- Email: [email protected]
Reversed-phase high-pressure liquid chromatography of arachidonic acid metabolites formed by cyclooxygenase and lipoxygenases
ANALYTICAL
BIOCHEMISTRY
148,
59-69 (1985)
Reversed-Phase High-Pressure Liquid Chromatography of Arachidonic Acid Metabolites Formed by Cyclooxygenase and Lipoxygenases WILLIAM Endocrine
Laboratory,
Royal Department
S. POWELL
Victoria Hospital, 687 Pine Avenue West, Montreal, of Medicine, McGill University, Montreal. Quebec,
Quebec Canada
H3A
IAI,
and the
Received December 3, 1984 High-pressure liquid chromatography is required to resolve the complex mixtures of arachidonic acid metaholites synthesized by many tissues. We have investigated some of the factors which affect the retention times of these substances in reversed-phase HPLC on columns of 5-pm octadecylsilyl silica. There are considerable differences in selectivity between mobile phases based on methanol and those based on acetonitrile, the latter being much better for cyclooxygenase products. The chromatographic behavior of peptidoleukotrienes (LTC, , LTDl , and LTE.,) is quite different from that of other arachidonic acid metabolites which do not contain amino acids. Addition of phosphoric acid to the mobile phase results in very long retention times for peptidoleukotrienes. Very low concentrations of trifluoroacetic acid have effects similar to that of phosphoric acid, but as its concentration is raised, the retention times of peptidoleukotrienes decrease, whereas those of other arachidonic acid metabolites are unaffected. Changing the concentration of acetonitrile in the mobile phase also affects the retention times of peptidoleukotrienes differently from those of other metabolites. This information has been used to devise simple linear gradients which separate most of the major cyclooxygenase and lipoxygenase products of arachidonic acid metabolism. 0 1985 Academic Press. Inc. KEY WORDS: HPLC, reversed-phase; arachidonic acid metabolites; prostaglandins; leukotrienes; eicosanoids, HETE.
Arachidonic acid (20:4),’ a major component of cellular lipids, is the precursor of a large number of extremely potent biological mediators, including prostaglandins, thromboxanes, and leukotrienes. The profile of
products formed by different tissues varies, depending on the enzymes present, but nearly all cells possess the enzymes required to metabolize 20:4 by at least one of these pathways. HPLC has proven to be a very useful technique for the separation and analysis of the complex mixtures of products which are often formed from 20:4. Several different approaches can be used to separate oxygenated 20:4 metabolites. Argentation HPLC will separate mixtures of closely related products and can be used to separate isotopically labeled metabolites from their unlabeled analogs (1,2). This method is useful as a second step in the purification of 20:4 metabolites, but the results obtained with complex mixtures can be difficult to interpret, because structurally unrelated products of greatly different polarities (e.g.
’ Abbreviations used 20:4, arachidonic acid; PG, prostaglandin; LT, leukotriene; LTB4, leukotriene B4 (5S, 12R-dihydroxy-6,8,10,14-ZEEZ-eicosatetraenoic acid); isomer- 1, 5S, 12R-dihydroxy-6,8,10,14-EEEZeicosatetraenoic acid; isomer-2, SS,12Sdihydroxy6,8,10,14-EEEZ-eicosatetraenoic acid, SS,129dh-20:4, 5S, 12Sdihydroxy-6,8,10,14-EZEZ-eicosatetraenoic acid; 5,6-dh-20:4, 5,6-dihydroxy-7,9,11,14-eicosatetraenoic acid; 5, I5-dh-20:4, 5,15-dihydroxy-6,8,11,13-eicosatetraenoic acid; HHT, l2-hydroxy-5,8, IO-heptadecatrienoic acid; 1I-HETE, 1I-hydroxy-5,8,12, ICeicosatetraenoic acid, 12-HETE, l2-hydroxy-5,8,10, ICeicosatetraenoic acid, 15-HETE, l5-hydroxy-5,8,11,13-eicosatetraenoic acid, TFA, trifluoroacetic acid, tR,retention time; PMNL, polymorphonuclear leukocytes: TXBz, thromboxane Bz. 59
0003-2697/85 $3.00 Copyright 0 1985 by Academic Press, Inc. All rights of reproduction in any form resewed.
60
WILLIAM
PGF,, and 15-hydroxy-5,811, I3-eicosatetraenoic acid (15-HETE)) can have identical retention times. Normal-phase HPLC will separate complex mixtures of 20:4 metabolites (3,4), but cannot be used for peptidoleukotrienes. This method is generally better than reversed-phase HPLC for separating closely related positional and stereo isomers, but in some cases recoveries can be rather low. The injection medium can also be a problem with normal-phase HPLC. It is necessary to add small amounts of polar solvents to dissolve residues containing polyhydroxylated 20:4 metabolites such as prostaglandins. However, the presence of even 2 or 3 ~1 of isopropanol in the injection medium, for example, can markedly affect the separation of the relatively nonpolar monohydroxy metabolites of 20:4 (5). For most applications we prefer to use reversed-phase HPLC, since recoveries of 20: 4 metabolites are generally very good and the injection medium is usually not a problem. In the present paper we have investigated some of the factors affecting the chromatographic behavior of both cyclooxygenase and lipoxygenase products in order to optimize the separation of closely related groups of metabolites. From this information, mobile phases have been developed which are capable of resolving complex mixtures containing prostaglandins, leukotrienes, and monohydroxy metabolites of 20:4. MATERIALS
AND
METHODS
Preparation of standards. A mixture of cyclooxygenase products was prepared by incubating [ 1-14C]20:4 (New England Nuclear) with a bovine lung homogenate for 10 min at 37°C (4). A mixture of monohydroxy, dihydroxy, and trihydroxy products resulting from the action of 5-lipoxygenase was prepared by incubating [1-*4C]20:4 with purified human polymorphonuclear leukocytes (PMNL) for 5 min at 37°C (6). 12-Hydroxy5,8,10,14-eicosatetraenoic acid ( 12-HETE) and 15-HETE were prepared using human
S. POWELL
platelets (7) and soybean lipoxygenase (Sigma Chemical Co.) (8) respectively. 1 1-Hydroxy5,8,12,14eicosatetraenoic acid ( 11-HETE) was prepared by chemical oxidation of 20:4 (9). All products were extracted using cartridges containing ODS silica (Cis Sep-Paks, Waters Associates) as previously described ( 10). Leukotrienes C4, D4, and E4 were kindly provided by Dr. J. Rokach of Merck Frosst, Pointe-Claire, Quebec. They were kept in water at -40°C in plastic or siliconized glass tubes. We experienced considerable losses in LTD4 and LTE4 with time, possibly due to adsorption to the tubes, and this appeared to be related to the number of times the samples were thawed. This was not nearly as serious a problem with LTC4. HPLC equipment. HPLC was carried out on a column (250 X 4.6 mm) of Ultrasphere ODS silica (5 pm particle size, Beckman Instruments). Two Model M45 pumps coupled to a Model 680 gradient controller (Waters Associates) were used to deliver the mobile phases. Products were detected using two Waters Model 48 1 variable-wavelength uv detectors and a Berthold radioactivity detector equipped with a cell containing a solid scintillator for heterogeneous counting. Water and aqueous stock solutions of 0.25 M trifluoroacetic acid (TFA) and 0.25 M phosphoric acid were purified by passing them through C,s Sep-Paks prior to use. When solvents were prepared, appropriate amounts of the stock solutions were added to give the desired concentrations of TFA or phosphoric acid. RESULTS
AND
DISCUSSION
The most commonly used mobile phases for the separation of 20:4 metabolites by reversed-phase HPLC have been water:methanol:acetic acid ( 11) and water:acetonitrile containing either acetic acid (12- 14) or phosphoric acid (15). The methanol:water system has been used mainly to separate mixtures of lipoxygenase products, whereas the system containing acetonitrile has been used for
LIQUID
CHROMATOGRAPHY
OF PROSTAGLANDINS
AND LEUKOTRIENES
61
cyclooxygenase products. The peptidoleukotrienes (LTC4, LTD4, and LTE4) behave differently from other 20:4 metabolites, since their retention times are strongly dependent on the pH of the mobile phase. Thus at a pH of about 6, these products have relatively short retention times (e.g., LTC4 < LTB& whereas at pH 4 (0.02% acetic acid), their retention times are considerably longer (e.g., LTC* > LTB4) (16). If the pH is further reduced by the addition of phosphoric acid (0.02%, pH 3.0) the retention times of peptidoleukotrienes are much longer than those of most other eicosanoids (17). This makes it possible to elute dihydroxy and monohydroxy 20:4 metabolites using a gradient containing phosphoric acid at low pH, and then to elute the peptidoleukotrienes with a third solvent at pH 5.5 (17). Selectivity D$erences between Mobile Phases Containing Methanol and Acetonitrile Mixtures of cyclooxygenase and lipoxygenase products were obtained by incubating [ 1-‘4C]20:4 with a bovine lung homogenate and a suspension of human PMNL, respectively. The products were extracted, and aliquots of the two crude extracts were combined with unlabeled 15-HETE, 11-HETE, and 12-[ 1-14C]HETE and chromatographed on a 5-pm Ultrasphere ODS column using a gradient made up of different concentrations of water:acetic acid (100:0.05) and methanol: acetic acid (100:0.05) (Fig. I). This mobile phase separates LTB4 from its two 6-trans isomers, but does not separate LTB4 from the double lipoxygenase product 5S, 124dihydroxyeicosatetraenoic acid (5s. 124dh-20: 4). Nor are the second 6-trans isomer of LTB4 and the other double lipoxygenase product, 5,15-dihydroxy-6,8,11,13-eicosatetraenoic acid (5,15-dh-20:4), separated under these conditions (Fig. 1). The retention times of the two 5,6-dh-20:4 (5,6-dh) isomers are between those of HHT and the HETEs. The methanol:water system separates 15-HETE, 11-HETE, 12-HETE, and 5-HETE quite well
40 60 TIME (min)
80
FIG. I. Reversed-phase HPLC of a mixture containing crude extracts of metabolites formed from [ I-“‘C]20:4 by suspensions of human PMNL and homogenates of bovine lung to which unlabeled 1 I-HETE and IS-HETE, along with 12-[ I-?Z]HETE, were added. The mobile phase consisted of a gradient prepared from water:acetic acid (100~0.05) and methanohacetic acid (100~0.05) (solvent G) as follows: 0 min, 56% G; 40 min, 62% G, 45 mitt, 70% G; 75 min, 70% G; 80 min, 75% G, 110 min, 82% G. The flow rate was 1.5 mi/min. 20h-B.,, 20hydroxyLTB,; 60-F,,, 6-oxoPGF,,; L F”.X; D2r ED,; F20, PGF,,; iso-l, first 6-truns isomer of LTB4; iso-2, second truns isomer of LTB4; 5,15-dh, 5,15-dh-20:4; 5,6dh, isomers of 5,6-dh-20:4; 15h, I5-HETE; 1 Ih, I IHETE; 12h, I2-HETE; 5h, 5-HETE.
