Address correspondence and reprint requests to Thad A. Rosenberger, Brain Physiology and Metabolism Section, National Institute on Aging, NIH, Building 10, Room 6 N202, Bethesda, MD 20892–1582, USA. E-mail: email@example.com
In a rat model of acute neuroinflammation, produced by a 6-day intracerebral ventricular infusion of bacterial lipopolysaccharide (LPS), we measured brain activities and protein levels of three phospholipases A2 (PLA2) and of cyclo-oxygenase-1 and -2, and quantified other aspects of brain phospholipid and fatty acid metabolism. The 6-day intracerebral ventricular infusion increased lectin-reactive microglia in the cerebral ventricles, pia mater, and the glial membrane of the cortex and resulted in morphological changes of glial fibrillary acidic protein (GFAP)-positive astrocytes in the cortical mantel and areas surrounding the cerebral ventricles. LPS infusion increased brain cytosolic and secretory PLA2 activities by 71% and 47%, respectively, as well as the brain concentrations of non-esterified linoleic and arachidonic acids, and of prostaglandins E2 and D2. LPS infusion also increased rates of incorporation and turnover of arachidonic acid in phosphatidylethanolamine, plasmenylethanolamine, phosphatidylcholine, and plasmenylcholine by 1.5- to 2.8-fold, without changing these rates in phosphatidylserine or phosphatidylinositol. These observations suggest that selective alterations in brain arachidonic acid metabolism involving cytosolic and secretory PLA2 contribute to early pathology in neuroinflammation.
A graded animal model of neuroinflammation can be produced by chronically infusing bacterial lipopolysaccharide (LPS) into the 4th cerebral ventricle of a rat (Hauss-Wegrzyniak et al. 1998b). After 14 days of infusion, the rat brain demonstrates activated OX-6-positive microglia and of glial fibrillary acidic protein (GFAP)-positive astrocytes, increased levels of cytokines, and increased expression of complement protein (Willard et al. 1999). With continuous infusion, their brains show a time-dependent loss of cholinergic markers and of hippocampal neurons (1 month), and display reduced performance on tests of memory (Hauss-Wegrzyniak et al. 1998a). Some of these changes can be reduced by pre-treatment with NSAIDs (Willard et al. 2000), suggesting that phospholipid metabolism involving AA contributes to the LPS-induced neuropathology. However, a detailed examination of brain fatty acid metabolism involving AA and PLA2 enzymes, prior to cholinergic cell loss, has not been made in this rat model.
To address the issue of altered brain fatty acid and phospholipid metabolism being a major participant in the evolution of neuroinflammatory events, we subjected rats to 6 days of intracerebral ventricular LPS infusion and measured brain activities of PLA2 and cyclo-oxygenase (COX) enzymes, brain concentrations of non-esterified and esterified fatty acids and of eicosanoids, and incorporation and turnover rates of AA in individual brain phospholipids. This study suggests that brain membrane metabolism is altered prior to the glial infiltration and cholinergic cell loss associated with this model and supports the contention that early alterations in brain fatty acid metabolism precedes clinical manifestation of diseases associated with neuro-inflammation. Tracer kinetic analysis was performed using an in vivo fatty acid method and model developed in our laboratory (Robinson et al. 1992; Washizaki et al. 1994; Rapoport 2001).
Materials and methods
[5, 6, 8, 9, 11, 12, 14, 15–3H]Arachidonic acid ([3H]AA, 240 Ci/mmol, ≥ 98% pure) was purchased from Moravek Biochemicals (Brea, CA, USA). Scintillation counting and gas chromatography confirmed the tracer-specific radioactivity. Phospholipid and neutral lipid standards were from Nu-Chek-Prep (Elysian, MN, USA) and ‘fatty acid free’ bovine serum albumin and LPS (Escherichia coli, serotype 055:B5, TCA extraction) were from Sigma Chemicals (St Louis, MO, USA). High-performance liquid chromatography (HPLC)-grade n-hexane and 2-propanol were from EM Science (Gibbstown, NJ, USA) and reagent-grade chloroform, methanol, and other chemicals were from Mallinckrodt (Paris, KY, USA). A scintillation cocktail (Ready-Safe, Beckman, Fullerton, CA, USA) containing 1.0% glacial acetic acid was used to determine radioactivity.
