Automated precolumn fluorescence labelling by carbodiimide activation of N-acetylaspartate and N-acetylaspartylglutamate applied to an HPLC brain tissue analysis

ANALYTICAL

BIOCHEMISTRY

196,

350-355

(19%)

Automated Precolumn Fluorescence Labelling by Carbodiimide Activation of A/-Acetylaspartate and A/-Acetylaspartylglutamate Applied to an HPLC Brain Tissue Analysis Jakob

Korf,l

Lammy

Veenma-van

der Duin,

Kor Venema,

and Johannes

H. Wolf2

Department of Biological Psychiatry, Groningen University, the Netherlands

Received

January

28, 1991

An automated method is described to couple carboxyl-containing metabolites to the fluorophore 2aminoanthracene in aqueous solution (containing 75% methanol) in the presence of N,N-dicyclohexylcarbodiimide. The reaction was optimized for N-acetylaspartate (N-AC-ASP) and N-acetylaspartylglutamate (N-AcAsp-Glu). The reactions occurred within 5 min at room temperature in the presence of 0.5-2 mM HCl. At concentrations of electrolytes exceeding 10 mM the coupling reaction became suboptimal. Derivatization was performed in a commercial precolumn derivatization unit. Additional tubing was needed to provide the reagents prior to reversed-phase HPLC and fluorescence detection. The assay is linear over at least three orders of magnitude; as little as 1 pmol could reproducibly be assayed in 100 rg wet weight brain tissue extracted with a mixture of methanol and 4 mM HCl (9:1, v/v). N-AC-ASP and N-AC-Asp-Glu levels in several brain regions and spinal cord were similar to those so far reported. The compounds could not be detected in peripheral tissue. The advantages, prospects and limitations of the present approach over existing methods to estimate water-soluble carboxylic acids is discussed. 0 1991

Academic

Press,

Inc.

Several compounds of biological interest have poor optical properties and only after derivatization they can conveniently be determined in tissue extracts or body

’ To whom correspondence should be addressed at Department of Biological Psychiatry, AZG/RUG, P.O. Box 30.001,9700 RB Groningen the Netherlands. ’ Present address: Gist Brocades, R&D/CMA, Postbus 1,260O MA Delft, The Netherlands.

fluids. Combining automated precolumn derivatization with reversed-phase HPLC-a relatively simple and powerful technique to separate and isolate compounds -and optical detectors allows to determine a variety of compounds with high sensitivity and specificity. Previously we have reported automated precolumn derivatization techniques for primary amines (including monoamines, amino acids, peptides, and some drugs) using o-phthaldialdehyde or naphthalenedialdehyde as a label in aqueous solutions and carboxylic acids (including free fatty acids and some drugs) using 4-bromomethyl7-methoxycoumarin as a label in aprotic solvents (l-5). Here we describe a derivatization technique for carbony1 groups containing compounds in aqueous solutions. These compounds are too hydrophylic to concentrate

in aprotic solvents, so derivatization

with the

coumarin label (2-4) is not feasible. The present derivatization technique is based on the coupling of the highly fluorescent label 2-aminoanthracene (2AA)3 to N-acetylaspartate (N-AC-ASP) and N-acetylaspartylglutamate (N-AC-Asp-Glu) in the presence of a carbodiimide (here N,N-dicyclohexylcarbodiimide, DCC, e.g., Ref (6,7)). N-AC-ASP and N-Ac-AspGlu have been identified in nervous tissue more than 2 decades ago and are supposed to be a metabolite (N-AcAsp) and a neuroactive (N-AC-Asp-Glu) compound (& ll), respectively. Thus far their function in the brain and spinal cord have been investigated only occasionally, which may be attributed, at least in part, to the lack of a simple assay. For example, in a relatively laborious

3 Abbreviations fluid; N-AC-ASP, tate glutamate;

used: 2AA, P-aminoanthracene; N-acetylaspartate; N-AC-Asp-Glu, DCC, N,N-dicyclohexylcarbodiimide.

350 All

Copyright 0 1991 rights of reproduction

CSF, cerebrospinal N-acetylaspar-

0003-2697/91 $3.00 by Academic Press, Inc. in any form reserved.

