Determination of Serotonin and Dopamine in Mouse Brain Tissue by High Performance Liquid Chromatography with Electrochemical Detection Suleiman Sasa and C. LeRoy Blank* Department of Chemistry, University of Oklahoma, 620 Parrington Oval, Norman, Okla. 730 19
High performance liquid chromatographywith electrochemical detection (LCEC) is shown to be readily applicable to the selective determination of serotonin and dopamine in tissue samples. Being less demanding than existent techniques with regard to reagent purity, the method is capable of quantitating single mouse brain parts. Whole mouse brain results were 805 f 12 (SEM) and 973 f 16 (SEM) ng/g wet tlssue for serotonin and dopamine, respectively ( n = 40). Individual analyses of cerebelium yielded values of 240 f 14 (SEM)and 19 f 2 (SEM) ng/g for these same species ( n = 6). Requiring only 10 mln between individual injections,the LCEC can quantitate as little as 0.1 pmol each of serotonin and dopamine In a single Injection.
The importance of the neurotransmitters, dopamine (3,4-dihydroxyphenethylamine,DA) and serotonin (5-hydroxytryptamine, 5-HT) to neurochemical investigations is well established. Their possible involvement in a variety of mental disorders is currently a subject of intense examination. Techniques presently available for the determination of these amines in tissue samples (1-15) all usually contain some basic similarities. These include homogenization, centrifugation, and some initial isolation of the amines. Generally, following physical separation of DA from 5-HT, or the division of the sample into two distinct samples, the quantitation of each species is separately performed. This latter step may or may not involve derivatization or chemical manipulation prior to actual detection. The only major differences between these techniques occur a t the isolation and quantitation stages. Isolation of DA and 5-HT has been accomplished by either liquid chromatographic or extraction methods. The chromatography has primarily utilized methodology similar to the cation exchange approach described by Taylor and Laverty ( 2 ) .Extractions have typically involved pH adjustment, extraction into, e.g., 1-butanol, and, after addition of heptane to the butanol, a back extraction into acidic aqueous solution. Such an extraction is described by Shore and Olin (13)for the determination of norepinephrine and 5-HT. DA, of course, can also be isolated by selective alumina adsorption ( 1 6 ) . The quantitation of the amines has been performed by fluorescence, gas chromatography, and radiochemical techniques. Although the fluorometric determination of 5-HT may be performed directly in 3 M HC1 (17),many workers seem to prefer derivatization with o-phthalaldehyde, as used by Maickel et al. ( 5 ) ,or with ninhydrin, as introduced by Vanable (6) to increase the sensitivity. DA, on the other hand, is normally converted to its fluorophor by an oxidative rearrangement known as the trihydroxyindole method (7), prior to fluorometric determination. Both 5-HT and DA determinations by fluorometry, however, require the use of spectroscopically pure reagents and various other precautions to avoid the possible introduction of fluorescing interferences. Analysis of individual brain parts also usually involves sample pooling to obtain the required sensitivity. 354
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
Gas chromatography with electron capture detection for the analysis of tissue samples containing 5-HT and DA has been suggested by Arnold and Ford (9) and others (10).Yet, the sensitivity limit was only ca. 50 nglg of wet tissue for weights of 500 mg. Analysis times were reported as five hours for six samples. The combined technique of gas chromatography with mass spectrometry (GCMS) was shown by Koslow et al. (11)to be applicable also to such determinations. While offering excellent selectivity and detection limits in the subpicomole range, GCMS is, undeniably, a very expensive technique to employ on a routine basis. A rather recent entry into the area of DA and 5-HT determinations is the radiochemical techniques (14,15).These use a radiolabeled cofactor and a purified enzyme preparation (e.g., catechol-0-methyl-transferase, COMT, and labeled S-adenosyl methionine for DA) to produce a radiolabeled derivative of the biogenic amine. After subsequent separation of the product, scintillation counting quantitates the amine directly. This, unfortunately, involves the use of purified enzymes which are not presently available commercially. The very recent introduction of high performance liquid chromatography with electrochemical detection (LCEC) has been shown to offer considerable selectivity, sensitivity, and low cost for the analysis of the catecholamines, norepinephrine and dopamine (12).Because of the importance of 5-HT and DA to neuroscience investigations, we decided to see if this apparatus might be directly applicable to this pair of compounds also.
EXPERIMENTAL Apparatus. Liquid Chromatograph. The high performance liquid chromatography with electrochemical detection (LCEC) setup has been adequately described by Kissinger et al. (18) and Refshauge e t al. (12).The major differences in the present system are [I]the cation exchange column, packed with DuPont Zipax SCX strong cation exchange resin, was 3-mm diameter by 750-mm length; [2] the chromatographic elution solvent was a pH 5.1 citrate/acetate buffer solution (vide infra); [3] the flow rate was ca. 1.3 ml/min; and, [4] the working electrode potential was set at +0.6 V vs. SCE. Homogenizer. The brain tissue samples were homogenized with an ultrasonic homogenizer (WZOOP,Ultrasonics, Inc.) equipped with a long probe tip for small volume work. A ground glass homogenizer (Kontes-Martin) was also briefly used in an investigation of the possible ill effects of ultrasonic treatment on 5-HT levels. Shaker. An Eberbach shaker, operated a t 280 cycles/min and immersed in an air bath at 15 "C, was employed. Vials were always placed on their sides with the long axis of the vials parallel to the direction of shaking. Animals. Mice. All mice used were adult males of the ARS-HA/ICR strain (Sprague-Dawley, Madison, Wis.) weighing 20 to 30 g a t the time of sacrifice. These were maintained on a 12-h dark/l2-h light cycle (lights on at 7:OO am-7:00 p.m,) and allowed access to food and water ad libitum. No mice were employed in the experiments until, a t least, one week after their arrival so that they could become accustomed to these conditions. Also, since daily fluctuations in the biogenic amine levels is well known (191,the animals were sacrificed only a t a fixed time each day (10:OO-10:30 a.m.). Reagents. 1-Butanol. Both reagent grade butanol from Mallinckrodt and butanol purified according to Shore and Olin (13)were employed.
n-Heptane. Baker Analyzed heptane and heptane purified according to Shore and Olin (13)were utilized. Biogenic Amines. Dopamine hydrochloride (DA) and serotonin creatinine sulfate monohydrate (5-HT) were obtained in the highest possible purity from Aldrich and Regis Chemical Co., respectively. All concentrations of these compounds are expressed as the free base. 3,4-Dihydroxybenzylamine. 3,4-Dihydroxybenzylamine hydrobromide (DHBA, the internal standard) was 98% pure from Aldrich. All solutions except those for brain part determinations were
M. Water. All H20 used was deionized. Stock Standard Solution. A solution containing ca. 90 pg/ml DA and 65 pg/ml 5-HT was prepared by dissolving the appropriate amount of their salts, accurately weighed, in 100 ml of 0.01 M HCl. The HCl was previously deaerated for 15 rnin with 02-free N2 (12). This solution was stable for up to one month if stored in the refrip erator at 4 "C. While designed specifically for the whole brain determinations, appropriate modifications for other samples are obvious. Working Standard Solution. On the day of use, a 1.00-mlpipet of the stock standard solution was diluted to 100 ml with deaerated 0.01 M HC1 to give a working standard containing ca. 900 ng/ml DA and 650 ng/ml5-HT. This is comparable to the whole brain concentrations (expressed as ngig wet tissue). M DHBA was prepared Internal Standard. A solution of ca. by dissolving the appropriate amount (this concentration need not be accurately known) of DHBAmHBr in deaerated 0.01 M HC1. This solution was stable for up to one month when stored at 4 "C. Different concentrations of DHBA may be needed for samples other than whole mouse brain. Ascorbic Acid. Prepared immediately before use, the ascorbic acid solution was obtained by dissolving 11 mg of ascorbic acid in 1 ml of deaerated 0.01 M HCl. Acetate-Citrate Buffer. This buffer (pH 5.1) was used as the mobile phase for the LCEC. It was prepared by dissolving 8.2 g sodium acetate, 2.1 ml acetic acid, 4.8 g sodium hydroxide, and 10.5 g citric acid in 1liter of water. Before use, it was filtered through a 0.45 pm Millipore filter to prolong pump life. a-Methyldopamine. The hydrobromide salt of 2-amino-3-(3,4dihydroxypheny1)propane (a-methyldopamine) was prepared according to Borgman et al. (20). Other Compounds. The remaining compounds, including those tested for possible interference in the procedure, were obtained in the highest possible purity from commercially available sources. Procedure. Whole Brains. The procedure described here is optimal (vide infra) for the determination of whole mouse brains having a typical weight of 450-500 mg. Appropriate modifications should be made for use with other samples. The mice were sacrificed by cervical dislocation; the brains were removed as rapidly as possible, frozen in liquid N2, and stored on dry ice. Each brain was weighed to the nearest mg and transferred to a 30-ml screw-cap vial. The following solutions were already present M DHBA, and 750 in the vial: 100 p1 of 0.1 M EDTA, 100 pl of pl of 0.025 M HC1. While the concentration of DHBA used need not be accurately known, the same volume of the same solution should be used throughout a given analysis for both brains and standards. The mixture was subjected to ultrasonic homogenization resulting in a homogenate pH of ca. 5.1-5.2. Salt saturation was accomplished by adding 3-4 g of NaC1. A 12.0-ml aliquot of butanol was added to the vial and the samples were shaken for 60 min. A 10.0-ml pipet of the butanol layer was then removed and placed in a second 30-ml screw-cap vial containing 17.0 ml heptane and 200 pl of 0.01 M HCl. The second bottle was shaken for 5 min and the aqueous layer was allowed 10 min t o separate. After removal of most of the nonaqueous layer, a 50-11sample of the HCl solution was injected into the LCEC apparatus. Working standards (0.500 ml) were treated exactly as the brain samples. The only differences were: [l] the initial solution, prior to homogenization, utilized 750 pl of the acetateicitrate buffer instead of the 0.025 M HC1 to make the pH comparable to that of a brain sample at this stage; [2] 10 pl of the ascorbic acid solution was added to the homogenization mixture to mimic brain ascorbic acid levels and to help prevent oxidation of the amines; [3] the shaking time for extraction into butanol was shortened to 20 min. Calculations of all results were exactly analogous to those reported by Refshauge et al. (12) for the catecholamines. Brain Parts (Cerebellum). Each tissue sample (or 50 p1 of the working standard plus 5 pl of a solution containing 2.75 mg/ml of ascorbic acid) was homogenized in a 30-ml screw-cap vial containing
T a b l e I. P e r c e n t Recovery of Amines as a Function of Initial S h a k i n g Time" Shaking time, min
5
Recovery, % DA1 47.8 f 0.9 48.3 f 1.9 43.7 & 0.2 40.6 f 1.5
20 40 60
47.2 f 1.0 48.8 i 1.9 46.9 f 0.3 43.1 f 3.9
42.0 f 0.7 42.8 f 0.8 40.9 f 0.2 42.8 f 2.2
Purified butanol and heptane used. All results expressed as mean of a t least six separate determinations f SEM.
