Arachidonate-derived Dihomoprostaglandin Production Observed in Endotoxin-stimulated Macrophage-like Cells*

Eicosanoids, including the prostaglandins, leukotrienes, hydroxyeicosatetraenoic acids, epoxyeicosatetraenoic acids, and related compounds, are biosynthetic, bioactive mediators derived from arachidonic acid (AA), a 20:4(n-6) fatty acid. We have developed a comprehensive and sensitive mass spectral analysis to survey eicosanoid release from endotoxin-stimulated RAW 264.7 macrophage-like cells that is capable of detecting over 70 diverse eicosanoids and eicosanoid metabolites, should they be present. We now address the question: Are biologically significant eicosanoids being overlooked? Herein, we illustrate a general approach to diverse isotope metabolic profiling of labeled exogenous substrates using mass spectrometry (DIMPLES/MS), demonstrated for one substrate (AA) and its resultant products (eicosanoids). RAW cells were incubated in medium supplemented with deuterium-labeled AA. When the cells are stimulated, two sets of eicosanoids are produced, one from endogenous AA and the other from the supplemented (exogenous) deuterium-labeled form. This produces a signature mass spectral “doublet” pattern, allowing for a comprehensive and diverse eicosanoid search requiring no previous knowledge or assumptions as to what these species may be, in contrast to traditional methods. We report herein observing unexpected AA metabolites generated by the cells, some of which may constitute novel bioactive eicosanoids or eicosanoid inactivation metabolites, as well as demonstrating differing metabolic pathways for the generation of isomeric prostaglandins and potential peroxisome proliferator-activated receptor activators. Unexpectedly, we report observing a series of 1a, 1b-dihomologue prostaglandins, products of adrenic acid (22:4(n-6)), resulting from the two-carbon elongation of AA by the RAW cells.

played a critical role in the biochemical study of eicosanoids. More recent advances in electrospray ionization-MS coupled to high-performance liquid chromatography have offered extremely sensitive and quantitative assays for most of the eicosanoids (2), without the need for chemical derivatization prior to analysis as was required by earlier gas chromatography electron ionization-MS methods (3). It is important to recognize, however, that while advances in high-performance LC-MS methods have offered extensive capabilities for surveying a number of different eicosanoid species in a single analysis, to date most efforts have focused on a specific eicosanoid or eicosanoid class, and additionally, with advanced knowledge and assumptions as to the identity of these species (4 -11). Recently, however, investigators have begun to explore the development of theoretical databases and algorithms based on virtual liquid chromatography-UV spectroscopy-tandem mass spectrometry (MS/MS) spectra and chromatograms for identifying potential lipid mediators without synthetic or authentic products as standards (12).
that occur with cellular metabolism, and no such strategy has yet to emerge. We have begun to address such a strategy using a single cell type, the RAW 264.7 macrophage-like cell, and a single stimulus, the high purity saccharolipid Kdo 2 -Lipid A component of the endotoxin lipopolysaccharide (14) for the production of eicosanoids and their downstream metabolites. Specifically, a set of dose-response and time-course studies was carried out measuring eicosanoid release from Kdo 2 -Lipid A-stimulated RAW cells. Eicosanoids 3 were detecting employing liquid chromatography coupled with MS and by operating the mass spectrometer in the ultrasensitive multiple reaction monitoring (MRM) mode, we had the capability to detect over 70 different eicosanoid and inactivation metabolite molecular species in a single analysis should they be present. Although this method does offer the capability for detecting a large number of diverse eicosanoid species, similarly it carries the limitations of the methods cited earlier (4 -11); namely, that assumptions have to be made in advance regarding the species that might be present, to provide MRM pairs for the detection scheme. With our goal being to conduct a global survey of the eicosanoids produced by the stimulated RAW cells, we now address the question: Are biologically significant eicosanoids being overlooked? That is, are there species for which we have no prior knowledge of or expectation of their presence and, hence, no available MRM pairs that would be required for their detection? In the present work, an MS-based general stable isotope exogenous substrate labeling strategy, which