FLUORESCENT

N-AC-Asp

procedure for the assay of N-AC-Asp-Glu (11) manual purification of the compounds has to precede an HPLC separation with UV detection. Recently, sensitive methods for determining N-AC-ASP in cerebrospinal fluid (CSF) and N-AC-Asp-Glu in brain slice superfusion fluid have been described utilizing gas chromatography and mass spectrometry (12,13). Here we present details of an automated derivatization technique based on carbodiimide activated coupling in aqueous solutions. This technique, in combination with reversed-phase HPLC, requires minima1 sample processing. Coupling of carbodiimide activated carboxylic acids to a label prior to liquid chromatography has already been described in nonaqueous solutions (15-20) but has, as far as we know, not been applied to biomedical assays. The usefulness of the present approach is illustrated by measuring levels of N-AC-ASP and N-AC-Asp-Glu in tissue of the central nervous system. MATERIALS

AND

METHODS

AND

N-AC-Asp-Glu

351

ASSAY

sample

pick up

/-

24 O mln

detection

FIG. 1. The time schedule of the assay. The sequence of the various steps are given in a circular way, visualizing simultaneous processes. By following the solid line the time course of the sample can be followed. In the inner circle the derivatization loop is indicated, which was reached by the sample via valve A. The inner circle i&he first column, which serves to separate the derivatives of N-AC-ASP and N-AC-Asp-Glu from excess of 2AA. The effluent was then applied to the analytical column and the fluorescence continuously monitored. In the mean time the precolumn is cleaned and prepared for the next sample.

Reagents and Solvents The label 2AA (practical grade, Sigma Chemical Co., St. Louis, MO) and the coupler DCC (20 mg/ml; Janssen Chemica, Beerse, Belgium, 99%) were dissolved in methanol (0.75 mg/ml). Quartz bidistilled water was used throughout. N-AC-ASP and N-AC-Asp-Glu were obtained from Sigma (St. Louis, MO) and standard solutions were made in methanol/4 mM HCl (l:l, v/v). Acetonitrile (Westburg, Leusden, the Netherlands), methanol (Merck, Darmstadt, FRG) and other chemicals used were of the purest analytical grade available. Two solvents were used: (A) methanol12 mM HCl in water (25:75, v/v) and (B) acetonitrile:2 InM HCl in water (90:10, v/v). Two mobile phases were used: the “regeneration solvent” consisting of 10% solvent A and 90% solvent B, and the “analytical solvent” consisting of 60% solvent A and 40% of solvent B. Equipment The sampler is a Promis model II (Spark, Emmen, the Netherlands), equipped with an automated precolumn derivatization unit with additional tubing for a second reagent. Elution gradients were delivered by two LKB pumps (Pharmacia, Bromma, Sweden), and controlled by an LKB 2152 LC-controller. The eluate of the HPLC was monitored with a Kontron spectrofluorometer (type SFM-23, Geneva, Switzerland), containing a Aow cell of 70 ~1 and two holographic concave grating monochromators. The excitation wavelength was set at 275 nm with a monochromator and the fluorescence light passed a Kodak Wratten filter 2A (transmission above 360 nm) before reaching a photomultiplier. The excitation light source was a 150-W high-pressure

xenon lamp. The eluted derivatives of N-AC-ASP and N-AC-Asp-Glu have a maximal excitation at 275 nm, whereas the emission was the highest at 420 nm; the spectra are identical with that of (unreacted) 2AA. Chromatograms were either recorded with an LKB 2210 two channel recorder or a LDC Milton Roy CI-10 integrator (Interscience, Breda, the Netherlands). The precolumn derivatization unit consists of a peristaltic pump for three tubes (Skalar, Breda, The Netherlands), two for transporting the reagents (silicone tubing), and one for the waste (solventflex). All other tubings were of Teflon. The mixture was led to a loop injector (50 ~1, Rheodyne 7010, Cotati, CA), which serves also as the reaction vessel. The samples were separated on two Spherosorb ODS-2 columns (5-pm particles; Chrompack, Middelburg, the Netherlands). Before passing the sample over the analytical column (length 20 cm, i.d. 3.0 mm) the sample was partially purified on another column (length 10 cm, i.d. 3.0 mm) to remove excess of reagent and reduce time of analysis. Principle of the Method The principle of the whole setup is visualized in Fig. 1. After introduction of the sample and reagents were transported to the reaction loop through valve A, using the peristaltic pump of the autosampler; a reaction time of about 5 min was routinely allowed. By switching valve A the derivatized sample is applied to the first column, which is then continuously eluted with the “analytical solvent.” Between 2 and 3 min derivatized N-AC-ASP and N-AC-Asp-Glu leave the first column and enters the analytical column. During passage of the N-AC-ASP and

