Chronic valproate treatment decreases the in vivo turnover of arachidonic acid in brain phospholipids: a possible common effect of mood stabilizers


Address correspondence and reprint requests to Dr Michael C. Chang, Division of Comparative Medicine, National Center for Research Resources, NIH, 6705 Rockledge Drive, Suite 6050, MSC 7965, Bethesda, MD 20892–7965, USA. E-mail:


Both (Li+) and valproic acid (VPA) are effective in treating bipolar disorder, but the pathway by which either works, and whether it is common to both drugs, is not agreed upon. We recently reported, using an in vivo fatty acid model, that Li+ reduces the turnover rate of the second messenger arachidonic acid (AA) by 80% in brain phospholipids of the awake rat, without changing turnover rates of docosahexaenoic or palmitic acid. Reduced AA turnover was accompanied by down-regulation of gene expression and protein levels of an AA-specific cytosolic phospholipase A2 (cPLA2). To see if VPA had the same effect on AA turnover, we used our in vivo fatty acid model in rats chronically administered VPA (200 mg/kg, i.p. for 30 days). Like Li+, VPA treatment significantly decreased AA turnover within brain phospholipids (by 28–33%), although it had no effect on cPLA2 protein levels. Thus, both mood stabilizers, Li+ and VPA have a common action in reducing AA turnover in brain phospholipids, albeit by different mechanisms.

Abbreviations used
ed: AA

arachidonic acid


choline glycerophospholipids


ethanolamine glycerophospholipids


inhibiting myo-inositol monophosphatase




phosphate-buffered saline


PLA2, phospholipase A2






valproic acid.

The treatment of mania in patients with bipolar disorder has advanced significantly over the last decade. Recently, VPA and other anticonvulsant medications (carbamazepine) have been used successfully to treat bipolar disorder (Emilien et al. 1995; Guay 1995). VPA is now a Federal Drug Administration (FDA)-approved antibipolar drug that is equally effective, when compared with Li+, in stabilizing mood in people with bipolar disorder (Guay 1995). While VPA and Li+ appear to treat the same syndrome, it is not clear that they work through a common pathway. For example, it has been demonstrated that Li+ and VPA differ in terms of their effects on brain neurotransmitter systems. VPA increases GABA levels in various brain regions via inhibition of GABA catabolism and activation of GABA synthesis (Chapman et al. 1982; Shukla 1987), while Li+ has effects on multiple neurotransmitters, increasing serotonin and dopamine (Waldmeier 1987). Given that both of these drugs with such different effects on CNS neurotransmitters have been shown to be effective in treating both phases of bipolar illness, mania as well as depression, the question arises as to what common mechanism might underlie their efficacy (Emilien et al. 1995; Guay 1995).

At therapeutic relevant brain concentrations, we have used an in vivo fatty acid method and model in awake rats to show that chronic Li+ treatment selectively decreased the turnover in brain phospholipids of arachidonic acid (AA) (Chang et al. 1996, 1999) without changing turnover of docosahexaenoic or palmitic acid. The effect was accompanied by reduced enzyme activity (Chang and Jones 1998), protein level and mRNA expression of the AA-specific enzyme, cytosolic phospholipase A2 (cPLA2) (Leslie 1997; Rintala et al. 1999).

If the efficacy of Li+ against bipolar disorder is related to reduced turnover of the important second messenger, AA, we hypothesized that VPA would have the same effect, but not necessarily through the same mechanism. We therefore decided to quantify the effect of chronically administered VPA on AA turnover in rat brain phospholipids, using the in vivo method with which we demonstrated Li+'s effect (Chang et al. 1996, 1999). We now report that chronic VPA significantly decreased AA turnover brain phospholipids of awake rats, without reducing the cPLA2 protein level. Thus, AA turnover appears to be a common target for both drugs.

Materials and methods


[5,6,8,9,11,12,14,15-3H(N)]AA, specific activity 180–220 Ci/mmol) was purchased from Moravek Biochemicals (Brea, CA, USA). Its radiochemical purity was verified as > 96% by thin layer chromatography. Unlabeled AA-CoA and unesterified fatty acids were purchased from Sigma Chemical Co. (St Louis, MO, USA). Phospholipid and neutral lipid standards were obtained from Avanti (Birmingham, AL, USA). BCA protein assay kit and SuperSignal® West Pico chemiluminescent substrate were obtained from Pierce (Rockford, IL, USA). NuPage gels and nitrocellulose membranes were from Invitrogen (Carlsbad, CA, USA). Mouse monoclonal antibody (1 : 2000) against cPLA2 and goat anti-mouse IgG (H + L)-HRP conjugate were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and BioRad (Hercules, CA, USA), respectively. Ponceau-S was from Sigma.

