Optimization of the Reaction Conditions for the Peroxyoxalate Chemiluminescence Detection System of Fluorescent Compounds in Ahigh-Performance Liquid Chromatography

ANALYTICAL SCIENCES OCTOBER 1996, VOL. 12 713

Optimization of the Reaction Conditions for the Peroxyoxalate Chemiluminescence Detection System of Fluorescent Compounds in aHigh-Performance Liquid Chromatography

Ryoya GoHDA, Kohsuke KIMOTO, Tomofumi SANTA, Takeshi FUKUSHIMA, Hiroshi HoMMA and Kazuhiro IMAIt

Faculty of Pharmaceutical Sciences, University of Tokyo, Hongo, Tokyo 113, Japan

Using an instrument of automatic fluid injection (batch method), we optimized such reaction conditions as pH, concentration and kinds of buffer for the peroxyoxalate chemiluminscence (PO-CL) detection of fluorescent compounds for high-performance liquid chromatography (HPLC). The results were compared with those obtained by HPLC. Consequently, the usefulness of the batch method for investigating the reaction conditions for PO-CL detection in HPLC was demonstrated. With the optimized condition using a phosphate buffer for the eluent in HPLC, the detection limits for Dns-Val and dipyridamole were 70 amol and 20 amol on column (S/N=2), respectively.

Keywords Peroxyoxalate chemiluminescence, batch method, high-performance liquid chromatography, bis[4-nitro-2- (3,6,9-t,rioxadecyloxycarbonyl)phenyl]oxalate, dipyridamole, dansylated valine

Since introducing the peroxyoxalate chemilumines- extrapolation from zero time to the first part of the decay cence (PO-CL) reaction to the detection system for high- curve based on the obtained results. In order to solve performance liquid chromatography (HPLC)', this this problem, we tried to employ an instrument of method has been widely used for the sensitive determi- automatic fluid injection (batch method) in order to nation of a number of compounds, such as catechol- optimize the reaction conditions for the PO-CL detection amines2-4, steroids and drugs.5 system in HPLC. Although several investigators have studied the In the present work, we first investigated the effects of mechanism of the PO-CL reaction, the exact mechanism the pH and buffer concentration on the chemilumines- of this reaction has not yet been clarified because of its cence intensity of a fluorescent drug, dipyridamole (an complexity. Rauhut and his co-workers6 have reported anti-platelet aggregation drug), and a dansylated valine that the key intermediate formed from oxalate ester and (Dns-Val) by both the batch method and HPLC. hydrogen peroxide is 1,2-dioxetane-3,4-dione. On the Comparing these results, we found that the batch method other hand, kinetic studies by Catherall et al.' have is useful to optimize the reaction conditions for the PO- indicated the occurrence of other key intermediates CL detection system in HPLC. Finally, we developed a during the CL reaction. Givens et al.8 suggested that the highly sensitive HPLC-PO-CL detection system by hydroperoxy oxalate ester was the likely reactive optimizing the reaction conditions for the PO-CL intermediate based on an investigation with bis(2,6- detection system determined by the batch method. difluorophenyl)oxalate (DFPO) using 19F-NMR. It was reported that the PO-CL reaction was affected by several factors, such as the pH, water content, kinds of Experimental oxalates, salts, temperature and organic solvents9-14,and especially by a catalyst, such as imidazole. Although an Chemicals optimization of the reaction conditions for the PO-CL Bis[4-nitro-2-(3,6,9-trioxadecyloxycarbonyl)phenyl]- detection system in HPLC is most desirable for the oxalate (TDPO), hydrogen peroxide (30%), nitric acid sensitive detection of fluorescent compounds, it is (61%) and boric acid were purchased from Wako Pure complicated and troublesome for these reasons. Chemical Industries Ltd. (Osaka, Japan). Dns-L-Val Although Imai et a1.15predicted the detection ranges for and dipyridamole were from Sigma (St. Louis, MO, the HPLC-PO-CL detection system of fluorescent USA). Imidazole was from Merck (Darmstadt, compounds using a manual method, the CL reaction Germany). Tris(hydroxymethyl)aminomethane, 3-(N curves required many points to be measured as well as an morpholino)propanesulfonic acid (MOPS) and 2- [4-(2-hydroxyethyl)-1-piperazinyl]ethanesulfonic acid (HEPES) were from Nacalai Tesque (Kyoto, Japan). 714 ANALYTICAL SCIENCES OCTOBER 1996, VOL. 12