352

KORF ET AL.

“analytical solvent.” are the same.

we-column

The valves A and B of Figs. 1 and 2

Tissue Extraction

sample

DCC

2AA

FIG. 2. * The components of the HPLC set-up are depicted in detail, showing the pumps delivering the “analytical” and the “regeneration” solvents to the two columns, precolumn derivatization unit with the peristaltic pump, the loop and the possible positions of the valves A and B. This figure follows the time schedule of Fig. 1. Further details are described in the text.

N-AC-Asp-Glu derivatives over the analytical column the first column is isocratically cleaned with the “regeneration solvent” by switching valve B at 3.50 min, and set back in the “sample position” by turning valve B again at 15 min to allow the entrance of the “analytical solvent”; after 24 min a new sample of derivatized material can be injected on a relatively clean first column. All chromatography was performed at ambient temperature. The Setup in Detail

The more precise setup of the system is shown in Fig. 2. In the lower part of the figure the derivatization unit is shown. The two reagents (DCC and 2AA), the excess of sample and reagents are pumped by the peristaltic (“roller”) pump of the autosampler (for about 40 s). The roller pump stops for about 5 min when the sample has entered the loop of valve A via A4-A3 after passing the tees. The amount that is sampled is the difference between the waste and the reagent line. The sample is degassed with N, during pickup. After the reaction has proceded the valve A changes position and the sample is now applied via A6-Al and B5-B6 to the first column (precolumn), which was already in the appropriate position (upper part of Fig. 2). The eluate of the precolumn is brought on the analytical column via B3-B4. After 3.50 min the position of valve B changes in such a way that the “regeneration solvent” enters the precolumn (via Bl-B6) and that the eluate of the precolumn, containing excess of reagents and other materials, goes to the waste loop (via B3-B2). At 15 min valve B is positioned so that the first column is conditioned with the

Rat brain and spinal tissue was obtained from male Wistar rats (200-300 g; Centraal Proefdieren Laboratorium, Groningen, the Netherlands), rapidly dissected into various regions and frozen on dry ice, as described previously (8). Peripheral rat tissue (heart, liver) was similarly treated. Tissue was dissected within 2 to 30 min after death. Frozen tissue was weighed and homogenized in methanol: 4 mM HC1(9:1, v/v) by a Potter-Elvehjem device at O”C, kept overnight at 4”C, and centrifuged for 2 min at 10,000 rpm (about 55OOg, Beckman Microfuge). Routinely about 100 mg of tissue per ml methanol: 4 mM HC1(9:1, v/v) was taken. After centrifugation 5 ~1 of the clear supernatant was diluted with 745 ~1 of the same methanol:HCl mixture in a sample vial and 750 ~14 mM HCl:methanol(9:1, v/v) was added to reach a final concentration of 4 mM HCl:methanol of 1:l (v/v). This solution was used for derivatization and HPLC. RESULTS