we coin "DIM-PLES/MS" for diverse isotope metabolic profiling of labeled exogenous substrates using mass spectrometry, approaching a global metabolite survey is illustrated. Herein we demonstrate the DIMPLES/MS approach using one substrate, arachidonic acid (AA), and its RAW cell generated eicosanoids and related metabolites. RAW cells are incubated in medium supplemented with deuterium-labeled arachidonic acid (AA-d 8 ) and then stimulated with Kdo 2 -Lipid A. Two sets of eicosanoid generation result, one set from endogenous AA, the other from the supplemented exogenous AA-d 8 . This results in a "doublet" pattern, resolvable with mass spectrometry, allowing for a sensitive and comprehensive eicosanoid search without any previous knowledge or assumptions as to what these species may be, in contrast to traditional methods. Additionally, the AA-d 8generated metabolites exhibit particular mass spectral patterns that may aid in determining possible functional groups and biosynthetic pathways, this being particularly useful when exploring for novel eicosanoids. Although DIMPLES/MS is demonstrated in this work for AA and eicosanoids, its utility should prove equally valuable for other substrates and the diverse analysis of their corresponding metabolites. Using DIMPLES/MS we report herein observing unidentified metabolic products of AA generated by Kdo 2 -Lipid A-stimulated RAW cells, which have the potential for being novel eicosanoids. If these unidentified species do exhibit particularly interesting and biologically relevant properties, the labor-intensive work required to determine their specific molecular details can be undertaken. Additionally, this approach reveals different detailed biosynthetic pathways for isomeric prostaglandins produced in these cells. Unexpectedly and of particular interest, we report observing a number of 22-carbon 1a,1b-dihomologue prostaglandins (dihomoprostaglandins), products of adrenic acid (22:4(n-6)) resulting from the two-carbon elongation of AA by the RAW cells. Although previously it has been observed that adrenic acid can serve as a COX substrate when added to cells, resulting in the production of dihomoprostaglandins (15)(16)(17), there has been no demonstration of its formation de novo from an arachidonate precursor.

EXPERIMENTAL PROCEDURES
Materials-Dulbecco's modified Eagle's medium, penicillin/streptomycin, the Quant-iT DNA assay kit and the Vybrant Cytotoxicity assay kit were purchased from Invitrogen. Heat-inactivated fetal calf serum was purchased from HyClone (Logan, UT). Kdo 2 -Lipid A (Lipid MAPS ID no. LMSL02000001; Avanti catalog no. 699500) was obtained from Avanti Polar Lipids (Alabaster, AL). All the standards used to compile our library of eicosanoid standards, including octa-deuterated AA (AA-d 8 ) used for supplementation, were purchased from Cayman Chemical (Ann Arbor, MI). Strata-X polymeric sorbent solid-phase extraction columns (catalogue no. 8B-S100-UBJ) were purchased from Phenomenex (Torrance, CA), and the vacuum manifold used with these extraction columns was purchased from Supelco (Bellefonte, PA). All reagents were chromatography grade and purchased from OmniSolv (Gibbstown, NJ).
Tissue Culture and AA-d 8 Supplementation-RAW 264.7 mouse macrophage-like cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal calf serum with low endotoxin content, 100 units/ml penicillin, and 100 g/ml streptomycin in an incubator at 37°C and 10% CO 2 . 4 ϫ 10 6 of these cells were plated in T-25 flasks with 4 ml of the same medium supplemented with and without 12.5 g/ml (40 M) of AA-d 8 for a 24-h period. Both sets of cells were then treated with 100 ng/ml of Kdo 2 -Lipid A or an equivalent amount of Me 2 SO-containing medium as a vehicle control for an additional 24 h. Thus four cell sample types were generated: Kdo 2 -Lipid A treated with and without AA-d 8 supplementation and non-treated with and without AA-d 8 supplementation. The cultured medium from each set was then collected, centrifuged for 3 min at 1000 ϫ g to remove any floating cells, supplemented with 10% methanol, and subjected to an eicosanoid extraction procedure.
Eicosanoid/Fatty Acid Extraction Procedure-Solid-phase extraction of eicosanoids and fatty acids was performed using Strata-X polymeric sorbent columns connected to a vacuum manifold. Columns were pre-washed using 2 ml of methanol followed by 2 ml of water. Samples were loaded onto columns, washed with 2 ml of 90/10 water/methanol, then eluted with 1 ml of 100% methanol. The methanol was then evaporated to dryness using a SpeedVac, and the remaining sample was reconstituted with 100 l of liquid chromatography buffer A.
