Search tips
Search criteria 


Logo of jbcThe Journal of Biological Chemistry
J Biol Chem. 2016 February 19; 291(8): 4069–4078.
Published online 2015 December 23. doi:  10.1074/jbc.M115.698001
PMCID: PMC4759183

Different Fatty Acids Compete with Arachidonic Acid for Binding to the Allosteric or Catalytic Subunits of Cyclooxygenases to Regulate Prostanoid Synthesis*


Prostaglandin endoperoxide H synthases (PGHSs), also called cyclooxygenases (COXs), convert arachidonic acid (AA) to PGH2. PGHS-1 and PGHS-2 are conformational heterodimers, each composed of an (Eallo) and a catalytic (Ecat) monomer. Previous studies suggested that the binding to Eallo of saturated or monounsaturated fatty acids (FAs) that are not COX substrates differentially regulate PGHS-1 versus PGHS-2. Here, we substantiate and expand this concept to include polyunsaturated FAs known to modulate COX activities. Non-substrate FAs like palmitic acid bind Eallo of PGHSs stimulating human (hu) PGHS-2 but inhibiting huPGHS-1. We find the maximal effects of non-substrate FAs on both huPGHSs occurring at the same physiologically relevant FA/AA ratio of ~20. This inverse allosteric regulation likely underlies the ability of PGHS-2 to operate at low AA concentrations, when PGHS-1 is effectively latent. Unlike FAs tested previously, we observe that C-22 FAs, including ω-3 fish oil FAs, have higher affinities for Ecat than Eallo subunits of PGHSs. Curiously, C-20 ω-3 eicosapentaenoate preferentially binds Ecat of huPGHS-1 but Eallo of huPGHS-2. PGE2 production decreases 50% when fish oil consumption produces tissue EPA/AA ratios of ≥0.2. However, 50% inhibition of huPGHS-1 itself is only seen with ω-3 FA/AA ratios of ≥5.0. This suggests that fish oil-enriched diets disfavor AA oxygenation by altering the composition of the FA pool in which PGHS-1 functions. The distinctive binding specificities of PGHS subunits permit different combinations of non-esterified FAs, which can be manipulated dietarily, to regulate AA binding to Eallo and/or Ecat thereby controlling COX activities.

Keywords: allosteric regulation, cyclooxygenase (COX), eicosanoid, fatty acid, prostaglandin, docosahexaenoic acid, eicosapentaenoic acid, half-sites, ω-3 fatty acid, palmitic acid


Cyclooxygenases (COXs),3 known more formally as prostaglandin endoperoxide synthases (PGHSs), catalyze the production of prostaglandin H2 (PGH2) from arachidonic acid (AA) in the committed step of prostaglandin biosynthesis (1,4). There are two PGHS isoforms, PGHS-1, which is generally expressed constitutively, and an inducible isoform PGHS-2 variously associated with cell growth, differentiation, and inflammation.

Both PGHSs catalyze the same two reactions. One is a bis-oxygenase reaction referred to as a cyclooxygenase (COX) reaction in which two molecules of O2 are introduced into the carbon skeleton of AA, the most common substrate, to form PGG2. A second reaction is a peroxidase reaction in which the 15-hydroperoxyl group of PGG2 undergoes a two-electron reduction to form PGH2 and water.

The COX activities of PGHSs are the targets of COX inhibitors called nonsteroidal anti-inflammatory drugs (NSAIDs) (5), and a subset of these drugs includes agents that are relatively selective for PGHS-2 and are often referred to as COX-2 inhibitors or coxibs. NSAIDs are the most widely used drugs in the United States. Unfortunately, the use of NSAIDs is associated with adverse gastrointestinal and cardiovascular effects estimated to be responsible for 20,000 deaths annually (5).

PGHSs are sequence homodimers that function as conformational heterodimers comprised of a regulatory allosteric monomer (Eallo) and a catalytic monomer (Ecat). Ecat has a bound heme, whereas Eallo does not, and maximal COX activity occurs with one heme per dimer (6,8).

Several experimental approaches have been developed to distinguish whether a ligand preferentially binds to Eallo versus Ecat (7,9). Fatty acids (FAs) that preferentially bind to Eallo of PGHS-1 increase the rate at which aspirin acetylates the enzyme. Ecat is the target of aspirin acetylation. Additionally, FAs that bind Eallo displace unreacted [1-14C]AA from Eallo resulting in the oxygenation of the displaced [1-14C]AA by Ecat. As detailed under “Experimental Procedures,” experiments to determine displacement of [1-14C]AA from Eallo involve first preincubating the enzyme at a high enzyme to [1-14C]AA ratio (e.g. 1 μm enzyme with 1 μm [1-14C]AA) and then adding FA to the reaction mixture and determining whether unreacted [1-14C]AA is converted to PG products. FAs that bind to Eallo of PGHS-2 characteristically activate AA oxygenation, promote inhibition by aspirin or celecoxib, displace [1-14C]AA from Eallo, and/or interfere with inhibition by naproxen. Agents that preferentially bind Ecat always inhibit AA oxygenation and are unable to displace [1-14C]AA from Eallo.

Previous studies have shown that the COX activities of both human (hu) PGHS-1 and huPGHS-2 can be allosterically modulated by many common fatty acids (FAs), including both those that are COX substrates and others that are not substrates. Additionally, huPGHS-1 appears to be allosterically inhibited by celecoxib (10), while huPGHS-2 is inhibited allosterically by some NSAIDs, including naproxen and flurbiprofen (7).

As noted above, agents that bind Eallo regulate not only COX activity but interactions of Ecat with NSAIDs and coxibs. For example, palmitic acid potentiates and celecoxib attenuates the response of huPGHS-1 to aspirin (8). Because of the functional interplay between FAs that bind Eallo and the substrates and COX inhibitors that bind Ecat, there are likely to be dietary effects on both total COX activity and the responses of PGHSs to NSAIDs. Some of these interactions may underlie adverse drug responses.