(the separation of these products can be improved by using a shallower gradient), but gives poor results with prostaglandins and TXB2. It should be noted that PGE2, although less polar than PGFz,, has a shorter retention time than the latter product in the presence of methanol. Similarly, 20hydroxyLTB, has a shorter retention time than 6-oxoPGF,, despite the fact that the latter product has an additional 0x0 group. Substitution of acetonitrile for methanol
62
WILLIAM
results in considerable changes in selectivity as illustrated by the chromatogram shown in Fig. 2. This system gives very good separation of cyclooxygenase products derived from bovine lung, including 6-oxoPGF1,, TXB2, PGE2, and PGD2. Unlike water: PGba, methanol:acetic acid, water:acetonitrile:acetic acid separates LTB4 from 5S, 124dh-20:4, and 5,15-dh-20:4 from the second 6-tram isomer of LTB4 (the latter two products can be completely separated using a mobile phase consisting of 40% acetonitrile and 0.05% acetic acid (6)). The acetonitrile system does not separate LTB4 from its two 6-tram iso-
L -2
I
20h-B4
E2 ,
lb0
0 TIME
1
(mid
FIG. 2. Reversed-phase HPLC of a mixture of 20~4 metabolites similar to that described in the legend to Fig. 1. The mobile phase consisted of a gradient prepared from water:acetic acid (100~0.05) and acetonitrile:acetic acid (100:0.05) (solvent H) as follows: 0 min, 24% H; 40 min, 33% H; 50 min, 44% H; 76 min, 44% H; 92 min, 50% H; 96 min, 54% H. The flow rate was 1.5 ml/min.
S. POWELL
mers, however. As with methanol, monohydroxy metabolites of 20:4 are Nell separated. The order of elution of 20:4 metabolites with acetonitrile appears to depend more on the polarity of the solute than with methanol, possibly because the latter solvent can form hydrogen bonds with the solute. For example, as shown in Fig. 2, PGF2, has a shorter retention time than PGE2 with acetonitrile, and 6-oxoPGF1, has a shorter retention time than the less polar 20-hydroxyLTB4. Reversed-Phase HPLC of Peptidoleukotrienes Efects of trijluoroacetic acid on retention times. Trifluoroacetic acid (TFA) has been used extensively for the separation of various peptides by reversed-phase HPLC (18). We have examined the effects of different concentrations of TFA on the retention times of leukotrienes Cd, Dq, and Ed. For comparison, we have also included the non-amino acidcontaining lipoxygenase products LTB4, 5HETE, and 12-HETE. Figure 3A shows the separation of these products using 70% acetonitrile containing 2.5 mM TFA (0.02%; apparent pH, 2.5). With these conditions, LTD4 and LTE4 are separated from one another, and have retention times identical to those of 12-HETE and 5-HETE, respectively. LTC, and LTB4 have shorter retention times. If the concentration of TFA is reduced to 0.5 mM, the retention times of the peptidoleukotrienes are considerably longer, whereas those of LTB4, 12-HETE, and 5HETE are unaffected (Fig. 3B). With this concentration of TFA, all three peptidoleukotrienes have retention times longer than that of 5-HETE. As can be seen from Fig. 3B, the low concentration of TFA does not adversely affect the shapes of the peaks. The effects of altering the concentration of TFA between 0.1 and 2.5 mM (0.0008 and 0.02%) on retention times (plotted on a logarithmic scale) are illustrated in Fig. 4. Even at the lowest concentration of TFA there is no appreciable deterioration in peak shapes, either for peptidoleukotrienes or non-amino
LIQUID
CHROMATOGRAPHY
0
OF PROSTAGLANDINS
63
AND LEUKOTRIENES
5 TIME
(min)
TIME
(mid
5h
5
15
10 TIME
4
20
hid
5
lb
1'5
20
TIME (mm)
FIG. 3. Reversed-phase HPLC of leukotrienes B4, Cq, D4, and E1, and 5-HETE and I2-HETE using a mobile phase consisting of (A) 2.5 mM (0.02%) TFA in 70% acetonitrile in water, (B) 0.5 mM TFA in 70% acetonitrile, and (C) 0.5 mM TFA in 55% acetonitrile. A gradient between 0.0008 and 0.02% TFA over 20 min (70% acetonitrile) was used to separate leukotrienes Bq, C,, D,,, and Ed, 5-HETE and [l“‘C 120:4 (D) . The flow rate was 2 ml/min in all cases. Solid lines, absorbance at 280 nm; dotted lines, absorbance at 235 nm; dashed line, radioactivity.
acid-containing products. The retention times of the peptidoleukotrienes increase very markedly as the concentration of TFA is lowered. LTD4 is more affected than LTE4, since at low concentrations of TFA its retention time is considerably longer than that of LTE4, whereas at higher concentrations it has a shorter retention time.
These results suggest that the pH of the mobile phase is not the only factor in determining the retention times of peptidoleukotrienes, since their retention times become shorter as the concentration of TFA is increased. The retention times are therefore reduced, rather than increased, as the pH is lowered. In addition to its effects on pH,
64
WILLIAM
0
1.0 [TFA] (mM)
2.0
FIG. 4. Effbcts of the concentration of TFA in 70% acetonitrile on the retention times of leukotrienes B4 (m), Cd (O), D4 (A), and E4 (0), and 5-HETE (A) and 12HETE (0).
TFA could affect the retention times of peptidoleukotrienes by its ion-pairing properties or by interacting with polar groups on the stationary phase. The ionization of TFA could be partially suppressed by high concentrations of organic solvents such as acetonitrile. Thus one might expect the concentration of acetonitrile in the mobile phase to affect the retention times of peptidoleukotrienes differently from those of 20:4 metabolites not containing amino acids. This is illustrated in Fig. 3C, which shows the separation of LTB4, LTC4, 5HETE, and 1ZHETE with 55% acetonitrile containing 0.5 mM TFA. With these conditions, the retention time of LTC4 is consid-
S. POWELL
erably shorter than that of 5-HETE. This is in contrast to the chromatogram shown in Fig. 3B (70% acetonitrile; 0.5 mM TFA), in which LTC4 has a longer retention time than does 5-HETE. The retention times of LTE4 (24.4 min) and LTD4 (41.7 min) are still longer than that of 5-HETE at an acetonitrile concentration of 55%, however. The effects of altering the acetonitrile concentration in mobile phases containing 2.5 mM TFA are shown in Fig. 5A. At a concentration of 70% acetonitrile, the retention times of LTD4 and LTE4 are identical to those of 12-HETE and 5-HETE, respectively. The retention times of all products increase with decreasing concentrations of acetonitrile, but those of the peptidoleukotrienes increase more slowly. Thus, at low concentrations of acetonitrile, the retention times of leukotrienes D4 and E4 are much shorter than those of monohydroxy metabolites of 20:4. At higher concentrations of acetonitrile, the retention time of LTD4 is shorter than that of LTE4, whereas at lower concentrations, the reverse is true. If one is interested primarily in the analysis of peptidoleukotrienes, a gradient between 0.0008% (0.1 mM) TFA and 0.02% (2.5 mM) TFA in 70% acetonitrile over 20 min can be used (Fig. 3D). With this system, the three peptidoleukotrienes have retention times in100
]I3 Y..