Experimental procedures

Animals and surgical preparation

The study was conducted according to National Institutes of Health guidelines (Publication no. 80–23) and approved by the NICHD Animal Care and Use Committee. Awake male Fischer 344 rats, weighing 180–200 g (Charles River Breeding Lab; Wilmington, MA, USA) were used. They were acclimatized for 1 week in our animal facility, in which temperature, humidity and light cycle are controlled, and had free access to food and water. Sodium valproate in 0.9% saline was administered by intraperitoneal injection once in the morning to rats at a dose of 200 mg/kg for 30 days (Li et al. 1993; Baf et al. 1994). A control group received 0.9% saline by the same route.

Thirty days after the start of treatment, rats were anesthetized with 1–3% halothane, and polyethylene catheters were inserted into a femoral artery and vein, as described previously (Grange et al. 1995; Chang et al. 1996). The animals were allowed to recover from surgery for 3.5 h with their hindquarters loosely wrapped and tied to a wooden block. During recovery, body temperature was monitored using a rectal probe and maintained by means of a heating element (YSI Indicating Temperature Controller; Yellow Springs Instrument Co., Yellow Springs, OH, USA). Arterial blood gases and pH (Blood Gas Analyzer model 238; Ciba-Corning, Medfield, MA, USA) were also monitored. Arterial blood showed no significant difference in mean pH, PCO2 or PO2 between groups.

Tracer infusion and blood sampling

[3H]AA was evaporated under nitrogen and resuspended in physiological saline containing 3% (w/v) fatty acid-free bovine serum albumin (Sigma) to a final concentration of 1.25 mCi/mL. The infusion was designed to maintain a stable plasma unesterified [3H]AA concentration. Infusions were for 5 min, but a steady-state condition was reached within 30 s (Washizaki et al. 1994; Grange et al. 1995). At the end of infusion, the rat was killed with an intra-arterial injection of sodium pentobarbital (32.4 mg) and the head was subjected to focused-beam microwave irradiation (5.5 kW, 3.0 s) (Cober Electronics, Norwalk, CT, USA) to stop metabolic activity (Washizaki 1994; Grange et al. 1995; Chang et al. 1996).

Sample processing

After microwaving, the brain was excised from the skull and processed for lipid extraction or quantification of acyl-CoA. Lipids were extracted by the method of Folch et al. (1957). Butylated hydroxytoluene (0.01%), was added to lipid extracts to prevent oxidation of unsaturated fatty acids.

Chromatographic analysis

Determination of radiolabel in phospholipids, neutral lipids and unesterified fatty acids

Brain and plasma lipids were extracted according to the method of Folch et al. (1957). Lipids in the organic extracts of brain were separated by thin layer chromatography on silica gel plates and identified by cochromatography with unlabeled standards as previously described (Chang et al. 1996; Chang et al. 1999). Bands were visualized with primulin (Sigma) under ultraviolet light. Each band was removed and analyzed for [3H] by liquid scintillation counting to determine the amount of tracer in various lipid classes.

Quantification of unlabeled fatty acids in plasma and brain unesterified fatty acids and in brain phospholipids

Plasma samples (30 µL) were extracted by the method of Folch et al. (1957) following the addition of heptadecanoic acid. Free fatty acids were methylated with diazomethane and analyzed by gas chromatography (Chang et al. 1996; Chang et al. 1999). Plasma VPA levels were determined by a gas chromatography assay method that employed a Stabilwax-DA capillary column (Restek, Bellefonte, PA, USA) without need for prior derivation (Carlin and Simmons 1997).

Brain-free fatty acids were separated by thin layer chromatography and placed in a mixture of toluene : methanol : sulfuric acid (3 mL, 50 : 50 : 2 v/v/v). Reactants were incubated for 1 h at 65°C (Akesson et al. 1970). After cooling, 1 mL of water was added to quench the reaction. Fatty acid methyl esters were extracted using petroleum ether and analyzed by GC.