Reversed-phase Liquid Chromatography-High-performance liquid chromatography was carried out using two Shimadzu (Columbia, MD) LC-10AD high performance pumps interfaced with a Shimadzu SCL-10A controller. Separation was performed using a 2.1-ϫ 250-mm Vydac (Hesperia, CA) reversed-phase C 18 column (catalogue no. 201TP52) equipped with a guard column (Vydac, catalogue no. 201GD52T) held at 35°C. LC buffer A was water/acetonitrile/formic acid: 63/37/ 0.02, v/v; buffer B was acetonitrile/isopropanol: 50/50, v/v. Gradient elution was achieved using 100/0, A/B at 0 min and linearly ramped to 80/20, A/B by 6 min; linearly ramped to 45/55, A/B by 6.5 min and held until 11 min, linearly ramped to 28/72, A/B by 11.5 min and held until 16 min, then linearly ramped back to 100/0, A:B by 18 min and held there until 25 min to achieve column re-equilibration. In some cases (details listed later) a second gradient was employed using 100/0, A/B at 0 min and linearly ramped to 25/75, A/B by 16 min and held until 20 min, then linearly ramped back to 100/0, A:B by 22 min and held there until 30 min to achieve column re-equilibration. The buffer flow rate was 0.3 ml/min for both gradients. 10 and 30 l of sample were injected onto the column using a Leap Technologies (Carrboro, NC) PAL autosampler. The liquid chromatography effluent was coupled to a mass spectrometer for further analysis.
Chiral Chromatography-Liquid normal phase chiral chromatography was carried out using the same pumping system described above for reversed-phase chromatography. Separation was carried out on a 4.6-ϫ 250-mm Chiral Technologies (West Chester, PA) derivatized amylose column (Chiralpak AD-H) equipped with a guard column (Chiralpak AD-H guard column) held at 35°C. Buffer A was hexane/anhydrous ethanol/water/formic acid: 96/4/0.08/0.02, v/v; buffer B was 100% anhydrous ethanol. This small amount of water in buffer A is miscible in the hexane/anhydrous ethanol mix and was found to be vital for satisfactory chiral separation and peak shape. Gradient elution was achieved using 100/0, A/B at 0 min and linearly ramped to 90/10, A/B by 13 min; linearly ramped to 75/25, A/B by 15 min and held until 25 min, then linearly ramped back to 100/0, A:B by 27 min and held there until 42 min to achieve column re-equilibration. The chiral chromatography effluent was coupled to a mass spectrometer for further analysis.
MS-All mass spectral analyses were performed using an Applied Bioscience (Foster City, CA) 4000 QTRAP hybrid triple quadrupole linear ion-trap mass spectrometer equipped with a Turbo V ion source and operated in full-scan, MS/MS, and multiple reaction monitoring (MRM) modes. Details regarding parameters specific to each mode of operation are provided below. For all experiments the Turbo V ion source was operated in negative electrospray mode (chiral chromatography utilized the ion source in chemical ionization mode; see below), and the QTRAP was set as follows: CUR ϭ 10 p.s.i., GS1 ϭ 40 p.s.i., GS2 ϭ 40 p.s.i., IS ϭ Ϫ4200 V, collisional activated dissociation ϭ HIGH, TEM ϭ 525°C, ihe ϭ ON, DP ϭ Ϫ30 V, EP ϭ Ϫ15 V and CXP ϭ Ϫ10 V. The voltage used for collisional activated dissociation varied according to molecular species and ranged from Ϫ20 to Ϫ45 V.