Cannula placement and animal and surgery
Surgery was performed following the Guide for the Care and Use of Laboratory Animals (NIH Publication no. 80–23). Alzet osmotic mini-pumps (Model 2002; 0.5 µL/h, Plastics One, Roanoke, VA, USA) were used to deliver either artificial cerebrospinal fluid (140 mm NaCl, 3.0 mm KCl, 2.5 mm CaCl2, 1.0 mm MgCl2, 1.2 mm Na2PO4, pH 7.4) or LPS dissolved in artificial cerebrospinal fluid (1.0 µg/ml) into the 4th ventricle of the rat, as described previously (Hauss-Wegrzyniak et al. 1998b). The animal was allowed to recover from anesthesia. Six days later, it was anesthetized with 3% halothane (Halocarbon, River Edge, NJ, USA), and polyethylene catheters (PE 50, Becton Dickinson, Sparks, MD, USA) filled with sodium heparin (100 IU) were implanted into the right femoral artery and vein. The skin was closed and 1% lidocaine was applied to the wound. The hindquarters of the rat were wrapped loosely in a fast-setting plaster body cast and taped to a wooden block. The animal was allowed to recover from anesthesia for 3–4 h, while body temperature was maintained at 36.5°C using a feedback-heating device (Yellow Springs Laboratories, Yellow Springs, OH, USA) equipped with a rectal thermometer.
Infusion of [3H]AA
The method and basis of tracer infusion have been reported elsewhere (Robinson et al. 1992; Washizaki et al. 1994; Rapoport 2001). With an infusion pump (Harvard Apparatus, South Natick, MA, USA), an awake rat was infused intravenously for 5 min at a rate of 0.4 mL/min, with 2.0 mL isotonic saline containing 1.75 mCi/kg body wt [3H]AA suspended in 0.06 mg bovine serum albumin (Sigma). Arterial blood samples (100 µL) were collected at fixed times during infusion to determine the radioactivity and concentrations of non-esterified fatty acids and lipids in whole blood and plasma. Five minutes after starting infusion, the rat was killed with sodium pentobarbital (100 mg/kg body wt, i.v.) and immediately subjected to head-focused high-energy microwave irradiation to stop brain metabolism (5.5 kW, 3.4 s; Cober Electronics, Stamford, CT, USA). Animals used for enzyme assays were killed with sodium pentobarbital and their brains were removed, frozen on dry ice, and stored at − 80°C until analyzed. Animals used for immunohistochemical analysis were anesthetized with sodium pentobarbital and killed with a cardiac perfusion of phosphate-buffered saline followed by perfusion of 4% paraformaldehyde. To avoid artifacts from the cannula implant (Ghirnikar et al. 1996), analyses were performed only on cortical and basal forebrain regions.
Astrocytes were visualized by immunohistochemical staining using polyclonal antiGFAP (Dako, Carpinteria, CA, USA) 1: 2000 coupled to an avidin–biotin complex (ABC; Vectastain Elite immunohistochemistry kit, Vector Laboratories, Burlingame, CA, USA) and reacted with peroxidase substrate (DAB, Vector Laboratories). Microglia were visualized using lectin–horseradish peroxidase conjugates (IB4–HRP) 1 : 50, H-2401-B4-05 (EY Laboratories, San Mateo, CA, USA) as previously described (Streit 1990). Sagittal brain sections (15-µm thick) were cut from fixed brain on a vibratome (Model 05330, Vibratome Co., St Louis, MO, USA) and counter-stained with hemotoxylin QS (Vector Laboratories). Sections were examined under a light microscope to identify morphological changes in astrocytes immunopositive for GFAP, and IB4–HRP-positive microglia. All procedures were performed at room temperature in a light-protected humidity box.
Enzyme assays and western blotting
Non-microwaved brain tissue was homogenized in 10 mm Tris buffer pH 7.8, containing 1% Nonidet P-40, 150 mm NaCl, 1 mm EDTA, 2 mm dithiothreitol, 10 µm phenylmethylsulphonyl fluoride (PMSF), 9.2 µm aprotinin, 0.11 µm leupeptin, and 10.2 µm pepstatin A, using a Polytron tissue homogenizer (Tekmar Institute, Cincinnati, OH, USA). Cytosolic fractions were obtained by centrifuging tissue homogenates at 18 000 g for 10 min at 4°C (Rintala et al. 1999), and protein concentration was determined (Bradford 1976) using bovine serum albumin as a standard.