Lipids in the organic phase of Folch-extracted brain were separated by thin layer chromatography. Fatty acids in brain phospholipids were converted to methyl esters using sodium methoxide and analyzed by gas chromatography (Washizaki et al. 1994).

Quantification of labeled and unlabeled arachidonyl-CoA

Arachidonyl-CoA was isolated from rat brain using a method developed in this laboratory (Deutsch et al. 1994; Rabin et al. 1997). Briefly, acyl-CoAs were separated from unesterified fatty acids by specific adsorption on an oligonucleotide purification cartridge (Applied Biosystems, Foster City, CA, USA). After elution, the acyl-CoA species were separated by high pressure liquid chromatography, and arachidonoyl-CoA mass and associated radioactivity were used to calculate arachidonoyl-CoA specific activity.

Immunoblotting of cPLA2

For western blotting rat whole brain was homogenized at 4°C in six volumes of buffer containing 25 mm Tris, 150 mm NaCl, 5 mm EGTA, 2 mm 1,4-dithiothreitol (DTT), 10 µm phenylmethylsulfonyl fluoride (PMSF), 9.2 µm aprotinin, 0.11 mm leupeptin and 10.2 µm pepstatin A. Protein concentrations were determined by BCA protein assay (Pierce, Rockford, IL, USA). Homogenate protein (20 µg) was loaded on NuPage gel and transferred to nitrocellulose membrane according to manufacturers' instructions. Homogenous transfer of the samples was verified by Ponceau-S staining of nitrocellulose membrane. After blocking with 5% milk, 0.05% Tween 20 in phosphate-buffered saline (PBS, pH 7.4) for 1 h at room temperature, the membrane was incubated overnight at 4°C using mouse monoclonal antibody (1 : 2000) against cPLA2. Biotinylated rat-adsorbed secondary antibody (1 : 2000) and horseradish peroxidase–streptavidin (1 : 1000) were used in subsequent steps. The immunoblots were visualized by using chemiluminescence reaction. The specificity of the cPLA2 band was verified by incubating cPLA2 antibody with an excess of cPLA2 protein (1 : 1000) overnight at room temperature and immunostained as described above. Image analysis of the blots was performed on optical density-calibrated images captured with a video camera (NIH image 1.6) and the area under each peak representing the integrated optical intensity of the band, was taken as a signal.


Operational equations (Eqns 1-5) to examine fatty acid flux (JFA) and turnover within individual brain phospholipids are presented in the Appendix.


Data are presented as mean ± SD for n experiments. Statistical analysis was evaluated by a two-tailed t-test for unpaired observations using InStat version 2.03 (1991; GraphPad Software, San Diego, CA, USA). Differences at p < 0.05 were taken as statistically significant.


Plasma fatty acids

The plasma concentration of VPA, determined from blood samples taken prior to the start of tracer infusion and 3 h after VPA i.p. injection, was 31 ± 6.3 µg/mL (mean ± SD; n = 6) in VPA-treated rats. VPA was not detected in vehicle-treated rats (n = 5).

Constant plasma specific activity of tracer was achieved within 30 s after the start of [3H]AA infusion. AA-specific activity in plasma at 5 min was 327.6 ± 94.5 and 337.1 ± 60.8 nCi/nmol in vehicle and VPA-treated rats (mean ± SD; p > 0.8), respectively. More than 92% of the tracer delivered over the 5 min infusion remained unmetabolized (Washizaki et al. 1994), and neither vehicle nor VPA treatment modified the distribution of the tracer in the organic, aqueous or protein fraction of plasma (data not shown).

Plasma concentrations of the different unesterified fatty acid molecular species (Table 1) were determined from the average of arterial blood aliquots sampled just before the start and at the end of tracer infusion. There was no significant difference in these two values, indicating no effect of tracer infusion on unesterified free fatty acid levels. A statistically significant 20% mean plasma concentration decrease was observed for AA in VPA-treated rats when compared with controls, suggesting a selective effect of VPA on whole body AA metabolism. No significant difference was seen in other plasma unesterified fatty acid levels or in total mass of plasma unesterified fatty acids between VPA and vehicle rats.

