A Novel Long Chain Polyunsaturated Fatty Acid, β-Oxa 21:3n-3, Inhibits T Lymphocyte Proliferation, Cytokine Production, Delayed-Type Hypersensitivity, and Carrageenan-Induced Paw Reaction and Selectively Targets Intracellular Signals¹

The n-3 polyunsaturated fatty acids (PUFA),⁷ eicosapentaenoic acid (20:5n-3) and docosahexaenoic acid (22:6n-3), have been demonstrated to suppress the inflammatory reaction in vivo (1), and depress in vitro leukocyte responses conducive with this property (2). However, they are also known to stimulate the neutrophil respiratory burst, a response that is amplified in the presence of proinflammatory cytokines (3). This property will limit their anti-inflammatory activities, and may in part explain some of the setbacks experienced with the applications of n-3 PUFA as anti-inflammatory agents. Our recent work has shown that the ability of PUFA to induce the respiratory burst is governed by specific structural elements, including chain length, number of double bonds, and series (n-3 vs n-6) (4, 5, 6). For example, by converting PUFA to the hydroperoxy derivative (OOH), a more selective effect on leukocyte function is achieved. The hydroperoxy PUFA are unable to stimulate the neutrophil respiratory burst, yet retain the ability to depress macrophage (4) and T lymphocyte cytokine production, and endothelial cell function (7). The OOH group is strongly polar, and oxygen has a larger atomic radius than either carbon or hydrogen. Thus, through either the increased polarity and/or altered conformation, the hydroperoxy derivative is endowed with selective biological activity. Although the hydroperoxy derivative displays more selective cellular actions than the parent PUFA, the OOH group is chemically unstable and can be readily converted to the less active hydroxy derivative either spontaneously or enzymatically (4, 7). The PUFA concentrations that are effective at evoking reactions are usually in the 5–30 μM range both in relation to their inhibitory and stimulatory actions (3). Interestingly, the hydroperoxy derivatives not only lack the neutrophil-stimulatory activity, but their ability to inhibit in vitro parameters of inflammation occurs at reduced concentrations of 0.2–5 μM (3). Based on these results (4, 7), a group of PUFA with an oxygen atom in the β position was synthesized (8). Because these compounds are devoid of the 1,4-cis-pentadiene structural units, structural units that are required for conversion by, for example, the 5-lipoxygenase or cyclooxygenase (COX), increased stability occurs (9).

We now demonstrate that β-oxa 21:3n-3, like the hydroperoxy PUFA, does not significantly activate the neutrophil respiratory burst, yet retains the ability to inhibit lymphocyte function measured both in vitro and in vivo. Its metabolism via the COX and lipoxygenase pathways was not required for it to mediate its actions. The mechanism of action of this molecule was highly specific, with the translocation of protein kinase C (PKC)-βI and PKC-ε and the activation of the extracellular signal-regulated protein kinase (ERK) module significantly inhibited, whereas activation of the c-Jun NH₂-terminal kinase (JNK) and p38 modules was not affected.

β-Oxa 21:3n-3 ((Z,Z,Z)-(octadeca-9,12,15-trienyloxy) acetic acid) was synthesized, as described previously (8, 9). Fatty acid methyl esters were synthesized, as described previously (10). The 22:6n-3 (Sigma, St. Louis, MO) and all fatty acid stocks (20 mM in chloroform or ethanol) were stored at −20°C. Fatty acid purity was determined using ¹H and ¹³C nuclear magnetic resonance spectroscopy, mass spectrometry, infrared spectroscopy, and microanalysis, as described previously (8). The purity was determined using these techniques, HPLC and TLC. The purity and absence of auto-oxidation products were determined at regular intervals using mass spectrometry and TLC.

Human leukocytes were isolated from the peripheral blood of healthy volunteers by a rapid single-step method, as described previously (11). The neutrophil preparation was >99% viable, as judged by their ability to exclude trypan blue. T lymphocytes were purified, as described previously (12). The T lymphocyte preparation consisted of >98% CD3⁺ cells, as determined by FACScan analysis, and viability was >99%, as determined by trypan blue dye exclusion.