T a b l e 11. P e r c e n t Recovery of Amines When E x t r a c t e d in Presence of E D T A a n d Ascorbic Acid" DA
5-HT
DHBA
48.8 f 0.3
52.9 f 0.8
50.5 f 0.8
a Purified butanol and heptane used. Initial shaking time = 20 min. All results expressed as mean for six determinations f
SEM.
10 p1 of 0.1 M EDTA, 50 pl of M DHBA, and 100 11.1 of 0.025 M HC1 (or 100 pl of citrate/acetate buffer for standards). The homogenate was salt saturated by the addition of ca. 1g NaCl and extracted with 2.20 ml of butanol. Shaking was performed 60 min for samples and 20 min for standards. A 1.50-ml pipet of the butanol layer was then transferred to another vial containing 2.5 ml of heptane and 200 1 1 of 0.01 M HCl and shaken for 5 min. After removal of most of the butanol/heptane layer, 50 pl of the aqueous layer was injected into the LCEC. Statistics. All uncertainties are expressed as the standard error of the mean (SEM), since this is the usual parameter used for such determinations. Statistical significance was examined by use of the student's t-test (21)
RESULTS AND DISCUSSION Extraction P r o c e d u r e . Because of t h e relative ease a n d butanol rapidity of the Shore and Olin (13) aqueous
-
-
aqueous extraction method, we decided t o utilize their basic technique. Yet, we are not utilizing fluorescence for detection, so a reexamination of some of t h e extraction parameters was deemed in order. Time of Initial Extraction. A solution containing 0.500 ml of t h e working standard, 1.75 ml of t h e acetate/citrate buffer, and 0.200 ml of DHBA was prepared. Samples of this solution, 50 pl, were injected into t h e LCEC to establish the peak heights prior t o extraction. T h e n 1.20 ml of t h e mixture ( p H 5.1) was transferred t o a 30-ml screw-cap vial containing 12.0 ml of purified butanol. Separate samples were shaken for 5, 20, 40, or 60 min. A 10.0-ml pipet of t h e butanol was transferred to a second vial containing 17.0 ml purified heptane and 0.500 m l of 0.01 M HC1. After shaking 5 rnin and allowing 10 rnin for t h e layers t o separate, a 50-pl aliquot of the HC1 layer was injected into t h e LCEC. T h e percent recovery was established by measuring the absolute peak heights, comparing t o the non-extracted samples, and correcting for dilution. The results are given in T a b l e I. T h e optimal time appears at 20 min. T h i s value was chosen for all subsequent investigations of standard solutions. Addition of Ascorbic Acid and EDTA. In an attempt t o minimize losses d u e t o oxidation, the addition of 10 p1 of ascorbic acid and 100 pl of 0.1 M EDTA, t o t h e above solution prior t o extraction was examined. T h e results are presented i n Table 11. Comparing this traditional ascorbate EDTA t r e a t m e n t t o t h e results of Table I points t o a better recovery for t h e
+
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
355
Table 111. Percent Recovery of Amines for Purified and Unpurified" Butanol and Heptane Extraction Solvents Solvents
DA
5-HT
DHBA
Non-purified Purified
48.1 f 0.3 52.8 f 0.8
46.6 f 0.3 50.5 f 0.8
41.4 f 0.3 44.8 f 0.2
Purified butanol and heptane prepared according to Shore and Olin (13).Unpurified solvents were reagent grade butanol and Baker Analyzed heptane. All results expressed as mean f SEM for five separate determinations. P < 0.02 for each compound.
Table IV. Resoltuion between Adjacent Chromatographic Peaks as a Function of Injection Volume Resolution Injection volume, pl
Solvent front/ DHBA
10
25 50
1.5 1.5 1.3
75
1.1
DHBA/DA
356
Recovery, % PH
DA
5-HT
DHBA
1.1
29.9 f 1.1 35.5 f 0.3 33.2 f 0.5 32.9 f 0.6
21.4 f 1.1 25.2 f 0.4 24.1 f 0.6 25.4 f 0.7
23.0 f 0.6 26.2 f 0.7 23.3 f 0.4 24.2 f 0.3
2.7 3.2 5.2
a All values are reported as mean f SEM for at least six separate determinations.
Table VI. Percent Recovery in Presence and Absence of Brain Tissuea Recovery, %
DAh-HT
1.5
1.3
1.4 1.3 0.87
1.2 1.2 1.1
addition case. In a separate experiment, 50 pl of 1M NaHS03 replaced the ascorbate as a possible antioxidant. However, the NaHS03 caused a significant loss in 5-HT recovery, as previously reported with metabisulfite by Welch and Welch (I). It should be pointed out that these data do not necessarily support the antioxidant properties of ascorbic acid and EDTA as being responsible for the improvement in extraction. The results may well indicate an effect of these species on the distribution coefficients of the amines. Nonetheless, the addition was included in all subsequent investigations because of the improved recoveries. Purified Butanol and Heptane. The use of repeatedly washed butanol and heptane (or spectroscopically pure solvents) is required for the fluorometric analyses (1, 13) to eliminate interferences. LCEC should not be sensitive to such components. Thus, we examined the effect of using simple reagent grade butanol and heptane for extraction instead of their tediously purified counterparts. Although the purified reagents (Table 111)do display a significantly better recovery for each of the components ( P < 0.02, n = 5 ) the relatively small difference (3-4% increase) was not deemed enough to justify the time-consuming purifications. Thus, reagent grade butanol and heptane were utilized for all the following investigations. Volume Injected into LCEC. Obviously, the largest possible volume would be desirable for the LCEC injection to maximize the response. But, if the volume becomes too large, band spreading will destroy the essential separation of components. Therefore, we investigated the resolution of adjacent peaks for various injected volumes of a mixture composed of 500 pl of the working standard, 100 ~1 of DHBA, and 10 pl of ascorbic acid. The results (Table IV) display adequate (>1)resolution for all pairs of peaks except the DHBADA combination a t 75 11. This lack of resolution a t larger volumes complicates the calculations of results, so 50 p1 was chosen as the optimal injection volume. Maximizing the LCEC Response. Because band spreading occurs a t injection volumes greater than 50 ~ 1it, is desirable to have this 50 pl contain the largest possible concentration of biogenic amines. A theoretical examination of the [l] aqueous butanol and [2] butanol/heptane aqueous extractions shows optimal results will be provided if the smallest
-
Table V. Percent Recovery as a Function of Initial Aqueous pH Value a
-
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
In presence of brain tissue In absence of brain tissue
DA
5-HT
DHBA
31.0 f 0.7
19.5 f 0.7
25.2 f 0.7
31.3 f 0.8
25.2 & 0.6
24.3 & 0.6
a All results in presence of brain tissue represent eight separate determinations. All results in absence of tissue represent thirteen separate determinations. Results expressed as mean f SEM. Recoveries significantly different only for 5-HT ( P < 0.001).