PLA2 activity was determined in cytosolic fractions using a cPLA2 assay Kit (Cayman, Ann Arbor, MI, USA) in the presence and absence of a cPLA2 and Ca2+-independent iPLA2 inhibitor, arachidonoyltriflouroketone (50 µm); of an iPLA2 inhibitor, bromoenol lactone (5 µm); of a specific sPLA2 inhibitor, thioetheramide-PC (50 µm; Farooqui et al. 1999). Activity was calculated by measuring absorbance at 414 nm using the DTNB extinction coefficient (10.66 mm). Activities in LPS-infused rats, reported as percentage control activity, were determined by measuring the rate, nmol/min/g protein, at which 1-O-hexadecyl-2-arachidonoylthio-2-deoxy-sn-glycero-3-phosphocholine (cPLA2 and sPLA2) or diheptanoyl thio-sn-glycero-3-phosphocholine (sPLA2 and iPLA2) was hydrolyzed (Reynolds et al. 1994).
Net brain COX activity was determined as previously described (Taniguchi et al. 1997). Extracts were diluted in buffer containing 10 mm phenol, 18.2 mm l-epinephrine, 4.6 mm glutathione, and 9.3 µm hematin. The assay was started with 1 mm AA, incubated for 10 min at 37°C, stopped using 250 µL HCl, and extracted in ethyl acetate. For western blotting, cytosolic protein (50 µg) was loaded on NuPage gels (Invitrogen, Carlsbad, CA, USA) and transferred to a nitrocellulose membrane as directed by the manufacturer. Western blotting was carried out as reported (Rintala et al. 1999), using an antibody to cPLA2 (monoclonal, 1 : 500), sPLA2 (polyclonal, 1 : 500; Cayman), or actin (1 : 15000, Sigma), followed by a secondary antibody conjugated with HRP (1 : 1000, Bio-Rad, Hercules, CA, USA). Immunoblots were visualized by a chemilluminescence reaction (Pierce, Rockford, IL, USA). The protein level was quantified by measuring the integrated optical density of the bands, after background subtraction by AlphaEase Stand Alone software (Alpha Innotech, San Leandro, CA, USA). Image analysis was performed on optical density-calibrated images captured with a video camera (Alpha Innotech).
Brain and plasma lipid extraction and chromatography
Total lipids from microwaved brains were extracted using n-hexane/2-propanol (3 : 2, by vol.) in a glass Tenbroeck homogenizer (Radin 1981). Plasma lipids were extracted by the method of Folch et al. (1957). Standards and lipid extracts in chloroform were applied to Whatman silica gel 60 A LK6 TLC plates and separated using chloroform/methanol/acetic acid/H2O (50 : 37.5 : 3 : 2, by vol; Jolly et al. 1997). Bands corresponding to ethanolamine glycerophospholipids, phosphatidylinositol (PtdIns), phosphatidylserine (PtdSer), or choline glycerophospholipids were scraped from the TLC plates. The plasmenylethanolamine (PlsEtn) and plasmenylcholine (PlsCho) fractions were isolated from the ethanolamine and choline glycerophospholipid fractions as described previously (Murphy et al. 1993). Neutral lipids were separated on silica gel 60 plates using the solvent system of heptane/diethyl ether/acetic acid (60 : 40 : 4, by vol; Breckenridge and Kuksis 1968). Prostaglandins were isolated from the brain lipid extracts on Unisil activated silicic acid (Clarkson Chem., Williamsport, PA, USA; Saunders and Horrocks 1984) and isolated by HPLC (Powell 1985). Gas chromatography was used to quantify esterified and non-esterified fatty acid and prostaglandin levels. Liquid scintillation counting was used to measure radioactivity. Extracts were stored in n-hexane/2-propanol (3 : 2, by vol) under N2 at − 80°C.
Quantification of labeled and unlabeled acyl–CoA
Long-chain acyl–CoA species were isolated from microwaved rat brain using oligonucleotide purification cartridges (Applied Biosystems, Foster City, CA, USA; Deutsch et al. 1994). Acyl–CoA concentrations and arachidonoyl–CoA radioactivity were measured using peak area analysis from HPLC chromatograms and liquid scintillation counting.