In the study reported here, we have documented details of the interactions of FAs that are not COX substrates, nsFAs, with Eallo. Additionally, we have determined the Eallo versus Ecat specificities of several polyunsaturated FAs that interact with PGHSs.

nsFAs act allosterically on PGHSs by binding to Eallo (7,9, 11,14). Interestingly, the binding of saturated and monounsaturated FAs (i.e. nsFAs) to Eallo of huPGHS-1 causes enzyme inhibition, whereas binding of several of these same FAs, notably palmitic acid (PA), to Eallo of huPGHS-2 markedly increases enzyme activity (7, 8). One goal of this study was to determine the relative concentrations of AA and nsFAs that elicit a maximal difference between PGHS-1 versus PGHS-2 activities. The results of these experiments lead us to a plausible explanation for how PGHS-2 can function at low AA concentrations, when PGHS-1 is effectively latent in cells co-expressing both isoforms (15).

We have also characterized interactions of Eallo versus Ecat of huPGHS-1 and huPGHS-2 with other FAs of potential physiologic importance that have not previously been examined in detail. These include C-18, C-20, and C-22 polyunsaturated FAs. For example, we tested C-20 and C-22 polyunsaturated ω-3 fish oil FAs, including 5,8,11,14,17-eicosapentaenoic acid (EPA), 7,10,13,16,19-cis-docosapentaenoic acid (DPA), and 4,7,10,13,16,19-cis-docosahexaenoic acid (DHA) that affect PG formation in cultured cells and in vivo (16, 17).

One surprising finding from our studies of polyunsaturated FAs is that C-22 FAs bind more tightly to Ecat than Eallo. Additionally, ω-3 fish oil FAs, when tested alone, are poorer inhibitors of purified huPGHSs than anticipated based on the magnitude of the effects of dietary fish oil on PG formation in vivo (17). The in vivo effects of fish oil may result from a combination of nsFAs and fish oil FAs working in tandem to inhibit PGHS-1.

Experimental Procedures


Complete protease inhibitor was from Roche Applied Science. Nickel-nitrilotriacetic acid Superflow resin and nickel-nitrilotriacetic acid were from Qiagen. Palmitic acid (16:0), oleic acid (18:1ω9), stearic acid (18:0), FLAG peptide, and FLAG affinity resin were from Sigma. AA, linoleic acid, adrenic acid, 11,14-cis, cis-eicosadienoic acid, EPA, DPA, and DHA were from Cayman Chemical (Ann Arbor, MI). Hemin was from Frontier Scientific, Logan, UT. [1-14C]AA (1.85 GBq/mmol) was from American Radiolabeled Chemicals. [5,8,11,14,17-3H]Eicosapentaenoic acid was prepared as described previously (18). Decyl maltoside (C10M), n-octyl β-d-glucopyranoside, and C10E6 were purchased from Anatrace (Maumee, OH). BCA protein reagent was from Pierce. Hexane, isopropyl alcohol, and acetic acid were HPLC grade from Thermo Fisher Scientific, Inc. Anti-PGHS-2 antibodies directed against the 18-amino acid insert unique to PGHS-2 were as described (19). Anti-FLAG antibodies were from LifeTein, South Plainfield, NJ. Horseradish peroxidase-conjugated secondary antibodies (goat anti-rabbit IgG and goat anti-mouse IgG) were from Bio-Rad.

Expression, Purification, and Assay of huPGHS-1 and huPGHS-2 Variants

Procedures for the expression and purification of native huPGHS-1 and native huPGHS-2 from insect cells were essentially as described previously (7,9). COX activities were typically determined using measurements of O2 consumption with an O2 electrode essentially as detailed previously (7). One unit of COX activity is defined as 1 μmol of O2 consumed per min at 37 °C in the standard assay mixture. The purity of the recombinant huPGHSs was determined by SDS-PAGE and Western blot analysis (8, 12). The average specific activities with 100 μm AA were 20 units/mg for purified huPGHS-1 and 40 units/mg for purified huPGHS-2. Radio-thin layer chromatography assays were performed as described previously using [1-14C]AA (12).

Measurements of FA Binding to Eallo

COX assays were performed at high enzyme/substrate ratios to quantify AA or EPA binding to Eallo of huPGHSs and FA-induced displacement of AA or EPA from Eallo. To measure AA or EPA binding, reaction mixtures (100 μl final volume) containing 1 μm [1-14C]AA or 1 μm [3H]EPA, 0.10–2.0 μm huPGHS-1 or huPGHS-2, 5 μm hematin, and 1 mm phenol in 0.1 m Tris-HCl, pH 8.0, were incubated at 37 °C for 1–8 min, and the products were separated and quantified by radio-reverse phase-HPLC as detailed previously (7, 8). The principle underlying this method is described in detail in earlier references (7, 9). Briefly, when [1-14C]AA and huPGHS-2 were incubated at high enzyme to substrate ratios, significant amounts of unreacted AA remain after the oxygenation reaction is largely complete. The amounts of unreacted AA remaining are directly proportional to the amount of added PGHS because much of the unreacted AA becomes bound to Eallo. This occurs because Eallo, although catalytically inert, has a 30-fold higher affinity (Kd ~0.25 μm) than Ecat (Kd ~7.5 μm) for AA. Non-substrate FAs (e.g. PA) bind Eallo (Kd ~ 7.5 μm) more tightly than Ecat (Kd ≥50 μm), and when added to the reaction mixture they displace the small amounts of unreacted [1-14C]AA from Eallo. The displaced [1-14C]AA can then be oxygenated by Ecat because non-substrate FAs do not compete effectively with AA for Ecat. Other agents that interact with Eallo are also able to displace AA from this subunit. FAs tested in this study that at a concentration of 5 μm do not displace [1-14C]AA are categorized as binding inefficiently to Eallo.

Statistical Analyses

Student's t tests were performed in Microsoft Excel. If the experiments had the same numbers of repetitions, probabilities were calculated with a Student's paired t test, with a two-tailed distribution. If the experiments had different numbers of repetitions, probabilities were calculated with a Student's unequal variance t test, with a two-tailed distribution.


Optimal Non-substrate FA/AA Ratio for Regulating huPGHSs

Most common saturated and monounsaturated FAs that are not COX substrates allosterically inhibit huPGHS-1 (8, 12), and several of these nsFAs, notably palmitate and oleate, allosterically stimulate huPGHS-2 (7, 9, 12). With both huPGHS isoforms, nsFAs function by binding Eallo and displacing AA from Eallo (7, 8). The Kd values for AA binding to Eallo of huPGHS-1 and huPGHS-2 are similar (~ 0.25 μm), whereas the Kd values for the binding of various nsFA to Eallo are all about 30-fold higher (~ 7.5 μm).