Fig. 1 Experimental apparatus of the batch method. L, Fig. 2 Flow diagram for the HPLC-PO-CL detection system. luminescencer (Luminescencer AB2000 (ATTO, Tokyo)); E, eluent (buffer/CH3CN (60/40, v/v)); CL reagent, TS, test solution (500 nM Dns-Val or dipyridamole in chemiluminogenic reagent (TDPO and H2O2 in CH3CN); buffer/CH3CN); CL reagent, chemiluminogenic reagent P1, pump for the eluent; P2, pump for the CL reagent; I, (0.25 mM TDPO and 25 mM H2O2 in CH3CN (60/40, v/v)); injector; C, analytical column (TSK-gel ODS-80TM (4.6 PMT, photomultiplier; P, pump; DP, data processor. mm i.d.X 150 mm)); CO, column oven (40°C); M, mixing device; D, chemiluminescence detector; DP, data processor.

Acetonitrile (HPLC grade), sodium dihydrogenphos- phate 2-hydrate and disodium hydrogenphosphate 12- hydrate were from Kanto Chemical (Tokyo, Japan). Water was purified by a Milli-Q reagent system (Millipore, Bedford, MA, USA).

Preparation of buffers The pHs of imidazole and Tris buffer were adjusted with HNO3, which did not quench the PO-CL reaction.16 The pHs of borate, MOPS and HEPES buffer were adjusted with NaOH. The phosphate buffers were obtained by mixing NaH2PO4 and Na2HPO4 buffer. Fig. 3 Chemiluminescence profile obtained by mixing the Batch method test solution and the CL reagent solution in batch method. A scheme of the experimental apparatus of the Test solution: 500 nM Dns-Val in 25 mM imidazole buffer batch method is shown in Fig. 1. The chemilumines- (pH 6.0)/CH3CN (60/40, v/v); CL reagent: 0.25 mM TDPO cence was measured by a Luminescencer AB2000 and 25 mM H2O2 in CH3CN. (ATTO, Tokyo, Japan). The obtained data were processed by a Macintosh LC 630 (Apple Computer, Inc., CA, USA). The peroxyoxalate chemiluminescence was produced by mixing the test solution (500 nM output was recorded by a Chromatocorder 12 (SIC, dipyridamole or Dns-Val in buffer/CH3CN (6/4, v/v)) Tokyo, Japan) (DP). and the chemiluminogenic reagent (CL reagent) (TDPO and H2O2 in acetonitrile) on a poly(tetrafluoroethylene) Preparation of standard solutions well. The test solution had been previously put into the Stock standard solutions of 10 nM dipyridamole and well; the CL reagent was pumped later. 100 nM Dns-Val were respectively prepared by dis- solving a few mg of dipyridamole and Dns-Val in water/ High-performance liquid chromatography acetonitrile (6/4, v/v) and successively diluting in A schematic flow diagram of HPLC is shown in Fig. 2. the same solution. A 10 µl (100 fmol and 1 pmol, The HPLC system consisted of a PU-880 pump (JASCO, respectively) aliquot of the mixture was subjected to Tokyo, Japan) for the eluent (P1), a PU-980 pump HPLC. (JASCO) for the CL reagent (P2), a 20 µl Rheodyne 7125 injector (Cotani, CA, USA) (I), a TSKgel ODS-80TM (4.6 mm i.d.X150 mm, 5 µm) (Tosoh Co., Tokyo, Japan) Results and Discussion (C) and a 825-CL Intelligent CL Detector (JASCO) (D), the cell of which comprised a wound poly(tetrafluoroeth- Effectof thepHand buffer concentration on the chemilumines- ylene)tube. The analytical column and mixing device cence intensity in the batch method (M) were set in a 860-CO column oven (JASCO) (CO) The effects of two parameters, pH and the buffer maintained at 40° C. The eluent was buffer/ acetonitrile concentration, on the chemiluminescence intensity of (60/40, v/v) at a flow rate of 1.0 ml/min. The CL dipyridamole and Dns-Val were first examined by the reagent was TDPO and H2O2in acetonitrile. The signal batch method. In this study, the pH was varied from 4.0 ANALYTICAL SCIENCES OCTOBER 1996, VOL. 12 715

Fig. 4 Effect of pH and imidazole concentration on the signal (S) values (A) and the signal/noise (S/N) values (B) by the batch method and the peak area (C) and the peak area/noise (D) by HPLC. [Batch Method] Test solution: 500 nM Dns-Val in imidazole buffer/CH3CN (60/40, v/v); CL reagent: 0.25 mM TDPO and 25 mM H2O2 in CH3CN. [HPLC] Sample: Dns-Val (1 pmol); eluent: imidazole buffer/CH3CN (60/40, v/v); CL reagent: 0.25 mM TDPO and 25 mM H2O2 in CH3CN; flow rate: eluent (1.0 ml/min), CL reagent (1.3 ml/min).