possible volumes of the initial and final aqueous solutions are used. For the initial aqueous solution (the homogenate), a minimum volume consists of 100 p1 of 0.1 M EDTA, 100 pl of 10-5 M DHBA, 750 11 of 0.025 M HCl (or acetate/citrate buffer for standards), and the brain (or 0.5-ml working standard aliquot). Selecting smaller volumes produces tissue samples which are too viscous to homogenize and manipulate. The 0.01 M HC1 solution in the final extraction was reduced from 500 to 200 p1. Smaller volumes are physically difficult to handle. It should be noted that this decrease in volume causes a definite decrease in the absolute percent recovery; but the amount of biogenic amine (nmol or ng) actually contained in the 50-pl LCEC injection is increased. p H Effects. Using the modifications noted immediately above, the effect of the p H of the initial aqueous solution of standards (measured just prior to extraction) was investigated. As can be seen in Table V, the percent recovery for each of the amines is fairly constant over the pH range of 1.1to 5.2. This result agrees well with the report of Shore and Olin (13).It also shows that if the standards or samples should become a bit too acidic prior to their initial extraction, the final result would not be adversely effected. T i m e of Shaking for Brain Samples. Shore and Olin (13) have concluded that the brain samples, as opposed to the standards, show a maximal recovery when the shaking time for the initial extraction into butanol is extended to 1h. We repeated this experiment at shaking times of 5,20,40, and 60 min. Indeed, our results also displayed significantly higher recoveries for brain tissue at the 60-min time. The shaking for the second extraction was also examined a t 5,20,40, and 60 min; but, this latter time did not significantly affect the results. So, 60-min initial shaking time and 5-min final shaking time was used for all tissue samples. Recoveries i n t h e Presence of Tissue. There is no a priori reason to believe that extraction of the biogenic amines is the
Table VII. Whole Mouse Brain Dopamine and Serotonin Concentrations Investigators
Technique of detection
DA, ng/g f SEM (number)
5-HT, ng/g f SEM (number)
Sasa and Blank (present report) Fleming et al. (22) Wiegand and Perry (23) Welch and Welch ( 1 ) Smith ( 2 4 ) Agrawal et al. (25) Weintraub et al. (26)
LCEC Fluorescence Fluorescence Fluorescence Fluorescence Fluorescence GCMS
973 f 16 (40) 770 f 6 (13) 870 f 46 (12) 700-900 (1000) 820 f 31 (5) 1330 f 98 (6) 800 & 30 ( 5 )
805 f 12a (40) 550 f 10 (10) 820 f 12 (3) 600-800 (1000) 660 & 13 ( 5 ) 410 f 16 (6) 670 f 40 (5)
The result originally obtained (623
9 ng/g) was corrected for percent recovery in presence of tissue as discussed in text.
same in the presence of a considerable amount of tissue as in its absence. A brain homogenate was prepared by sonicating ten mouse brains in a mixture of 1.00 ml of 0.1 M EDTA, 7.50 ml of 0.025 M HC1, and 100 ml of deionized water (to replace the usual DHBA). A 1.00-ml pipet of the homogenate was transferred to each of ten separate vials. To five of these was added a 50-11 M DHBA, 4.4 X pipet of a solution containing 2.8 X M DA, and 3.0 X 10-6 M 5-HT. Five additional vials were prepared each containing 100 pl of 0.1 M EDTA, 10 pl of 11 mg/ml ascorbic acid, 900 pl of the citrate/acetate buffer, and 50 pl of the same solution of DHBA, DA, and 5-HT. After salt saturation, each vial was subjected to the extraction procedure outlined above. The DHBA, DA, and 5-HT mixture, appropriately diluted, was injected to establish peak heights prior to extraction. Samples containing only the amines (no tissue) were used to determine percent recovery in the absence of tissue. Percent recovery in the presence of tissue was determined by subtracting the samples containing only brains from those containing both brain and standard addition and comparing to the unextracted results. As presented in Table VI, DHBA and DA produced results completely comparable in both cases. However, 5-HT yielded significantly lower recovery in the presence of brain tissue than in its absence (19.5% compared to 25.2%, P < 0.001, n = 9). Thus, when using the procedure outlined for tissue samples, the final 5-HT results must be multiplied by 1.29 to obtain the proper results. Shaking Temperature. At the outset, we decided that shaking a t 15 "C would minimize losses. Thus, we examined the percent recovery of the three major components with shaking a t 15 "C and room temperature (21-25 "C). Samples were run both in the presence and absence of brain tissue. In all cases, the cooler shaking temperature produced significantly better recoveries. Individual results were relatively some 11-22% better. Whole Mouse Brain Determinations. Using the modifications listed above (see Procedure, Experimental section for summarized description), the determination of DA and 5-HT in whole mouse brain was undertaken. A typical chromatogram for such samples is presented in Figure 1.The results for a total of 40 samples is given in Table VII, along with the reported values of other investigators. As can be readily seen, the levels for both DA and 5-HT are well within the range of previously reported determinations. Yet, the present method yields both values simultaneously; no sample splitting, derivitization, or enzymatic reactions are required. Interferences. There are, in fact, a relatively small number of components which could feasibly interfere with the analysis outlined. Such species [l]must be carried through the extraction procedure, [2] must have a percent recovery and endogenous concentration sufficient to yield a final molar concentration at least within a factor of 100 of the determined
I
I
DA
I 1
L 0 10 t m e minutes
Figure 1. LCEC chromatogram of a typical whole mouse brain determination; 50 pI of final HCI solution containing ca. 14 ng DHBA, 27 ng DA, and 18 ng 5-HT; flow rate a. 1.3 ml/min; electrode potential 0.60
+
V vs. SCE
compounds, [3] must be retained by the cation exchange resin, and [4] must be electrochemically active on the carbon paste electrode a t +0.6 V vs. SCE in this medium. The most likely candidates are, thus, the basic metabolites in the catecholamine and indoleamine biosynthetic pathways. We also investigated neutral metabolites, acidic metabolites, and ascorbic acid. Ascorbic acid fulfills all the requirements for interference except one. It is not retained by the column and, thus, forms a major portion of the solvent front ( t =~2.5 min). Aromatic amino acids, including tyrosine tryptophan, 3,4-dihydroxyphenylalamine (DOPA), and 5-hydroxytryptophan, are all in the zwitterion form a t the p H of the eluting solvent. Thus, they elute on the solvent front. Tyrosine and typtophan, incidentally, are also not observed because of their inability to be oxidized at this potential. Both 3,4-dihydroxyphenethyl alcohol and 3,4-dihydroxyphenylglycol (neutral metabolites) elute under the solvent front. 3,4-Dihydroxyphenylaceticacid and 3,4-dihydroxymandelic acid, both acidic metabolites, elute under the solvent front, also. Biogenic amines related to DA and 5-HT comprise another group examined. Norepinephrine (NE) is slightly separated from the solvent front. But because of the large tissue concentrations of ascorbic acid, NE is effectively buried and, unfortunately, not determinable. Epinine (N-methyldopamine) is completely separated (tR = 18.5 min) from the 5-HT peak ( t R = 9.3 min) and the procedure could be utilized diANALYTICAL CHEMISTRY, VOL. 49,
NO. 3,
MARCH 1977
357
Table VIII. Effect of Homogenization Method on Whole Mouse Brain Determination of Dopamine and Serotonina Method of homogenization
DA, ngk
5-HT, nglg
Ultrasonic Ground glass
1023 f 60 985 f 30
837 f 44 810 f 23
a All values reported as mean f SEM for five separate determinations. There is no significant difference between methods of homogenization for either DA or 5-HT. 5-HT results corrected for percent recovery in presence of tissue (see text).
Table IX. Dopamine and Serotonin Concentrations in Mouse Brain Cerebelluma DA, nglg
5-HT, nglg
19 f 2
240 i 14
a Results expressed as mean f SEM for six separate determinations. 5-HT results corrected for percent recovery in presence of tissue (see text).
rectly to determine this species in appropriate samples. Tryptamine is not detectable a t the potential chosen; but chromatograms run a t +0.94 V vs. SCE display a very broad peak for tryptamine at t R = 24.5 min. Epinephrine (E) elutes f t R = 5.2 min) almost directly between DHBA ( t R = 4.0 min) and DA ( t =~6.2 min), but is rather poorly resolved ( R = 0.77 and 0.67, respectively). However, E is normally present in brain tissue in only very small amounts(27). Even if present in equimolar concentrations with DA, E would contribute only ca. 6% to the measured peak height for the former. Thus, as long as E cannot be visually detected in the chromatography of other samples as a peak or noticeable shoulder, it will not present any major problems. Quantitative separation of DA and E can, if necessary, be accomplished by either diluting the citrate/acetate elution solvent or utilizing the column conditions of Refshauge (28). Methoxylated metabolites form the other major group examined. 3-Methoxy-4-hydroxyphenylalanine(3-0-methylDOPA) elutes with the solvent front as do the other amino acids mentioned previously. 5-Methoxytryptamine is not electrochemically active a t the potential chosen; however, it does exhibit a broad, poorly resolved peak ( t =~28.0) at a potential of +0.94 V vs. SCE. Metabolites which are observed under the given conditions include normetanephrine ( t =~ 4.0 min), 3-methoxy-4-hydroxyphenethylamine( t R = 46 min), 4-methoxy-3-hydroxyphenethylamine ( t =~ 68 rnin), and metanephrine ( t =~23 min). Of these, only normetanephrine, which is isographic with DHBA, is of concern. The other species are clearly separated from the components of interest. Also, their endogenous concentrations are generally so low as to allow ignoring them in routine determinations. Normetanephrine (NM), as the other methoxylated catecholamines, is only partially oxidized (ca. 15%)a t the potential chosen. Additionally, NM is present in brain tissue a t relatively low concentrations (65 f 4 ng/g, personal observation). Determinations in which DHBA was omitted showed no discernible NM peak at the sensitivity employed (less than 1%of the usual DHBA peak). Thus, N M should not normally cause any significant interference. Since both NM and E overlap the chosen internal standard, other species were investigated for possible use in samples 358
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH i s 7 7
where either of these may pose problems. a-Methyldopamine ( t =~18.5min) was found to be quite adequate for this purpose. But, one should remember that this will increase the time required for individual determinations. Also, amethyldopamine is isographic with epinine. In short, there are no major known interferences in the procedure if the comments above are properly considered. We are presently (to be published) applying the technique to the analysis of such samples as serum and other body fluids and tissues. Method of Homogenization. Alliger (29) reported that long exposure of biological samples to ultrasonic irradiation caused a destruction of the indole nucleus of tryptophan. Since 5-HT also contains an indole moiety, we compared ultrasonic homogenization to that performed with a ground glass apparatus. As presented in Table VIII, there was no significant effect on the results. Mouse Brain Cerebellum Determination. T o give a feeling for the sensitivity of the LCEC technique in routine investigations, the determination of DA and 5-HT in mouse brain cerebellum was undertaken. This particular brain part was chosen because it is easily accessible for dissection and it also shows the smallest concentration for DA (to be published) of any of the usual parts examined. The combined results for a typical determination are presented in Table IX. The average tissue weight was 53 mg, so the actual amount of DA determined in each sample was ca. 1.0 ng (6.6 pmol), while that for 5-HT was 9.8 ng (51 pmol). Although we could not find any references to DA and 5-HT determinations in mouse brain cerebellum, the results do agree fairly well with those reported for rat brains (30,31). Routine Application. With the present single column setup, it has been possible to run as many as 30 samples during a normal eight-hour day. This required that the tissue samples were ready to be weighed a t the start of the day. However, it also included sufficient time to clean all appropriate glassware while the chromatography was being run. When sacrificing and tissue removal were included, 20 samples could be handled in the same time period. The number of samples per day, however, may be increased significantly by utilizing the dual parallel LCEC apparatus described by Blank (32);this costs only a moderate amount more than a single setup, but effectively halves the time required for each chromatogram. Sensitivity and Range of t h e LCEC. In the course of development, it was necessary to show the LCEC peaks to be linear over a reasonable range. The linearity, displayed as amount of injected compound, was exhibited from 0.1 pmol to 5 nmol for each of DHBA, DA, and 5-HT.