Fatty acids were presented to cells in the form of micelles in one of two ways. On the day of use, fatty acids were prepared in the presence of DL-α-dipalmitoyl phosphatidylcholine (DPC; Sigma) as the carrier using a weight ratio of 4:1 of DPC to fatty acid, as described (4). This procedure gave rise to a clear solution that was added directly to the cells. Control cells received an equivalent amount of DPC. In some experiments, fatty acids were presented to cells in ethanol. In these experiments, fatty acids were diluted in sterile water to form self micelles and then diluted in HBSS. The final concentration of ethanol was 0.01–0.1% (v/v). Control cells received an equivalent amount of ethanol. The fatty acids were used immediately following preparation. Fifty microliters containing 2 × 10⁵ T lymphocytes were incubated with 50 μl fatty acid (0.5–30 μM) in 96-well U-bottom plates (Linbro; Flow Laboratories, McLean, VA) for the indicated periods of time. Cells were then stimulated with a variety of agonists for 48 or 72 h (see below).

Purified T lymphocytes were stimulated with either PHA (2 μg/ml; Murex Diagnostics, Dartford, U.K.) and PMA (10 ng/ml; Sigma) for 48 h, PMA (1 ng/ml) and A23187 (0.1 μM; Sigma) for 72 h diluted in RPMI 1640 containing 5% heat-inactivated blood group AB serum, or anti-CD3 Ab (1/500 dilution; CLB, Amsterdam, The Netherlands) and anti-CD28 Ab (25 ng/ml; Immunotech, Brea, CA) for 48 h diluted in Iscove’s medium (Sigma) containing 5% heat-inactivated blood group AB serum, in a humid atmosphere of 5% CO₂ in air at 37°C. Six hours before harvesting, 1 μCi [methyl-³H]thymidine (Amersham Life Sciences, Arlington Heights, IL) diluted in RPMI 1640 (5% AB serum) (PHA-PMA and PMA-A23187) or diluted in Iscove’s medium (5% AB serum) (anti-CD3-CD28 Abs) was added to the cultures. The supernatant was removed for cytokine estimation, the cells were harvested, and the incorporated radioactivity was measured using a Wallac liquid scintillation beta counter (Wallac 1409; Turku, Finland).

TNF-β, IFN-γ, and IL-2 levels were determined by ELISA, as described previously (13). Briefly, immobilized goat anti-mouse IgG (Cappel, Aurora, OH) was used to capture an anti-TNF-β, anti-IFN-γ, or anti-IL-2 mAb (Boehringer Mannheim, Indianapolis, IN). After addition of the supernatants, the wells were incubated with polyclonal rabbit anti-TNF-β, anti-IFN-γ (Boehringer Mannheim), or anti-IL-2 (Endogen, Woburn, MA) Ab. Detection was achieved using a HRP-conjugated goat anti-rabbit IgG Ab (BioSource International, Camarillo, CA), using hydrogen peroxide as the substrate and 2,2′-azino-di[3-ethylbenzthiazoline sulfate] (Boehringer Mannheim) as the chromogen.

Superoxide production by human neutrophils in response to fatty acid treatment was measured by the reduction of the probe lucigenin (N,N′-dimethyl-9,9′-biacridinium dinitrate; Sigma), as described previously (14). This method provides a direct and specific measure of agonist-induced superoxide production (15).

PKC translocation was determined, as described previously (16). Briefly, T lymphocytes (1 × 10⁷; 1 × 10⁶/ml) were treated for 30 min with 20 μM β-oxa 21:3n-3, 20 μM 22:6n-3, or an equivalent amount of DPC, and then stimulated with PHA (2 μg/ml) and PMA (10 ng/ml) for 5 min. The cells were harvested and sonicated (3 × 10 s) (Ystrom systems, setting 3), and particulate fractions were extracted by sonication with 2% Triton X-100. PKC activity was assayed as described (16), and the activity was expressed as Ca²⁺/phosphatidylserine-dependent histone phosphorylation/minute.