Reaction Conditions

Several aspects of the reaction conditions were investigated using 75 and 15 ng of N-AC-ASP or N-Ac-AspGlu, equivalent to 430 and 50 pmol in the sample loop, respectively. Results are shown in Figs. 3A to 3D. (i) The relationship between the concentration of DCC at a concentration of 0.75 mg 2AA/ml methanol and a reaction time of 2 min is shown in Fig. 3A. Maximal responses were obtained at a DCC concentration of 10 mg/ml. (ii) The time course of the reaction at the conditions chosen as mentioned under (i) appeared to be maximal between 2 and 10 min (Fig. 3B). Thereafter a gradual decline of the fluorescence was seen, possibly due to the formation of other fluorescent reaction products. We observed not only that the peaks eluted at about 6 min of the N-AC-Asp-Glu derivative and at 8 min of the fluorescent N-AC-ASP derivative decreased, but that other peaks with longer retention times appeared or increased. (iii) The optimal amount of HCl to be added to the reaction mixture for the formation of the derivatives with short retention times was about 1 mM HCl (Fig. 3C). This pH optimum suggests that of the N-AC-ASP and N-AC-Asp-Glu molecules the ,& and y-carboxylic groups are protonized and that these groups are primarily labeled with 2AA. At longer reaction times at least two carboxylic groups (in N-AC-Asp-Glu) may become labeled, whereas at still longer times the a-carbonyl groups may become also derivatized.

FLUORESCENT

N-AC-ASP

AND

N-AC-Asp-Glu

I

‘0 mg DCClml

353

ASSAY

I

I

I

I

I

10

20

30 reaction

40

50

J

60

time (min)

I 1500 t

OO% HP

mM HCI

FIG. 3. A variety of reaction conditions were tested on the magnitude of the peaks in the HPLC chromatograms. A shows the dependence of the reaction on the concentration of DCC at a reaction time of 5 min; B, the time course of the reaction; C, the concentration of HCl added to the sample before pickup; D, the dependency upon the concentration of water with or without saline (artificial cerebrospinal Auid) or human lumbar cerebrospinal fluid: maximal derivatizations were obtained between 0 to 2.5% saline or CSF added to standards of N-AC-ASP or N-AC-Asp-Glu added to methanol with 25% water. (-) N-AC-Asp-Glu; (---1 N-AC-ASP. In all cases similar optimal conditions for both N-AC-ASP and N-AC-Asp-Glu were found.

(iv) The interference of electrolytes or CSF in the derivatization reaction is given in Fig. 3D. It appears, that the amount of pure H,O in the loop does not interfere with the reaction, but that addition of increasing amounts of real CSF or artificial CSF rapidly disturbs derivatization. The results indicate that in the injection loop maximally 100 gg wet weight tissue per 50 ~1 can be handled, corresponding to about 100 mg tissue/ml extraction solvent. Chromatograms,

Sensitivity,

and Tissue Analysis

Linearity of the assay and Linearity and recovery. the recovery of standards added to the tissue extracts were determined. The assays of N-AC-ASP and N-AcAsp-Glu are linear from 1 pmol up to at least 1 nmol per injection (Fig. 4). The amount of tissue was linearly related to peak height when less than 100 pg of tissue was present in the injection/reaction loop. Moreover standards of N-AC-ASP or N-AC-Asp-Glu, added to this or a smaller amount of tissue, was completely recovered. Accordingly more than 95% of N-AC-ASP and N-Ac-AspGlu added to 25 mg of brain tissue (internal standard) was recovered. Under optimal reaction conditions Chromatograms. the chromatograms shown in Fig. 5 were made: from left to right, blank, standards in water (80 pmol of N-AcAsp and 50 pmol N-AC-Asp-Glu), extract of 20 mg of the rat pons, and two chromatograms of the same tissue,

but with standard addition of 3 and 5 nmol of N-AcAsp-Glu and N-AC-ASP, respectively, to a homogenate of 1 mg tissue in 10 ~1 of the methanol/water extraction medium. It appeared that the N-AC-Asp-Glu derivative exhibited in most elution solvents a double peak, found with standards, tissue, and tissue with standards added. Brain tissue. Tissue was dissected within 2 to 30 min after death. In a few cases we determined how critical postmortem delay is for brain and spinal cord N-AC-ASP and N-AC-Asp-Glu levels. No alterations were seen when the cadavers were kept at room temperature from

pm01 FIG. 4. height.

Linearity

of the standard

concentration

curves

versus

peak

KORF

ET

AL.

DISCUSSION

L I I 1, 062

FIG. 6. From left to pmol N-AC-Asp-Glu tissue of the rat pons; curve l+); pons tissue

0’

+2

right: chromatograms of blank; standards of 50 (curves 1) and 85 pmol N-AC-ASP (curves 2); pons tissue + added N-AC-Asp-Glu (50 pmol, + added N-AC-ASP (85 pmol, curve 2f).