ACKNOWLEDGMENT The authors thank Dave Wassil for the preparation of a methyldopamine.
LITERATURE CITED (1) A. S.Welch and B. L. Welch, Anal. Biochem., 30, 161 (1969). (2) K. M. Taylor and R. Laverty, J. Neurochem., 16, 1367 (1969). (3) G. B. Ansell and M. F. Beeson, Anal. Biochem., 23, 169 (1968). (4) M. K. Shellenberger and J. H. Gordon, Anal. Biochem., 39, 356 (1971).
(5) R. P. Maickel, R. H. Cox, J. Saillant, and F. P. Miller, Int. J. Neuropharmacol., 7, 725 (1968). (6) J. W. Vanable, Jr., Anal. Biochem., 6, 393 (1963). (7) U. S.von Euler, Pharmacol. Rev., 11, 262 (1959). (8) D. F. Bogdanski, A. Pletscher, B.B. Brodie, and S. Udenfriend, J. Pharmacol. Exp. Ther., 117, 82 (1956). (9) E . L. Arnold and R. Ford, Anal. Chem., 45, 85 (1973). (IO) i. L. Martin and G. B. Ansell, Biochem. fharmacol., 22, 521 (1973). (1 1) S. H. Koslow, F. Cattabeni, and E. Costa, Science, 176, 177 (1972). (12) C. Refshauge, P. T. Kissinger, R. Dreiling, L. Blank, R. Freeman, andR. N. Adams, L i f e Sci., 14, 311-(1974). (13) P. A. Shore and J. S. Olin, J. Pharmacol. f x p . Ther., 122, 295 (1958). (14) A. C. Cuello, R. Hiley, and L. L. Iversen, J. Neurochem., 21, 1337 (1973). (15) J. M. Saavedra, M. Brownstein, and J. Axelrod, J. Pharmacol. f x p . Ther., 186, 508 (1973).
(16) A. H. Antone and D. F. Sayre, J. Pharmacol. Exp. Ther., 138, 360 (1966). (17) S. Udenfriend, D. F. Bogdanski, and D. F. Weissbach, Science, 122, 972 (1955). (18) P. T. Kissinger. C. Refshauge,R. Dreiling, and R. N. Adams, Anal. Left.,8, 465 (1973). (19) P. Albrecht, M. B. Visscher,J. J. Bittner, and F. Halberg, Proc. SOC.Exp. Biol. Med., 92, 702 (1956). (20) R. J. Borgman, M. R. Baylor, J. J. McPhillips,and R. E. Stitzel, J. Med. Chem, 17, 427 (1974). (21) Louis Levine, "Biology of the Gene", 2d ed., C. V. Mosby Co., St. Louis, Mo., 1973. (22) R. M. Flemina, W . G. Clark, G. D. Fenster, and J. E . Towne, Anal. Chem., 37, 292 (1963). (23) R. G. Wiegand and J. E. Perry, Biochem. Pharmacol., 7, 181 (1961). (24) C. B.Smith, J. Pharmacol. Exp. Ther., 142, 343 (1963). (25) H. C. Agrawai, S.N. Glisson, and W. A. Himwich, lnt. J. Neuropharmacol., 7, 97 (1968).
(26) S. T. Weintraub, W. B. Stavinoha,R. L. Pike, W. W. Morgan, A. T. Modak, S. H.Koslow, and L. Blank, Life Sci., 17, 1423 (1975). (27) R. D. Ciavanelio, R. E. Barchas, G. S. Byers, D. W. Stemmle, and J. D. Barchas, Nature (London),221, 363 (1969). (28) C. Refshauge, Ph.D. Thesis, University of Kansas, Lawrence, Kan., 1974. (29) H. Alliger, Am. Lab., No. 7, 75 (1975). (30) R. H.Cox and J. L. Perhach,Jr., J. Neurochem., 20, 1777 (1973). (31) R. P. Maickel in ':Methods of Neurochemistry", Voi. 2, R. Fried, Ed,, Marcel Dekker, New York, 1972. (32) C. L. Blank, J. Chromatogr., 117, 35 (1976).
RECEIVEDfor review September 7,1976. Accepted December lg7" This work was by grant number MH26866-01, awarded by NIMH/DHEW. 8 j
Separation of Five Major Alkaloids in Gum Opium and Quantitation of Morphine, Codeine, and Thebaine by lsocratic Reverse Phase High Performance Liquid Chromatography C. Y. WU" and J. J. Wittick Merck Chemical Manufacturing Division, Rahway, N.J. 07065
Two similar isocratic liquld chromatographic (LC) systems have been developed for the determination of morphine, codeine, and thebaine in gum opium. One Is for the quantitatlve determination of morphine alone and the other for the simultaneous quantltative determination of codelne and thebaine. Two 30-cm X 4.0-mm i.d., p-Bondapak C18/Porasll columns are used for both systems which differ only In the composition of the mobile phase used. The composltlonof the mobile phase used for the morphine determination is 0.1 M NaH2P04in 5 % CH3CN/H20 and for the codelne and thebaine determination Is 0.1 M NaH2P04In 25% CH3CN/H20. Both systems are very reproduclble and sultable for routlne anaiysls. The preclslon of the method for morphine is 0.8% relative standard deviation and for codelne and thebaine 1.3 and 3.3 % relatlve standard deviation, respectively.
In spite of the fact that opium has been associated with human activities for centuries, chemists are still searching for better analytical methods to measure the alkaloids content of this important pharmaceutical raw material. Progress was hampered primarily because there are more than twenty alkaloids present in gum opium and suitable techniques for separating this complex mixture were lacking. However, with the development of a variety of chromatographic techniques, these difficulties have been overcome. Tedious prior purification of a specific alkaloid before quantitation is no longer necessary and simultaneous quantitative analysis for several alkaloids in the complex mixture has become a reality. Steady progress in the use of various chromatographic techniques for the analysis of opium and related mixtures has been demonstrated in the literature (1-18). In 1973 Wu et al. (7), using a synthetic mixture of five of the major opium alkaloids, successfully demonstrated that the HPLC technique is potentially useful for such analysis. In the same year, Wittwer (16) reported a gradient HPLC system which he claimed to be suitable for the determination of opium alkaloids. However, only limited data were reported. Later, Beaseley et
al. (15)described a procedure using normal phase liquid-solid adsorption chromatography with gradient elution. The method is lengthy and, unlike reverse phase chromatography, separation by normal phase liquid-solid adsorption chromatography is highly susceptible to the water content both of the adsorbent and of the mobile phase (19), making the technique difficult to use for routine analysis in our opinion. Techniques for opium alkaloids analysis other than chromatography are also available. For details refer to the literature (20-25). The objective of this paper is to introduce a practical, specific, relatively simple and rapid method with the capability of yielding precise quantitation for morphine, codeine, and thebaine in gum opium. Two essentially identical isocratic liquid chromatographic systems are described, one for morphine and the other for codeine and thebaine. The chief difference in these two systems is in the composition of the mobile phase. A microparticulate, w-Bondapak C l g column was selected and used because it provided the high column efficiency which is necessary for the analysis of a coplex mixture such as gum opium. This column also provided excellent reproducibility over a long period of time, making it especially suitable for routine analysis.
EXPERIMENTAL Apparatus. A DuPont Liquid Chromatograph Model 830, a Schoeffel SF 770 Spectroflow Monitor equipped with a deuterium lamp, an Autolab Vidar 6300 Digital Integrator, and a HewlettPackard strip chart recorder Model 7123A were properly connected and used. Samples were introduced through an injection valve purchased from DuPont Instruments (Wilmington, Del., Catalogue No. 204590). Reagents a n d Solutions. Pure morphine sulfate pentahydrate, pure codeine phosphate, and pure thebaine standards were obtained within the Merck Chemical Manufacturing Division. Reagent grade calcium hydroxide powder, 85%phosphoric acid, glacial acetic acid, sodium chloride, sodium perchlorate, dimethyl sulfoxide, and methylene chloride were used. Acetonitrile (UV) "Distilled in Glass" was purchased from Burdick & Jackson, (Muskegon, Mich.). Presaturated calcium hydroxide solution was prepared fresh daily by ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
359
High performance liquid chromatographywith electrochemical detection (LCEC) is shown to be readily applicable to the selective determination of serotonin and dopamine in tissue samples. Being less demanding than existent techniques with regard to reagent purity, the method is capable of quantitating single mouse brain parts. Whole mouse brain results were 805 f 12 (SEM) and 973 f 16 (SEM) ng/g wet tlssue for serotonin and dopamine, respectively ( n = 40). Individual analyses of cerebelium yielded values of 240 f 14 (SEM)and 19 f 2 (SEM) ng/g for these same species ( n = 6). Requiring only 10 mln between individual injections,the LCEC can quantitate as little as 0.1 pmol each of serotonin and dopamine In a single Injection.