We performed a series of measurements to determine what combination of nsFA concentrations and nsFA/AA ratios would cause the biggest difference between the rates of AA oxygenation by huPGHS-1 versus huPGHS-2. We first tested the effect of PA on AA oxygenation (Fig. 1). Both maximal relative inhibition of huPGHS-1 and maximal relative activation of huPGHS-2 occurred under very similar conditions, at a molar ratio of PA/AA of ~20 and at relatively low AA concentrations (~0.5–2.0 μm AA). With 20 μm PA and 1 μm AA, we observed a 45% inhibition of huPGHS-1 and 215% of the starting huPGHS-2 activity, a 4.8 ratio of huPGHS-2 to huPGHS-1 activities. The fact that maximal effects of PA on PGHS-1 and PGHS-2 occur under the same conditions in vitro implies that the enzymes can be coordinately regulated in cells.

Effects of different ratios of palmitic acid to arachidonic acid on the rates of oxygenation of arachidonic acid by huPGHS-1 versus huPGHS-2. The indicated concentrations of PA and AA were used in a standard O2 electrode assay mixture and either purified ...

When oleic acid and stearic acid were tested at a ratio of nsFA/AA of 20, oleate and stearate showed patterns similar to those seen with PA, although the magnitudes of the differences between huPGHS-1 and huPGHS-2 were less (Fig. 2). The ratios of huPGHS-2/huPGHS-1 activities with palmitic, oleic, and stearic acid were 4.9, 3.7, and 2.2, respectively. The differences are a consequence of oleate and stearate being poorer activators than PA of huPGHS-2.

Comparison of the effects of palmitic, oleic, and stearic acids on the rates of oxygenation of arachidonic acid by huPGHS-1 versus huPGHS-2. Palmitic (PA), oleic (OA), or stearic (SA) acids each at a concentration of 20 μm were combined with 1 ...

Interactions of EPA with huPGHS-1 Versus huPGHS-2

EPA is poor substrate and a weak inhibitor of AA oxygenation by both recombinant huPGHS-1 (Table 1) (8) and ovine (o) PGHS-1 (11, 12). The Km and Vmax values for EPA with huPGHS-1 are 11 μm and 3.3 units/mg, respectively, compared with Km and Vmax values for AA with huPGHS-1 of 5 μm and 19 units/mg, respectively. Although EPA inhibits huPGHS-1, EPA does not inhibit AA oxygenation by huPGHS-2 at EPA/AA ratios of less than five (Table 1) (7, 11, 12).

Comparison of the relative rates of oxygenation by huPGHS-1 versus huPGHS-2 of various FAs alone and in combination with AA

In the absence of AA, [3H]EPA is oxygenated by huPGHS-2, and residual [3H]EPA becomes bound to Eallo of huPGHS-2. This [3H]EPA can be displaced by 2.5 μm PA (Table 2) (7). The Kd value for EPA binding to Eallo of huPGHS-2 calculated from the data in Table 2 is 0.28 μm, which is essentially identical to the Kd for AA (7). Also like AA (9), EPA fails to bind to Eallo of Y385F R120A/Native huPGHS-2 (data not shown).

Oxygenation of EPA at high huPGHS-2 to EPA ratios

In marked contrast to what is observed with huPGHS-2, insignificant amounts of [3H]EPA remain after 1 μm [3H]EPA is incubated for 4 min with 5 μm huPGHS-1 (Fig. 3). The results indicate that EPA fails to efficiently bind Eallo of huPGHS-1. However, EPA can bind to Ecat as a substrate, albeit a poor substrate, of huPGHS-1. According to our model for the functioning of PGHS conformational heterodimers (7,9), the Kd value for substrate binding to Ecat-, EPA in this case, is the same as the Km value (i.e. 11 μm).

Unreacted EPA remaining after incubation of [3H]EPA with excess huPGHS-1 or huPGHS-2. [5,6,8,9,11,12,14,15,17,18-3H]EPA (1 μm) was incubated at 37 °C for 4 min with 50 units (~5 μm) of purified huPGHS-1 (A) or 60 units ...

Consistent with the concept that EPA binds poorly to Eallo of huPGHS-1, we found that C-15 to C-19 saturated FAs as well as oleic acid at a relatively low concentration of 10 μm inhibit EPA oxygenation by huPGHS-1. Inhibition occurs even when EPA is present at high concentrations (100 μm) under conditions that optimize EPA oxygenation (Fig. 4A) (20, 21). nsFAs can exert an inhibitory effect on AA oxygenation by binding Eallo of huPGHS-1, but this occurs only at high nsFA/AA ratios (i.e. ≥5) (8). In a related experiment (Fig. 4B), EPA alone caused 12% inhibition of AA oxygenation by huPGHS-1, whereas a mixture of nsFAs caused 16% inhibition. An additive response would yield 28% inhibition, a level only less than the 35% inhibition caused by the combination of EPA plus nsFAs. Although this is not a large change, it suggests that nsFAs bound to Eallo enhance the binding of EPA versus AA to Ecat when EPA and AA are present at low concentrations, 1 and 2 μm, respectively. as tested in Fig. 4B. Overall, our results indicate that the most abundant nsFAs inhibit EPA oxygenation by huPGHS-1, that EPA preferentially binds Ecat versus Eallo of huPGHS-1, and that nsFAs bound to Eallo of huPGHS-1 function in combination with EPA bound to Ecat to inhibit PGHS-1.

Inhibition by non-substrate fatty acids of EPA oxygenation by huPGHS-1. Measurements of COX activity were performed as described under “Experimental Procedures” using an O2 electrode assay. All reactions were performed in the presence ...

Although the inhibitory effect of EPA is augmented by nsFAs in the case of huPGHS-1, the stimulatory effects of nsFAs on AA oxygenation by huPGHS-2 are attenuated in the presence of EPA (Fig. 5). For example, PA causes a 62% increase in the rate of AA oxygenation by huPGHS-2, but the stimulatory effect of PA is reduced to 26% in the presence of EPA. Similarly, the stimulatory effect of OA plus EPA is 16% versus a 30% stimulatory effect of OA on AA alone (i.e. a net 53% decrease in stimulation). EPA and AA bind about 30 times more tightly than nsFAs to Eallo (Table 2) (7). Thus, increasing the concentration of EPA has the effect of displacing nsFAs from Eallo of huPGHS-2. Because nsFAs are better activators of AA oxygenation than EPA, the net effect of added EPA is a decreased rate of AA oxygenation by huPGHS-2 in the presence of PA and oleic acid. Conversely, increasing the concentration of nsFAs reduces EPA and AA binding to Eallo of huPGHS-2 and stimulates COX activity and thus can enhance PGE2 formation.