Methylation of esterified, non-esterified acids, and prostaglandins
Esterified fatty acids in the different phospholipid classes were methylated with 0.5 m methanolic potassium hydroxide at 37°C for 30 min. The reaction was stopped with methylformate and the fatty acid methyl esters were extracted with n-hexane. The hydroxyl groups of the individual prostaglandins were acetylated using anhydrous pyridine/acetic anhydride (2 : 1, by vol) at 37°C for 15 min followed by the addition of H2O and extracted with petroleum ether. The non-esterified fatty acids and acetylated prostaglandins were methylated using 2% sulfuric acid in toluene/methanol (1 : 1, by vol) at 65°C for 2 h. The reaction was terminated with H2O and extracted with petroleum ether.
Gas chromatography of fatty acid and prostaglandin methyl esters
Fatty acid and prostaglandin methyl esters were quantified using a gas chromatograph (Trace 2000, ThermoFinnigan, Houston, TX, USA) equipped with a capillary column (SP 2330; 30 m × 0.32 mm i.d., Supelco, Bellefonte, PA, USA) and a flame ionization detector. Sample runs were initiated at 90°C with a temperature gradient to 230°C over 20 min. Fatty acid methyl ester and prostaglandin methyl ester standards were used to establish relative retention times and response factors. The internal standard, methyl heptadecanoate, and the individual fatty acids were quantified by peak area analysis. The detector response was linear, with correlation coefficients of 0.998 or greater within the sample concentration range for all standards.
Radioactivity of a brain phospholipid i of interest, nCi/g, was calculated by correcting its net brain radioactivity for its intravascular radioactivity (Grange et al. 1995). Blood samples taken at the time of death, T = 5 after starting tracer infusion, were extracted and analyzed to make this correction. Unidirectional incorporation coefficients, ml/s/g, of [3H]AA from plasma into phospholipids i were calculated as follows,
where t is time after beginning of infusion, and (nCi/ml) is the plasma concentration of radiolabeled AA during infusion. Rates of incorporation of non-esterified AA from plasma into brain phospholipid i,Jin,i, and from brain arachidonoyl–CoA into brain phospholipid i, JFA,i, were calculated as follows,
cpl (nmol/ml) is the concentration of unlabeled non-esterified AA in plasma. λarachidonoyol−CoA represents the steady-state-specific activity of arachidonoyl–CoA relative to that of plasma during [3H]AA infusion,
where the numerator is the specific activity of brain arachidonoyl–CoA and the denominator is the specific activity of plasma AA. The fractional turnover rate of AA within phospholipid i, FFA,i (%/h), is defined as,
The half-life of the FA in i is defined as,
In addition to the above calculations, we determined the steady-state specific activity of non-esterified brain AA relative to plasma specific activity during [3H]AA infusion,
Data and statistics
Integrals of plasma radioactivity were determined by trapezoidal integration (SigmaPlot, SPSS Science, Chicago, IL). Unpaired t-tests (Instat® Ver. 3.05, GraphPad, San Diego, CA) were used to compare means between LPS-infused and control rats, where statistical significance was taken as p ≤ 0.05. Data are presented as means ± SD.
Physical and immunohistochemical changes
As previously reported (Hauss-Wegrzyniak et al. 1998b), during the 6 days of LPS infusion rats had a lower rate of weight gain compared with the controls, 10.5 ± 8.3 g/day versus 31.3 ± 6.3 g/day (p < 0.05). Immunohistochemical analyses of sagittal brain sections showed that lectin staining of microglia was increased in the LPS-treated animals, but limited to the cerebral ventricles, pia mater, and the glial membrane of the cortex (Figs 1a and b). No lectin staining was found in the parenchyma of the brain. Immunohistochemical analysis using GFAP-confirmed dense staining around the outer edges of the cortex (Figs 1c and d). At higher magnification, we did not find astrocyte hypertrophy compared to control, but rather an elongation of astrocyte processes (Fig. 1d, insert). No astrocyte hypertrophy or hyperplasia was evident in all regions analyzed.
Brain PLA2 activities and protein levels
LPS infusion resulted in a 46% increase in net brain PLA2 enzymic activity (Fig. 2a). A subtraction assay (Reynolds et al. 1994) showed that the increase in enzymic activity represented a 71% increase in cPLA2 activity and a 47% increase in sPLA2 activity. In contrast, the brain activity of the calcium-independent iPLA2, which was determined using 1-O-hexadecyl-2-arachidonoylthio-2-deoxy-sn-glycero-3phosphocholine and diheptanoyl thio-sn-glycero-3-phosphocholine as substrate, was not altered significantly by LPS infusion. Western blot analysis showed that the increases in cPLA2 and sPLA2 activities were not accompanied by increases in their respective protein levels (Fig. 2b).