2 min up to 6 h. Therefore variations in postmortem delay, necessary because of the duration of the dissection, were not taken into account. The distribution of N-AC-ASP and N-AC-Asp-Glu in 12 areas of the brain and spinal cord of 2 to 6 rats were determined and compared to the reported N-Ac-Asp-Glu values published in Ref. (5). The distribution of N-AC-Asp is rather uniform throughout the central nervous system: the mean levels (a range, 2 to 6 determinations) in, e.g., the hippocampus, cerebellum, pons, and striatum were in our study 5.3 f 0.8,6.3 f 0.3,7.9 + 0.7, and 6.8 f 0.5 nmol/mg wet weight tissue, respectively, as compared to the levels reported by others 54 & 3, 42 k 4, 35 -t 4, and 43 t 4 nmol/mg protein (e.g., Ref. (9)). There were marked regional differences in the levels of N-Ac-Asp-Glu: for instance the levels in the spinal cord are about lo-fold to that in the forebrain. In Fig. 6 the present levels and those of Ref. 9 are shown. The linear correlation coefficient was 0.99 (N = 7), despite the different way of expressing tissue levels: pmol/g wet weight tissue in the present study versus nmol/mg protein. The present method was also applied to human lumbar CSF. Fresh human CSF (obtained from the Department of Neurology, courtesy of Dr. A. W. Teelken) was centrifuged, diluted with the above methanol:HCl mixture (l:lO, v/v) and assayed. Several other protocols for CSF were applied, most of them resulted in high recoveries of added N-AC-ASP or N-AC-Asp-Glu (serving as internal standards). None of these protocols, however, exhibited a peak in the chromatogram at the place of either N-AC-ASP or N-AC-Asp-Glu.

New aspects in biomedical applications of HPLC analysis as presented in the present report include: first, 2AA as label for fluorometric determinations; second, a carbodiimide (here DCC) as coupler for the direct labeling of N-acetylated compounds in (nearly) aqueous solutions; and third, the automation of this reaction. We have chosen 2AA for labeling, because it is in the series of polybenzenes the smallest and most efficiently fluorescing amine, so that the chromatographic properties of the derivatized acids are determined as strong as possible by the analytes and not by the label. Moreover the label is suitable for chemiluminescence detection. Disadvantages of the label are its relative lipophylicityreducing the maximum concentration of the label in the derivatization procedure-and, in contrast to such labels as o-phthaldialdehyde, naphthalenedialdehyde, or bromomethylmethoxycoumarin, the high fluorescence of unreacted 2AA. In assays where lipophilicity may become a problem, more hydrophylic derivatives, such as sulfonated aminoanthracenes, may be considered. The second disadvantage is the presence of excess unreacted label. To obtain acceptable short analysis times, two columns were needed: the first column to separate the N-AC-ASP and N-AC-Asp-Glu derivatives from the unreacted 2AA and contaminants, which was-because of the short retention times-rather easy, and the second for the separation of the analytes. Although in this way clean chromatograms were obtained, still relatively long analysis times were necessary. The sensitivity of the present assay compares favourably with the HPLC method of Koller and co-workers

N-AC-Asp-Glu

4zo.5/// 00 -

V Oo

/

I

5

I

I

I

IO

1

15

20

25

nmol/mg

protein

FIG. 6. Correlation between N-AC-Asp-Glu levels in various brain regions of the rat determined with the present method and those reported by Blakely and Coyle (Ref. (9)). Note that the levels are expressed differently, in the present study as Fmol/g wet weight, in the reference study as nmol/mg protein. Linear correlation coefficient, 0.99.