PKC translocation was performed as described previously (17). Briefly, after extracting PKC with 2% Triton X-100, 100 μg denatured protein was separated by 12% SDS-PAGE transferred to nitrocellulose (Schleicher & Schuell, Keene, NH), and PKC isozymes were detected using isozyme-specific Abs (PKCα (C-20); βI (C-16); βII (C-18); ε (C-15); θ (C-18); Santa Cruz Biotechnology, Santa Cruz, CA), and detected by ECL, according to the manufacturer’s instructions.

ERK activity was assayed as described previously (18). Briefly, T lymphocytes (1 × 10⁷; 1 × 10⁶/ml) were pretreated with β-oxa 21:3n-3 (20 μM) or an equivalent amount of DPC and stimulated with PHA (2 μg/ml) and PMA (10 ng/ml) for 30 min. The cells were sonicated and centrifuged, and the supernatant was adsorbed onto phenyl-Sepharose CL4B (Pharmacia, Piscataway, NJ). ERK was batch eluted, and ERK activity was assayed by measuring the amount of ³²P incorporated into myelin basic protein (18). The kinase activity in fractions prepared in this manner has previously been demonstrated to be indistinguishable from that obtained using immunoprecipitated ERK (18), and was almost undetectable in samples prepared from cells that had been pretreated with the mitogen-activated protein/ERK kinase (MEK) inhibitor, PD98059 (18).

A solid-phase assay was used to assay JNK activity, as described previously (19). Briefly, T lymphocytes (1 × 10⁷; 1 × 10⁶/ml) were pretreated with β-oxa 21:3n-3 (20 μM) or an equivalent amount of DPC and then stimulated with PMA (10 ng/ml) and A23187 (1 μM) for 30 min. Cells were lysed, and the lysate protein was added to 25 μl GST-jun 1–79(1–79) coupled to glutathione-Sepharose beads, supplemented with 15 mM MgCl₂ and 10 μM ATP. The samples were incubated for 2 h at 4°C with gentle rocking and were resolved by 12% SDS-PAGE, and incorporated radioactivity was detected and quantitated using an Instant Imager (Packard Instruments, Meriden, CT).

p38 activity was measured as described previously (19). Briefly, T lymphocytes (1 × 10⁷; 1 × 10⁶/ml) were treated with β-oxa 21:3n-3 (20 μM) or an equivalent amount of DPC and stimulated as described for JNK. T cells were lysed, and the lysate was precleared with protein A-Sepharose. Anti-p38 Ab (Santa Cruz Biotechnology) was added, and tubes were gently rocked for 90 min at 4°C. The Ag-Ab complexes were precipitated by the addition of protein A-Sepharose (20 μl/sample) and then washed. p38 activity was determined by measuring the incorporation of ³²P into myelin basic protein (19). Phosphorylated myelin basic protein was resolved by 16% SDS-PAGE and detected, and the amount of incorporated radioactivity was quantitated as described above.

The DTH response was induced in 12-wk-old female BALB/c mice (Animal Resource Center, Perth, Australia), as described previously (20). Briefly, mice were injected with SRBCs (100 μl of 10% hematocrit; Sigma). Six days later, mice were treated with β-oxa 21:3n-3, 22:6n-3 (50 mg/kg), indomethacin (IM, 30 mg/kg; Sigma), or an equivalent amount of DMSO carrier via i.p. injection 1 h before being injected intradermally in the right hind footpad with SRBC (25 μl of 40% hematocrit) or into the left footpad with diluent (25 μl). The DTH response was determined 24 h postchallenge, and was calculated by comparing the thickness between the diluent vs SRBC-injected footpads.

Carrageenan-induced paw edema was induced, as described previously (21). Briefly, mice were administered β-oxa 21:3n-3 (5–100 mg/kg), 22:6n-3 (100 mg/kg), or prednisolone (20 mg/kg, Sigma) i.p. 1 h before inoculation with type IV carrageenan (1 ml/kg of a 1% solution; Sigma) into the right hind paw. Edema was assessed by measuring hind paw thickness at intervals during a 24-h period following carrageenan administration.