Effect of combinations of EPA and non-substrate FAs on AA oxygenation by huPGHS-2. Measurements of COX activity were performed as described under “Experimental Procedures” in a standard O2 electrode assay containing the indicated combinations ...

Inhibition of huPGHS-2-mediated AA Oxygenation by DHA and DPA

We previously reported that DHA is a poor substrate for huPGHS-1 and huPGHS-2 and that DHA is converted only to monohydroxy docosahexaenoic acids by huPGHS-2 (7, 12, 13). Under optimal conditions in the presence of 100 μm H2O2, we determined a Vmax = 4.1 units/mg and Km = 27 μm for DHA oxygenation (data not shown). huPGHS-2 undergoes a suicide inactivation when oxygenating polyunsaturated FA substrates (2, 22). We determined that 1 nmol (~150 μg) of huPGHS-2 can convert 100 nmol of DHA to product under these conditions.

The rates of oxygenation of 100 μm DPA by both huPGHS-1 and huPGHS-2 in the presence of 100 μm H2O2 were 70–80% that of DHA (Table 1) (12). DHA and DPA are also about equally effective inhibitors of AA oxygenation by both isoforms. For example, in the case of DHA and huPGHS-2, an ~50% reduction in O2 consumption occurred at a 5:1 ratio of DHA (100 μm)/AA (20 μm), and a similar 50% inhibition of the conversion of [1-14C]AA to products occurs at this DHA/AA ratio (Table 1) (12). The correlation between inhibition of O2 uptake and that of [1-14C]AA consumption implies that the inhibition of O2 uptake by DHA is primarily due to inhibition of AA oxygenation. A similar level of inhibition of O2 consumption and conversion of [1-14C]AA to products occurs at a 5:1 ratio of DPA (100 μm)/AA (20 μm) (Table 1). Thus, our data indicate that DHA and DPA are poor substrates and similarly effective but modest inhibitors of AA oxygenation by both huPGHS-1 and huPGHS-2.

As noted above, AA and EPA bind equally well to Eallo of huPGHS-2 with Kd values of ~0.25 μm (Table 2) (7). We assumed the same would be true for DHA. To test this, we examined the ability of DHA to displace [1-14C]AA from Eallo of huPGHS-2. As shown in Table 3, even with 10 μm DHA, which constitutes an ~100-fold excess over the ~0.12 μm bound AA (i.e. 0.135–0.015 μm [1-14C]AA remaining for control conditions minus 5 μm PA conditions) was unable to displace unreacted [1-14C]AA from Eallo of huPGHS-2. Under these conditions, 5 μm PA, which has a relatively high Kd value for Eallo (~7.5 μm) (7), displaced most if not all of the bound AA. There is sufficient enzyme available under the conditions used in the experiment with 10 μm DHA to convert all of the available DHA to product so the fact that AA levels are not decreased by DHA is because AA is not displaced from Eallo and not because DHA prevents the binding of AA to Ecat. DPA, like DHA, was also ineffective in displacing [1-14C]AA from Eallo (Table 3). Likewise, the C-22 ω6 FA adrenic acid had surprisingly little effect on AA binding to Eallo. These results suggest that ω3- and ω6-polyunsaturated FAs having 22 carbon atoms fail to bind tightly to Eallo. We also examined linoleic acid, which contains 18 carbons. Linoleate displaced [1-14C]AA from Eallo of huPGHS-2 (Table 3) but was no more potent on a molar basis than PA (Tables 1 and and33).

RP-HPLC analysis of the displacement of AA from Eallo of huPGHS-2 by unlabeled DHA, DPA, adrenic acid, linoleic acid, or PA

Effects of nsFAs on the Interactions of DHA and EPA with huPGHS-2

As shown in Table 1, DHA and DPA when tested alone with either huPGHS-1 or huPGHS-2 are comparably effective in inhibiting both isoforms. DHA inhibits huPGHS-1 with approximately the same potency of EPA. However, DHA is a more potent inhibitor than EPA of huPGHS-2 when each ω-3 FA is tested alone with AA as the primary substrate.

In earlier studies, we observed that PA had no effect on the rate of DHA oxygenation by huPGHS-2 (7, 13) as can also be seen in Fig. 6. However, we observed that the modest inhibitory effect of DHA on AA oxygenation by PGHS-2 is slightly augmented in the presence of common nsFAs (Fig. 6). Specifically, a mixture of the most common nsFAs stimulated AA oxygenation by 27% in the absence of DHA compared with 21% in the presence of DHA. Thus, DHA modestly interfered with the stimulation of AA oxygenation by nsFAs that function by binding Eallo of huPGHS-2. Because DHA does not appear to bind to Eallo efficiently, we interpret the results in Fig. 6 as indicating that nsFAs promote the binding of DHA versus AA to Ecat of huPGHS-2. The small decrease in AA oxygenation observed with the concentrations of DHA plus nsFA tested in Fig. 6 may not be biologically significant.

Effect of combinations of DHA and non-substrate FAs on AA oxygenation by huPGHS-2. Measurements of COX activity were performed as described under “Experimental Procedures” in a standard O2 electrode assay containing the indicated combinations ...


Each PGHS isoform is encoded by a single gene (23, 24). Ovine PGHS-1 was originally found to be a homodimer (25), and it is now clear that PGHS-1 and PGHS-2 are each sequence homodimers composed of subunits having identical primary structures (2, 4). There is some heterogeneity involving differences in levels of N-glycosylation that leads to subunits with molecular masses that differ by about 2 kDa and are distinguishable by SDS-PAGE (4, 19, 26,28). However, N-glycosylation of PGHSs is involved in folding and trafficking (28, 29) and deglycosylation of native oPGHS-1 does not alter enzyme activity (29). Furthermore, even removal of the near C-terminal targeting domain of PGHS-2 does not significantly affect enzyme activity (28, 30).