Doublet Peak-picking Software-Raw mass spectral generated data files (ABI*.wiff) were acquired at an m/z resolution of 0.1 atomic mass unit and then converted to mzXML format (18) using the mzStar utility (www.sashimi.sourceforge.net/). The mzXML files were processed with a custom in-house program to integrate peaks in each full mass scan over a 0.6-atomic mass unit window and generate a formatted array of quantitated data for each mass spectral sample run. A set of customized Excel Visual Basic macros was then used to load a pair of array files (AA-d 8 supplemented and control), perform data normalization and subtractive analyses, view peaks of interest, and generate a list of detected mass spectral "doublets," according to user-defined criteria. Fig. 1A. The uptake of AA-d 8 by RAW cells as a function of time was determined by drawing an aliquot of medium from a plate of cells in 4 ml of medium supplemented with 12.5 g/ml (ϳ40 M) AA-d 8 at 0, 0.5, 2, 5, and 18 h after initial supplementation, carrying out eicosanoid/fatty acid extraction, and measuring the relative amount of AA-d 8 ; this was also done for a sample identical to the above, however, containing no cells (control). A comparison of these two measurements permitted a determination of the amount of AA-d 8 remaining in the cellcontaining medium at the different time periods, and the decrease could be attributed to uptake by the cells (Fig. 1B). LC-MRM-MS/MS (MRM pair of 311/267) was used to determine the relative amount of AA-d 8 in the medium, as defined by the area under its chromatographic peak, and its retention time was consistent with a standard. Triplicate samples were measured to determine the error shown in Fig. 1B. Note that at 18 h only a few percent of the initial AA-d 8 remained in the medium indicating that by the time of Kdo 2 -Lipid A stimulation (after a 24-h incubation period) virtually all the AA-d 8 was incorporated into the cells.

AA-d 8 Incorporation-A timeline depicting the supplementation of AA-d 8 to plated cells, Kdo 2 -Lipid A stimulation, medium collection and extraction, and lastly LC-MS analysis is shown in
AA/AA-d 8 Mass Spectral Doublet-The media from the four cell sample types (Kdo 2 -Lipid A stimulation with and without AA-d 8 (control), and no stimulation with and without AA-d 8 ) were separately collected, extracted for eicosanoids/fatty acids, and resuspended in 100 l of LC buffer A. A separate LC-MS analysis was performed for each, injecting 30 l of sample onto the LC column and operating the mass spectrometer in full scan mode (m/z 280 -400). The spectrum obtained for the stimulated/control sample (without AA-d 8 ) produced a number of peaks at various m/z values, most noticeably a very intense peak at m/z 351 accompanied by its C 13 isotopic peaks at 352 and 353 ( Fig. 2A, solid line plot) that had an LC retention time consistent with that of a prostaglandin D 2 (PGD 2 ) standard. The intensity of this peak was consistent with our previous Kdo 2 -Lipid A stimulated RAW cell results (14), in which PGD 2 has repeatedly been an eicosanoid produced in high abundance, typically 50 -100 ng/10 6 cells.
A spectrum obtained for the stimulated/AA-d 8 sample ( Fig.  2A, dashed line plot) contained peaks at m/z 351, 352, as 353 as before, but additionally a peak pattern offset 4 -8 atomic mass units up from the 351 peak. Overlaying these two spectra ( Fig.  2A) produces a conspicuous "doublet" pattern characterized by the peaks at 351, 352, and 353 present in both the AA-d 8 and control samples, and the higher offset peaks at 355-359 present only in the AA-d 8 sample and noticeably absent in the control sample. The higher offset peaks have a peculiar distribution shape, different from their non-deuterated counterpart, and we speculate that this is partly due to the actual number of deuterium atoms present on the deuterated-AA molecule used for  supplementation; Fig. 1C shows the molecular structure of deprotonated AA-d 8 and the location of its deuterium atoms. A direct-infusion (sample was not run through an LC column) mass spectral analysis of the AA-d 8 stock solution used for the supplementation shows a portion of the molecules contained not only the maximal 8 deuterium atoms, but 7 (AA-d 7 ) and 6 (AA-d 6 ) as well in the ratio of ϳ65%, 30, and 5%, respectively (Fig. 1D). This ratio was determined using the integrated area under the three peaks, and any C 13 contribution was ignored in this estimation. It is additionally noted that when the deuterated-AA was placed in cell medium (with serum) for a 48-h period, this produced no noticeable change in its deuterium atom containing distribution (data not shown).