Liver and kidney biochemical parameters and protein levels were determined using standard enzyme chemistry assays (Synchron Clinical System CX5CE; Beckman Coulter, Fullerton, CA).

In vitro experiments were conducted in sextuplicate using cells from at least three different donors. Statistical significance was evaluated using a two-tailed unpaired Student’s t test or ANOVA. A value of p < 0.05 was considered significant.

The structure of β-oxa 21:3n-3 in comparison with 22:6n-3 is shown in Fig. 1,A. The chemical structure shows the oxygen atom inserted in the β position, which is the site at which the enzymes of the β-oxidation pathway attack and oxidize natural fatty acid molecules. Although β-oxa 21:3n-3 has been demonstrated not to be β oxidized (9), β-oxa 21:3n-3 can be metabolized via ω-oxidation. Unlike the hydroperoxy PUFA, the change in the chemically synthesized β-oxa PUFA is in the carboxyl end of the molecule. Like the hydroperoxy PUFA, the β-oxa PUFA, β-oxa 21:3n-3 was found to be a poor stimulator of the neutrophil respiratory burst compared with 22:6n-3 over a broad concentration range (Fig. 1 B).

When T lymphocytes were pretreated for 24 h with β-oxa 21:3n-3 before being stimulated with PHA-PMA (stimulation index; 166 ± 41, n = 7), there was marked inhibition of the lymphoproliferative response (p < 0.001). The data in Fig. 2,A show that the IC₅₀ for β-oxa 21:3n-3 was 1.9 μM compared with an IC₅₀ of 5.2 μM for 22:6n-3. This implies that β-oxa 21:3n-3 is more effective at inhibiting T lymphocyte proliferation than 22:6n-3. β-oxa 21:3n-3 was also quite effective in inhibiting cytokine production (Fig. 2,B). The IC₅₀ for TNF-β, IFN-γ, and IL-2 production was 3, 4, and 4 μM, respectively. This biological effect of the fatty acid was not due to toxicity, as the viability of the T lymphocytes was not affected after this incubation period, as judged by their ability to exclude trypan blue. A 30-min preincubation time with β-oxa 21:3n-3 or 22:6n-3 fatty acid was also effective at suppressing lymphocyte proliferation. Thus, preincubation of human T lymphocytes with β-oxa 21:3n-3 (0.5–30 μM) for 30 min reduced the PHA-PMA-induced lymphocyte proliferation and cytokine production in a concentration-dependent manner (Fig. 3 and data not shown). Significant inhibition was obtained at concentrations ≥ 7.5 μM. With the shorter pretreatment times, β-oxa 21:3n-3 inhibited lymphocyte proliferation, TNF-β, IFN-γ, and IL-2 production with IC₅₀ values of 16, 15, 15, and 18 μM, respectively. Although little or no inhibition was observed when the fatty acid was added together with PHA-PMA, significant inhibition of lymphocyte proliferation and cytokine production was apparent after 10- to 20-min preincubation (data not shown).

When T lymphocytes were stimulated with human anti-CD3 and anti-CD28 mAbs, β-oxa 21:3n-3 was found to be a strong inhibitor of lymphocyte proliferation (82 ± 5.1% inhibition, p < 0.001). Therefore, the data demonstrate that the inhibitory effects of the fatty acid are not specific to one particular means of T cell activation.

To further expand our knowledge of the relationship between structure and biological activity, the terminal carboxyl group of β-oxa 21:3n-3 was altered to yield a methyl ester. The carboxyl group was found to be critical for inhibition of lymphocyte proliferation and cytokine production by β-oxa 21:3n-3. The methyl ester was significantly less inhibitory than the parent fatty acid (data not shown). This reduced activity may be due to the carboxyl group being important or required for binding to plasma membrane fatty acid-binding proteins and subsequent transport into the cell, and conversion into its acyl CoA derivative.