X-ray crystal structures of PGHSs suggest that the enzymes are structural homodimers and that each monomer has a bound heme. When crystallized with ligands that bind the COX active sites, both COX sites are typically occupied. When crystallized with sub-stoichiometric amounts of flurbiprofen, only molecules of oPGHS-1 having flurbiprofen in both monomers are observed in the structure (31). Thus, PGHSs tend to crystallize only in symmetric forms in which both of the heme-binding sites and both COX active sites are fully occupied. There are a few exceptions. AA and EPA bind in different orientations in the two monomers (32, 33), and PA is bound in only one of the two monomers (7) of muPGHS-2; additionally, celecoxib and certain other non-steroidal anti-inflammatory drugs are partially bound in only one of the two monomers of oPGHS-1 (10, 31).

There is now considerable evidence dating back to studies by Kulmacz and co-workers (6,14, 31, 34,38) in the mid-1980s that PGHSs function in solution as conformational heterodimers composed of Ecat and Eallo monomers. The work reported here serves to identify the Eallo versus Ecat specificities of several physiologically important FAs that had not previously been examined in detail. Fig. 7 is a diagram that summarizes the subunit preferences for a large number of nsFAs and COX substrates and inhibitors that are able to interact with the Eallo and/or Ecat COX sites of huPGHS-1 and huPGHS-2. The physiological and pharmacological significance of these many potential interactions have yet to be studied. However, these biochemical interactions are likely to have important biological consequences. This contention is based on findings that have indicated that even relatively subtle differences in COX activities and responses to NSAIDs are important clinically (5, 39). For example, incomplete (50–75%) suppression of PG formation correlates with analgesic and anti-inflammatory responses (40,42).

Isoform-specific interactions of COX substrates, nsFAs, and COX inhibitors with huPGHS-1 and huPGHS-2. Both PGHS isoforms are sequence homodimers that function as conformational heterodimers composed of an allosteric (Eallo) and a catalytic (Ecat) subunit. ...

Differential Coordinate Regulation of PGHSs by nsFAs

Previous studies had shown that huPGHS-1 and huPGHS-2 are allosterically regulated, but in different directions, by nsFAs (7,9). We determined that the biggest differential between the two activities, about 5-fold, occurs at concentrations of AA of 0.5–2 μm and a PA/AA ratio of about 20.

Although often described as “low,” the effective concentration of AA, which is the “free” AA available to COXs at their site(s) action, is unknown (43). Addition of exogenous AA at concentrations of 0.5–1 μm to murine 3T3 cells expressing muPGHS-1 versus muPGHS-2 are sufficient to elicit PG formation, and the maximal ratio of PGHS-2 versus PGHS-1 activity occurs with 1–2.5 μm exogenous AA (44).

Although there is limited information on the topic (see LIPID MAPS Lipidomics Gateway), in RAW264.7 cells treated with endotoxin to stimulate COX-2 formation, the most abundant non-esterified FAs in the cells are palmitate, oleate, and stearate, and the relative ratios of non-esterified PA to AA range from 5 to 60. Maximal ratios of nsFA to AA occur during a period 4–8 h after endotoxin treatment when PGHS-2 expression peaks and maximal PGHS-2-derived PGD2 synthesis occurs.

Our experiments indicate that at normal nsFA/AA levels and low AA concentrations (1–2.5 μm), huPGHS-1 has, on a molar basis, only a fifth of the activity as huPGHS-2 toward AA. We propose that it is this reciprocal allosteric regulation of the PGHS isoforms by nsFAs that permits PGHS-2 to function, although PGHS-1 is essentially latent when PGHS-1 and PGHS-2 are co-expressed in cells (15).

EPA Interactions with Eallo and Ecat of huPGHSs

huPGHS-1 and huPGHS-2 interact differently with EPA. EPA binds Ecat but fails to efficiently bind Eallo of huPGHS-1. The affinities of AA and EPA for Ecat of huPGHS-1 are similar, but the Vmax for EPA is less than 20% that for AA. Consequently, EPA is a modest competitive inhibitor of AA oxygenation by huPGHS-1. The inhibitory potency of EPA toward huPGHS-1 is similar to those observed for DHA and DPA. Common nsFAs, including palmitate, oleate, and stearate, that allosterically inhibit huPGHS-1 by binding Eallo have effects additive with EPA, which binds Ecat. Interestingly, the ability of EPA to inhibit huPGHS-1 is augmented somewhat in the presence of nsFAs perhaps because nsFAs potentiate the binding of EPA versus AA to Ecat of huPGHS-1.

The degree of inhibition of PGHS-1-mediated PGE2 formation by normal colonic tissue and of urinary PGE2 levels in animals fed fish oil supplements is much greater than predicted based simply on the EPA/AA ratio in tissue phospholipids (17). Thus, 50% inhibition of PGE2 formation occurs with a tissue EPA/AA ratio of 0.2, whereas 50% inhibition of PGH2 synthesis by huPGHS-1 occurs at an EPA/AA ratio of five (Table 1). This apparent augmentation of EPA-mediated inhibition of PGE2 formation could reflect a high level of nsFAs in the milieu in which PGHS-1 operates in vivo and/or could be attributable to a PGHS-1 substrate pool having a high EPA/AA ratio. AA and EPA appear to be handled similarly during cellular uptake and metabolism (45,47), and a relative excess of dietary EPA could lead to relatively high levels of EPA versus AA in substrate pools available to PGHSs. Dietary fish oil also raises the levels of other ω-3 FAs, including DPA and DHA, that may further augment PGHS-1 inhibition.

AA and EPA each bind with similar affinities to Ecat (Km = Kd = 5–10 μm) of huPGHS-2, and we have shown here that EPA binds to Eallo with the same affinity as AA (Kd ~0.25 μm (7)). Paradoxically, huPGHS-2 uses AA in preference to EPA when examined under Michaelis-Menten conditions (i.e. [E] [dbl greater-than sign] [S]) at concentrations of EPA that are up to five times that of AA (9, 11, 12). The simplest explanation is that EPA bound to Eallo of huPGHS-2 behaves in the same way as AA itself, both EPA and AA promote the use of AA by Ecat. It is only when EPA concentrations are much greater than AA that EPA is oxygenated by huPGHS-2. The products are PGH3 and monohydroxy acids and are formed in approximately equal amounts (11, 13).

EPA interferes with the binding of nsFAs to Eallo of huPGHS-2 and thus interferes with the activation of huPGHS-2 by nsFAs. This process, along with displacing AA from phospholipid precursors, is an indirect way that EPA can interfere with AA oxygenation by huPGHS-2.