The importance of the neurotransmitters, dopamine (3,4-dihydroxyphenethylamine,DA) and serotonin (5-hydroxytryptamine, 5-HT) to neurochemical investigations is well established. Their possible involvement in a variety of mental disorders is currently a subject of intense examination. Techniques presently available for the determination of these amines in tissue samples (1-15) all usually contain some basic similarities. These include homogenization, centrifugation, and some initial isolation of the amines. Generally, following physical separation of DA from 5-HT, or the division of the sample into two distinct samples, the quantitation of each species is separately performed. This latter step may or may not involve derivatization or chemical manipulation prior to actual detection. The only major differences between these techniques occur a t the isolation and quantitation stages. Isolation of DA and 5-HT has been accomplished by either liquid chromatographic or extraction methods. The chromatography has primarily utilized methodology similar to the cation exchange approach described by Taylor and Laverty ( 2 ) .Extractions have typically involved pH adjustment, extraction into, e.g., 1-butanol, and, after addition of heptane to the butanol, a back extraction into acidic aqueous solution. Such an extraction is described by Shore and Olin (13)for the determination of norepinephrine and 5-HT. DA, of course, can also be isolated by selective alumina adsorption ( 1 6 ) . The quantitation of the amines has been performed by fluorescence, gas chromatography, and radiochemical techniques. Although the fluorometric determination of 5-HT may be performed directly in 3 M HC1 (17),many workers seem to prefer derivatization with o-phthalaldehyde, as used by Maickel et al. ( 5 ) ,or with ninhydrin, as introduced by Vanable (6) to increase the sensitivity. DA, on the other hand, is normally converted to its fluorophor by an oxidative rearrangement known as the trihydroxyindole method (7), prior to fluorometric determination. Both 5-HT and DA determinations by fluorometry, however, require the use of spectroscopically pure reagents and various other precautions to avoid the possible introduction of fluorescing interferences. Analysis of individual brain parts also usually involves sample pooling to obtain the required sensitivity. 354
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
Gas chromatography with electron capture detection for the analysis of tissue samples containing 5-HT and DA has been suggested by Arnold and Ford (9) and others (10).Yet, the sensitivity limit was only ca. 50 nglg of wet tissue for weights of 500 mg. Analysis times were reported as five hours for six samples. The combined technique of gas chromatography with mass spectrometry (GCMS) was shown by Koslow et al. (11)to be applicable also to such determinations. While offering excellent selectivity and detection limits in the subpicomole range, GCMS is, undeniably, a very expensive technique to employ on a routine basis. A rather recent entry into the area of DA and 5-HT determinations is the radiochemical techniques (14,15).These use a radiolabeled cofactor and a purified enzyme preparation (e.g., catechol-0-methyl-transferase, COMT, and labeled S-adenosyl methionine for DA) to produce a radiolabeled derivative of the biogenic amine. After subsequent separation of the product, scintillation counting quantitates the amine directly. This, unfortunately, involves the use of purified enzymes which are not presently available commercially. The very recent introduction of high performance liquid chromatography with electrochemical detection (LCEC) has been shown to offer considerable selectivity, sensitivity, and low cost for the analysis of the catecholamines, norepinephrine and dopamine (12).Because of the importance of 5-HT and DA to neuroscience investigations, we decided to see if this apparatus might be directly applicable to this pair of compounds also.
EXPERIMENTAL Apparatus. Liquid Chromatograph. The high performance liquid chromatography with electrochemical detection (LCEC) setup has been adequately described by Kissinger et al. (18) and Refshauge e t al. (12).The major differences in the present system are [I]the cation exchange column, packed with DuPont Zipax SCX strong cation exchange resin, was 3-mm diameter by 750-mm length; [2] the chromatographic elution solvent was a pH 5.1 citrate/acetate buffer solution (vide infra); [3] the flow rate was ca. 1.3 ml/min; and, [4] the working electrode potential was set at +0.6 V vs. SCE. Homogenizer. The brain tissue samples were homogenized with an ultrasonic homogenizer (WZOOP,Ultrasonics, Inc.) equipped with a long probe tip for small volume work. A ground glass homogenizer (Kontes-Martin) was also briefly used in an investigation of the possible ill effects of ultrasonic treatment on 5-HT levels. Shaker. An Eberbach shaker, operated a t 280 cycles/min and immersed in an air bath at 15 "C, was employed. Vials were always placed on their sides with the long axis of the vials parallel to the direction of shaking. Animals. Mice. All mice used were adult males of the ARS-HA/ICR strain (Sprague-Dawley, Madison, Wis.) weighing 20 to 30 g a t the time of sacrifice. These were maintained on a 12-h dark/l2-h light cycle (lights on at 7:OO am-7:00 p.m,) and allowed access to food and water ad libitum. No mice were employed in the experiments until, a t least, one week after their arrival so that they could become accustomed to these conditions. Also, since daily fluctuations in the biogenic amine levels is well known (191,the animals were sacrificed only a t a fixed time each day (10:OO-10:30 a.m.). Reagents. 1-Butanol. Both reagent grade butanol from Mallinckrodt and butanol purified according to Shore and Olin (13)were employed.
n-Heptane. Baker Analyzed heptane and heptane purified according to Shore and Olin (13)were utilized. Biogenic Amines. Dopamine hydrochloride (DA) and serotonin creatinine sulfate monohydrate (5-HT) were obtained in the highest possible purity from Aldrich and Regis Chemical Co., respectively. All concentrations of these compounds are expressed as the free base. 3,4-Dihydroxybenzylamine. 3,4-Dihydroxybenzylamine hydrobromide (DHBA, the internal standard) was 98% pure from Aldrich. All solutions except those for brain part determinations were
M. Water. All H20 used was deionized. Stock Standard Solution. A solution containing ca. 90 pg/ml DA and 65 pg/ml 5-HT was prepared by dissolving the appropriate amount of their salts, accurately weighed, in 100 ml of 0.01 M HCl. The HCl was previously deaerated for 15 rnin with 02-free N2 (12). This solution was stable for up to one month if stored in the refrip erator at 4 "C. While designed specifically for the whole brain determinations, appropriate modifications for other samples are obvious. Working Standard Solution. On the day of use, a 1.00-mlpipet of the stock standard solution was diluted to 100 ml with deaerated 0.01 M HC1 to give a working standard containing ca. 900 ng/ml DA and 650 ng/ml5-HT. This is comparable to the whole brain concentrations (expressed as ngig wet tissue). M DHBA was prepared Internal Standard. A solution of ca. by dissolving the appropriate amount (this concentration need not be accurately known) of DHBAmHBr in deaerated 0.01 M HC1. This solution was stable for up to one month when stored at 4 "C. Different concentrations of DHBA may be needed for samples other than whole mouse brain. Ascorbic Acid. Prepared immediately before use, the ascorbic acid solution was obtained by dissolving 11 mg of ascorbic acid in 1 ml of deaerated 0.01 M HCl. Acetate-Citrate Buffer. This buffer (pH 5.1) was used as the mobile phase for the LCEC. It was prepared by dissolving 8.2 g sodium acetate, 2.1 ml acetic acid, 4.8 g sodium hydroxide, and 10.5 g citric acid in 1liter of water. Before use, it was filtered through a 0.45 pm Millipore filter to prolong pump life. a-Methyldopamine. The hydrobromide salt of 2-amino-3-(3,4dihydroxypheny1)propane (a-methyldopamine) was prepared according to Borgman et al. (20). Other Compounds. The remaining compounds, including those tested for possible interference in the procedure, were obtained in the highest possible purity from commercially available sources. Procedure. Whole Brains. The procedure described here is optimal (vide infra) for the determination of whole mouse brains having a typical weight of 450-500 mg. Appropriate modifications should be made for use with other samples. The mice were sacrificed by cervical dislocation; the brains were removed as rapidly as possible, frozen in liquid N2, and stored on dry ice. Each brain was weighed to the nearest mg and transferred to a 30-ml screw-cap vial. The following solutions were already present M DHBA, and 750 in the vial: 100 p1 of 0.1 M EDTA, 100 pl of pl of 0.025 M HC1. While the concentration of DHBA used need not be accurately known, the same volume of the same solution should be used throughout a given analysis for both brains and standards. The mixture was subjected to ultrasonic homogenization resulting in a homogenate pH of ca. 5.1-5.2. Salt saturation was accomplished by adding 3-4 g of NaC1. A 12.0-ml aliquot of butanol was added to the vial and the samples were shaken for 60 min. A 10.0-ml pipet of the butanol layer was then removed and placed in a second 30-ml screw-cap vial containing 17.0 ml heptane and 200 pl of 0.01 M HCl. The second bottle was shaken for 5 min and the aqueous layer was allowed 10 min t o separate. After removal of most of the nonaqueous layer, a 50-11sample of the HCl solution was injected into the LCEC apparatus. Working standards (0.500 ml) were treated exactly as the brain samples. The only differences were: [l] the initial solution, prior to homogenization, utilized 750 pl of the acetateicitrate buffer instead of the 0.025 M HC1 to make the pH comparable to that of a brain sample at this stage; [2] 10 pl of the ascorbic acid solution was added to the homogenization mixture to mimic brain ascorbic acid levels and to help prevent oxidation of the amines; [3] the shaking time for extraction into butanol was shortened to 20 min. Calculations of all results were exactly analogous to those reported by Refshauge et al. (12) for the catecholamines. Brain Parts (Cerebellum). Each tissue sample (or 50 p1 of the working standard plus 5 pl of a solution containing 2.75 mg/ml of ascorbic acid) was homogenized in a 30-ml screw-cap vial containing
T a b l e I. P e r c e n t Recovery of Amines as a Function of Initial S h a k i n g Time" Shaking time, min
5
Recovery, % DA1 47.8 f 0.9 48.3 f 1.9 43.7 & 0.2 40.6 f 1.5
20 40 60
47.2 f 1.0 48.8 i 1.9 46.9 f 0.3 43.1 f 3.9
42.0 f 0.7 42.8 f 0.8 40.9 f 0.2 42.8 f 2.2
Purified butanol and heptane used. All results expressed as mean of a t least six separate determinations f SEM.