Table 1.  Effect of chronic valproate on plasma unesterified fatty acids and brain acyl-CoA concentrations in awake rats
 Plasma unesterified fatty acid
Brain unesterified fatty acid
Brain acyl-CoA
Fatty acidVehicle (n = 6)Valproate† (n = 7)Vehicle (n = 8)Valproate† (n = 7)Vehicle (n = 6)Valproate† (n = 7)
  1. Data are means ± SD. Statistical significance was determined by unpaired t-statistics. *p < 0.05 mean different from vehicle mean. N.D., not detected and n.d., not determined. †Rats were injected (i.p.) daily for 30 days with sodium valproate (200 mg/kg).

Palmitate (16 : 0)148.8 ± 22.9135.9 ± 15.021.1 ± 20.729.3 ± 22.110.9 ± 2.112.2 ± 2.0
Stearate (18 : 0)42.4 ± 2.946.0 ± 3.248.6 ± 37.249.4 ± 27.1N.D.N.D.
Oleate (18 : 1 n9)99.8 ± 16.1111.6 ± 9.3n.d.n.d.14.8 ± 2.616.6 ± 4.1
Linoleate (18 : 2 n6)118.2 ± 18.0130.1 ± 22.62.1 ± 1.42.7 ± 0.90.8 ± 0.20.9 ± 0.2
Linolenate (18 : 3 n3)11.1 ± 2.311.9 ± 2.5N.D.N.D.N.D.N.D.
Arachidonate (20 : 4 n6)26.1 ± 4.620.9 ± 1.2*5.6 ± 2.47.9 ± 6.71.7 ± 0.31.8 ± 0.2
Docosahexaenoate (22 : 6 n3)15.5 ± 2.313.8 ± 1.80.6 ± 0.71.4 ± 0.41.5 ± 0.41.5 ± 0.3
Σn − 6/n − 35.4 ± 0.75.9 ± 0.57.9 ± 4.610.0 ± 6.01.7 ± 0.31.9 ± 0.4

Brain unesterified fatty acid and acyl-CoA pools

Brain unesterified fatty acid concentrations are similar to those values previously reported (Contreras et al. 2000). However, unesterified oleic acid levels could not be determined due to considerable amount of peak splitting in the chromatogram at the retention time for methyl oleate. Brain unesterified fatty acid concentrations for palmitate, AA and docosahexaenoate were 21.1 ± 20.7, 5.6 ± 2.4 and 0.6 ± 0.7 nmol/g (mean ± SD; n = 7) in vehicle-treated rats and were not significantly altered by chronic VPA treatment (Table 1).

Mean concentrations of the different molecular species of brain acyl-CoA from brains of vehicle and VPA rats are also summarized in Table 1. In both groups, the concentrations of the saturated acyl-CoA species, palmitoyl-CoA and oleoyl-CoA, were higher than for the polyunsaturated molecular species. Brain concentrations of arachidonoyl-CoA averaged 1.7 ± 0.3 and 1.8 ± 0.2 nmol/g in vehicle and VPA treated animals, respectively. VPA treatment did not affect the size of the arachidonyl-CoA pool nor the proportion of other molecular acyl-CoA species. The mean concentration of the acyl-CoA pool did not differ significantly between the VPA and vehicle rats (29.6 ± 4.8 nmol/g compared with 33.1 ± 6.3 nmol/g).

Steady-state brain arachidonoyl-CoA specific activity in vehicle-treated rats relative to plasma (λ in Equation 2) was < 4% demonstrating a marked dilution of plasma-derived AA (96%) in the acyl-CoA pool (Purdon et al. 1997). The mean values for λ for AA, determined using Equation 2 (Appendix), significantly increased from 0.035 ± 0.006 in vehicle rats to 0.054 ± 0.008 in VPA rats. This observation suggests that VPA reduced recycling of AA (de-esterification and re-esterification).

[3H]AA incorporation into brain phospholipids

Most brain 3H radioactivity following 5 min of infusion of [3H]AA was found in the organic (lipid) compartment, which accounted for 91.8% ± 2.9% and 92.6% ± 3.0% of non-volatile radioactivity in vehicle and VPA-treated rats, respectively. In the brain organic fraction, phospholipids were the most labeled (84–86%), particularly choline glycerophospholipids (ChoGpl) and phosphatidylinositol (PtdIns) and to a lesser extent in ethanolamine glycerophospholipids (EtnGpl) and phosphatidylserine (PtdSer). Unidirectional incorporation coefficients (inline image) into phospholipids (Appendix, Equation 3) are summarized in Table 2. Chronic VPA treatment significantly increased k* for total phospholipid by 30% when compared to vehicle. Indeed, VPA treatment significantly increased k* for each phospholipid class, with the greatest change occurring in EtnGpl (38%) (Table 2).