Table 1 Conditions which gave the maximal S and S/ N values in various buffers by the batch method

to 8.0, and the concentration of buffer was varied from (0.25 mM TDPO and 25 mM H202 in acetonitrile, 5 mM to 100 mM. The buffers used were imidazole, 250 µl), a typical chemiluminescence profile was obtained borate, Tris, MOPS, HEPES and phosphate. Using the (Fig. 3). The N value means an integral value of blank test solution (500 nM dipyridamole or Dns-Val in buffer/ emission for 3 s, from 1 to 4 s after mixing the test acetonitrile (6/4, v/v), 192 µl) and the CL reagent solution and the CL reagent. An integral value of the 716 ANALYTICAL.SCIENCES OCTOBER 1996, VOL. 12 chemiluminescence of Dns-Val for the same 3 s minus the We then investigated the concentrations of oxalate and N value is the S value. H2O2, the ratio between the test solution and the CL Figures 4A) and 4B) showed the effect of the pH and reagent and the cell volume by the batch method when a imidazole concentration on the S value and the S/N phosphate buffer was used for the test solution. So far, value of Dns-Val by the batch method. A slight the CL reagent conditions (0.25 mM TDPO and 25 mM difference was observed in the two figures. This H2O2 in acetonitrile) were optimum for the imidazole difference was attributed to the different effect of the pH buffer. and the concentration of the buffer on the S and Nvalues. Thus, the conditions which gave the maximal S value Development of a highly sensitive HPLC PO-CL detection did not always correspond to the conditions which gave system using a phosphate buffer as an eluent by the batch the maximal S/N value in the same buffer. Similar method phenomena were observed in all buffers. The condi- Concentration of TDPO and H2O2. It was reported tions which gave the maximal S and S/ N values in that the concentrations of oxalate and H2O2 affect the various buffers by the batch method using 0.25 mM chemiluminescence intensity.13,1' In the case of TCPO, TDPO and 25 mM H2O2in acetonitrile as the CL reagent when the concentration of TCPO was below 10 mM, the are depicted in Table 1. Among the six buffers inves- peak height was proportional to the TCPO concentra- tigated, the maximal S value was obtained using the tion.13 We first investigated the optimal concentration imidazole buffer, and the maximal S/N value was of oxalate and H2O2 for Dns-Val detection by the batch obtained using the phosphate buffer. method. We fixed the molar ratio at between TDPO and H2O2 of 1 to 100; the concentrations ofTDPO and H202 Effectof thepHand buffer concentrationon the chemilumines- were varied from 0.50 mM and 50 mM to 2.0 mM and cence intensity in HPLC 200 mM. Also, the pH and concentration of phosphate The effects of the pH and concentration of buffer on buffer were examined. The results are shown in Fig. 5. the peak area of Dns-Val and dipyridamole were also The S values gradually increased along with an increase examined by HPLC to compare with the results of the in the concentrations of TDPO and H2O2;however, when batch method. The reaction conditions for the PO-CL the concentrations of TDPO and H2O2 were 2.0 mM and detection in HPLC were set up as follows. The eluent 200 mM, the maximal S value decreased. It was con- was buffer/acetonitrile (60/40, v/v) at a flow rate of sidered that high concentrations of TDPO and H2O2 1.0 ml/min and the CL reagent was 0.25 mM TDPO and caused a promotion of the PO-CL reaction; as a result, an 25 mM H2O2 in acetonitrile at a flow rate of 1.3 ml/min. increase in initial intensity and a decrease in the lifetime In order to correspond with the integral time on the batch resulted. In all conditions, the maximal S value was method, we set up the detection time of chemilumines- obtained when the concentrations of TDPO and H2O2 cence to be from 1 to 4 s after mixing the test solution and were 1.0 mM and 100 mM for the CL reagent using 5 mM the CL reagent by changing cell volume (about 130 µl) phosphate buffer (pH 5.0) for the eluent. and the length from the mixing device to the inlet of the Concentration of H2O2. It was reported that the peak cell. height increased with increasing H2O2 concentration; Figures 4C) and 4D) show the effect of the pH and however, when the concentration was higher than about imidazole concentration on the peak area and peak 1 mM, the signal intensity remained constant for the area/noise of Dns-Val by HPLC. Here, the noise was TCPO system.`' We investigated the suitable concen- measured as the width of the base line in HPLC. The tration of H2O2 to produce the maximal S value of Dns- conditions which gave the maximal peak area corre- Val. The concentration of TDPO was fixed 1.0 mM. sponded to the conditions which gave the maximal peak The concentration of H2O2 was varied from 50 mM to area/noise. In other buffers, we observed the same 400 mM. Also, the effects of the pH and the phosphate relationship between the peak area and the peak area/ buffer concentration were examined. Although the S noise; the results for dipyridamole were similar in shape value gradually increased along with an increase in the to those of Dns-Val in six buffers. concentration of H2O2, when the concentration of H2O2 was above 200 mM, the maximal S value decreased. It Comparison between the optimal conditions obtained by the was also considered that a high concentration of H2O2 batch method and HPLC promoted the PO-CL reaction because of an increase in We also compared the results of the batch method with the water content. The water content is known to those of HPLC (Fig. 4). As a result, the conditions change the velocity of PO-CL reactions.18 Under all which gave the maximal S value corresponded to the conditions, the maximal S value was obtained when the conditions which gave the maximal peak area. As concentration of TDPO and H2O2 was 1.0 mM and mentioned above, the conditions which gave the maximal 100 mM for the CL reagent using a 5 mM phosphate peak area corresponded to the conditions which gave the buffer (pH 5.0) for the eluent. maximal peak area/ noise. Consequently, we could Ratio of the eluent and CL reagent solution. The easily find the optimum reaction condition for the PO- changes in the flow rates of the eluent and CL reagent CL detection system in HPLC by finding the conditions solution in HPLC can vary many of the controlling which gave the maximal S value on the batch method. parameters of the final solution: pH, water content, ANALYTICAL SCIENCES OCTOBER 1996, VOL. 12 717