The DTH response (20) was used as a model of a T cell-mediated inflammatory process (22). Mice were sensitized with SRBC, and 5 days later rechallenged with the Ag. One hour before challenge, the animals received one injection of β-oxa 21:3n-3 or 22:6n-3 i.p., and the inflammatory reaction was examined 24 h after antigenic challenge. Although β-oxa 21:3n-3 was found to significantly inhibit the DTH response, 22:6n-3 only suppressed the response slightly (Fig. 4,A). In fact, β-oxa 21:3n-3 was significantly more effective than IM. In addition, the effect of β-oxa 21:3n-3 and 22:6n-3 on carrageenan-induced paw edema was examined (21). Mice were treated with β-oxa 21:3n-3, 22:6n-3, or prednisolone 1 h before inoculation with a 1% solution of carrageenan, and the reaction was examined 3–24 h after induction of inflammation. β-oxa 21:3n-3 significantly inhibited carrageenan-induced paw edema in a dose-dependent manner (Fig. 4 B). In contrast, 22:6n-3 only inhibited the inflammatory response slightly over the 24-h period. β-Oxa 21:3n-3 was nearly as effective as prednisolone, a known inhibitor in this model (23). Although the time course of the effects of prednisolone and the PUFA was different, with prednisolone acting early (3 h) and decreasing after 6 h, the effects of β-oxa 21:3n-3 were dose dependently increased and were sustained over 24 h.

As β-oxa 21:3n-3 is an engineered molecule, it was important to determine that any in vivo inhibitory effects were not due to toxicity. To assess this possibility, rats were administered 100 mg/kg/day β-oxa 21:3n-3 for 4 days by gavage. After 4 days, blood was removed and biochemical parameters associated with liver and kidney function were measured (Table I). All parameters were found to be within normal ranges (24), indicating no liver or kidney damage. In addition, the animals did not display any external signs of discomfort during the 4-day period.

From the above studies, it is evident that β-oxa 21:3n-3 was a strong inhibitor of T lymphocyte responses both in vitro and in vivo. We next examined the possible mechanism by which the β-oxa PUFA may have mediated its inhibitory activity. Alterations in the expression of cell surface molecules involved in lymphocyte activation could result in reduced lymphocyte responses. To examine this possibility, T lymphocytes were treated with β-oxa 21:3n-3 (20 μM) for 30 min, and the expression of CD3, CD4, and CD8 was determined by FACScan analysis. The β-oxa fatty acid did not significantly alter the cell surface receptor expression of CD3 (94 ± 2.5, percentage of control ± SEM), CD4 (91.6 ± 0.3), or CD8 (99.25 ± 0.24) compared with control cells. This suggests that processes postreceptor binding were likely to have been affected by the fatty acid. Evidence that the inhibitory effects were mediated at a postreceptor level was examined using the agonists PMA and A23187. These agents bypass cell surface receptors and stimulate T cells by activating PKC and Ca²⁺/calmodulin-dependent effectors (25, 26). β-Oxa 21:3n-3 was found to significantly inhibit lymphocyte proliferation, TNF-β, IFN-γ, and IL-2 production (Fig. 5) in response to PMA-A23187 (stimulation index, 43.6 ± 11.9, n = 3). These results support the conclusion that modulation of cell surface receptors was not a mechanism involved in the inhibition of lymphocyte function by β-oxa 21:3n-3.

PHA-PMA, anti-human CD3-CD28 mAb, and PMA-A23187-induced proliferation of T lymphocytes requires activation of PKC (25, 26). We thus examined whether β-oxa 21:3n-3 affected PKC activation by determining its translocation to the particulate fraction. Pretreatment of T lymphocytes with β-oxa 21:3n-3 (20 μM) for 30 min, followed by stimulation with PHA-PMA was found to significantly reduce total PKC translocation (data not shown). Several PKC isozymes have been demonstrated to regulate the activity of the MAP kinases and the production of lymphokines. PKC-α and PKC-ε have been shown to regulate ERK (27), whereas PKC-θ regulates JNK (28, 29). To further characterize the reduction in PKC translocation, the ability of β-oxa 21:3n-3 and 22:6n-3 to inhibit individual isozyme translocation was examined. β-oxa 21:3n-3 was found to significantly inhibit the translocation of PKC-βI and PKC-ε, but not PKC-α, PKC-βII, or PKC-θ (Fig. 6). Interestingly, 22:6n-3 did not inhibit the translocation of any of the PKC isozymes examined.