30 -
P
&-=
103-
-.....,
. . . . “-h. .-a..
. ...b ._...,,_,,_,,_ * .............Q D4 ~... a.-....on... x.,, -----O % ‘b... ““. ....__~,,, <:
\
I’
,
,
,
50
60 MeCN
(%I
r
aJ
70
‘5’0 MeCN
(%)
5. Effects of the concentration of acetonitrile on the retention times of leukotrienes B4 (m), C4 (O), D, (A), and E4 (0), and S-HETE (A) and 12-HETE (0) in the presence of (A) 2.5 mM TFA and (B) 2.5 mM phosphoric acid. FIG.
LIQUID
CHROMATOGRAPHY
OF PROSTAGLANDINS
termediate between those of 5-HETE and 20: 4. The calcium ionophore A23 187 does not interfere with the chromatography. The peaks for the peptidoleukotrienes are made sharper by the TFA gradient, especially that of LTD4, whereas the peak shapes and retention times of the other products, such as 20:4, are relatively unaffected. This has the advantage of increasing the sensitivity for peptidoleukotrienes over that obtained with an isocratic system (cf. Fig. 3B). Eflects of phosphoric acid on retention times. In general, the retention times of peptidoleukotrienes in the presence of phosphoric acid were found to be considerably longer than in the presence of a comparable concentration of TFA. With a mobile phase consisting of 2.5 mM phosphoric acid in 70% acetonitrile, all three peptidoleukotrienes have retention times longer than that of 5-HETE (Fig. 6A). Unlike the situation with TFA, the retention times of the peptidoleukotrienes are not appreciably affected by lowering the concentration of phosphoric acid to 0.1 mM (data not shown). The effects of phosphoric acid are strongly influenced by the concentration of acetonitrile in the mobile phase. As the concentration of acetonitrile is lowered from 70 to 60%, there are relatively small increases in the retention times of leukotrienes Cq, Dq, and Ed, com-
’
65
AND LEUKOTRIENES
pared to the much larger increases in the retention times of the other products (Fig. 5B). Thus, at a concentration of 70% acetonitrile (Fig. 6A), LTC, has a retention time longer than that of 5-HETE, whereas at 60% acetonitrile (Fig. 6B), its retention time is shorter than that of both 12-HETE and 5HETE. At 70% acetonitrile, the retention time of LTE4 is about 2.5 times that of 5HETE, whereas at 50% acetonitrile it is about the same (Fig. 5B). Peptidoleukotrienes give sharp peaks, comparable to those of non-amino acid-containing 20:4 metabolites, when either TFA or phosphoric acid is added to the mobile phase (Figs. 3 and 6). With mobile phases containing either of these acids, it is not necessary to wash the column with EDTA for HPLC of peptidoleukotrienes, as is the case when mobile phases containing acetic acid are used ( 19). The low concentrations of TFA used in this study do not appear to have a deleterious effect on the performance of the column. Considerably higher concentrations of TFA (0.1%) are routinely used in the reversedphase HPLC of peptides (18). For analytical applications, both TFA and phosphoric acid thus have advantages over acetic acid. This is not so clear for preparative applications, however. Phosphoric acid is not volatile, whereas, although it is volatile, TFA is quite
A
1 0 z
B 5h
1.
5
10
20
15 TIME
25
(mid
FIG. 6. Reversed-phase HPLC of leukotrienes Bq, C4, D4, and E,, and S-HETE and 12-HETE with mobile phases consisting of (A) 2.5 mM phosphoric acid in 70% aqueous acetonitrile and (B) 2.5 mM phosphoric acid in 60% aqueous acetonitrile. The flow rate was 2 ml/min.
3
66
WILLIAM
S. POWELL
a strong acid. Thus when the solvent is evaporated, the concentration of TFA, even though it is initially very low, increases and the pH of the solution just prior to complete evaporation can be quite low. This problem can be alleviated by the addition of a small amount of ammonium hydroxide or trimethylamine prior to removal of the solvent.
eluted as a group after the monohydroxy products. This should be possible with mobile phases containing either phosphoric acid or very low concentrations of TFA. Finally, since we wanted as simple a system as possible, we used linear gradients between two solvent mixtures. The solvents we initially selected were water:acetonitrile:TFA (75.04:25:0.0008, solvent A) and methanol:acetonitrile:water:TFA (60:40:0.1:0.002, solvent B). These were prepared by adding the appropriate amounts of the stock solution of 2% TFA to water: acetonitrile (75:25) and methanol:acetonitrile (60:40). The apparent pH of solvent A was 3.6, whereas the TFA concentrations in solvents A and B corresponded to 0.1 and 0.25 mM, respectively. Chromatography of a mixture of cyclooxygenase and lipoxygenase products using a gradient between 5% solvent B in solvent A and 100% solvent B over 80 min gave good separation of nearly all the major products, except for LTC4 and 12HETE, which had identical retention times (Table 1). With these conditions, the divalent cation ionophore A23187, which is often used to stimulate the release of lipoxygenase products, appeared as a broad peak between LTE4 and LTD4. Arachidonic acid was observed as a sharp peak with about the same retention time as A23187 (Table 1).
Separation of Cyclooxygenase and Lipoxygenase Products Using Solvent Gradients We attempted to take advantage of the selectivity differences between methanol and acetonitrile to develop a mobile phase that was capable of separating all the major cyclooxygenase and lipoxygenase products of 20:4 metabolism. In order to separate cyclooxygenase products, the first part of the gradient should contain a high ratio of acetonitrile to methanol. As illustrated in Figs. 1 and 2, neither methanol nor acetonitrile can separate LTB4 and its three major 5,12dihydroxy isomers (isomers 1 and 2, and 5S, 124dh-20:4). However, it is possible that this could be accomplished by using a mixture of these two solvents. From Figs. 3 and 6 it can be seen that peptidoleukotrienes can have retention times either longer or shorter than those of HETEs. To avoid confusion, it would be preferable if these three products TABLE RETENTION
1
TIMES OF 20~4 METABOLITES
USING
SOLVENT
Gradient
LTB,
I2-HETE
5-HETE
LTC,
LTE,
5-100% B in A/80 min* 5-10096 D in C/80 min’ 5-80% D in C/63 min’
43.2 42. I 42.7
58.2 57.7 57.8
59.7 59.2 59.4
58.1 65.7 66.0
65.0 74.7 74.6
o The retention times for 6-oxoF’GF,. (10 min), 20-hydroxyLTB, with all three systems. The retention time for FGBl (34 min) was b Solvent A was water:acetonitrile:TFA (75.04:25:0.0008) and The flow rate was 2 ml/min. ’ Solvent C was water:acetonitrile:phosphoric acid (75.4:25:0.01) (60:40:0.4:0.01). The flow rate was 2 ml/min. d LTD, was not elated with 100% solvent D (Ia > 105 min). solvent D to 80%.