Brain COX activities, protein levels, and eicosanoid levels
As illustrated in Fig. 3, LPS infusion did not increase the net brain COX activity, or the protein level of either COX-1 or COX-2, compared to controls. Western blot analysis of COX-1 and COX-2 protein levels from whole-brain extracts following 24 h of LPS infusion and northern blot analysis of COX-2 mRNA at 6 and 24 h following cannula implant were not significantly different from control samples (data not shown). These data suggest that changes in COX expression are not evident in this model or are extremely localized and below the level of detection of the techniques used to examine whole brain extracts. However, LPS infusion did increase the brain concentrations of PGE2 and PGD2 by 1.5- and twofold, respectively, without affecting the concentrations of PGF1α, PGF2α or TxB2(Fig. 4).
Plasma and brain lipid concentrations
There was no statistically significant difference in the mean plasma concentration of any non-esterified fatty acid, between LPS-infused and control animals. Additionally, the brain long chain acyl–CoA concentrations did not differ significantly between the two groups (Table 1). On the other hand, significant twofold increases in the concentrations of non-esterified brain linoleic acid and AA were produced by LPS infusion, whereas the concentrations of the other non-esterified brain fatty acid, including docosahexaenoic acid, were unaffected (Table 1). The relative specific activity (λarachidonoyl-CoA) of brain arachidonoyl–CoA compared to plasma AA during steady-state [3H]AA infusion (Eq. 4) did not differ between the groups, whereas the relative specific activity (λAA) of brain non-esterified AA (Eq. 7) was decreased twofold (Table 2).
Table 1. Concentrations of plasma and brain non-esterified fatty acids and brain acyl–CoAs in control and LPS-infused rats
Plasma non-esterified fatty acid (nmol/ml)
Brain non-esterified fatty acid (nmol/g)
Brain acyl–CoA (nmol/g)
Values are means ± SD (n = 6); *p < 0.05, differs from control mean. nd, not detected.
As illustrated in Table 3, there was no difference in the mean brain concentration of any of the phospholipid classes between the LPS-infused and control rats. Furthermore, there was no difference in the concentration of any esterified fatty acid in any phospholipid class between the two groups (Table 4). The brain concentrations of phospholipids and their fatty acid compositions are comparable to previously published values (Contreras et al. 2001).
Table 3. Brain phospholipid concentrations in control and LPS-infused rats
Control µmol/g brain
LPS-infused µmol/g brain
Values are means ± SD (n = 6). PtdEtn, phosphatidylethanolamine; PlsEtn, plasmenylethanolamine; PtdIns, phosphatidylinositol; PtdSer, phosphatidylserine; PtdCho, phosphatidylcholine; PlsCho, plasmenylcholine.
12.67 ± 2.97
12.63 ± 0.69
7.29 ± 0.42
7.83 ± 0.23
2.87 ± 0.06
2.81 ± 0.10
6.15 ± 0.29
6.56 ± 0.44
19.18 ± 0.33
20.37 ± 0.72
1.25 ± 0.32
1.27 ± 0.08
Table 4. Esterified fatty acid concentrations in individual brain phospholipids of control and LPS-infused rats
Control nmol/g brain
Control nmol/g brain
Control nmol/g brain
Values are means ± SD (n = 6). Abbreviations: See Table 3 legend.