It is important to note that, while the majority of the supplemented deuterated AA contains 8 deuterium atoms, the majority of the generated PGD 2 contains 6 deuterium atoms (PGD 2 ϩd 6 , m/z ϭ 357, Fig. 2A). The formation of a ketone group on C-11 of PGD 2 (Fig. 2B) requires the loss of a deuterium atom from this position of its AA-d 8 precursor (Fig. 1C); however, because the peak at m/z ϭ 357 in Fig. 2A (PGD 2 ϩd 6 ) is so strongly present it is likely that another deuterium atom is being lost from AA-d 8 during its conversion into PGD 2 , i.e. from C-8, C-12, C-15, or any of the other four remaining loca-tions. It should be noted, however, that, because the peak at m/z 358 in Fig. 2A is present at a higher level than can be accounted for solely from a C 13 isotopic contribution, PGD 2 ϩd 7 is also formed (if the peak at m/z ϭ 358 was solely due to a C 13 contribution one would expect its intensity to be ϳ20% the intensity observed in the peak at m/z ϭ 357). Therefore, the loss of this second deuterium atom appears to not be an absolute occurrence, as is the C-11 deuterium atom loss, but a loss that occurs a significant portion of the time during the AA-d 8 conversion to PGD 2 .
Comparison of the two non-stimulated cell sample types (with/without AA-d 8 supplementation) shows a similar pattern as described above, the doublet pattern in the AA-d 8supplemented sample compared with a major single peak at m/z 351, accompanied by its C 13 isotopic peaks, in the non-supplemented sample. These were at basal levels in both samples, roughly 100-fold below stimulated intensity levels, a trend that has been previously observed (14). This would indicate that the AA-d 8 supplementation per se does not noticeably up-regulate eicosanoid production in the cell. Fig. 3 (A-D) shows the profiles generated during an LC-full mass scan for some of the other known eicosanoids produced by the AA-d 8 supplemented, Kdo 2 -Lipid A-stimulated cells. It should be noted that for some of these metabolites (i.e. PGF 2␣ and 11-HETE) the most abundant AA-d 8 -generated products contain 8 deuterium atoms, whereas for others (PGD 2 , PGE 2 , and 11␤-PGF 2␣ ) 6 deuterium atoms are contained in the most abundant product. The significance of these differences is addressed under "Discussion." Chiral chromatography (19) coupled to MRM mass spectrometry showed the 11-HETE species to be exclusively 11(R)-HETE (Fig. 4, A and B).
AA/AA-d 8 Doublet Recognition Software-In-house customized software has been developed for efficiently processing and comparing control and AA-d 8 (Table 1, discussed below). One of the hits was a species with m/z of 381 and LC-RT of ϳ6.5 min (Fig. 5A), which did not match any entry in our "Library of Eicosanoid Standards" ( Table 2). We have compiled a comprehensive library of eicosanoid standards, which is available in the supplemental material. The library contains information on each standard listed in Table 2 that includes LC-RT, systematic nomenclature and LIPID MAPS ID number (20), and an MS/MS fragmentation spectrum. If other reversed-phase liquid chromatography conditions are employed, these retention times can still be useful, because the order of elution should not drastically change; e.g. PGE 2 elutes prior to PGD 2 and 15-HETE elutes prior to 11-HETE, which elutes prior to 5-HETE. An overlay of the corresponding full-scan mass spectra obtained for the m/z of 381 metabolite discussed above is shown in Fig. 5B.
Tandem Mass Spectral Analysis of an Unidentified, Unknown Compound-Having recognized an unidentified, unknown compound at m/z 381 and LC-RT Ϸ 6.5 m, a tandem mass spectrum (collision energy of Ϫ40 V) was obtained and is shown in Fig. 5C. The fragment ions observed in the MS/MS spectrum are used to generate MRM pairs, allowing the mass spectrometer to be operated in a highly sensitive detection mode. An 18-atomic mass unit loss from the parent ion (M-H) Ϫ is typically indicative of a water loss (hydroxyl group). Note the fragment peaks at 363, 345, and 327 in Fig. 5C suggesting the possibility of 3-hydroxyl groups contained on the parent ion. Interpretation of this spectrum is addressed in more detail under "Discussion." Recognized Unknowns-A number of other hits were identified using our AA/AA-d 8 doublet recognition software for which there were no matches, based on m/z and LC-RT, in our eicosanoid library. A tandem mass spectrum was obtained for each of these unknowns, and similarly this information was used to generate MRM pairs. A partial list of these unknowns, including their most intense product ions and LC retention times, are listed in Table 1. Care was taken to ensure these species were not simply known compounds with water loses or adducted ions.