GRADIENTS
LTD, 75. I d
93.3
A23187
20~4
69.6 69.8 70.1
69.8 71.0 71.2
(12.5 min), TXBl (19 min), and PGE, (22.5 min) were the same determined only for the mobile phase containing TFA. solvent B was methanol:acetonitrile:water:TFA (60:40:0. I :0.002). and solvent D was methanol:acetonitrile:water:phosphoric It could subsequently
be eluted by lowering the concentration
acid of
LIQUID
CHROMATOGRAPHY
OF PROSTAGLANDINS
The data in Fig. 4 would suggest that the low concentrations of TFA employed for the 80-min gradient discussed above (Table 1) would result in LTC4 having a longer retention time than the monohydroxy metabolites of 20:4. This is not the case, presumably due to the fact that the low concentration of acetonitrile present during the early part of the gradient tends to shorten the retention time of LTC, relative to the monohydroxy products (see Fig. 5A and Figs. 3B and C). The effects of low acetonitrile concentrations on the retention times of peptidoleukotrienes could be minimized by using a steeper gradient. We therefore repeated the chromatography described above, using identical conditions except that the gradient was over 40 min instead of 80 min. Under these conditions LTC4 has a retention time considerably longer than that of 5-HETE (Fig. 7). The cyclooxygenase products, 6-OXOPGF,,, TXB2, PGF*, , PGE2, and PGD2, are well resolved in the early part of the gradient. LTB4 and its three major 5,12-dihydroxy isomers are all separated from one another in the middle section of the gradient. The double lipoxygenase product, 5,15-dh-20:4, which absorbs at 235 nm but not at 280 nm, has a retention time intermediate between the two tram isomers of LTB4, but is not completely separated from either of them. The elution positions of PGB2 (23 min) and [1-‘4C]20:4 (42 min) are indicated in the chromatogram shown in Fig. 7, although they were run separately. The steeper gradient thus results in a considerably shorter analysis time with improved resolution of peptidoleukotrienes from monohydroxyeicosatetraenoic acids. The ionophore A23 187 appears as a broad peak between LTC4 and LTE4, however, and, if present in large amounts, could interfere with the detection of these substances. Results similar to those described above were obtained for 20:4 metabolites lacking amino acids with mobile phases containing acetic acid (data not shown). Acetic acid did not give good results with peptidoleuko-
67
AND LEUKOTRIENES
lb
2’0 TIME hid
3’0
40
FIG. 7. HPLC of a mixture of cyclooxygenase and Iipoxygenase products similar to that described in the legend to Fig. 1, except that leukotrienes C., (100 ng), D4, and E, were added. The theoretical amounts of LTD., and LTE., injected were 200 ng, but the actual amounts were probably much less due to losses incurred during storage as discussed above. A gradient between (i) 95% solvent A and 5% solvent B and (ii) 100% solvent B over 40 min was used. The flow rate was 2 ml/min. Although the sample did not contain PGB2 and [I“‘C]20:4, their elution positions with identical conditions are indicated by arrows.
trienes, however, since the retention times were not as reproducible and the peaks were broader. This could probably have been im-
68
WILLIAM
proved if the column had been washed with EDTA, however (cf. (19)). Gradients similar to those described above in which phosphoric acid (1 mM, 0.0 1%) was substituted for TFA were also investigated (Table 1). If a gradient of 5% solvent C (water:acetonitrile:phosphoric acid (75.4:25: 0.0 1)) in solvent D (methanol:acetonitrile: water:phosphoric acid (60:40:0.4:0.01)) to 100% solvent D over a period of 80 min was used, LTC, and LTE4 had retention times longer than that of 5-HETE, but LTD4 was not eluted from the column (ta > 105 min). LTD4 could subsequently be eluted by lowering the concentration of solvent D to 80%, however. This would appear to be another difference between phosphoric acid and TFA. In the presence of TFA, LTD4 can be eluted after the concentration of the organic solvent reaches 100% (Fig. 7), whereas with phosphoric acid, a high concentration of organic solvent prevents elution of LTD4. This is even more apparent if a shorter time for the gradient (5 to 100% D over 40 min) is used (data not shown). Under these conditions, peptidoleukotrienes are not eluted from the column. They can subsequently be eluted by lowering the concentration of solvent D to 75 or 80%. Thus with phosphoric acid, the mobile phase at the end of the gradient should include some water. If an identical rate of increase of solvent D is used, but the gradient stops at 80% D instead of 100% D, all of the peptidoleukotrienes are eluted (Table 1). The retention times of 20:4 (71.2 min) and A23 187 (70.1 min, broad peak) under these conditions were intermediate between those of LTC, and LTE4. One of the problems with the mobile phases shown in Fig. 7 and in Table 1 is that the peptidoleukotrienes as a group are not separated from 20:4 and A23 187. If A23 187 or radiolabeled 20:4 are not used this would not matter. For some applications, however, it would be preferable if the peptidoleukotrienes eluted later than A23187 and 20:4. This can be accomplished by using a gradient between 5% solvent F (methanokacetonitrile:
S. POWELL
water:TFA (60:40:0.08:0.00 16)) in solvent E (water:acetonitrile:phosphoric acid (76:25: 0.025)) and 80% solvent F over 32 min (Fig. 8). The presence of phosphoric acid in the early part of the gradient lengthens the retention times of the peptidoleukotrienes without affecting those of 20:4 and A23187. With these conditions the latter substances have retention times of about 41 min, earlier than those of LTC4 (43.6 min), LTE4 (48.8 min), and LTD4 (59.7 min). The separation of other lipoxygenase and cyclooxygenase products is similar to that obtained using a gradient between 5 and 100% solvent D in solvent C (Fig. 7).
s 4L
TtME
(mid
FIG. 8. HPLC of a mixture of to that described in the legend to between (i) 95% solvent E and 80% solvent F over 32 min was 2 ml/min. Although the sample its elution position with identical by an arrow.
20~4 metabolites similar Fig. 7. A linear gradient 5% solvent F and (ii) used. The flow rate was did not contain PGBZ, conditions is indicated
LIQUID
CHROMATOGRAPHY
OF PROSTAGLANDINS
69
ACKNOWLEDGMENTS
CONCLUSIONS
Mobile phases containing TFA or phosphoric acid offer a number of advantages over those containing acetic acid for the HPLC of peptidoleukotrienes. On the other hand, the chromatographic properties of 20: 4 metabolites not containing amino acids are virtually the same with TFA, phosphoric acid, and acetic acid. Samples suspected of containing peptidoleukotrienes can be chromatographed using different concentrations of TFA and the same concentration of acetonitrile. Peaks due to peptidoleukotrienes should be shifted due to the change in the concentration of TFA, whereas the retention times of peaks due to products not containing amino acids should be unaffected (Fig. 3), except at very low concentrations (CO.00 1%) of TFA, in which case they may be slightly longer. For analysis of peptidoleukotrienes, a TFA gradient with a constant concentration of acetonitrile in water results in elution of these products between monohydroxyeicosatetraenoic acids and 20:4 (Fig. 3D). This system has the advantage of much shorter analysis times and sharper peaks for peptidoleukotrienes, thereby increasing sensitivity. For analysis of more complex mixtures of 20:4 metabolites, simple linear gradient systems employing mobile phases composed of water:acetonitrile:methanol and either TFA or phosphoric acid can be used. The analysis time is quite rapid, and nearly all of the major products are well separated from one another. If it is desired to achieve greater separation of the products in any area of the chromatogram, a more complex gradient, composed of a combination of shallow and steep segments (cf. Figs. 1 and 2) could be used, but this might result in overlap between LTC4 and monohydroxy metabolites of 20:4.
AND LEUKOTRIENES
This work was supported by grants from the Medical Research Council of Canada and the Quebec Heart Foundation. The author is the holder of a Scientist award from the MRC. The author is grateful to Fran&e Gravelle for expert technical assistance. REFERENCES 1. Powell, W. S. (1981) Anal. Biochem. 115, 267-277. 2. Powell, W. S. (1983) Anal. B&hem. 128, 93-103. 3. Borgeat, P., Picard S., Vallerand, P., and Sirois, P. (1981) Prosiaglandins Med. 6, 557-570. 4. Powell, W. S. (1980) Prostaglandins 20, 947-957. 5. Powell, W. S. (1985) in Biochemistry of Arachidonic Acid Metabolism (Lands, W. E. M., ed.), KluwerNijhoff Publishing, Boston, in press. Powell, W. S. (1984) J. Biol. Chem. 259, 30823089. Hamberg, M., and Samuelsson, B. (1974) Proc. Natl. Acad.
Sci. USA 71, 3400-3404.
Hamberg, M., and Samuelsson, B. (1967) J. Biol. Chem.
242, 5329-5335.
Boeynaems, J. M., Brash, A. R., Oates, J. A., and Hubbard, W. C. (1980) Anal. Biochem. 104,259267. 10. Powell, W. S. (1982) in Methods in Enzymology (Lands, W. E., and Smith, W. L., eds.), Vol. 86, pp. 467-477, Academic Press, New York. 11. Borgeat, P., and Samuelsson, B. (1979) Proc. Natl. Acad. Sci. USA 76, 2148-2152.
12. Alam, I., Ohuchi, I., and Levine, L. (1979) Anal. Biochem.
93, 339-345.
13. Henke, D. C., Kouzan, S., and Eling, T. E. (1984) Anal.
Biochem.
140, 87-94.
14. Peters, S. P., Schulman, E. S., Liu, M. C., Hayes, E. C., and Lichtenstein, L. M. (1983) 1. Zmmunol. Methods
64, 335-343.
15. VanRollins, M., Aveldano, M. I., Sprecher, H. W., and Horrocks, L. A. (1982) in Methods in Enzymology (Lands, W. E., and Smith, W. L., eds.), Vol. 86, pp. 5 18-530, Academic Press, New York. 16. Mathews, W. R., Rokach, J., and Murphy, R. C. (1981) Ana/. Biochem. 118, 96-101. 17. Borgeat, P. (1984) Prostaglandins 27, 356-358. 18. Bennett, H. P. J., Browne, C. A. B., and Solomon, S. (198 1) Biochemistry 20, 4530-4538. 19. Metz, S. A., Hall, M. E., Harper, T. W., and Murphy, R. C. (1982) J. Chromatogr. 233, 193201.