Each of the MAP kinases, ERK, JNK, and p38, have been demonstrated to play central roles in T cell proliferation and cytokine production (18, 27, 28, 30). Thus, the effects of the β-oxa fatty acid on the activity of ERK, JNK, and p38 were examined. Pretreatment of T lymphocytes with β-oxa 21:3n-3 (20 μM) for 30 min was found to significantly inhibit ERK activation, but not the activation of JNK or p38 (Fig. 7). Metabolism of natural fatty acids, particularly 20:4n-6 via the COX and lipoxygenase enzymes, is believed to be one mechanism by which natural fatty acids mediate their activities in vivo (1, 2). It was thus important to determine whether metabolism of the β-oxa PUFA was required for its inhibitory activity in vitro. T lymphocytes were pretreated with IM (100 μM), a COX inhibitor, or nordihydroguaiaretic acid (NDGA, 10 μM), a lipoxygenase inhibitor, before being treated with β-oxa 21:3n-3 (20 μM) and stimulated with PHA-PMA. The results (percentage of inhibition ± SEM) show that neither IM (diluent, 73.9 ± 5.9; IM, 76 ± 4.5) nor NDGA (diluent, 73.9 ± 1.8; NDGA, 62.9 ± 5.2) prevented β-oxa 21:3n-3 from inhibiting lymphocyte proliferation. Thus, the inhibitory effects of β-oxa 21:3n-3 on T lymphocyte function are independent of its metabolism via COX or lipoxygenase enzymes. These results are consistent with the ability of n-3 and n-6 fatty acids to mediate their biological effects toward T lymphocyte function in an eicosanoid-independent manner (31, 32). In addition, pretreatment with the antioxidant, vitamin E (100 μM), for 30 min did not prevent the β-oxa 21:3n-3-mediated inhibition (percentage of inhibition ± SEM) of lymphocyte proliferation (diluent, 73 ± 2.8; vitamin E, 70 ± 1.9), suggesting that lipid peroxidation is not involved in the inhibition of lymphocyte proliferation. This is consistent with previous results in which the activity of n-3 fatty acids was not abrogated by vitamin E treatment (33).

Compared with 22:6n-3, β-Oxa 21:3n-3 was found to be a very poor stimulator of the neutrophil respiratory burst, but more active in depressing T lymphocyte function. β-oxa 21:3n-3 significantly inhibited T lymphocyte proliferation and cytokine production in vitro. Lymphocyte proliferation, TNF-β, IFN-γ, and IL-2 production were inhibited in a dose-dependent manner and under optimal conditions; the IC₅₀ value for β-Oxa 21:3n-3 was 1.9 μM, more active than 22:6n-3 (IC₅₀ of 5.2 μM). But the neutrophil respiratory burst was poorly stimulated by concentrations of β-oxa-21:3n-3 approaching 30 μM compared with 22:6n-6, which induces a marked respiration burst at 5–10 μM (3).

The inhibition of T cell responses was reflected in the ability of β-oxa 21:3n-3 to inhibit the DTH response, a reaction that requires activated T lymphocytes and secretion of Th1-type cytokines, such as TNF-β and IFN-γ (34). In fact, β-oxa 21:3n-3 was more effective than 22:6n-3 and IM at inhibiting footpad swelling. The ability of the fatty acid to reduce footpad swelling is most likely to be due to a decrease in IFN-γ production. β-Oxa 21:3n-3 was an effective inhibitor of both TNF-β and IFN-γ production from mitogen-stimulated T lymphocytes in vitro, cytokines that are involved in the effector phase of the DTH response. β-Oxa 21:3n-3 was also a strong inhibitor of IL-2, which is a propagator of T cell activation, associated with an inflammatory response. In addition, β-oxa 21:3n-3 was a strong inhibitor of carrageenan-induced inflammation. It is recognized that this inflammatory process also involves a mononuclear cell infiltrate (35) and these cells may be the target of β-oxa-21:3n-3. In preliminary animal studies, β-oxa-21:3n-3 was preferentially incorporated into lipids in the liver and white blood cells. The presence of β-oxa-21:3n-3 in white blood cells is consistent with its ability to target immune cell-mediated inflammation. In comparison, under similar experimental conditions, 22:6n-3 was poor at inhibiting this type of inflammation.