T a b l e 11. P e r c e n t Recovery of Amines When E x t r a c t e d in Presence of E D T A a n d Ascorbic Acid" DA
5-HT
DHBA
48.8 f 0.3
52.9 f 0.8
50.5 f 0.8
a Purified butanol and heptane used. Initial shaking time = 20 min. All results expressed as mean for six determinations f
SEM.
10 p1 of 0.1 M EDTA, 50 pl of M DHBA, and 100 11.1 of 0.025 M HC1 (or 100 pl of citrate/acetate buffer for standards). The homogenate was salt saturated by the addition of ca. 1g NaCl and extracted with 2.20 ml of butanol. Shaking was performed 60 min for samples and 20 min for standards. A 1.50-ml pipet of the butanol layer was then transferred to another vial containing 2.5 ml of heptane and 200 1 1 of 0.01 M HCl and shaken for 5 min. After removal of most of the butanol/heptane layer, 50 pl of the aqueous layer was injected into the LCEC. Statistics. All uncertainties are expressed as the standard error of the mean (SEM), since this is the usual parameter used for such determinations. Statistical significance was examined by use of the student's t-test (21)
RESULTS AND DISCUSSION Extraction P r o c e d u r e . Because of t h e relative ease a n d butanol rapidity of the Shore and Olin (13) aqueous
-
-
aqueous extraction method, we decided t o utilize their basic technique. Yet, we are not utilizing fluorescence for detection, so a reexamination of some of t h e extraction parameters was deemed in order. Time of Initial Extraction. A solution containing 0.500 ml of t h e working standard, 1.75 ml of t h e acetate/citrate buffer, and 0.200 ml of DHBA was prepared. Samples of this solution, 50 pl, were injected into t h e LCEC to establish the peak heights prior t o extraction. T h e n 1.20 ml of t h e mixture ( p H 5.1) was transferred t o a 30-ml screw-cap vial containing 12.0 ml of purified butanol. Separate samples were shaken for 5, 20, 40, or 60 min. A 10.0-ml pipet of t h e butanol was transferred to a second vial containing 17.0 ml purified heptane and 0.500 m l of 0.01 M HC1. After shaking 5 rnin and allowing 10 rnin for t h e layers t o separate, a 50-pl aliquot of the HC1 layer was injected into t h e LCEC. T h e percent recovery was established by measuring the absolute peak heights, comparing t o the non-extracted samples, and correcting for dilution. The results are given in T a b l e I. T h e optimal time appears at 20 min. T h i s value was chosen for all subsequent investigations of standard solutions. Addition of Ascorbic Acid and EDTA. In an attempt t o minimize losses d u e t o oxidation, the addition of 10 p1 of ascorbic acid and 100 pl of 0.1 M EDTA, t o t h e above solution prior t o extraction was examined. T h e results are presented i n Table 11. Comparing this traditional ascorbate EDTA t r e a t m e n t t o t h e results of Table I points t o a better recovery for t h e
+
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
355
Table 111. Percent Recovery of Amines for Purified and Unpurified" Butanol and Heptane Extraction Solvents Solvents
DA
5-HT
DHBA
Non-purified Purified
48.1 f 0.3 52.8 f 0.8
46.6 f 0.3 50.5 f 0.8
41.4 f 0.3 44.8 f 0.2
Purified butanol and heptane prepared according to Shore and Olin (13).Unpurified solvents were reagent grade butanol and Baker Analyzed heptane. All results expressed as mean f SEM for five separate determinations. P < 0.02 for each compound.
Table IV. Resoltuion between Adjacent Chromatographic Peaks as a Function of Injection Volume Resolution Injection volume, pl
Solvent front/ DHBA
10
25 50
1.5 1.5 1.3
75
1.1
DHBA/DA
356
Recovery, % PH
DA
5-HT
DHBA
1.1
29.9 f 1.1 35.5 f 0.3 33.2 f 0.5 32.9 f 0.6
21.4 f 1.1 25.2 f 0.4 24.1 f 0.6 25.4 f 0.7
23.0 f 0.6 26.2 f 0.7 23.3 f 0.4 24.2 f 0.3
2.7 3.2 5.2
a All values are reported as mean f SEM for at least six separate determinations.
Table VI. Percent Recovery in Presence and Absence of Brain Tissuea Recovery, %
DAh-HT
1.5
1.3
1.4 1.3 0.87
1.2 1.2 1.1
addition case. In a separate experiment, 50 pl of 1M NaHS03 replaced the ascorbate as a possible antioxidant. However, the NaHS03 caused a significant loss in 5-HT recovery, as previously reported with metabisulfite by Welch and Welch (I). It should be pointed out that these data do not necessarily support the antioxidant properties of ascorbic acid and EDTA as being responsible for the improvement in extraction. The results may well indicate an effect of these species on the distribution coefficients of the amines. Nonetheless, the addition was included in all subsequent investigations because of the improved recoveries. Purified Butanol and Heptane. The use of repeatedly washed butanol and heptane (or spectroscopically pure solvents) is required for the fluorometric analyses (1, 13) to eliminate interferences. LCEC should not be sensitive to such components. Thus, we examined the effect of using simple reagent grade butanol and heptane for extraction instead of their tediously purified counterparts. Although the purified reagents (Table 111)do display a significantly better recovery for each of the components ( P < 0.02, n = 5 ) the relatively small difference (3-4% increase) was not deemed enough to justify the time-consuming purifications. Thus, reagent grade butanol and heptane were utilized for all the following investigations. Volume Injected into LCEC. Obviously, the largest possible volume would be desirable for the LCEC injection to maximize the response. But, if the volume becomes too large, band spreading will destroy the essential separation of components. Therefore, we investigated the resolution of adjacent peaks for various injected volumes of a mixture composed of 500 pl of the working standard, 100 ~1 of DHBA, and 10 pl of ascorbic acid. The results (Table IV) display adequate (>1)resolution for all pairs of peaks except the DHBADA combination a t 75 11. This lack of resolution a t larger volumes complicates the calculations of results, so 50 p1 was chosen as the optimal injection volume. Maximizing the LCEC Response. Because band spreading occurs a t injection volumes greater than 50 ~ 1it, is desirable to have this 50 pl contain the largest possible concentration of biogenic amines. A theoretical examination of the [l] aqueous butanol and [2] butanol/heptane aqueous extractions shows optimal results will be provided if the smallest
-
Table V. Percent Recovery as a Function of Initial Aqueous pH Value a
-
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
In presence of brain tissue In absence of brain tissue
DA
5-HT
DHBA
31.0 f 0.7
19.5 f 0.7
25.2 f 0.7
31.3 f 0.8
25.2 & 0.6
24.3 & 0.6
a All results in presence of brain tissue represent eight separate determinations. All results in absence of tissue represent thirteen separate determinations. Results expressed as mean f SEM. Recoveries significantly different only for 5-HT ( P < 0.001).