BIOCHEMISTRY
148,
59-69 (1985)
Reversed-Phase High-Pressure Liquid Chromatography of Arachidonic Acid Metabolites Formed by Cyclooxygenase and Lipoxygenases WILLIAM Endocrine
Laboratory,
Royal Department
S. POWELL
Victoria Hospital, 687 Pine Avenue West, Montreal, of Medicine, McGill University, Montreal. Quebec,
Quebec Canada
H3A
IAI,
and the
Received December 3, 1984 High-pressure liquid chromatography is required to resolve the complex mixtures of arachidonic acid metaholites synthesized by many tissues. We have investigated some of the factors which affect the retention times of these substances in reversed-phase HPLC on columns of 5-pm octadecylsilyl silica. There are considerable differences in selectivity between mobile phases based on methanol and those based on acetonitrile, the latter being much better for cyclooxygenase products. The chromatographic behavior of peptidoleukotrienes (LTC, , LTDl , and LTE.,) is quite different from that of other arachidonic acid metabolites which do not contain amino acids. Addition of phosphoric acid to the mobile phase results in very long retention times for peptidoleukotrienes. Very low concentrations of trifluoroacetic acid have effects similar to that of phosphoric acid, but as its concentration is raised, the retention times of peptidoleukotrienes decrease, whereas those of other arachidonic acid metabolites are unaffected. Changing the concentration of acetonitrile in the mobile phase also affects the retention times of peptidoleukotrienes differently from those of other metabolites. This information has been used to devise simple linear gradients which separate most of the major cyclooxygenase and lipoxygenase products of arachidonic acid metabolism. 0 1985 Academic Press. Inc. KEY WORDS: HPLC, reversed-phase; arachidonic acid metabolites; prostaglandins; leukotrienes; eicosanoids, HETE.
Arachidonic acid (20:4),’ a major component of cellular lipids, is the precursor of a large number of extremely potent biological mediators, including prostaglandins, thromboxanes, and leukotrienes. The profile of
products formed by different tissues varies, depending on the enzymes present, but nearly all cells possess the enzymes required to metabolize 20:4 by at least one of these pathways. HPLC has proven to be a very useful technique for the separation and analysis of the complex mixtures of products which are often formed from 20:4. Several different approaches can be used to separate oxygenated 20:4 metabolites. Argentation HPLC will separate mixtures of closely related products and can be used to separate isotopically labeled metabolites from their unlabeled analogs (1,2). This method is useful as a second step in the purification of 20:4 metabolites, but the results obtained with complex mixtures can be difficult to interpret, because structurally unrelated products of greatly different polarities (e.g.
’ Abbreviations used 20:4, arachidonic acid; PG, prostaglandin; LT, leukotriene; LTB4, leukotriene B4 (5S, 12R-dihydroxy-6,8,10,14-ZEEZ-eicosatetraenoic acid); isomer- 1, 5S, 12R-dihydroxy-6,8,10,14-EEEZeicosatetraenoic acid; isomer-2, SS,12Sdihydroxy6,8,10,14-EEEZ-eicosatetraenoic acid, SS,129dh-20:4, 5S, 12Sdihydroxy-6,8,10,14-EZEZ-eicosatetraenoic acid; 5,6-dh-20:4, 5,6-dihydroxy-7,9,11,14-eicosatetraenoic acid; 5, I5-dh-20:4, 5,15-dihydroxy-6,8,11,13-eicosatetraenoic acid; HHT, l2-hydroxy-5,8, IO-heptadecatrienoic acid; 1I-HETE, 1I-hydroxy-5,8,12, ICeicosatetraenoic acid, 12-HETE, l2-hydroxy-5,8,10, ICeicosatetraenoic acid, 15-HETE, l5-hydroxy-5,8,11,13-eicosatetraenoic acid, TFA, trifluoroacetic acid, tR,retention time; PMNL, polymorphonuclear leukocytes: TXBz, thromboxane Bz. 59
0003-2697/85 $3.00 Copyright 0 1985 by Academic Press, Inc. All rights of reproduction in any form resewed.
60
WILLIAM
PGF,, and 15-hydroxy-5,811, I3-eicosatetraenoic acid (15-HETE)) can have identical retention times. Normal-phase HPLC will separate complex mixtures of 20:4 metabolites (3,4), but cannot be used for peptidoleukotrienes. This method is generally better than reversed-phase HPLC for separating closely related positional and stereo isomers, but in some cases recoveries can be rather low. The injection medium can also be a problem with normal-phase HPLC. It is necessary to add small amounts of polar solvents to dissolve residues containing polyhydroxylated 20:4 metabolites such as prostaglandins. However, the presence of even 2 or 3 ~1 of isopropanol in the injection medium, for example, can markedly affect the separation of the relatively nonpolar monohydroxy metabolites of 20:4 (5). For most applications we prefer to use reversed-phase HPLC, since recoveries of 20: 4 metabolites are generally very good and the injection medium is usually not a problem. In the present paper we have investigated some of the factors affecting the chromatographic behavior of both cyclooxygenase and lipoxygenase products in order to optimize the separation of closely related groups of metabolites. From this information, mobile phases have been developed which are capable of resolving complex mixtures containing prostaglandins, leukotrienes, and monohydroxy metabolites of 20:4. MATERIALS
AND
METHODS
Preparation of standards. A mixture of cyclooxygenase products was prepared by incubating [ 1-14C]20:4 (New England Nuclear) with a bovine lung homogenate for 10 min at 37°C (4). A mixture of monohydroxy, dihydroxy, and trihydroxy products resulting from the action of 5-lipoxygenase was prepared by incubating [1-*4C]20:4 with purified human polymorphonuclear leukocytes (PMNL) for 5 min at 37°C (6). 12-Hydroxy5,8,10,14-eicosatetraenoic acid ( 12-HETE) and 15-HETE were prepared using human
S. POWELL
platelets (7) and soybean lipoxygenase (Sigma Chemical Co.) (8) respectively. 1 1-Hydroxy5,8,12,14eicosatetraenoic acid ( 11-HETE) was prepared by chemical oxidation of 20:4 (9). All products were extracted using cartridges containing ODS silica (Cis Sep-Paks, Waters Associates) as previously described ( 10). Leukotrienes C4, D4, and E4 were kindly provided by Dr. J. Rokach of Merck Frosst, Pointe-Claire, Quebec. They were kept in water at -40°C in plastic or siliconized glass tubes. We experienced considerable losses in LTD4 and LTE4 with time, possibly due to adsorption to the tubes, and this appeared to be related to the number of times the samples were thawed. This was not nearly as serious a problem with LTC4. HPLC equipment. HPLC was carried out on a column (250 X 4.6 mm) of Ultrasphere ODS silica (5 pm particle size, Beckman Instruments). Two Model M45 pumps coupled to a Model 680 gradient controller (Waters Associates) were used to deliver the mobile phases. Products were detected using two Waters Model 48 1 variable-wavelength uv detectors and a Berthold radioactivity detector equipped with a cell containing a solid scintillator for heterogeneous counting. Water and aqueous stock solutions of 0.25 M trifluoroacetic acid (TFA) and 0.25 M phosphoric acid were purified by passing them through C,s Sep-Paks prior to use. When solvents were prepared, appropriate amounts of the stock solutions were added to give the desired concentrations of TFA or phosphoric acid. RESULTS
AND
DISCUSSION
The most commonly used mobile phases for the separation of 20:4 metabolites by reversed-phase HPLC have been water:methanol:acetic acid ( 11) and water:acetonitrile containing either acetic acid (12- 14) or phosphoric acid (15). The methanol:water system has been used mainly to separate mixtures of lipoxygenase products, whereas the system containing acetonitrile has been used for
LIQUID
CHROMATOGRAPHY
OF PROSTAGLANDINS
AND LEUKOTRIENES
61
cyclooxygenase products. The peptidoleukotrienes (LTC4, LTD4, and LTE4) behave differently from other 20:4 metabolites, since their retention times are strongly dependent on the pH of the mobile phase. Thus at a pH of about 6, these products have relatively short retention times (e.g., LTC4 < LTB& whereas at pH 4 (0.02% acetic acid), their retention times are considerably longer (e.g., LTC* > LTB4) (16). If the pH is further reduced by the addition of phosphoric acid (0.02%, pH 3.0) the retention times of peptidoleukotrienes are much longer than those of most other eicosanoids (17). This makes it possible to elute dihydroxy and monohydroxy 20:4 metabolites using a gradient containing phosphoric acid at low pH, and then to elute the peptidoleukotrienes with a third solvent at pH 5.5 (17). Selectivity D$erences between Mobile Phases Containing Methanol and Acetonitrile Mixtures of cyclooxygenase and lipoxygenase products were obtained by incubating [ 1-‘4C]20:4 with a bovine lung homogenate and a suspension of human PMNL, respectively. The products were extracted, and aliquots of the two crude extracts were combined with unlabeled 15-HETE, 11-HETE, and 12-[ 1-14C]HETE and chromatographed on a 5-pm Ultrasphere ODS column using a gradient made up of different concentrations of water:acetic acid (100:0.05) and methanol: acetic acid (100:0.05) (Fig. I). This mobile phase separates LTB4 from its two 6-trans isomers, but does not separate LTB4 from the double lipoxygenase product 5S, 124dihydroxyeicosatetraenoic acid (5s. 124dh-20: 4). Nor are the second 6-trans isomer of LTB4 and the other double lipoxygenase product, 5,15-dihydroxy-6,8,11,13-eicosatetraenoic acid (5,15-dh-20:4), separated under these conditions (Fig. 1). The retention times of the two 5,6-dh-20:4 (5,6-dh) isomers are between those of HHT and the HETEs. The methanol:water system separates 15-HETE, 11-HETE, 12-HETE, and 5-HETE quite well
40 60 TIME (min)
80
FIG. I. Reversed-phase HPLC of a mixture containing crude extracts of metabolites formed from [ I-“‘C]20:4 by suspensions of human PMNL and homogenates of bovine lung to which unlabeled 1 I-HETE and IS-HETE, along with 12-[ I-?Z]HETE, were added. The mobile phase consisted of a gradient prepared from water:acetic acid (100~0.05) and methanohacetic acid (100~0.05) (solvent G) as follows: 0 min, 56% G; 40 min, 62% G, 45 mitt, 70% G; 75 min, 70% G; 80 min, 75% G, 110 min, 82% G. The flow rate was 1.5 mi/min. 20h-B.,, 20hydroxyLTB,; 60-F,,, 6-oxoPGF,,; L F”.X; D2r ED,; F20, PGF,,; iso-l, first 6-truns isomer of LTB4; iso-2, second truns isomer of LTB4; 5,15-dh, 5,15-dh-20:4; 5,6dh, isomers of 5,6-dh-20:4; 15h, I5-HETE; 1 Ih, I IHETE; 12h, I2-HETE; 5h, 5-HETE.