Interactions of C-18 and C-22 Polyunsaturated FAs with huPGHSs

We found that DHA is oxygenated with about 5% of the efficiency of AA by huPGHS-2. Similar results were obtained by Malkowski and co-workers (33). However, studies by Kulmacz and co-workers (48) have indicated that DHA has Vmax and Km values comparable with those of AA; this latter study takes into consideration a higher rate of substrate turnover-dependent inactivation with DHA versus AA of huPGHS-2. DPA has not previously been compared as a substrate or inhibitor of purified huPGHS-1 and huPGHS-2. We find that this ω-3 fish oil FA behaves like its homolog DHA with both huPGHS isoforms.

A surprising observation of our studies of DPA and DHA is that these C-22 polyunsaturated FAs bind preferentially to Ecat rather than Eallo of huPGHS-2. Thus, C-22 polyunsaturated FAs such as DPA and DHA probably interfere with AA oxygenation mainly by competing with AA for binding to Ecat and not Eallo. This also appears to be the case with C-22 adrenic acid, an ω-6 FA. Although linoleic acid, a C-18 FA, does displace [1-14C]AA from Eallo of huPGHS-2 (Table 2), it is no more effective than PA on a molar basis, and thus, like C-22, FAs appears to bind relatively more tightly to Ecat than Eallo. We conclude that only certain C-20 ω-6 FAs (i.e. AA and dihomo-γ-linolenic acid) bind to Eallo of huPGHS-1, whereas both ω-3 and ω-6 C-20 FAs (i.e. AA, dihomo-γ-linolenic acid and EPA) bind to Eallo of the huPGHS-2 isoform. It is unclear why DHA fails to efficiently bind to Eallo, although it binds to Ecat, albeit with a lower affinity than its 20 carbon homolog EPA. We speculate that in solution Eallo is constricted in such a way that the carboxyl group of DHA cannot interact well with Arg-120 of Eallo (33).

Based on studies of huPGHS-2 inhibition by DHA and DPA, we conclude that in the absence of other interfering FAs (or other ligands), DHA and DPA would be expected to cause 50% inhibition of PGHS-1 and PGHS-2 only when present at a 5-fold molar excess of AA. This is consistent with a 5-fold difference in Km (i.e. the Kd for Ecat) between AA and DHA. A 5-fold higher concentration of free DHA (or DPA) does not occur in cells, except in special cases like retina or testes that have significantly higher levels of DHA than AA (49) or in cells cultured with high concentrations of ω-3 FAs (16). Accordingly, our results would suggest that the biologic effects of DHA and DPA are not mediated primarily via effects on PGHSs but rather through other mechanisms.

In earlier studies, we had found that PA, linoleate, DHA, and 2-arachidonoylglycerol were ineffective in preventing cross-linking between monomers of huPGHS-2, whereas EPA and AA and most COX inhibitors did prevent cross-linking (12). These experiments were performed with huPGHS-2 variants having sulfhydryl groups juxtaposed in adjoining subunits of huPGHS-2 (e.g. at positions 127 and 541). We can now interpret these cross-linking data to imply that ligands only interfere with cross-linking if simultaneously bound to Ecat and Eallo and that this only occurs with FAs and COX inhibitors that bind with submicromolar affinities to Eallo (e.g. AA and EPA). A corollary to this is that ligands having a preference for binding Ecat versus Eallo are those that do not interfere with cross-linking. DHA, PA and oleate all fail to interfere with cross-linking (12). The FAs that do interfere with cross-linking between huPGHS-2 monomers include all C-20 carbon FAs with three or more double bonds and 18:3ω6 and 18:4ω3. We speculate that these latter FAs all bind Eallo of huPGHS-2 with high affinities.

Author Contributions

L. D. and H. Z. contributed equally to this manuscript. L. D., H. Z., C. Y., Y. H. H., and D. K. performed experiments and contributed to the data analyses. L. D., H. Z., and W. S. were involved in experimental design, data analysis, and writing of this manuscript.


We thank Drs. Michael Malkowski, David L. DeWitt, Zora Djuric, Robert Murphy, and Gilad Rimon for carefully reading and commenting on this manuscript.

*This work was supported in part by National Institutes of Health Grants GM68848, CA130810, and HL117798. W. L. S. serves a consultant for Cayman Chemical Co., products from which were used in this study. The other authors have no conflicts of interest. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

3The abbreviations used are:

arachidonic acid
4,7,10,13,16,19-cis-docosahexaenoic acid
7,10,13,16,19-cis-docosapentaenoic acid
5,8,11,14,17-cis-eicosapentaenoic acid
fatty acid
nonsteroidal anti-inflammatory drug
non-substrate FAs
palmitic acid
prostaglandin endoperoxide H synthase
ovine PGHS-1.