Some of the compounds listed in Table 1 were initially considered unidentified, however, upon further examination these were determined to be COX products of adrenic acid (22:4(n-6)), with the adrenic acid apparently resulting from the twocarbon elongation of AA by the RAW cells. The presence of deuterated adrenic acid and adrenic acid products (Fig. 6, A and  B) strongly suggests their arachidonate origin. The tentative identities of these dihomoprostaglandins are listed in Table 1 (synthetic standards and additional approaches will be used to verify their identities). This unexpected and particularly interesting result is discussed in further detail below.
Semi-quantitative Comparison of AA, Adrenic Acid, and Their Products-The full-scan mass spectra obtained for AA, adrenic acid, and their products permitted a semi-quantitative comparison of these species. The intensity of product was obtained by integrating the area under the mono-isotopic parent ion ([M-H] Ϫ ) peak in full-scan mass spectra. To determine if this measurement was appropriately correlated to product amount, equal molar amounts of PGF 2␣ and dihomo-PGF 2␣ standards were loaded onto the reversed-phase liquid chromatography column, and it was observed that the areas corresponding to their mono-isotopic parent ion peak in their full- scan mass spectra were indeed similar (within 10%). A semiquantitative comparison of AA, adrenic acid, and their corresponding products obtained from Kdo 2 -Lipid A-stimu-lated RAW cells is shown in Fig. 7 The number above each pair is the ratio of adrenic acid to AA (products). It should be noted that no dihomo-15d-⌬ 12,14 -PGJ 2 was observed.    (Table 2). Primary product ions are listed for each unknown in descending order of intensity, determined through tandem mass spectral analysis (CE used is listed). b Initially unidentified compounds were later determined to be cyclooxygenase products of adrenic acid (22:4(n-6)) and tentatively identified as specific dihomoprostaglandins, resulting from the two-carbon elongation of AA (20:4(n-6)) at the carboxyl end by the RAW cells. Refer to text for more details.

DIMPLES/MS and Unexpected
Metabolites-A strategy has been described for diverse isotope metabolic profiling of labeled substrates using mass spectrometry (DIMPLES/MS), devoid of any previous knowledge and assumptions as to what these metabolites may be. Details are provided by using deuterium-labeled AA (AA-d 8 ) and profiling its eicosanoid metabolites, however, we believe this strategy can be extended to other labeled exogenous substrates and their metabolic products.
When RAW macrophage-like cells are incubated in medium supplemented with AA-d 8 , over time the endogenous fatty acids esterified to the sn-2 position of cell membrane glycerophospholipids are exchanged for the exogenous deuterated AA analog, however it appears that the endogenous AA is not completely replaced. When the cells are subjected to stimulation by lipopolysaccharide-like Kdo 2 -Lipid A, our results show that the incorporated AA-d 8 behaves in a similar fashion to the endogenous AA of the cell, a cascade commences with the activation of phospholipase A 2 , which hydrolyzes the sn-2 acyl ester bond resulting in a non-indiscriminating release of both free AA and AA-d 8 . Both can then act as substrates for downstream enzymes, as is demonstrated by the production of PGD 2 and PGD 2 ϩ[d 5 -7 ] ( Fig. 2A) as well as other known eicosanoids (Fig.  3, A-D), producing a conspicuous doublet pattern that is both observable with mass spectrometry and amenable to efficient processing of data sets using customized recognition software. Unexpectedly, dihomoprostaglandins, COX products of adrenic acid (22:4(n-6)), were also observed to be produced by the stimulated RAW, cells and a more detailed discussion regarding this finding is provided at the end of this section.