Table 2.  Effect of chronic valproate on brain arachidonate incorporation coefficients (k*) and net incorporation rates from plasma unesterified arachidonate (J1) and brain arachidonyl-CoA pool (JFA)
 k* (ml/g/s × 105)J1 (nmol/g/s × 104)JFA (nmol/g/s × 102)
Brain compartmentVehicle (n = 6)Valproate (n = 7)Vehicle (n = 6)Valproate (n = 7)Vehicle (n = 6)Valproate (n = 7)
  1. Data are mean ± SD values. PtdIns, phosphatidylinositol; ChoGpl, choline glycerophospholipids; EtnGpl, ethanolamine glycerophospholipids; PtdSer, phosphatidylserine. Statistical significance was determined by unpaired t-statistics. *p < 0.05 and **p < 0.01 mean different from control mean.

Total phospholipids27.9 ± 2.936.1 ± 5.8*72.9 ± 15.475.5 ± 12.721.5 ± 7.314.4 ± 3.5*
PtdIns10.0 ± 1.112.9 ± 2.3*26.1 ± 5.226.9 ± 5.27.7 ± 2.55.1 ± 1.4*
ChoGpl12.4 ± 1.116.1 ± 2.5**32.4 ± 7.433.6 ± 5.39.6 ± 3.56.4 ± 1.5*
EtnGpl2.3 ± 0.33.1 ± 0.5**5.9 ± 1.06.5 ± 1.11.7 ± 0.51.2 ± 0.3*
PtdSer1.7 ± 0.32.3 ± 0.3**4.5 ± 1.04.7 ± 0.81.3 ± 0.40.9 ± 0.2

Incorporation rates and turnover

Calculated rates of unlabeled AA incorporation into brain phospholipids i from plasma (J1,i), and from brain arachidonoyl-CoA (JFA,i), are presented in Table 2. Values of JFA (Appendix, Equation 1) exceeded the flux from unesterified AA in plasma (J1) by 29-fold and 19-fold in vehicle and VPA rats, respectively. Calculated net flux of AA from brain arachidonoyl-CoA into phospholipids was reduced significantly by 33%, from 21.5 ± 7.3 nmol/g/s −14.4 ± 3.5 nmol/g/s of brain, in VPA rats compared with vehicle controls. The change reflected the net fluxes of AA incorporation into individual phospholipids, which were significantly decreased (28–33%) by chronic VPA treatment.

Table 3 summarizes concentrations of unlabeled brain fatty acids in total phospholipids and in each of the four main classes for vehicle and VPA rats. Brain esterified AA concentrations for total and individual phospholipid classes were not significantly changed by VPA treatment. In addition, chronic VPA did not significantly alter concentrations of other unlabeled fatty acid molecular species in total and individual brain phospholipids.

Table 3.  Fatty acid concentrations (nmol/g) of brain phospholipid classes from vehicle- and chronic valproate-treated rats

Fatty acid
(n = 6)
(n = 7)
(n = 6)
(n = 7)
(n = 6)
(n = 7)
(n = 6)
(n = 7)
  1. Values represent means ± SD. PtdIns, phosphatidylinositol; ChoGpl, choline glycerophospholipids; EtnGpl, ethanolamine glycerophospholipids; PtdSer, phosphatidylserine. Statistical significance was determined by unpaired t-statistics.

16 : 0779 ± 87861 ± 14615994 ± 152413737 ± 13645323 ± 5475544 ± 543408 ± 119397 ± 104
18 : 02381 ± 1212030 ± 3425360 ± 4684549 ± 32316522 ± 151116946 ± 21124884 ± 6734094 ± 874
18 : 1659 ± 299689 ± 2408244 ± 7936745 ± 53714764 ± 257216023 ± 26092371 ± 5021928 ± 605
20 : 41952 ± 6882085 ± 4932344 ± 2552244 ± 36013995 ± 408714820 ± 4364559 ± 219539 ± 129
22 : 6169 ± 42157 ± 431679 ± 1481456 ± 14618160 ± 252418397 ± 17582403 ± 3072281 ± 333

Calculated fractional turnover rates (Fi) (Appendix, Equation 4) of AA in brain phospholipids are summarized in Table 4. Chronic VPA significantly decreased the turnover rate of AA in total phospholipid from 14.1 ± 3.7%/h −9.3 ± 2.6%/h. This reflected the fact that the turnover rates for PtdIns, ChoGpl, EtnGpl and PtdSer were decreased significantly by 33%, 33%, 28% and 32%, respectively, in VPA-treated rats compared with vehicle controls.