possible volumes of the initial and final aqueous solutions are used. For the initial aqueous solution (the homogenate), a minimum volume consists of 100 p1 of 0.1 M EDTA, 100 pl of 10-5 M DHBA, 750 11 of 0.025 M HCl (or acetate/citrate buffer for standards), and the brain (or 0.5-ml working standard aliquot). Selecting smaller volumes produces tissue samples which are too viscous to homogenize and manipulate. The 0.01 M HC1 solution in the final extraction was reduced from 500 to 200 p1. Smaller volumes are physically difficult to handle. It should be noted that this decrease in volume causes a definite decrease in the absolute percent recovery; but the amount of biogenic amine (nmol or ng) actually contained in the 50-pl LCEC injection is increased. p H Effects. Using the modifications noted immediately above, the effect of the p H of the initial aqueous solution of standards (measured just prior to extraction) was investigated. As can be seen in Table V, the percent recovery for each of the amines is fairly constant over the pH range of 1.1to 5.2. This result agrees well with the report of Shore and Olin (13).It also shows that if the standards or samples should become a bit too acidic prior to their initial extraction, the final result would not be adversely effected. T i m e of Shaking for Brain Samples. Shore and Olin (13) have concluded that the brain samples, as opposed to the standards, show a maximal recovery when the shaking time for the initial extraction into butanol is extended to 1h. We repeated this experiment at shaking times of 5,20,40, and 60 min. Indeed, our results also displayed significantly higher recoveries for brain tissue at the 60-min time. The shaking for the second extraction was also examined a t 5,20,40, and 60 min; but, this latter time did not significantly affect the results. So, 60-min initial shaking time and 5-min final shaking time was used for all tissue samples. Recoveries i n t h e Presence of Tissue. There is no a priori reason to believe that extraction of the biogenic amines is the
Table VII. Whole Mouse Brain Dopamine and Serotonin Concentrations Investigators
Technique of detection
DA, ng/g f SEM (number)
5-HT, ng/g f SEM (number)
Sasa and Blank (present report) Fleming et al. (22) Wiegand and Perry (23) Welch and Welch ( 1 ) Smith ( 2 4 ) Agrawal et al. (25) Weintraub et al. (26)
LCEC Fluorescence Fluorescence Fluorescence Fluorescence Fluorescence GCMS
973 f 16 (40) 770 f 6 (13) 870 f 46 (12) 700-900 (1000) 820 f 31 (5) 1330 f 98 (6) 800 & 30 ( 5 )
805 f 12a (40) 550 f 10 (10) 820 f 12 (3) 600-800 (1000) 660 & 13 ( 5 ) 410 f 16 (6) 670 f 40 (5)
The result originally obtained (623
9 ng/g) was corrected for percent recovery in presence of tissue as discussed in text.
same in the presence of a considerable amount of tissue as in its absence. A brain homogenate was prepared by sonicating ten mouse brains in a mixture of 1.00 ml of 0.1 M EDTA, 7.50 ml of 0.025 M HC1, and 100 ml of deionized water (to replace the usual DHBA). A 1.00-ml pipet of the homogenate was transferred to each of ten separate vials. To five of these was added a 50-11 M DHBA, 4.4 X pipet of a solution containing 2.8 X M DA, and 3.0 X 10-6 M 5-HT. Five additional vials were prepared each containing 100 pl of 0.1 M EDTA, 10 pl of 11 mg/ml ascorbic acid, 900 pl of the citrate/acetate buffer, and 50 pl of the same solution of DHBA, DA, and 5-HT. After salt saturation, each vial was subjected to the extraction procedure outlined above. The DHBA, DA, and 5-HT mixture, appropriately diluted, was injected to establish peak heights prior to extraction. Samples containing only the amines (no tissue) were used to determine percent recovery in the absence of tissue. Percent recovery in the presence of tissue was determined by subtracting the samples containing only brains from those containing both brain and standard addition and comparing to the unextracted results. As presented in Table VI, DHBA and DA produced results completely comparable in both cases. However, 5-HT yielded significantly lower recovery in the presence of brain tissue than in its absence (19.5% compared to 25.2%, P < 0.001, n = 9). Thus, when using the procedure outlined for tissue samples, the final 5-HT results must be multiplied by 1.29 to obtain the proper results. Shaking Temperature. At the outset, we decided that shaking a t 15 "C would minimize losses. Thus, we examined the percent recovery of the three major components with shaking a t 15 "C and room temperature (21-25 "C). Samples were run both in the presence and absence of brain tissue. In all cases, the cooler shaking temperature produced significantly better recoveries. Individual results were relatively some 11-22% better. Whole Mouse Brain Determinations. Using the modifications listed above (see Procedure, Experimental section for summarized description), the determination of DA and 5-HT in whole mouse brain was undertaken. A typical chromatogram for such samples is presented in Figure 1.The results for a total of 40 samples is given in Table VII, along with the reported values of other investigators. As can be readily seen, the levels for both DA and 5-HT are well within the range of previously reported determinations. Yet, the present method yields both values simultaneously; no sample splitting, derivitization, or enzymatic reactions are required. Interferences. There are, in fact, a relatively small number of components which could feasibly interfere with the analysis outlined. Such species [l]must be carried through the extraction procedure, [2] must have a percent recovery and endogenous concentration sufficient to yield a final molar concentration at least within a factor of 100 of the determined
I
I
DA
I 1
L 0 10 t m e minutes
Figure 1. LCEC chromatogram of a typical whole mouse brain determination; 50 pI of final HCI solution containing ca. 14 ng DHBA, 27 ng DA, and 18 ng 5-HT; flow rate a. 1.3 ml/min; electrode potential 0.60
+
V vs. SCE
compounds, [3] must be retained by the cation exchange resin, and [4] must be electrochemically active on the carbon paste electrode a t +0.6 V vs. SCE in this medium. The most likely candidates are, thus, the basic metabolites in the catecholamine and indoleamine biosynthetic pathways. We also investigated neutral metabolites, acidic metabolites, and ascorbic acid. Ascorbic acid fulfills all the requirements for interference except one. It is not retained by the column and, thus, forms a major portion of the solvent front ( t =~2.5 min). Aromatic amino acids, including tyrosine tryptophan, 3,4-dihydroxyphenylalamine (DOPA), and 5-hydroxytryptophan, are all in the zwitterion form a t the p H of the eluting solvent. Thus, they elute on the solvent front. Tyrosine and typtophan, incidentally, are also not observed because of their inability to be oxidized at this potential. Both 3,4-dihydroxyphenethyl alcohol and 3,4-dihydroxyphenylglycol (neutral metabolites) elute under the solvent front. 3,4-Dihydroxyphenylaceticacid and 3,4-dihydroxymandelic acid, both acidic metabolites, elute under the solvent front, also. Biogenic amines related to DA and 5-HT comprise another group examined. Norepinephrine (NE) is slightly separated from the solvent front. But because of the large tissue concentrations of ascorbic acid, NE is effectively buried and, unfortunately, not determinable. Epinine (N-methyldopamine) is completely separated (tR = 18.5 min) from the 5-HT peak ( t R = 9.3 min) and the procedure could be utilized diANALYTICAL CHEMISTRY, VOL. 49,
NO. 3,
MARCH 1977
357
Table VIII. Effect of Homogenization Method on Whole Mouse Brain Determination of Dopamine and Serotonina Method of homogenization
DA, ngk
5-HT, nglg
Ultrasonic Ground glass
1023 f 60 985 f 30
837 f 44 810 f 23
a All values reported as mean f SEM for five separate determinations. There is no significant difference between methods of homogenization for either DA or 5-HT. 5-HT results corrected for percent recovery in presence of tissue (see text).
Table IX. Dopamine and Serotonin Concentrations in Mouse Brain Cerebelluma DA, nglg
5-HT, nglg
19 f 2
240 i 14
a Results expressed as mean f SEM for six separate determinations. 5-HT results corrected for percent recovery in presence of tissue (see text).
rectly to determine this species in appropriate samples. Tryptamine is not detectable a t the potential chosen; but chromatograms run a t +0.94 V vs. SCE display a very broad peak for tryptamine at t R = 24.5 min. Epinephrine (E) elutes f t R = 5.2 min) almost directly between DHBA ( t R = 4.0 min) and DA ( t =~6.2 min), but is rather poorly resolved ( R = 0.77 and 0.67, respectively). However, E is normally present in brain tissue in only very small amounts(27). Even if present in equimolar concentrations with DA, E would contribute only ca. 6% to the measured peak height for the former. Thus, as long as E cannot be visually detected in the chromatography of other samples as a peak or noticeable shoulder, it will not present any major problems. Quantitative separation of DA and E can, if necessary, be accomplished by either diluting the citrate/acetate elution solvent or utilizing the column conditions of Refshauge (28). Methoxylated metabolites form the other major group examined. 3-Methoxy-4-hydroxyphenylalanine(3-0-methylDOPA) elutes with the solvent front as do the other amino acids mentioned previously. 5-Methoxytryptamine is not electrochemically active a t the potential chosen; however, it does exhibit a broad, poorly resolved peak ( t =~28.0) at a potential of +0.94 V vs. SCE. Metabolites which are observed under the given conditions include normetanephrine ( t =~ 4.0 min), 3-methoxy-4-hydroxyphenethylamine( t R = 46 min), 4-methoxy-3-hydroxyphenethylamine ( t =~ 68 rnin), and metanephrine ( t =~23 min). Of these, only normetanephrine, which is isographic with DHBA, is of concern. The other species are clearly separated from the components of interest. Also, their endogenous concentrations are generally so low as to allow ignoring them in routine determinations. Normetanephrine (NM), as the other methoxylated catecholamines, is only partially oxidized (ca. 15%)a t the potential chosen. Additionally, NM is present in brain tissue a t relatively low concentrations (65 f 4 ng/g, personal observation). Determinations in which DHBA was omitted showed no discernible NM peak at the sensitivity employed (less than 1%of the usual DHBA peak). Thus, N M should not normally cause any significant interference. Since both NM and E overlap the chosen internal standard, other species were investigated for possible use in samples 358
ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH i s 7 7
where either of these may pose problems. a-Methyldopamine ( t =~18.5min) was found to be quite adequate for this purpose. But, one should remember that this will increase the time required for individual determinations. Also, amethyldopamine is isographic with epinine. In short, there are no major known interferences in the procedure if the comments above are properly considered. We are presently (to be published) applying the technique to the analysis of such samples as serum and other body fluids and tissues. Method of Homogenization. Alliger (29) reported that long exposure of biological samples to ultrasonic irradiation caused a destruction of the indole nucleus of tryptophan. Since 5-HT also contains an indole moiety, we compared ultrasonic homogenization to that performed with a ground glass apparatus. As presented in Table VIII, there was no significant effect on the results. Mouse Brain Cerebellum Determination. T o give a feeling for the sensitivity of the LCEC technique in routine investigations, the determination of DA and 5-HT in mouse brain cerebellum was undertaken. This particular brain part was chosen because it is easily accessible for dissection and it also shows the smallest concentration for DA (to be published) of any of the usual parts examined. The combined results for a typical determination are presented in Table IX. The average tissue weight was 53 mg, so the actual amount of DA determined in each sample was ca. 1.0 ng (6.6 pmol), while that for 5-HT was 9.8 ng (51 pmol). Although we could not find any references to DA and 5-HT determinations in mouse brain cerebellum, the results do agree fairly well with those reported for rat brains (30,31). Routine Application. With the present single column setup, it has been possible to run as many as 30 samples during a normal eight-hour day. This required that the tissue samples were ready to be weighed a t the start of the day. However, it also included sufficient time to clean all appropriate glassware while the chromatography was being run. When sacrificing and tissue removal were included, 20 samples could be handled in the same time period. The number of samples per day, however, may be increased significantly by utilizing the dual parallel LCEC apparatus described by Blank (32);this costs only a moderate amount more than a single setup, but effectively halves the time required for each chromatogram. Sensitivity and Range of t h e LCEC. In the course of development, it was necessary to show the LCEC peaks to be linear over a reasonable range. The linearity, displayed as amount of injected compound, was exhibited from 0.1 pmol to 5 nmol for each of DHBA, DA, and 5-HT.