Fig. 5 Effect of pH and phosphate concentration on the signal value (8) by the batch method. Test solution: 500 nM Dns-Val in phosphate buffer/CH3CN (60/40, v/v); CL reagent: A) 0.25 mM, 25 mM, B) 0.50 mM, 50 mM, C) 1.0 mM, 100 mM, D) 2.0 mM, 200 mM (TDPO, H2O2) in CH3CN.

concentration of TDPO and H2O2. In addition to these changed instead of changing the cell volume in HPLC. parameters, the sample concentrations are dependent on The results are shown in Fig. 6. On the other hand, the flow rates of the eluent and the CL reagent, and affect the cell volume in HPLC was actually varied from 0 to 130 µl peak heights,12,17,19,2° We thus investigated the ratio by adjusting the winding of the poly(tetrafluoroethylene) between the test solution and the CL reagent by the batch tube so as to correspond with the integral time of the method instead of the investigating the flow rate in batch method (Fig. 7). Compared with these results, it HPLC. The concentration of the CL reagent was was ascertained that we could investigate the optimal cell 1.0 mM TDPO and 100 mM H2O2. We fixed the volume of the detector in HPLC by the batch method. volume of the test solution at 192 µl and varied the Highly sensitive detection of Dns- Val and d4 yridamole by volume of the CL reagent from 100 to 250 µl. With HPLC-PO-CL detection system. Through these inves- larger the volume of the CL reagent, the S values of Dns- tigation by the batch method, the conditions for Dns-Val Val increased. Under all conditions, the maximal S in HPLC were chosen as follows. The eluent was a value was obtained when the volume of the CL reagent 5 mM phosphate buffer (pH 5.0)/acetonitrile (60/40, was 250 µl using a 5 mM phosphate buffer (pH 5.0). v/v) at a flow rate of 1.0 ml/min. The CL reagent was Cell volume. The cell volume was reported to affect 1.0 mM TDPO and 100 mM H202 in acetonitrile the peak heights and widths in HPLC.1012 In addition, at a flow rate of 1.3 ml/min. Dns-Val (1 fmol) was zero-dead-volume PO-CL detection in liquid chromatog- separated on the TSK gel ODS-80TM column and raphy was also reported) In this work, we investigated detected under the optimum conditions (obtained as the influence of the cell volume on the peak area by the described above). The detection limit (S/N=2) for batch method. In this method, the integral time was Dns-Val was 70 amol (Fig. 8A)), which was superior to 718 ANALYTICAL SCIENCES OCTOBER 1996, VOL. 12

Fig. 6 Effect of the detection time on the signal (S) and the noise. Marks: (0), signal; (S), noise. Test solution: 500 nM Dns-Val in 5 mM phosphate buffer (pH 5.0)/ CH3CN (60/40, v/v); CL reagent: 1.0 mM TDPO, 100 mM H2O2 in CH3CN. Fig. 8 Chromatograms of Dns-Val (A) and dipyridamole (B). Eluent: 5 mM phosphate buffer (pH 5.0)/CH3CN (60/40, v/ v); CL reagent: (A) 1.0 mM TDPO, 100 mM H2O2 in CH3CN, (B) 0.50 mM TDPO, 50 mM H2O2 in CH3CN; flow rate: eluent (1.0 ml/min), CL reagent (1.3 ml/min).

which was comparable to the value (10 amol) obtained using imidazole buffer in a previous study.21

In conclusion, the usefulness of the batch method for optimizing the PO-CL detection system was demonstrated. A further study on its application for other fluorescent compounds is in progress.