FLUORESCENT

N-AC-ASP

(11) by a factor of 100. Routinely we need less than 100 or 10 pg of fresh brain tissue for the assay of N-Ac-AspGlu and N-AC-ASP, respectively, and only 10 pg spinal cord tissue per injection. In addition our assay requires minimal sample pretreatment. Nevertheless these differences, the levels of N-AC-ASP and N-AC-Asp-Glu in brain and spinal tissue are rather similar. The present method is somewhat less sensitive than the recently published gas chromatography mass/spectrometric methods for N-AC-ASP and N-AC-Asp-Glu (12,13). Considering the cost and the relative ease, the present assay is a useful alternative to the latter assays. The double peak for N-AC-Asp-Glu was seen in standards, tissue extracts and N-AC-Asp-Glu added to brain or blank tissue. The ratio of the two adjacent peaks changed hardly by a change in the-relatively short-reaction time, pH, or temperature. No such double peaks were seen with N-AC-ASP, ruling out the possibility that the labeling procedure itself caused the double peak, because of, e.g., impurities of 2AA, or that the a-carboxylic group is or is not consistently labeled. The peaks of NAC-Asp-Glu disappear at far longer reaction times (between 30 to 180 min; data not shown). Moreover, the relative peak height of N-AC-ASP is about twice that of a single peak of N-AC-Asp-Glu, and near the sum of the two N-AC-Asp-Glu peaks. These observations together favor the idea, that the o-carboxylic group of either aspartate or glutamate of N-AC-Asp-Glu can be labeled, but that under the present conditions and there is no consistent simultaneous labeling of both the w-carboxylic groups of N-AC-Asp-Glu. The present automated procedure for derivatization was possible due to a slight adaptation of the Promis sampler. Our approach may also be useful for other carboxylic acids and for other labels, so in addition to fluerescence, electrochemical techniques, and chemoluminescence can be considered for detection.

AND

N-AC-Asp-Glu

355

ASSAY

ACKNOWLEDGMENTS This study was supported by the Netherlands Foundation of Technical Sciences (STW) and the Dutch Organization for Pure and Applied Medical Sciences. The help of Dr. D. Jaarsma and Mr. F. Postema, who dissected the brains, is greatly acknowledged. Mrs. Wieke van der Meer edited the final version of the manuscript.

REFERENCES K., Leever, 1. Venema, (1983) J. ChFOTTUltOgF.

W., Bakker,

2. Wolf, J. H., and Korf, 3. Wolf, J. H., Veenma-van matogr.

487,

J. O., Haayer,

J. (1988) J. Chromatogr. der Duin, L., and Korf,

matogr. 533, 171. Khorana, H. G. (1953)

New

Chem.

Rev. 53,145.

Organic

Reactions,

N. F. (1962)

436, 437. J. (1989) J. Chro502,

423.

J. (1990)

Vol.

J. Chro-

12, p. 205, Wiley,

York.

8. Curatolo, 9.

J.

496.

4. Wolf, J. H., and Korf, J. (1990) J. Chromatogr. 5. Koning, H., Wolf, J. H., Venema, K., and Korf, 6. 7. Albertson,

G., and Korf,

260,371.

A., D’Arcangello, Neurochem. 12,339. Blakely, B. D., and Coyle,

10. D’Arcangelo, 11. Koller,

P. Lino, J. T. (1988)

P., and Brancati,

K. J., Zaczek,

A., and Brancati,

A. (1965)

Int. Rev. Neurobiol.

A. (1990)

R., and Coyle,

Neurosci.

30,39.

114,82.

Lett.

J. T. (1984)

J.

J. Neurochem.

43,1136. 12. Swahn,

C.-G.

13. Zollinger,

M.,

(1990) Amsler,

J. New-o&em.

54,

1584.

U., and Brauchli,

J. (1990)

J. Chromatog.

532.27. 15. Gorog, 417.

S., Herenyi,

16. Kobayashi, M., and 122. 17. Kasal, Y., Tanimura,

B., and Low, Ichishima,

M.

(1988)

E. (1990)

T. T., and Tamura,

J. Chromatogr. And.

353,

Biochem.

Z. (1975)

Anal.

189, Chem.

47,34. 18. Goto, J., Goto, N., Hikichi, A., Nishimaki, (1980) Anal. Chim. Actu 120, 187. 19. Goto,

J., Goto,

20. Lingeman,

N., and Nambara

(1982)

H., Hulshoff, A., Underberg, F. B. J. M. (1984) J. Chromatogr. 290,

T., and Nambara, J. ChrOmatOgr.

239,559.

W. J. W., and Offerman, 215.

T.