Palmitate (16 : 0)
1572 ± 460
1393 ± 123
1036 ± 87
1005 ± 82
175 ± 17
187 ± 11
Stearate (18 : 0)
5757 ± 526
5066 ± 240
1784 ± 201
2098 ± 232
1108 ± 38
1061 ± 43
Oleate (18 : 1n-9)
2318 ± 282
2246 ± 415
2807 ± 106
2575 ± 185
176 ± 10
214 ± 15
Arachidonate (20 : 4n-6)
1151 ± 188
1381 ± 190
862 ± 63
904 ± 189
1195 ± 41
1134 ± 59
Docosahexaenoate (22 : 6n-3)
1473 ± 364
2082 ± 363
1476 ± 112
1472 ± 239
189 ± 18
178 ± 13
Palmitate (16 : 0)
112 ± 10
120 ± 19
9487 ± 175
10 113 ± 284
430 ± 84
436 ± 75
Stearate (18 : 0)
2885 ± 131
3032 ± 157
3088 ± 93
3329 ± 170
348 ± 96
347 ± 50
Oleate (18 : 1n-9)
1205 ± 136
1352 ± 195
4097 ± 182
4542 ± 304
176 ± 34
251 ± 19
Arachidonate (20 : 4n-6)
295 ± 25
338 ± 63
725 ± 23
809 ± 85
63 ± 9
72 ± 18
Docosahexaenoate (22 : 6n-3)
1482 ± 30
1499 ± 36
576 ± 34
635 ± 101
115 ± 23
111 ± 26
AA incorporation and turnover in individual brain phospholipids
Mean arterial plasma radioactivity profiles during intravenous infusion of [3H]AA did not differ between LPS-infused and control rats. A steady-state radioactivity was achieved 120 s following the start of intravenous [3H]AA infusion in both LPS-infused and control rats, and did not differ between the groups. Mean steady-state values equaled 120 ± 22 and 128 ± 19 nCi/nmol in the two groups, respectively.
Table 5 shows that LPS infusion did not significantly affect calculated incorporation coefficients k* ( Eq. 1) of AA from plasma in PtdIns or PtdSer, whereas infusion increased k* significantly in PtdEtn, PlsEtn, PtdCho, and PlsCho. Because derived parameters for each phospholipid were calculated by multiplying k* by common constant factors (Eqs 1–6) , the same pattern of significance was found for the rates of incorporation of unlabeled AA from the precursor arachidonoyl–CoA pool into phospholipids, given as JFA , i ( Eq. 3) , and for turnover FFA , i of AA in the phospholipids ( Eq. 5) . Thus, turnover of AA was increased in PtdEtn, PlsEtn, PtdCho, and PlsCho by 1.5- to 2.8-fold, while it was unchanged in PtdIns or PtdSer ( Table 5 ).
Table 5. Brain incorporation coefficients ( ) of arachidonic acid, net incorporation rates from brain arachidonoyl–CoA ( JFA , i ) , and turnover rates in different phospholipids in control and LPS-infused rats
(ml/g/s, × 10 −5 )
(nmol/g/s, × 10 −2 )
Values represent means ± SD (n = 6). *p < 0.05 differs from control mean). Abbreviations: See Table 3 legend.
Infusion of LPS into the 4th cerebral ventricle of rats for 6 days, a model for acute neuroinflammation, increased the lectin-reactive microglia in the cerebral ventricles, pia mater, and the glial membrane of the cortex. LPS infusion also resulted in morphological changes of GFAP-positive astrocytes in the cortical mantel and areas surrounding the cerebral ventricles. Thus the reported presence of hypertrophied GFAP-positive astrocytes in the brain of LPS-infused animals found following 14 days of LPS infusion (Hauss-Wegrzyniak et al. 1998a) is not evident at the early 6-day time point and is consistent with the time-dependent injury associated with this model. Regardless, LPS infusion significantly increased the brain concentrations of linoleic acid and AA, without changing the concentration of other non-esterified fatty acids or acyl–CoA species. The brain concentrations of PGE2 and PGD2, derived from AA via COX-1 and COX-2 (Shimizu and Wolfe 1990; Fitzpatrick and Soberman 2001), were increased significantly by LPS infusion, whereas the concentrations of TxB2, PGF1α, and PGF2α were unaffected compared with control. Because the control brain concentrations of the eicosanoids are comparable to values found in microwaved brain using similar techniques (Anton et al. 1983), suggest that the increased levels brain of PGE2 and PGD2 following LPS infusion is a result of a shift in the saturation kinetics of COX-1 and/or COX-2 due to the increased brain AA levels. Although net brain COX activity and the net brain protein levels of the two enzymes were not significantly changed by LPS infusion, we cannot rule out localized changes in brain COX expression in periventricular regions in immediate contact with the CSF (Kaufmann et al. 1996).