(the separation of these products can be improved by using a shallower gradient), but gives poor results with prostaglandins and TXB2. It should be noted that PGE2, although less polar than PGFz,, has a shorter retention time than the latter product in the presence of methanol. Similarly, 20hydroxyLTB, has a shorter retention time than 6-oxoPGF,, despite the fact that the latter product has an additional 0x0 group. Substitution of acetonitrile for methanol
62
WILLIAM
results in considerable changes in selectivity as illustrated by the chromatogram shown in Fig. 2. This system gives very good separation of cyclooxygenase products derived from bovine lung, including 6-oxoPGF1,, TXB2, PGE2, and PGD2. Unlike water: PGba, methanol:acetic acid, water:acetonitrile:acetic acid separates LTB4 from 5S, 124dh-20:4, and 5,15-dh-20:4 from the second 6-tram isomer of LTB4 (the latter two products can be completely separated using a mobile phase consisting of 40% acetonitrile and 0.05% acetic acid (6)). The acetonitrile system does not separate LTB4 from its two 6-tram iso-
L -2
I
20h-B4
E2 ,
lb0
0 TIME
1
(mid
FIG. 2. Reversed-phase HPLC of a mixture of 20~4 metabolites similar to that described in the legend to Fig. 1. The mobile phase consisted of a gradient prepared from water:acetic acid (100~0.05) and acetonitrile:acetic acid (100:0.05) (solvent H) as follows: 0 min, 24% H; 40 min, 33% H; 50 min, 44% H; 76 min, 44% H; 92 min, 50% H; 96 min, 54% H. The flow rate was 1.5 ml/min.
S. POWELL
mers, however. As with methanol, monohydroxy metabolites of 20:4 are Nell separated. The order of elution of 20:4 metabolites with acetonitrile appears to depend more on the polarity of the solute than with methanol, possibly because the latter solvent can form hydrogen bonds with the solute. For example, as shown in Fig. 2, PGF2, has a shorter retention time than PGE2 with acetonitrile, and 6-oxoPGF1, has a shorter retention time than the less polar 20-hydroxyLTB4. Reversed-Phase HPLC of Peptidoleukotrienes Efects of trijluoroacetic acid on retention times. Trifluoroacetic acid (TFA) has been used extensively for the separation of various peptides by reversed-phase HPLC (18). We have examined the effects of different concentrations of TFA on the retention times of leukotrienes Cd, Dq, and Ed. For comparison, we have also included the non-amino acidcontaining lipoxygenase products LTB4, 5HETE, and 12-HETE. Figure 3A shows the separation of these products using 70% acetonitrile containing 2.5 mM TFA (0.02%; apparent pH, 2.5). With these conditions, LTD4 and LTE4 are separated from one another, and have retention times identical to those of 12-HETE and 5-HETE, respectively. LTC, and LTB4 have shorter retention times. If the concentration of TFA is reduced to 0.5 mM, the retention times of the peptidoleukotrienes are considerably longer, whereas those of LTB4, 12-HETE, and 5HETE are unaffected (Fig. 3B). With this concentration of TFA, all three peptidoleukotrienes have retention times longer than that of 5-HETE. As can be seen from Fig. 3B, the low concentration of TFA does not adversely affect the shapes of the peaks. The effects of altering the concentration of TFA between 0.1 and 2.5 mM (0.0008 and 0.02%) on retention times (plotted on a logarithmic scale) are illustrated in Fig. 4. Even at the lowest concentration of TFA there is no appreciable deterioration in peak shapes, either for peptidoleukotrienes or non-amino
LIQUID
CHROMATOGRAPHY
0
OF PROSTAGLANDINS
63
AND LEUKOTRIENES
5 TIME
(min)
TIME
(mid
5h
5
15
10 TIME
4
20
hid
5
lb
1'5
20
TIME (mm)
FIG. 3. Reversed-phase HPLC of leukotrienes B4, Cq, D4, and E1, and 5-HETE and I2-HETE using a mobile phase consisting of (A) 2.5 mM (0.02%) TFA in 70% acetonitrile in water, (B) 0.5 mM TFA in 70% acetonitrile, and (C) 0.5 mM TFA in 55% acetonitrile. A gradient between 0.0008 and 0.02% TFA over 20 min (70% acetonitrile) was used to separate leukotrienes Bq, C,, D,,, and Ed, 5-HETE and [l“‘C 120:4 (D) . The flow rate was 2 ml/min in all cases. Solid lines, absorbance at 280 nm; dotted lines, absorbance at 235 nm; dashed line, radioactivity.
acid-containing products. The retention times of the peptidoleukotrienes increase very markedly as the concentration of TFA is lowered. LTD4 is more affected than LTE4, since at low concentrations of TFA its retention time is considerably longer than that of LTE4, whereas at higher concentrations it has a shorter retention time.
These results suggest that the pH of the mobile phase is not the only factor in determining the retention times of peptidoleukotrienes, since their retention times become shorter as the concentration of TFA is increased. The retention times are therefore reduced, rather than increased, as the pH is lowered. In addition to its effects on pH,
64
WILLIAM
0
1.0 [TFA] (mM)
2.0
FIG. 4. Effbcts of the concentration of TFA in 70% acetonitrile on the retention times of leukotrienes B4 (m), Cd (O), D4 (A), and E4 (0), and 5-HETE (A) and 12HETE (0).
TFA could affect the retention times of peptidoleukotrienes by its ion-pairing properties or by interacting with polar groups on the stationary phase. The ionization of TFA could be partially suppressed by high concentrations of organic solvents such as acetonitrile. Thus one might expect the concentration of acetonitrile in the mobile phase to affect the retention times of peptidoleukotrienes differently from those of 20:4 metabolites not containing amino acids. This is illustrated in Fig. 3C, which shows the separation of LTB4, LTC4, 5HETE, and 1ZHETE with 55% acetonitrile containing 0.5 mM TFA. With these conditions, the retention time of LTC4 is consid-
S. POWELL
erably shorter than that of 5-HETE. This is in contrast to the chromatogram shown in Fig. 3B (70% acetonitrile; 0.5 mM TFA), in which LTC4 has a longer retention time than does 5-HETE. The retention times of LTE4 (24.4 min) and LTD4 (41.7 min) are still longer than that of 5-HETE at an acetonitrile concentration of 55%, however. The effects of altering the acetonitrile concentration in mobile phases containing 2.5 mM TFA are shown in Fig. 5A. At a concentration of 70% acetonitrile, the retention times of LTD4 and LTE4 are identical to those of 12-HETE and 5-HETE, respectively. The retention times of all products increase with decreasing concentrations of acetonitrile, but those of the peptidoleukotrienes increase more slowly. Thus, at low concentrations of acetonitrile, the retention times of leukotrienes D4 and E4 are much shorter than those of monohydroxy metabolites of 20:4. At higher concentrations of acetonitrile, the retention time of LTD4 is shorter than that of LTE4, whereas at lower concentrations, the reverse is true. If one is interested primarily in the analysis of peptidoleukotrienes, a gradient between 0.0008% (0.1 mM) TFA and 0.02% (2.5 mM) TFA in 70% acetonitrile over 20 min can be used (Fig. 3D). With this system, the three peptidoleukotrienes have retention times in100
]I3 Y..