We are grateful to ATTO Co., Ltd. (Tokyo, Japan) for the use of the Luminescencer AB2000.

Fig. 7 Effect of the cell volume on the peak area and the noise Marks: (0), peak area; (•), noise. Sample: Dns-Val References (1 pmol); eluent: 5 mM phosphate buffer (pH 5.0)/CH3CN (60/40, v/v); CL reagent: 1.0 mM TDPO,100 mM H2O2 in CH3CN; flow rate: eluent (1.0 ml/min), CL reagent (1.3 ml/ 1. S. Kobayashi and K. Imai, Anal. Chem., 52, 424 (1980). min). 2. S. Higashidate and K. Imai, Analyst [London], 17, 1863 (1992). 3. P. R. Prados, S. Higashidate and K. Imai, Biomed. Chromatogr., 8, 1(1994). that obtained using an imidazole buffer in a previous 4. K. Imai, S. Higashidate, P. R. Prados, T. Santa, S. study (sub-fmol).21 Similarly, the conditions for Adachi-Akabane and T. Nagao, Biol. Pharm. Bull., 17, 907 dipyridamole in HPLC were obtained by the batch (1994). method. The eluent was 5 mM phosphate buffer 5. P. J. M. Kwakman and U. A. Th. Brinkman, Anal. Chim. Acta, 266, 175 (1992). (pH 5.0)/acetonitrile (60/40, v/v) at a flow rate of 6. M. M. Rauhut, D. Sheehan, R. A. Clarke and A. M. 1.0 ml/ min. The CL reagent was 0.50 mM TDPO and Semsel, Photochem. Photobiol., 4,1097 (1965). 50 mM H2O2 in acetonitrile at a flow rate of 1.3 ml/min. 7. C. L. R. Catherall, T. F. Palmer and R. B. Cundal, J. Dppyridamole (100 amol) was separated on the TSK gel Chem. Soc. Faraday Trans 2, 80, 823 (1984). ODS-80TM column and detected under the optimum 8. H. P. Chokshi, M. Barbush, R. G. Carlson, R. S. Givens, conditions obtained, as described above. The detection T. Kuwana and R. L. Schowen, Biomed. Chromatogr., 4, limit (S/N=2) for dipyridamole was 20 amol (Fig. 8B)), 96 (1990). ANALYTICAL SCIENCES OCTOBER 1996, VOL. 12 719

9. S. Kobayashi, J. Sekino, K. Honda and K. Imai, Anal. 16. K. Honda, J. Sekino and K. Imai, Anal. Chem., 55, 940 Biochem., 112, 99 (1981). (1983). 10. G. J. de Jong, N. Lammers, F. J. Spruit, U. A. Th. 17. S. Uzu, K. Imai, K. Nakashima and S. Akiyama, Analyst Birkman and R. W. Frei, Chromatographia, 18, 129 [London], 116, 1353 (1991). (1984). 18. J. E. Dubois, M. Carabedian and B. C. R. Ancian, Acad. 11. M. Sugiura, S. Kanda and K. Imai, Anal. Chim. Acta, 251, Sci. Ser. C, 290, 369 (1980). 247 (1992). 19. S. Kobayashi and K. Imai, Anal. Chem., 52,1548 (1980). 12. N. Hanaoka, J. Chromatogr., 503, 155 (1990). 20. G. J. de Jong and P. J. M. Kwankman, J. Chromatogr., 13. N. Hanaoka, R. S. Givens, R. L. Schowen and T. 492, 319 (1983). Kuwana, Anal. Chem., 60, 2193 (1988). 21. M. Sugiura, S. Kanda and K. Imai, Biomed. Chromatogr., 14. J. Cepas, M. Silva and D. Perez-Bendito, Anal. Chem., 67, 7, 149 (1993). 4376 (1995). 15. K. Imai, A. Nishitani and Y. Tsukamoto, Chromatogra- (Received May 13, 1996) phia, 24, 77 (1987). (Accepted June 21, 1996)