation of the immune response with consequent up-regulation of cytokine expression is a patten common in infectious diseases (51)
, including malaria. The pathogenesis of PbA-induced CM is dependent on interferon- (52
53
54)
. For many years it was believed that TNF was similarly crucial (55
, 56)
, though it is now believed that lymphotoxin-, which can act in part through TNF receptors, is the critical effector cytokine (57)
. We observed increased expression of the genes for interferon-, TNF, TNF-R1, TNF-R2, and lymphotoxin- in the brain in PbA infection (Table 2)
. These changes were always greater in PbA infection than in PbK(Dm), though sometimes there were significant, smaller increases in expression in PbK(Pm). Interferon- has been shown to increase the rate of lactate production and ATP turnover in rat enterocytes (58)
and primary cultured astrocytes (59)
, similar to changes observed in PbA infection (15)
. The effects of lymphotoxin- on metabolic activity in the brain have not been studied. TNF, however, can stimulate glycolysis and/or lactate production in a variety of cell types (60
, 61)
, including astrocytes. Injection of TNF into rabbit brain leads to reductions in cerebral oxygen uptake and cerebral blood flow, accompanied by an increase in CSF lactate (62)
. TNF has also been reported to inhibit glutamate uptake by astrocytes (27)
; inhibition of glutamate transport has been shown to result in increased glutamatergic activity in the brain (17)
. However, we found no significant change in glutamate uptake rates in the brains of PbA infected mice compared with controls, suggesting that decreased glutamate transport was not causative of the observed increased glutamate/glutamine cycling in these brains. Indeed, it has been suggested that TNF may in fact be neuroprotective after ischemia (63)
. An investigation of outcomes from experimental head injuries in cytokine gene knockout mice showed no change in the pathophysiology of the head injury in mice lacking genes for proinflammatory (TNF, Il-6, LT-) cytokines, but increased post-traumatic mortality rates in these mice, suggesting that cytokines are mediating a protective effect by mechanisms not yet understood (64)
.
Increased production of lactate in PbA infection could reflect either reduced oxygen delivery, decreased oxygen use, or decreased lactate clearance rates (related to perfusion). The evidence for the former possibility in human CM is equivocal (12
, 65)
. An acquired intrinsic defect in cellular respiration has been termed "cytopathic hypoxia" (66)
. A proposed key mechanism is activation of poly(ADP-ribose) polymerase-1 (PARP-1), leading to depletion of NAD+/NADH and thus inhibition of mitochondrial respiration. This can be triggered by exposure to a combination of TNF-, IL-1ß, and interferon- (67)
in a model system in vitro. Several features of this mechanism fit well with published observations in human and murine CM. For example, cerebral venous pO2 in CM patients is above normal, providing no evidence for the diminution of oxygen delivery to the tissue that would be expected if there were substantial vascular obstruction (12)
. PARP expression is seen in the brains of malaria victims (68)
, and total NAD+/NADH level is reduced in the brain in murine CM (15)
. Finally, local overexpression of the TNF- and IL-1ß genes in the CNS occurs in murine (Table 2
; 69
) and human (70)
CM.
The evidence for altered cerebral blood flow rates is less clear. There is currently no information available on cerebral blood flow in murine CM. In humans, blood flow has been investigated with varied conclusions (12
, 65)
, but the true situation is difficult to interpret as classic methods for measuring cerebral blood flow, such as transcranial Doppler, are arguably not sufficiently sensitive to detect small vessel occlusion (48)
. It is possible that small vessel occlusion may occur, resulting in localized areas of lactate buildup due to a mixture of decreased lactate clearance and decreased oxygen delivery.
In summary, there are distinct metabolic, bioenergetic and genetic alterations seen in the brain in PbA infection (CM) vs. those seen in PbK infection without CM. In PbA infection, we speculate that the changes are consistent with a theory of cytopathic hypoxia. In PbK infection, the changes are consistent with benign hypermetabolism, whose cause remains the subject of speculation.
|
ACKNOWLEDGMENTS |
|---|
Received for publication July 10, 2003. Accepted for publication November 21, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
inhibits glutamate uptake by primary human astrocytes. Implications for the pathogenesis of HIV-1 dementia. J. Biol. Chem. 271,15303-15306 receptor-mediated signalling in pathology and anti-malarial immunity induced by Plasmodium berghei infection. Eur. J. Immunol. 30,1646-1655[CrossRef][Medline]
can prevent experimental cerebral malaria and its associated overproduction of tumour necrosis factor. Proc. Natl. Acad. Sci. USA 86,5572-5574, not systemic tumor necrosis factor , is the principle mediator of murine cerebral malaria. J. Exp. Med. 195,1371-1377 and interleukin-1 enhance glucose utilization by astrocytes: involvement of phospholipase A2. Mol. Pharmacol. 48,550-558[Abstract]
on cerebral oxygen uptake, cerebrospinal fluid lactate, and cerebral blood flow in the rabbit: role of nitric oxide. J. Clin. Invest. 95,1086-1091
. Cell. Mol. Neurobiol. 18,695-701[CrossRef][Medline]