DIMPLES/MS Reveals Differing Mechanistic Origins and Metabolic
Pathways of Isomeric Prostaglandins-The full scan mass spectral peak patterns generated with the AA-d 8 -supplemented samples can provide insight into the origins of AA-derived metabolites and also into details on their molecular structures. An example of this is demonstrated with the different peak patterns observed for 11␤-PGF 2␣ and PGF 2␣ (Fig. 3, B and C, respectively). Although these two compounds are stereoisomers of one another having identical fragmentation patterns (fortunately, 11␤-PGF 2␣ has a shorter LC-RT than PGF 2␣ allowing them to be resolved in our analysis), they have significantly different origins. 11␤-PGF 2␣ is an enzymatic product of PGD 2 (action of PGD 2 11-ketoreductase), whereas PGF 2␣ forms directly from PGH 2 (action of PGH 2 9,11-endoperoxide reductase) (21). The peak pattern observed for 11␤-PGF 2␣ , with ([Mϩd 6 ]-H) Ϫ being the most intense peak of the deuterated components and resembling the pattern obtained for PGD 2 ( Fig. 2A), supports that it does indeed originate from PGD 2 . The peak pattern observed for PGF 2␣ is markedly different, with ([Mϩd 8 ]-H) Ϫ being the most intense peak of the deuterated components. Originating directly from PGH 2 requires no deuterium loss at C11; a loss that is required if it were to originate from PGD 2 and discussed earlier. The mechanism for the formation PGE 2 ( Fig. 2A) is similar to that for PGD 2 except the deuterium atom loss associated with the pentane ring ketogroup occurs at C9 instead of C11 and, here too, for PGE 2 the As stated earlier, chiral chromatography showed the 11-HETE produced by the RAW cells to be exclusively 11(R)-HETE (Fig. 4, A and B). It has been reported that 11(R)-HETE is an abortive COX product, resulting from the AA-to-prostaglandin G 2 pathway (see below) terminating before completion. It should also be noted that 11(R)-HETE (no 11(S)-HETE) would be the exclusive product of this termination process. Fig.  3D shows the full mass scan peak pattern for 11(R)-HETE generated from an AA-d 8 -supplemented, Kdo 2 -Lipid A-stimulated sample. Here, as with PGF2 ␣ , ([Mϩd 8 ]-H) Ϫ is the most intense peak of its deuterated components. For unidentified AA metab-olites, such observed differences (and/or similarities) can lend valuable insight regarding their production pathways, and we therefore now briefly review how eicosanoids are synthesized from AA. Action by the enzyme prostaglandin H synthase (also known as cyclooxygenase or COX) on AA is the first step in all prostaglandin biosynthesis. COX first abstracts one hydrogen atom from the C-13 of AA (refer to Fig. 1C) to form an arachidonate radical, which then reacts sequentially with two oxygen molecules, first creating an endoperoxide between C-9 and C-11 and then forming a hydroperoxide at C-15, thus forming the intermediate prostaglandin G 2 . Next, COX acts as a peroxidase on prostaglandin G 2 , reducing its C-15 hydroperoxide to a hydroxyl group, forming PGH 2 (22,23). PGH 2 can then serve as a substrate for a number of downstream enzymes: PGD 2 isomerase, PGE 2 isomerase, PGF 2␣ reductase, thromboxane synthase, or prostacyclin synthase. This leads us to conclude that the different deuterated profiles we observe for PGD 2 and PGE 2 compared with PGF2 ␣ and 11(R)-HETE are the result of processes that occur after the synthesis of PGH 2 , i.e. due to the action of downstream enzymes.
Observing the deuterated peak pattern in the full scan mass spectrum of the unidentified compound our software recognized at m/z 381, LC-RT 6.5 m (Fig. 5, A and B) with ([Mϩd 8 ]-H) Ϫ being the most intense peak of its deuterated components suggested this compound to be more similar to PGF 2␣ and 11(R)-HETE than to PGD 2 or PGE 2 . This would indicate this compound does not contain a cyclopentane keto-group as seen in PGE 2 and PGD 2 (also observed in the enzymatic and nonenzymatic products of PGD 2 ). As stated earlier, the MS/MS fragments observed at 363, 345, and 327 ( Fig. 5C) suggest the loss of 1-, 2-, and 3-hydroxyl groups, respectively. The most abundant MS/MS fragment at 337, 44 atomic mass units less than the parent ion mass, is a peak that is very characteristic of PGF 2␣ involving a charge-site remote loss of C 2 H 4 O from its cyclopentane ring (24). Another very abundant PGF 2␣ fragment occurs at 193, a 160-atomic mass unit loss resulting from the charge-site remote loss of 44 atomic mass units described above, accompanied by the loss of the methyl terminus side chain and cleaving of the bond between C-13 and C-14 (C 7 H 16 O) (2,24). A 160-atomic mass unit loss from 381 results in a fragment at 221, strongly evident in Fig. 5C. These pronounced similarities between PGF 2␣ and the 381 unknown, as well as their 28-atomic mass unit mass difference, led us to believe this unknown was PGF 2␣ with a C 2 H 4 attachment at the molecule's carboxyl end, such a molecule describes 1a,1b-dihomo-9S,11R,15S-trihydroxy-5Z,13E-prostadienoic acid (dihomo-PGF 2␣ ). Agreement in LC-RT and MS/MS fragmentation with a dihomo-PGF 2␣ standard con-firmed this to be the case. Structural determination for some of the unknowns listed in Table 1 (i.e. UNK 335-1 and UNK 353-1) will require additional experimental approaches.