The changes in brain lipid metabolism described in this study are comparable in part to changes reported in other LPS-induced inflammation models. Based on these latter models, Balsinde et al. (2002) has argued that LPS-induced cytokine release will first activate cPLA2 and release AA. These events in turn will activate type IIA sPLA2 and promote its movement from the inner to outer cell membrane leaflet. In rat mesangial and glioma cells, the production of cytokine IL-1 increases Ca2+ influx and promotes the phosphorylation of cPLA2 by protein kinase C and/or p42 MAP kinase, thereby increasing cPLA2 activity and the formation of lyso-phospholipid and the release of AA (Gronich et al. 1994; Ozaki et al. 1994; Bankers-Fulbright et al. 1996). Increased concentrations of AA and lyso-phospholipids promote the activation of type IIA secretory sPLA2 (Laposata et al. 1985; Sun and Hu 1995; Tong et al. 1995; Kuwata et al. 1998). Our evidence of increased activation of both cPLA2 and sPLA2 is consistent with this general scheme. LPS infusion increased net brain PLA2 enzyme activity, which could be ascribed to increased activities of cPLA2 and sPLA2, which are both calcium-dependent (Dennis 1994; Clark et al. 1995). In contrast, the brain activity of the calcium-independent iPLA2 was unchanged. The fact that the protein levels of cPLA2 and sPLA2 were not altered suggests that LPS infusion increased PLA2 activity by modulating intracellular calcium concentrations rather than altering gene transcription. It remains to be shown if LPS infusion altered the phosphorylation state of cPLA2, as this enzyme depends on phosphorylation for full activation (Kramer and Sharp 1997).
Awake LPS-infused rats, demonstrated a significant twofold decrease in the steady-state specific activity of brain non-esterified AA, relative to the plasma-specific activity, given as the ratio λAA (Table 2, Eq. 7). In contrast, the steady-state-specific activity of arachidonoyl–CoA normalized to plasma AA-specific activity, λarachidonoyl–CoA, was not altered. Non-esterified brain AA is the precursor for the formation of arachidonoyl–CoA via the action of acyl–CoA synthetase with the consumption of 2 mol of ATP per mole of AA (Purdon et al. 2002). In control brain, the approximate equivalence of λAA and λarachidonoyl–CoA (Table 2) is consistent with this precursor–product relation, and with evidence that the actual brain precursor AA pool is in rapid equilibrium with tracer AA in plasma (Washizaki et al. 1994). However, the twofold decline in λAA, accompanied by an approximately twofold increase in the net brain non-esterified brain AA (Table 1), implies that the AA released during LPS infusion was not in equilibrium with labeled AA in plasma, and that it constituted a brain pool that was not an immediate precursor for the synthesis of arachidonoyl–CoA. Thus, it is likely that there are two ‘functional’ non-esterified AA compartments in brain. One, normally larger, that can rapidly exchange with labeled AA in plasma (Washizaki et al. 1994), and a second, into which AA is massively released from phospholipids following the LPS-induced activation of PLA2. The two non-esterified brain AA compartments, suggested by our observations, are consistent with evidence found in cell culture that exogenous labeled AA can be converted to prostaglandins only if it is first incorporated into phospholipids and subsequently released by PLA2 (Neufeld et al. 1985; Capriotti et al. 1988).
LPS-infused rats demonstrated significantly increased rates of AA incorporation and turnover in brain PtdCho, PlsCho, PtdEtn, and PlsEtn but not in PtdIns or PtdSer. These changes may reflect the phospholipid and fatty acid selectivity of cPLA2 and sPLA2 (Dennis 1994; Clark et al. 1995), both of whose activities were increased by LPS infusion (Fig. 2). Another possibility is that a 39 kDa plasmalogen-selective PLA2 also was activated by LPS infusion (Yang et al. 1996). The fact that both PlsEtn and PlsCho are plasmalogens, which contain a vinyl-ether linkage in the sn-1 position of their glycerol backbone and turn over rapidly in brain (Rosenberger et al. 2002), suggests that this PLA2 also may be activated by LPS-infusion.
Post-mortem studies of the AD brain have demonstrated that PlsEtn and PtdCho concentrations are reduced by 20% and 7.5%, respectively, in temporal cortex, and that PlsEtn is reduced in the hippocampus, with no alteration in the proportion of PtdIns or PtdSer (Nitsch et al. 1992; Guan et al. 1999). Electron spray mass spectrometry confirmed the loss of PlsEtn and showed it to be correlated with pre-morbid dementia severity (Han et al. 2001). These phospholipid changes may underlie the reported reduced ‘critical temperature’ of phospholipid membranes in affected brain regions in AD (Gershfeld and Ginsberg 1995; Ginsberg et al. 1995; Farooqui et al. 1997) and are consistent with in vivo magnetic resonance spectroscopic evidence of increased breakdown of brain phospholipids (Pettegrew et al. 1995). Our studies in the acute (6-day) LPS neuroinflammatory model suggest that some of the changes in phospholipids and membrane stability in the AD brain may be directly related to the inflammatory process.