Conversion of β-oxa 21:3n-3 into its methyl ester derivative inhibited lymphocyte proliferation and cytokine production by <10%. The carboxyl group is required for binding to fatty acid-binding proteins and esterification into membrane phospholipids. This would thus limit the degree of incorporation of the fatty acid into membrane phospholipids. The reduced biological activity of the methyl ester is consistent with previously published results from our laboratory, in which the carboxyl group has been shown to be important for fatty acid activity (36).

The mechanisms by which the β-oxa PUFA inhibited T lymphocyte responses were partly identified. The effects could not be explained merely on a basis of a change to functional T cell surface receptors used by the agonists, as the fatty acid did not cause a significant change in the expression of these molecules. Furthermore, the fatty acid was still effective in inhibiting T lymphocyte responses induced by PMA and A23187, agonists that act at a postsurface receptor level. Further evidence that β-oxa 21:3n-3 acted on intracellular targets was provided by the finding that the fatty acid inhibited the agonist-induced translocation of PKC from the cytosol to the particulate fraction. Thus, in this manner, this β-oxa PUFA mimics the effects of hydroperoxy fatty acids on PKC activation (7). PUFA have previously been shown to interact directly with PKC and activate this kinase. It is thus conceivable that the altered conformation of β-oxa 21:3n-3 could interact with PKC and prevent the interaction of PKC with 1,2-diacylglycerol, and hence PKC translocation. Alternatively, the β-oxa PUFA may be incorporated in the sn-2 position of phosphatidylinositol-containing phospholipids, and in this manner generate unusual 1,2-diacylglycerol molecules, which may lack the ability to induce PKC translocation.

The activities of ERK, JNK, and p38 in T cells are stimulated by PMA or PMA and Ca²⁺ ionophore, suggesting a role of PKC in regulating the activities of the MAP kinases. Indeed, PKC-α, PKC-ε (27), and PKC-θ (28) have been demonstrated to regulate the activities of ERK and JNK, respectively, in T cells, and studies in smooth muscle cells have demonstrated that PKC-δ is involved in regulating p38 activity (37). The ability of β-oxa 21:3n-3 to partially inhibit ERK, but not JNK activity is consistent with suppression of PKC-ε, but not PKC-θ translocation.

The inhibition of ERK activity by β-oxa 21:3n-3 is consistent with its ability to suppress lymphocyte function. Members of the ERK cascade, including ras, raf, MEK, and ERK, have been demonstrated to be critical for the production of TNF-β in Jurkat T cells (18). Thus, dominant-negative mutants of ras, raf, or ERK1 suppressed PHA-PMA-stimulated production/secretion of TNF-β. In addition, ERK has been demonstrated to be involved in production/secretion of IL-2 by Jurkat T cells (38). The importance of the ERK cascade in the production of other T cell cytokines has also been demonstrated. Thus, the MEK1/MEK2 inhibitor, PD98059, has been reported by Egerton et al. (39) to inhibit the production of IL-3, IL-4, IL-5, IL-10, GM-CSF, and IFN-γ (39). A hierarchy was found to exist, with IL-5, IL-10, GM-CSF, and IFN-γ being more severely affected than IL-3 and IL-4 when ERK activation was inhibited (39). β-Oxa 21:3n-3 did not differentiate between TNF-β, IFN-γ, and IL-2. Although ERK action was inhibited, a direct comparison with the data obtained with PD98059 (39) is not possible because different cytokines were examined between this study and the study by Egerton et al. (39).