1. Schneider C., Pratt D. A., Porter N. A., and Brash A. R. (2007) Control of oxygenation in lipoxygenase and cyclooxygenase catalysis. Chem. Biol. 14, 473–488 [PMC free article] [PubMed]
2. Tsai A. L., and Kulmacz R. J. (2010) Prostaglandin H synthase: resolved and unresolved mechanistic issues. Arch. Biochem. Biophys. 493, 103–124 [PMC free article] [PubMed]
3. Rouzer C. A., and Marnett L. J. (2011) Endocannabinoid oxygenation by cyclooxygenases, lipoxygenases, and cytochromes P450: cross-talk between the eicosanoid and endocannabinoid signaling pathways. Chem. Rev. 111, 5899–5921 [PMC free article] [PubMed]
4. Smith W. L., Urade Y., and Jakobsson P. J. (2011) Enzymes of the cyclooxygenase pathways of prostanoid biosynthesis. Chem. Rev. 111, 5821–5865 [PMC free article] [PubMed]
5. Grosser T., Yu Y., and Fitzgerald G. A. (2010) Emotion recollected in tranquility: lessons learned from the COX-2 saga. Annu. Rev. Med. 61, 17–33 [PubMed]
6. Kulmacz R. J., and Lands W. E. (1984) Prostaglandin H synthase. Stoichiometry of heme cofactor. J. Biol. Chem. 259, 6358–6363 [PubMed]
7. Dong L., Vecchio A. J., Sharma N. P., Jurban B. J., Malkowski M. G., and Smith W. L. (2011) Human cyclooxygenase-2 is a sequence homodimer that functions as a conformational heterodimer. J. Biol. Chem. 286, 19035–19046 [PMC free article] [PubMed]
8. Zou H., Yuan C., Dong L., Sidhu R. S., Hong Y. H., Kuklev D. V., and Smith W. L. (2012) Human cyclooxygenase-1 activity and its responses to COX inhibitors are allosterically regulated by nonsubstrate fatty acids. J. Lipid Res. 53, 1336–1347 [PMC free article] [PubMed]
9. Dong L., Sharma N. P., Jurban B. J., and Smith W. L. (2013) Pre-existent asymmetry in the human cyclooxygenase-2 sequence homodimer. J. Biol. Chem. 288, 28641–28655 [PMC free article] [PubMed]
10. Rimon G., Sidhu R. S., Lauver D. A., Lee J. Y., Sharma N. P., Yuan C., Frieler R. A., Trievel R. C., Lucchesi B. R., and Smith W. L. (2010) Coxibs interfere with the action of aspirin by binding tightly to one monomer of cyclooxygenase-1. Proc. Natl. Acad. Sci. U.S.A. 107, 28–33 [PubMed]
11. Wada M., DeLong C. J., Hong Y. H., Rieke C. J., Song I., Sidhu R. S., Yuan C., Warnock M., Schmaier A. H., Yokoyama C., Smyth E. M., Wilson S. J., FitzGerald G. A., Garavito R. M., Sui de X., Regan J. W., and Smith W. L. (2007) Specificities of enzymes and receptors of prostaglandin pathways with arachidonic acid and eicosapentaenoic acid derived substrates and products. J. Biol. Chem. 282, 22254–22266 [PubMed]
12. Yuan C., Sidhu R. S., Kuklev D. V., Kado Y., Wada M., Song I., and Smith W. L. (2009) Cyclooxygenase allosterism, fatty acid-mediated cross-talk between monomers of cyclooxygenase homodimers. J. Biol. Chem. 284, 10046–10055 [PMC free article] [PubMed]
13. Sharma N. P., Dong L., Yuan C., Noon K. R., and Smith W. L. (2010) Asymmetric acetylation of the cyclooxygenase-2 homodimer by aspirin and its effects on the oxygenation of arachidonic, eicosapentaenoic, and docosahexaenoic acids. Mol. Pharmacol. 77, 979–986 [PubMed]
14. Kudalkar S. N., Nikas S. P., Kingsley P. J., Xu S., Galligan J. J., Rouzer C. A., Banerjee S., Ji L., Eno M. R., Makriyannis A., and Marnett L. J. (2015) 13-Methylarachidonic acid is a positive allosteric modulator of endocannabinoid oxygenation by cyclooxygenase. J. Biol. Chem. 290, 7897–7909 [PMC free article] [PubMed]
15. Reddy S. T., and Herschman H. R. (1994) Ligand-induced prostaglandin synthesis requires expression of the TIS10/PGS-2 prostaglandin synthase gene in murine fibroblasts and macrophages. J. Biol. Chem. 269, 15473–15480 [PubMed]
16. Norris P. C., and Dennis E. A. (2012) ω-3 fatty acids cause dramatic changes in TLR4 and purinergic eicosanoid signaling. Proc. Natl. Acad. Sci. U.S.A. 109, 8517–8522 [PubMed]
17. Jiang Y., Djuric Z., Sen A., Ren J., Kuklev D., Waters I., Zhao L., Uhlson C. L., Hong Y. H., Murphy R. C., Normolle D. P., Smith W. L., and Brenner D. E. (2014) Biomarkers for personalizing ω-3 fatty acid dosing. Cancer Prev. Res. 7, 1011–1022 [PMC free article] [PubMed]
18. Kuklev D. V., Hankin J. A., Uhlson C. L., Hong Y. H., Murphy R. C., and Smith W. L. (2013) Major urinary metabolites of 6-keto-prostaglandin F2α in mice. J. Lipid Res. 54, 1906–1914 [PMC free article] [PubMed]
19. Mbonye U. R., Yuan C., Harris C. E., Sidhu R. S., Song I., Arakawa T., and Smith W. L. (2008) Two distinct pathways for cyclooxygenase-2 protein degradation. J. Biol. Chem. 283, 8611–8623 [PubMed]
20. Kulmacz R. J., Pendleton R. B., and Lands W. E. (1994) Interaction between peroxidase and cyclooxygenase activities in prostaglandin-endoperoxide synthase. Interpretation of reaction kinetics. J. Biol. Chem. 269, 5527–5536 [PubMed]
21. Liu J., Seibold S. A., Rieke C. J., Song I., Cukier R. I., and Smith W. L. (2007) Prostaglandin endoperoxide H synthases: peroxidase hydroperoxide specificity and cyclooxygenase activation. J. Biol. Chem. 282, 18233–18244 [PubMed]
22. Song I., Ball T. M., and Smith W. L. (2001) Different suicide inactivation processes for the peroxidase and cyclooxygenase activities of prostaglandin endoperoxide H synthase-1. Biochem. Biophys. Res. Commun. 289, 869–875 [PubMed]
23. Tanabe T., and Tohnai N. (2002) Cyclooxygenase isozymes and their gene structures and expression. Prostaglandins Other Lipid Mediat. 68, 95–114 [PubMed]
24. Kang Y. J., Mbonye U. R., DeLong C. J., Wada M., and Smith W. L. (2007) Regulation of intracellular cyclooxygenase levels by gene transcription and protein degradation. Prog. Lipid Res. 46, 108–125 [PMC free article] [PubMed]
25. Van der Ouderaa F. J., Buytenhek M., Nugteren D. H., and Van Dorp D. A. (1977) Purification and characterisation of prostaglandin endoperoxide synthetase from sheep vesicular glands. Biochim. Biophys. Acta 487, 315–331 [PubMed]