Table 4.  Effect of chronic valproic acid on arachidonate turnover in brain phospholipids
  Turnover (%/h)
 λFA-CoATotal phospholipidsPtdInsChoGplEtnGplPtdSer
  1. Data are mean ± SD. PtdIns, phosphatidylinositol; ChoGpl, choline glycerophospholipids; EtnGpl, ethanolamine glycerophospholipids; PtdSer, phosphatidylserine. Statistical significance was determined by unpaired t-statistics. *p < 0.05 and **p < 0.01 mean different from control mean.

Vehicle (n = 6)0.035 ± 0.00614.1 ± 3.713.4 ± 2.815.5 ± 4.40.5 ± 0.212.1 ± 3.6
Valproate (n = 7)0.054 ± 0.008**9.3 ± 2.6*7.9 ± 2.0*9.9 ± 2.4*0.3 ± 0.1*7.3 ± 1.6*

Immunoblot for cPLA2 protein in rat brains are presented in Fig. 1. Image analysis of blots indicated that the brain cPLA2 protein level was not significantly changed by chronic VPA treatment.

Figure 1.

Autoradiogram of immunoblot (a) and levels of cPLA2 (b) protein in brains of control and chronic valproate-treated rats. Values represent means ± SD (n = 4 in both groups).


The therapeutic modes of action of mood stabilizing drugs are not agreed upon. Li+ has been demonstrated to alter the turnover of phosphoinositide and reduce brain myoinositol (Allison and Stewart 1971; Hallcher and Sherman 1980; Sherman et al. 1981) by inhibiting myo-inositol monophosphatase (IMPase), implicating this enzyme as its therapeutic target of action in bipolar disorder (Hallcher and Sherman 1980; Sherman et al. 1981). Other drugs used in the treatment of bipolar disorder, VPA and carbamazepine, have been reported to have diverse effects on IMPase activity, suggesting that they may not work through a pathway common to that of Li+ (Vadal and Parthasarathy 1995). In addition, using proton magnetic resonance spectroscopy, Moore et al. (1999) reported a reduction of brain myo-inositol in bipolar patients after 5 days of Li+ treatment, but several weeks elapsed before a clinical response was evident. Taken together, these studies do not support the hypothesis that the critical therapeutic effect of Li+ is to reduce phosphoinositide signaling and deplete brain myo-inositol.

There also is evidence from studies in the awake rat that Li+ attenuates brain AA signaling while reducing expression of an AA-specific cPLA2 (Chang et al. 1996, 1999; Chang and Jones 1998; Rintala et al. 1999). Our finding that another mood stabilizer, VPA, also decreases AA turnover in brain phospholipids suggests the final common action of both drugs, perhaps accounting for the therapeutic efficacy, and possibly of other mood stabilizers, is to decrease AA turnover and its second messenger action in the central nervous system.

The Li+ effect on turnover is associated with down-regulation of the AA-specific cPLA2, but this did not occur with VPA. On the other hand, VPA may reduce AA turnover by being directly incorporated into brain phospholipids. Although Aly and Abdel-Latif (1980) did not detect incorporation of [3H]VPA in brain phospholipids 30 min after its i.p. injection, tracer incorporation was detected in phospholipids and neutral lipids in GT1–7 neurons (Siafaka-Kapadai et al. 1998). If VPA is incorporated into membrane lipids, it may disorder the lipid bilayer (Perlman and Goldstein 1984; Siafaka-Kapadai et al. 1998) and inhibit binding of cPLA2 or other phospholipases to their substrates, thereby reducing AA hydrolysis. Such incorporation might also explain VPA's ability to reduce AA metabolism in rat brain microvessels and platelets (Szupera et al. 2000). Clearly, this issue has to be further explored.