Arachidonic acid can be metabolized into eicosanoids via lipoxygenase and COX enzymes. Many of the in vivo effects of 20:4n-6 are thought to be mediated by the production of eicosanoids. Indeed, some of the beneficial anti-inflammatory effects of n-3 fatty acids are believed to be due to a reduction in the level of 20:4n-6-derived eicosanoids in favor of n-3 metabolites (1, 2). Recently, it has been demonstrated that acetylated COX-2 produces aspirin-triggered-15-epi-lipoxin A₄ (40). These molecules are the carbon 15 epimers of native lipoxins. When the substrate for the acetylated COX-2 is 20:5n-3, two novel metabolites, 15R-hydroxyeicosapentaenoic acid (15R-HEPE) and 18R-HEPE, are produced. These two molecules can be metabolized by neutrophils into further novel lipids such as 5-series 15R-lipoxin(LX)₅ and 5,12,18R-triHEPE. These products have been demonstrated to be potent inhibitors of neutrophil migration and infiltration in dorsal air pouches (41), chemotaxis (42), and neutrophil degranulation, as measured by elastase release (43).

The generation of secondary metabolites from 20:5n-3, which can then be further metabolized by cells involved in the acute inflammatory response, provides one possible mechanism by which n-3 PUFA may exert their anti-inflammatory effects. Although the 15R-LX₅ and 5,12,18R-triHEPE have at present only been demonstrated to inhibit neutrophil-driven inflammatory response in vivo (40), our engineered PUFA target both lymphocytes and neutrophils, as demonstrated by the two models of in vivo inflammation. Thus, through chemical modification, the beneficial activities of n-3 PUFA can be retained, yet the proinflammatory activity eliminated. It would be interesting to determine whether β-oxa 21:3n-3 is a substrate for the acetylated COX-2, and how such a product will differ in action from β-oxa 21:3n-3.

Neither IM nor NDGA abrogated the ability of the fatty acid to inhibit lymphocyte proliferation, indicating that the inhibitory action of the fatty acid was due to β-oxa 21:3n-3, and its metabolism is not required for biological activity. In addition, pretreatment with vitamin E did not block the activity of the PUFA, suggesting that lipid peroxidation is not involved.

Activated T lymphocytes with a Th1 cytokine profile have been implicated in the pathogenesis of several autoimmune diseases (44, 45, 46, 47, 48). Dietary supplementation with n-3 fatty acids has been used as a means of treating rheumatoid arthritis (44, 45), multiple sclerosis (46), insulin-dependent diabetes (47), psoriasis, and atopic dermatitis (48). However, such treatments have generally provided only modest improvements in disease severity (44, 45, 46, 47, 48). The less than expected amelioration following n-3 PUFA treatment may in part be explained by the ability of n-3 fatty acids to activate the neutrophil respiratory burst, which is likely to cause tissue damage. Based on our knowledge of the structure-function relationship of fatty acids toward biological activity, novel chemically engineered fatty acids were made in an attempt to synthesize molecules that displayed differential activities, that is, fatty acids that did not activate neutrophils, but still depressed T lymphocyte function. The engineered PUFA contained an oxygen atom substituted for the β-methylene group (8) to mimic the selective properties of hydroperoxyeicosatetraenoic acid (4). This work illustrates that PUFA with desirable biological activities can be generated by modifying specific structural elements. By introducing an oxygen atom into the β-position, preferential anti-inflammatory properties were achieved, giving rise to molecules with very poor ability to stimulate oxygen radical production from human neutrophils, but still possess an ability to inhibit T lymphocyte function. This skewing of the fatty acid toward anti-inflammatory activity was similar to that seen with metabolites of 20-4n-6, such as hydroperoxyeicosatetraenoic acid, hydroxyeicosatetraenoic acid, and lipoxins (43). β-Oxa 21:3n-3 may be useful as an antiinflammatory agent, demonstrating increased biological activity and selectivity both in vitro and in vivo compared with the natural long chain PUFA. This selectivity could be due in part to its ability to selectively target the PKC-ε, ERK1/ERK2 module that is required for T lymphokine production.

We are grateful to Louise Furphy for technical assistance in aspects of this work. We thank the staff of the Central Specimen Collection, Core Laboratory, Women’s and Children’s Hospital for analyzing the liver and kidney biochemical parameters.