Detection of Arachidonate-derived Dihomoprostaglandins-In addition to dihomo-PGF 2␣ , a number of other dihomoprostaglandins were observed to be produced by the Kdo 2 -Lipid A-stimulated cells. The tentative identities of these other dihomoprostaglandins (listed in Table 1) were determined using (i) standards when available, (ii) estimation of the increased LC-RT that results from the additional C 2 H 4 attachment (actual LC-RT values were obtained using PGF 2␣ /dihomo-PGF 2␣ and PGE 1 /dihomo-PGE 1 standards, and these were used to extrapolate for the other compounds), and (iii) MS/MS fragmentation patterns. To correlate the LC-RT of the dihomoprostaglandins with their corresponding AA-generated counterparts, an entirely linear LC gradient was employed; this was done to avoid inconsistencies that might occur due to abrupt ramps in buffer during an LC run. The MS/MS fragmentation spectra also confirmed that the elongation of AA occurs at the carboxyl end of the molecule which, from a biosynthetic perspective, is expected. These COX products of adrenic acid (7Z,10Z,13Z,16Z-docosatetraenoic acid) appear to result from the two-carbon elongation of AA prior to the esterification of the adrenic acid to the sn-2 position of the cell membrane glycerophospholipid pool. This is strongly suggested by the observation of deuterated adrenic acid and dihomoprostaglandins, which originate from the supplemented AA-d 8 . along with non-deuterated adrenic acid dihomologue prostaglandins produced from the endogenous AA (Fig. 6, A and B). These findings demonstrate some additional utility of the DIMPLES/MS methodology. The proposed steps in the AA elongation process, and the subsequent production of dihomoprostaglandins by stimulated cells, are depicted in Fig. 8. It should be noted that neither dinor (18 carbons), tetranor (16 carbons), nor tetrahomo (24 carbons) prostaglandins were observed to be produced by the stimulated RAW cells. Although it has been observed previously that adrenic acid added to cells can serve as a COX substrate resulting in the production of dihomoprostaglandins (15)(16)(17), there has been no demonstration of its formation de novo from an arachidonate precursor. As such, the question arises as to why the elongation of AA occurs. The addition of adrenic acid to cell culture was found previously to inhibit the conversion of AA to prostaglandins (25). It is possible, therefore, that the elongation process occurs as a regulatory mechanism, a means of limiting the production of AA derived prostaglandins? Additionally, the dihomoprostaglandins may have direct actions of their own and further investigation of these actions is warranted.
It is also possible that the 2-carbon elongation may be taking place on the already formed 20-carbon prostaglandins; however, the observation of AA-generated adrenic acid (Fig. 6A) raises some doubts about this scenario. Additionally, the observed plentiful generation of some dihomoprostaglandins compared with the meager production of others (discussed in more detail below), raises further questions regarding a prostaglandin two-carbon elongation scheme. The action of an elongase should be fairly non-discriminatory considering its broad, nonspecific class of action. As such, if the elongase was FIGURE 7. Semi-quantitative comparison of AA, adrenic acid, and their metabolites obtained from Kdo 2 -Lipid A-stimulated RAW cells. Full-scan mass spectra obtained for AA, adrenic acid, and their metabolites permitted a semi-quantitative comparison of these species (refer to text for more detail regarding this measurement). The number above each pair is the ratio of adrenic acid to AA (metabolites). It should be noted that no dihomo-15d-⌬ 12,14 -PGJ 2 was observed.