ACKNOWLEDGMENT The authors thank Dave Wassil for the preparation of a methyldopamine.
LITERATURE CITED (1) A. S.Welch and B. L. Welch, Anal. Biochem., 30, 161 (1969). (2) K. M. Taylor and R. Laverty, J. Neurochem., 16, 1367 (1969). (3) G. B. Ansell and M. F. Beeson, Anal. Biochem., 23, 169 (1968). (4) M. K. Shellenberger and J. H. Gordon, Anal. Biochem., 39, 356 (1971).
(5) R. P. Maickel, R. H. Cox, J. Saillant, and F. P. Miller, Int. J. Neuropharmacol., 7, 725 (1968). (6) J. W. Vanable, Jr., Anal. Biochem., 6, 393 (1963). (7) U. S.von Euler, Pharmacol. Rev., 11, 262 (1959). (8) D. F. Bogdanski, A. Pletscher, B.B. Brodie, and S. Udenfriend, J. Pharmacol. Exp. Ther., 117, 82 (1956). (9) E . L. Arnold and R. Ford, Anal. Chem., 45, 85 (1973). (IO) i. L. Martin and G. B. Ansell, Biochem. fharmacol., 22, 521 (1973). (1 1) S. H. Koslow, F. Cattabeni, and E. Costa, Science, 176, 177 (1972). (12) C. Refshauge, P. T. Kissinger, R. Dreiling, L. Blank, R. Freeman, andR. N. Adams, L i f e Sci., 14, 311-(1974). (13) P. A. Shore and J. S. Olin, J. Pharmacol. f x p . Ther., 122, 295 (1958). (14) A. C. Cuello, R. Hiley, and L. L. Iversen, J. Neurochem., 21, 1337 (1973). (15) J. M. Saavedra, M. Brownstein, and J. Axelrod, J. Pharmacol. f x p . Ther., 186, 508 (1973).
(16) A. H. Antone and D. F. Sayre, J. Pharmacol. Exp. Ther., 138, 360 (1966). (17) S. Udenfriend, D. F. Bogdanski, and D. F. Weissbach, Science, 122, 972 (1955). (18) P. T. Kissinger. C. Refshauge,R. Dreiling, and R. N. Adams, Anal. Left.,8, 465 (1973). (19) P. Albrecht, M. B. Visscher,J. J. Bittner, and F. Halberg, Proc. SOC.Exp. Biol. Med., 92, 702 (1956). (20) R. J. Borgman, M. R. Baylor, J. J. McPhillips,and R. E. Stitzel, J. Med. Chem, 17, 427 (1974). (21) Louis Levine, "Biology of the Gene", 2d ed., C. V. Mosby Co., St. Louis, Mo., 1973. (22) R. M. Flemina, W . G. Clark, G. D. Fenster, and J. E . Towne, Anal. Chem., 37, 292 (1963). (23) R. G. Wiegand and J. E. Perry, Biochem. Pharmacol., 7, 181 (1961). (24) C. B.Smith, J. Pharmacol. Exp. Ther., 142, 343 (1963). (25) H. C. Agrawai, S.N. Glisson, and W. A. Himwich, lnt. J. Neuropharmacol., 7, 97 (1968).
(26) S. T. Weintraub, W. B. Stavinoha,R. L. Pike, W. W. Morgan, A. T. Modak, S. H.Koslow, and L. Blank, Life Sci., 17, 1423 (1975). (27) R. D. Ciavanelio, R. E. Barchas, G. S. Byers, D. W. Stemmle, and J. D. Barchas, Nature (London),221, 363 (1969). (28) C. Refshauge, Ph.D. Thesis, University of Kansas, Lawrence, Kan., 1974. (29) H. Alliger, Am. Lab., No. 7, 75 (1975). (30) R. H.Cox and J. L. Perhach,Jr., J. Neurochem., 20, 1777 (1973). (31) R. P. Maickel in ':Methods of Neurochemistry", Voi. 2, R. Fried, Ed,, Marcel Dekker, New York, 1972. (32) C. L. Blank, J. Chromatogr., 117, 35 (1976).
RECEIVEDfor review September 7,1976. Accepted December lg7" This work was by grant number MH26866-01, awarded by NIMH/DHEW. 8 j
Separation of Five Major Alkaloids in Gum Opium and Quantitation of Morphine, Codeine, and Thebaine by lsocratic Reverse Phase High Performance Liquid Chromatography C. Y. WU" and J. J. Wittick Merck Chemical Manufacturing Division, Rahway, N.J. 07065
Two similar isocratic liquld chromatographic (LC) systems have been developed for the determination of morphine, codeine, and thebaine in gum opium. One Is for the quantitatlve determination of morphine alone and the other for the simultaneous quantltative determination of codelne and thebaine. Two 30-cm X 4.0-mm i.d., p-Bondapak C18/Porasll columns are used for both systems which differ only In the composition of the mobile phase used. The composltlonof the mobile phase used for the morphine determination is 0.1 M NaH2P04in 5 % CH3CN/H20 and for the codelne and thebaine determination Is 0.1 M NaH2P04In 25% CH3CN/H20. Both systems are very reproduclble and sultable for routlne anaiysls. The preclslon of the method for morphine is 0.8% relative standard deviation and for codelne and thebaine 1.3 and 3.3 % relatlve standard deviation, respectively.
In spite of the fact that opium has been associated with human activities for centuries, chemists are still searching for better analytical methods to measure the alkaloids content of this important pharmaceutical raw material. Progress was hampered primarily because there are more than twenty alkaloids present in gum opium and suitable techniques for separating this complex mixture were lacking. However, with the development of a variety of chromatographic techniques, these difficulties have been overcome. Tedious prior purification of a specific alkaloid before quantitation is no longer necessary and simultaneous quantitative analysis for several alkaloids in the complex mixture has become a reality. Steady progress in the use of various chromatographic techniques for the analysis of opium and related mixtures has been demonstrated in the literature (1-18). In 1973 Wu et al. (7), using a synthetic mixture of five of the major opium alkaloids, successfully demonstrated that the HPLC technique is potentially useful for such analysis. In the same year, Wittwer (16) reported a gradient HPLC system which he claimed to be suitable for the determination of opium alkaloids. However, only limited data were reported. Later, Beaseley et
al. (15)described a procedure using normal phase liquid-solid adsorption chromatography with gradient elution. The method is lengthy and, unlike reverse phase chromatography, separation by normal phase liquid-solid adsorption chromatography is highly susceptible to the water content both of the adsorbent and of the mobile phase (19), making the technique difficult to use for routine analysis in our opinion. Techniques for opium alkaloids analysis other than chromatography are also available. For details refer to the literature (20-25). The objective of this paper is to introduce a practical, specific, relatively simple and rapid method with the capability of yielding precise quantitation for morphine, codeine, and thebaine in gum opium. Two essentially identical isocratic liquid chromatographic systems are described, one for morphine and the other for codeine and thebaine. The chief difference in these two systems is in the composition of the mobile phase. A microparticulate, w-Bondapak C l g column was selected and used because it provided the high column efficiency which is necessary for the analysis of a coplex mixture such as gum opium. This column also provided excellent reproducibility over a long period of time, making it especially suitable for routine analysis.
EXPERIMENTAL Apparatus. A DuPont Liquid Chromatograph Model 830, a Schoeffel SF 770 Spectroflow Monitor equipped with a deuterium lamp, an Autolab Vidar 6300 Digital Integrator, and a HewlettPackard strip chart recorder Model 7123A were properly connected and used. Samples were introduced through an injection valve purchased from DuPont Instruments (Wilmington, Del., Catalogue No. 204590). Reagents a n d Solutions. Pure morphine sulfate pentahydrate, pure codeine phosphate, and pure thebaine standards were obtained within the Merck Chemical Manufacturing Division. Reagent grade calcium hydroxide powder, 85%phosphoric acid, glacial acetic acid, sodium chloride, sodium perchlorate, dimethyl sulfoxide, and methylene chloride were used. Acetonitrile (UV) "Distilled in Glass" was purchased from Burdick & Jackson, (Muskegon, Mich.). Presaturated calcium hydroxide solution was prepared fresh daily by ANALYTICAL CHEMISTRY, VOL. 49, NO. 3, MARCH 1977
359