Consistent with our evidence that VPA reduced AA turnover by some 35% are reports that VPA can inhibit the synthesis of prostaglandins, leukotrienes and epoxides (Horrobin 1978; Kis et al. 1999; Szupera et al. 2000).). VPA has been shown to inhibit the metabolic conversion of AA to 12-hydroxyheptadecatrienoic acid, prostaglandin D2, prostaglandin E2, prostaglandin F, 6-keto prostaglandin F and thromboxane B2 COX and lipoxygenase products, in platelets and brain microvessels (Kis et al. 1999; Szupera et al. 2000). Based on the prominent role of these bioactive products in inflammation, it is not surprising that VPA treatment in rats reduces paw edema (Raza et al. 1996).

The clinical efficacy of VPA in both bipolar disorder and epilepsy has been ascribed to enhanced central GABAergic neurotransmission (Emilien et al. 1995; Guay 1995), but its effects on brain membranes also may be relevant to its pharmacological action. VPA is reported to inhibit degradation of phosphatidylcholine in neuroblastoma cells, inhibit sterol and phospholipid synthesis and decrease membrane order in neurons (Roberti et al. 1989; Vorhees et al. 1991; Bolaños and Medina 1993, 1994). A decrease by VPA in the ratio of phosphatidylserine/phosphatidylethanolamine synthesis and in membrane order may contribute to its anticonvulsive action by increasing the threshold for the firing state in neurons of epileptic patients (Bolaños and Medina 1993, 1994, 1997). Our studies now suggest another mechanism of action of chronic VPA, inhibition of AA recycling and its consequences.

A reduced brain level of unesterified AA was not evident in VPA-treated rats, despite 35% decreases in the turnover rates of AA in several brain phospholipids. This lack of a significant difference between means may have been due to difficulty in measuring unesterified brain fatty acids, evident by the large standard deviations of many of the means (Table 1). Nevertheless, it is possible that a reduced AA turnover would not be accompanied by a reduction in unesterified brain AA concentration after a steady-state has been established, if the rate of loss of AA from its compartment declined in proportion to the reduced rate of entry associated with reduced turnover (Chang et al. unpublished calculations). On the other hand, it is evident in Table 1 that VPA treatment produced a selective and statistically significant reduction in the plasma unesterified concentration of AA (where the mean had a small variance) compared with the other fatty acids, suggesting a widespread effect on AA metabolism within the body.

In conclusion, chronic VPA significantly decreased the net rates of incorporation and turnover of AA in brain phospholipids of the awake rat. Consistent with the fact that chronic Li+ selectively decreases the turnover of AA within brain phospholipids by up to 80% (Chang et al. 1996), these results support the hypothesis that both drugs are effective in bipolar disorder because of this common effect on AA.


Brain radioactivity (nCi/g brain) for each lipid was corrected for residual blood radioactivity by subtracting the product of cerebral blood volume (2.0 × 10−2 mL/g) and the corresponding blood 3H concentration in the lipid (nCi/mL). Blood samples at time of death were extracted and chromatographed to make the vascular correction.

Incorporation of AA into brain phospholipids and neutral lipids takes place from the rapidly turning-over brain acyl-CoA precursor pool, not directly from plasma. The net rate of incorporation (JFA) of the unlabeled fatty acid into brain phospholipid and other stable brain compartments, from the precursor pool, is calculated using the model of Robinson et al. (1992) as,


where inline image is concentration (nCi/g) of labeled fatty acid in stable compartment i at the end of experiment (t = T), inline image and inline image represent conctenrations (nmol/mL) of labeled and unlabeled unesterified fatty acid in plasma, respectively. A dilution factor, λ, is determined experimentally as the steady state ratio specific activity of the brain acyl-CoA pool, to that of plasma unesterified fatty acid during infusion of the tracer.


Unidirectional incorporation coefficients (inline image) into compartment i for [3H]AA were calculated as follows:


where inline image is in units of mL/s/g, inline image (in units of nCi/g) is brain radioactivity of i at T = 5 min.

Rates of unlabeled fatty acid incorporation into individual phospholipid classes from plasma unesterified AA (J1,i) and from brain arachidonyl-CoA (JFA,i) were estimated as inline image and inline image (Equation 4), respectively (Washizaki et al. 1994). By definition inline image, inline image, and inline image for total phospholipids.

Based on the incorporation rates, and the concentrations of AA esterified to phospholipids, fractional turnover rates (Fi,%/h) of AA within various phospholipids i were calculated as follows:


where unlabeled cbr,i is the amount of AA (in nmol) in a given phospholipid class i per gram of brain. This calculation assumes that each phospholipid turns over as a single homogenous pool.