The constitutive PLA2 activity in brain is mainly calcium-independent (iPLA2) in nature (Yang et al. 1999) and is decreased in brains of AD patients (Gattaz et al. 1995; Ross et al. 1998; Talbot et al. 2000). In contrast, the immunoreactivity of cPLA2 is increased in affected regions (Stephenson et al. 1996). According to the present study, iPLA2 activity in not modified by the 6-day LPS infusion. The lack of changes in iPLA2 activity in these studies may be explained by regional alterations in phospholipid metabolism or cell-type specific expression of the different PLA2 isoforms. It is accepted that cPLA2 is expressed in the astrocytes (Stephenson et al. 1994), while iPLA2 is for the most part expressed in the neuron (Talbot et al. 2000). Because significant cholinergic cell loss has not been identified in this model until 14 days of LPS infusion (Hauss-Wegrzyniak et al. 1998a), it is not surprising that no changes were found in the neuronal-localized iPLA2 activity following the acute 6-day LPS infusion. Thus we suspect that the changes in AA metabolism found in these studies represent an early stage of inflammation in which the glial response associated with neuroinflammation is being consolidated, but does not represent a stage with significant neuronal cell loss.
Non-esterified AA, released from phospholipids by PLA2, has been reported elsewhere to contribute to neuroinflammation in a number of models (Chilton et al. 1996). Among its many effects, non-esterified AA can modulate the activities of protein kinases A and C, alter ion channels and electrical excitability, influence long-term potentiation, inhibit neurotransmitter uptake, and enhance synaptic transmission (Katsuki and Okuda 1995). High concentrations of non-esterified AA can damage mitochondrial cell membranes, interfere with ATP synthesis (Atsumi et al. 1997), and enhance glutamate-induced excitotoxicity (Katsuki and Okuda 1995). Additionally, prostaglandins produced by the initial actions of COX-1 and COX-2 (Shimizu and Wolfe 1990) can have multiple second messenger effects. For example, they can modulate cerebral blood flow (Fitzpatrick and Soberman 2001), immune function, and inflammation (Katsuki and Okuda 1995). Our observed increases in brain PGE2 and PGD2 and a decrease in the specific radioactivity of the brain non-esterified AA pool (λAA) following 6 days of LPS infusion point to altered availability and recycling of brain AA.
Increased expression in rat brain of a number of neuroinflammatory markers has been noted immediately after (Quan et al. 1994) and following 14 days of LPS infusion (Hauss-Wegrzyniak et al. 1999). More prolonged LPS infusion (1–2 months) has been shown to intensify these changes, and to provoke the loss of hippocampal CA3 pyramidal neurons and of cholinergic neurons in the basal forebrain (Willard et al. 1999, 2000). Thus, we propose that altered lipid metabolism in the acute phase of LPS infusion contributes to the evolution and intensity of the neuropathology involving neuroinflammatory events. Our data suggest time-dependent and specific targets, cPLA2 and sPLA2, and possibly COX, for preventive therapy in human diseases associated with neuroinflammation. The reported efficacy of early treatment with NSAIDs in subjects at risk for AD (Stewart et al. 1997; in t' Veld et al. 2001) supports a neuroinflammatory contribution involving AA and altered phospholipid metabolism to AD neuropathology.
In summary, we found increased activity of the brain cPLA2 and sPLA2, increased levels of brain non-esterified AA and eicosanoids PGE2 and PGD2, and increased incorporation and turnover of AA within certain brain phospholipids, following 6 days of intracerebral ventricular LPS infusion, consistent with altered brain fatty acid metabolism being a major participant in early neuroinflammation. Future studies of the effects of long-term LPS infusion on phospholipid metabolism and neuronal death, with and without treatment with NSAIDs or inhibitors of PLA2, may help to characterize the contribution altered AA metabolism plays in the progressive neuropathology in brain disorders associated with inflammation. Results from such studies could help to develop and evaluate different therapies to target specific pathways involved in altered brain phospholipid and AA metabolism.