26. Habib A., Créminon C., Frobert Y., Grassi J., Pradelles P., and Maclouf J. (1993) Demonstration of an inducible cyclooxygenase in human endothelial cells using antibodies raised against the carboxyl-terminal region of the cyclooxygenase-2. J. Biol. Chem. 268, 23448–23454 [PubMed]
27. Wada M., Saunders T. L., Morrow J., Milne G. L., Walker K. P., Dey S. K., Brock T. G., Opp M. R., Aronoff D. M., and Smith W. L. (2009) Two pathways for cyclooxygenase-2 protein degradation in vivo. J. Biol. Chem. 284, 30742–30753 [PMC free article] [PubMed]
28. Yuan C., and Smith W. L. (2015) A cyclooxygenase-2-dependent prostaglandin E2 biosynthetic system in the Golgi apparatus. J. Biol. Chem. 290, 5606–5620 [PMC free article] [PubMed]
29. Otto J. C., DeWitt D. L., and Smith W. L. (1993) N-Glycosylation of prostaglandin endoperoxide synthases-1 and -2 and their orientations in the endoplasmic reticulum. J. Biol. Chem. 268, 18234–18242 [PubMed]
30. Mbonye U. R., Wada M., Rieke C. J., Tang H.-Y., Dewitt D. L., and Smith W. L. (2006) The 19-amino acid cassette of cyclooxygenase-2 mediates entry of the protein into the endoplasmic reticulum-associated degradation system. J. Biol. Chem. 281, 35770–35778 [PubMed]
31. Sidhu R. S., Lee J. Y., Yuan C., and Smith W. L. (2010) Comparison of cyclooxygenase-1 crystal structures: cross-talk between monomers comprising cyclooxygenase-1 homodimers. Biochemistry 49, 7069–7079 [PMC free article] [PubMed]
32. Kiefer J. R., Pawlitz J. L., Moreland K. T., Stegeman R. A., Hood W. F., Gierse J. K., Stevens A. M., Goodwin D. C., Rowlinson S. W., Marnett L. J., Stallings W. C., and Kurumbail R. G. (2000) Structural insights into the stereochemistry of the cyclooxygenase reaction. Nature 405, 97–101 [PubMed]
33. Vecchio A. J., Simmons D. M., and Malkowski M. G. (2010) Structural basis of fatty acid substrate binding to cyclooxygenase-2. J. Biol. Chem. 285, 22152–22163 [PMC free article] [PubMed]
34. Kulmacz R. J., and Lands W. E. (1985) Stoichiometry and kinetics of the interaction of prostaglandin H synthase with anti-inflammatory agents. J. Biol. Chem. 260, 12572–12578 [PubMed]
35. Swinney D. C., Mak A. Y., Barnett J., and Ramesha C. S. (1997) Differential allosteric regulation of prostaglandin H synthase 1 and 2 by arachidonic acid. J. Biol. Chem. 272, 12393–12398 [PubMed]
36. Yuan C., Rieke C. J., Rimon G., Wingerd B. A., and Smith W. L. (2006) Partnering between monomers of cyclooxygenase-2 homodimers. Proc. Natl. Acad. Sci. U.S.A. 103, 6142–6147 [PubMed]
37. Prusakiewicz J. J., Duggan K. C., Rouzer C. A., and Marnett L. J. (2009) Differential sensitivity and mechanism of inhibition of COX-2 oxygenation of arachidonic acid and 2-arachidonoylglycerol by ibuprofen and mefenamic acid. Biochemistry 48, 7353–7355 [PMC free article] [PubMed]
38. Duggan K. C., Hermanson D. J., Musee J., Prusakiewicz J. J., Scheib J. L., Carter B. D., Banerjee S., Oates J. A., and Marnett L. J. (2011) (R)-Profens are substrate-selective inhibitors of endocannabinoid oxygenation by COX-2. Nat. Chem. Biol. 7, 803–809 [PMC free article] [PubMed]
39. Yu Y., Cheng Y., Fan J., Chen X.-S., Klein-Szanto A., Fitzgerald G. A., and Funk C. D. (2005) Differential impact of prostaglandin H synthase 1 knockdown on platelets and parturition. J. Clin. Invest. 115, 986–995 [PubMed]
40. Muth-Selbach U. S., Tegeder I., Brune K., and Geisslinger G. (1999) Acetaminophen inhibits spinal prostaglandin E2 release after peripheral noxious stimulation. Anesthesiology 91, 231–239 [PubMed]
41. Lapicque F., Vergne P., Jouzeau J. Y., Loeuille D., Gillet P., Vignon E., Thomas P., Velicitat P., Türck D., Guillaume C., Gaucher A., Bertin P., and Netter P. (2000) Articular diffusion of meloxicam after a single oral dose: relationship to cyclo-oxygenase inhibition in synovial cells. Clin. Pharmacokinet. 39, 369–382 [PubMed]
42. Burian M., and Geisslinger G. (2005) COX-dependent mechanisms involved in the antinociceptive action of NSAIDs at central and peripheral sites. Pharmacol. Ther. 107, 139–154 [PubMed]
43. Brash A. R. (2001) Arachidonic acid as a bioactive molecule. J. Clin. Invest. 107, 1339–1345 [PMC free article] [PubMed]
44. Shitashige M., Morita I., and Murota S. (1998) Different substrate utilization between prostaglandin endoperoxide H synthase-1 and -2 in NIH3T3 fibroblasts. Biochim. Biophys. Acta 1389, 57–66 [PubMed]
45. Surette M. E., and Chilton F. H. (1998) The distribution and metabolism of arachidonate-containing phospholipids in cellular nuclei. Biochem. J. 330, 915–921 [PubMed]
46. McIntosh A. L., Huang H., Atshaves B. P., Wellberg E., Kuklev D. V., Smith W. L., Kier A. B., and Schroeder F. (2010) Fluorescent n-3 and n-6 very long chain polyunsaturated fatty acids: three-photon imaging in living cells expressing liver fatty acid-binding protein. J. Biol. Chem. 285, 18693–18708 [PMC free article] [PubMed]
47. Jump D. B., Depner C. M., and Tripathy S. (2012) ω-3 fatty acid supplementation and cardiovascular disease. J. Lipid Res. 53, 2525–2545 [PMC free article] [PubMed]
48. Liu W., Cao D., Oh S. F., Serhan C. N., and Kulmacz R. J. (2006) Divergent cyclooxygenase responses to fatty acid structure and peroxide level in fish and mammalian prostaglandin H synthases. FASEB J. 20, 1097–1108 [PubMed]
49. Kingma P. B., Bok D., and Ong D. E. (1998) Bovine epidermal fatty acid-binding protein: determination of ligand specificity and cellular localization in retina and testis. Biochemistry 37, 3250–3257 [PubMed]

Articles from The Journal of Biological Chemistry are provided here courtesy of American Society for Biochemistry and Molecular Biology