30 -
P
&-=
103-
-.....,
. . . . “-h. .-a..
. ...b ._...,,_,,_,,_ * .............Q D4 ~... a.-....on... x.,, -----O % ‘b... ““. ....__~,,, <:
\
I’
,
,
,
50
60 MeCN
(%I
r
aJ
70
‘5’0 MeCN
(%)
5. Effects of the concentration of acetonitrile on the retention times of leukotrienes B4 (m), C4 (O), D, (A), and E4 (0), and S-HETE (A) and 12-HETE (0) in the presence of (A) 2.5 mM TFA and (B) 2.5 mM phosphoric acid. FIG.
LIQUID
CHROMATOGRAPHY
OF PROSTAGLANDINS
termediate between those of 5-HETE and 20: 4. The calcium ionophore A23 187 does not interfere with the chromatography. The peaks for the peptidoleukotrienes are made sharper by the TFA gradient, especially that of LTD4, whereas the peak shapes and retention times of the other products, such as 20:4, are relatively unaffected. This has the advantage of increasing the sensitivity for peptidoleukotrienes over that obtained with an isocratic system (cf. Fig. 3B). Eflects of phosphoric acid on retention times. In general, the retention times of peptidoleukotrienes in the presence of phosphoric acid were found to be considerably longer than in the presence of a comparable concentration of TFA. With a mobile phase consisting of 2.5 mM phosphoric acid in 70% acetonitrile, all three peptidoleukotrienes have retention times longer than that of 5-HETE (Fig. 6A). Unlike the situation with TFA, the retention times of the peptidoleukotrienes are not appreciably affected by lowering the concentration of phosphoric acid to 0.1 mM (data not shown). The effects of phosphoric acid are strongly influenced by the concentration of acetonitrile in the mobile phase. As the concentration of acetonitrile is lowered from 70 to 60%, there are relatively small increases in the retention times of leukotrienes Cq, Dq, and Ed, com-
’
65
AND LEUKOTRIENES
pared to the much larger increases in the retention times of the other products (Fig. 5B). Thus, at a concentration of 70% acetonitrile (Fig. 6A), LTC, has a retention time longer than that of 5-HETE, whereas at 60% acetonitrile (Fig. 6B), its retention time is shorter than that of both 12-HETE and 5HETE. At 70% acetonitrile, the retention time of LTE4 is about 2.5 times that of 5HETE, whereas at 50% acetonitrile it is about the same (Fig. 5B). Peptidoleukotrienes give sharp peaks, comparable to those of non-amino acid-containing 20:4 metabolites, when either TFA or phosphoric acid is added to the mobile phase (Figs. 3 and 6). With mobile phases containing either of these acids, it is not necessary to wash the column with EDTA for HPLC of peptidoleukotrienes, as is the case when mobile phases containing acetic acid are used ( 19). The low concentrations of TFA used in this study do not appear to have a deleterious effect on the performance of the column. Considerably higher concentrations of TFA (0.1%) are routinely used in the reversedphase HPLC of peptides (18). For analytical applications, both TFA and phosphoric acid thus have advantages over acetic acid. This is not so clear for preparative applications, however. Phosphoric acid is not volatile, whereas, although it is volatile, TFA is quite
A
1 0 z
B 5h
1.
5
10
20
15 TIME
25
(mid
FIG. 6. Reversed-phase HPLC of leukotrienes Bq, C4, D4, and E,, and S-HETE and 12-HETE with mobile phases consisting of (A) 2.5 mM phosphoric acid in 70% aqueous acetonitrile and (B) 2.5 mM phosphoric acid in 60% aqueous acetonitrile. The flow rate was 2 ml/min.
3
66
WILLIAM
S. POWELL
a strong acid. Thus when the solvent is evaporated, the concentration of TFA, even though it is initially very low, increases and the pH of the solution just prior to complete evaporation can be quite low. This problem can be alleviated by the addition of a small amount of ammonium hydroxide or trimethylamine prior to removal of the solvent.
eluted as a group after the monohydroxy products. This should be possible with mobile phases containing either phosphoric acid or very low concentrations of TFA. Finally, since we wanted as simple a system as possible, we used linear gradients between two solvent mixtures. The solvents we initially selected were water:acetonitrile:TFA (75.04:25:0.0008, solvent A) and methanol:acetonitrile:water:TFA (60:40:0.1:0.002, solvent B). These were prepared by adding the appropriate amounts of the stock solution of 2% TFA to water: acetonitrile (75:25) and methanol:acetonitrile (60:40). The apparent pH of solvent A was 3.6, whereas the TFA concentrations in solvents A and B corresponded to 0.1 and 0.25 mM, respectively. Chromatography of a mixture of cyclooxygenase and lipoxygenase products using a gradient between 5% solvent B in solvent A and 100% solvent B over 80 min gave good separation of nearly all the major products, except for LTC4 and 12HETE, which had identical retention times (Table 1). With these conditions, the divalent cation ionophore A23187, which is often used to stimulate the release of lipoxygenase products, appeared as a broad peak between LTE4 and LTD4. Arachidonic acid was observed as a sharp peak with about the same retention time as A23187 (Table 1).
Separation of Cyclooxygenase and Lipoxygenase Products Using Solvent Gradients We attempted to take advantage of the selectivity differences between methanol and acetonitrile to develop a mobile phase that was capable of separating all the major cyclooxygenase and lipoxygenase products of 20:4 metabolism. In order to separate cyclooxygenase products, the first part of the gradient should contain a high ratio of acetonitrile to methanol. As illustrated in Figs. 1 and 2, neither methanol nor acetonitrile can separate LTB4 and its three major 5,12dihydroxy isomers (isomers 1 and 2, and 5S, 124dh-20:4). However, it is possible that this could be accomplished by using a mixture of these two solvents. From Figs. 3 and 6 it can be seen that peptidoleukotrienes can have retention times either longer or shorter than those of HETEs. To avoid confusion, it would be preferable if these three products TABLE RETENTION
1
TIMES OF 20~4 METABOLITES
USING
SOLVENT
Gradient
LTB,
I2-HETE
5-HETE
LTC,
LTE,
5-100% B in A/80 min* 5-10096 D in C/80 min’ 5-80% D in C/63 min’
43.2 42. I 42.7
58.2 57.7 57.8
59.7 59.2 59.4
58.1 65.7 66.0
65.0 74.7 74.6
o The retention times for 6-oxoF’GF,. (10 min), 20-hydroxyLTB, with all three systems. The retention time for FGBl (34 min) was b Solvent A was water:acetonitrile:TFA (75.04:25:0.0008) and The flow rate was 2 ml/min. ’ Solvent C was water:acetonitrile:phosphoric acid (75.4:25:0.01) (60:40:0.4:0.01). The flow rate was 2 ml/min. d LTD, was not elated with 100% solvent D (Ia > 105 min). solvent D to 80%.
GRADIENTS
LTD, 75. I d
93.3
A23187
20~4
69.6 69.8 70.1
69.8 71.0 71.2
(12.5 min), TXBl (19 min), and PGE, (22.5 min) were the same determined only for the mobile phase containing TFA. solvent B was methanol:acetonitrile:water:TFA (60:40:0. I :0.002). and solvent D was methanol:acetonitrile:water:phosphoric It could subsequently
be eluted by lowering the concentration
acid of
LIQUID
CHROMATOGRAPHY
OF PROSTAGLANDINS
The data in Fig. 4 would suggest that the low concentrations of TFA employed for the 80-min gradient discussed above (Table 1) would result in LTC4 having a longer retention time than the monohydroxy metabolites of 20:4. This is not the case, presumably due to the fact that the low concentration of acetonitrile present during the early part of the gradient tends to shorten the retention time of LTC, relative to the monohydroxy products (see Fig. 5A and Figs. 3B and C). The effects of low acetonitrile concentrations on the retention times of peptidoleukotrienes could be minimized by using a steeper gradient. We therefore repeated the chromatography described above, using identical conditions except that the gradient was over 40 min instead of 80 min. Under these conditions LTC4 has a retention time considerably longer than that of 5-HETE (Fig. 7). The cyclooxygenase products, 6-OXOPGF,,, TXB2, PGF*, , PGE2, and PGD2, are well resolved in the early part of the gradient. LTB4 and its three major 5,12-dihydroxy isomers are all separated from one another in the middle section of the gradient. The double lipoxygenase product, 5,15-dh-20:4, which absorbs at 235 nm but not at 280 nm, has a retention time intermediate between the two tram isomers of LTB4, but is not completely separated from either of them. The elution positions of PGB2 (23 min) and [1-‘4C]20:4 (42 min) are indicated in the chromatogram shown in Fig. 7, although they were run separately. The steeper gradient thus results in a considerably shorter analysis time with improved resolution of peptidoleukotrienes from monohydroxyeicosatetraenoic acids. The ionophore A23 187 appears as a broad peak between LTC4 and LTE4, however, and, if present in large amounts, could interfere with the detection of these substances. Results similar to those described above were obtained for 20:4 metabolites lacking amino acids with mobile phases containing acetic acid (data not shown). Acetic acid did not give good results with peptidoleuko-
67
AND LEUKOTRIENES
lb
2’0 TIME hid
3’0
40
FIG. 7. HPLC of a mixture of cyclooxygenase and Iipoxygenase products similar to that described in the legend to Fig. 1, except that leukotrienes C., (100 ng), D4, and E, were added. The theoretical amounts of LTD., and LTE., injected were 200 ng, but the actual amounts were probably much less due to losses incurred during storage as discussed above. A gradient between (i) 95% solvent A and 5% solvent B and (ii) 100% solvent B over 40 min was used. The flow rate was 2 ml/min. Although the sample did not contain PGB2 and [I“‘C]20:4, their elution positions with identical conditions are indicated by arrows.
trienes, however, since the retention times were not as reproducible and the peaks were broader. This could probably have been im-
68
WILLIAM
proved if the column had been washed with EDTA, however (cf. (19)). Gradients similar to those described above in which phosphoric acid (1 mM, 0.0 1%) was substituted for TFA were also investigated (Table 1). If a gradient of 5% solvent C (water:acetonitrile:phosphoric acid (75.4:25: 0.0 1)) in solvent D (methanol:acetonitrile: water:phosphoric acid (60:40:0.4:0.01)) to 100% solvent D over a period of 80 min was used, LTC, and LTE4 had retention times longer than that of 5-HETE, but LTD4 was not eluted from the column (ta > 105 min). LTD4 could subsequently be eluted by lowering the concentration of solvent D to 80%, however. This would appear to be another difference between phosphoric acid and TFA. In the presence of TFA, LTD4 can be eluted after the concentration of the organic solvent reaches 100% (Fig. 7), whereas with phosphoric acid, a high concentration of organic solvent prevents elution of LTD4. This is even more apparent if a shorter time for the gradient (5 to 100% D over 40 min) is used (data not shown). Under these conditions, peptidoleukotrienes are not eluted from the column. They can subsequently be eluted by lowering the concentration of solvent D to 75 or 80%. Thus with phosphoric acid, the mobile phase at the end of the gradient should include some water. If an identical rate of increase of solvent D is used, but the gradient stops at 80% D instead of 100% D, all of the peptidoleukotrienes are eluted (Table 1). The retention times of 20:4 (71.2 min) and A23 187 (70.1 min, broad peak) under these conditions were intermediate between those of LTC, and LTE4. One of the problems with the mobile phases shown in Fig. 7 and in Table 1 is that the peptidoleukotrienes as a group are not separated from 20:4 and A23 187. If A23 187 or radiolabeled 20:4 are not used this would not matter. For some applications, however, it would be preferable if the peptidoleukotrienes eluted later than A23187 and 20:4. This can be accomplished by using a gradient between 5% solvent F (methanokacetonitrile:
S. POWELL
water:TFA (60:40:0.08:0.00 16)) in solvent E (water:acetonitrile:phosphoric acid (76:25: 0.025)) and 80% solvent F over 32 min (Fig. 8). The presence of phosphoric acid in the early part of the gradient lengthens the retention times of the peptidoleukotrienes without affecting those of 20:4 and A23187. With these conditions the latter substances have retention times of about 41 min, earlier than those of LTC4 (43.6 min), LTE4 (48.8 min), and LTD4 (59.7 min). The separation of other lipoxygenase and cyclooxygenase products is similar to that obtained using a gradient between 5 and 100% solvent D in solvent C (Fig. 7).
s 4L
TtME
(mid
FIG. 8. HPLC of a mixture of to that described in the legend to between (i) 95% solvent E and 80% solvent F over 32 min was 2 ml/min. Although the sample its elution position with identical by an arrow.
20~4 metabolites similar Fig. 7. A linear gradient 5% solvent F and (ii) used. The flow rate was did not contain PGBZ, conditions is indicated
LIQUID
CHROMATOGRAPHY
OF PROSTAGLANDINS
69
ACKNOWLEDGMENTS
CONCLUSIONS
Mobile phases containing TFA or phosphoric acid offer a number of advantages over those containing acetic acid for the HPLC of peptidoleukotrienes. On the other hand, the chromatographic properties of 20: 4 metabolites not containing amino acids are virtually the same with TFA, phosphoric acid, and acetic acid. Samples suspected of containing peptidoleukotrienes can be chromatographed using different concentrations of TFA and the same concentration of acetonitrile. Peaks due to peptidoleukotrienes should be shifted due to the change in the concentration of TFA, whereas the retention times of peaks due to products not containing amino acids should be unaffected (Fig. 3), except at very low concentrations (CO.00 1%) of TFA, in which case they may be slightly longer. For analysis of peptidoleukotrienes, a TFA gradient with a constant concentration of acetonitrile in water results in elution of these products between monohydroxyeicosatetraenoic acids and 20:4 (Fig. 3D). This system has the advantage of much shorter analysis times and sharper peaks for peptidoleukotrienes, thereby increasing sensitivity. For analysis of more complex mixtures of 20:4 metabolites, simple linear gradient systems employing mobile phases composed of water:acetonitrile:methanol and either TFA or phosphoric acid can be used. The analysis time is quite rapid, and nearly all of the major products are well separated from one another. If it is desired to achieve greater separation of the products in any area of the chromatogram, a more complex gradient, composed of a combination of shallow and steep segments (cf. Figs. 1 and 2) could be used, but this might result in overlap between LTC4 and monohydroxy metabolites of 20:4.
AND LEUKOTRIENES
This work was supported by grants from the Medical Research Council of Canada and the Quebec Heart Foundation. The author is the holder of a Scientist award from the MRC. The author is grateful to Fran&e Gravelle for expert technical assistance. REFERENCES 1. Powell, W. S. (1981) Anal. Biochem. 115, 267-277. 2. Powell, W. S. (1983) Anal. B&hem. 128, 93-103. 3. Borgeat, P., Picard S., Vallerand, P., and Sirois, P. (1981) Prosiaglandins Med. 6, 557-570. 4. Powell, W. S. (1980) Prostaglandins 20, 947-957. 5. Powell, W. S. (1985) in Biochemistry of Arachidonic Acid Metabolism (Lands, W. E. M., ed.), KluwerNijhoff Publishing, Boston, in press. Powell, W. S. (1984) J. Biol. Chem. 259, 30823089. Hamberg, M., and Samuelsson, B. (1974) Proc. Natl. Acad.
Sci. USA 71, 3400-3404.
Hamberg, M., and Samuelsson, B. (1967) J. Biol. Chem.
242, 5329-5335.
Boeynaems, J. M., Brash, A. R., Oates, J. A., and Hubbard, W. C. (1980) Anal. Biochem. 104,259267. 10. Powell, W. S. (1982) in Methods in Enzymology (Lands, W. E., and Smith, W. L., eds.), Vol. 86, pp. 467-477, Academic Press, New York. 11. Borgeat, P., and Samuelsson, B. (1979) Proc. Natl. Acad. Sci. USA 76, 2148-2152.
12. Alam, I., Ohuchi, I., and Levine, L. (1979) Anal. Biochem.
93, 339-345.
13. Henke, D. C., Kouzan, S., and Eling, T. E. (1984) Anal.
Biochem.
140, 87-94.
14. Peters, S. P., Schulman, E. S., Liu, M. C., Hayes, E. C., and Lichtenstein, L. M. (1983) 1. Zmmunol. Methods
64, 335-343.
15. VanRollins, M., Aveldano, M. I., Sprecher, H. W., and Horrocks, L. A. (1982) in Methods in Enzymology (Lands, W. E., and Smith, W. L., eds.), Vol. 86, pp. 5 18-530, Academic Press, New York. 16. Mathews, W. R., Rokach, J., and Murphy, R. C. (1981) Ana/. Biochem. 118, 96-101. 17. Borgeat, P. (1984) Prostaglandins 27, 356-358. 18. Bennett, H. P. J., Browne, C. A. B., and Solomon, S. (198 1) Biochemistry 20, 4530-4538. 19. Metz, S. A., Hall, M. E., Harper, T. W., and Murphy, R. C. (1982) J. Chromatogr. 233, 193201.