Gas-Liquid Chromatographic Determination of Vitamin D3 in Tuna Liver and Vitamin D3 Resin Oils1,2

J. Nutr. Sci. Vitaminol., 22, 215-224, 1976

GAS-LIQUID CHROMATOGRAPHIC DETERMINATION OF VITAMIN D3 IN TUNA LIVER AND VITAMIN D3 RESIN OILS1,2

Tadashi KOBAYASHI, Atsuko ADACHI, and Keiko FURUTA3

Department of Hygienic Chemistry, Kobe Women's College of Pharmacy, Higashinada-ku, Kobe 658, Japan (Received January 14, 1976)

Summary Gas-liquid chromatographic (GLC) determination of vitamin D3 in tuna liver and vitamin D3 resin oils was investigated and a routine method slightly modified from the previously reported methods (1, 2) was established. Since both tuna liver and vitamin D3 resin oils contained large amounts of sterols, digitonin-Celite column chromato graphy according to SHEPPARDet al. (4) was used to remove the sterol influence from the unsaponifiable matters of the oils. After collecting the eluate and evaporating the solvent, the residue was subjected to thin-layer chromatography (TLC) using Kieselgel GF254as an adsorbent and a mix ture of n-hexane-ethyl acetate (4: 1) as a developing solvent. The scraped zones corresponding to vitamin D3 and pre-D3 were trimethylsilylated and then applied to the GLC using 1.5% OV-17 packed on Shimalite W (80 100 mesh) as a stationary phase. Trimethylsilylation of the gas chro matograms was an essential procedure, because the peaks of unknown substances in tuna liver oils and lumisterol in vitamin D3 resin oils could not be separated from the peak of pyro-D3 without trimethylsilylation. When the proposed method was applied to the samples, satisfactory results were obtained.

It was reported in the previous papers (1, 2) that the two kinds of routine methods for GLC determination of vitamin D4 in multivitamin preparations were established. The first method involving alkali saponification, isolation of the unsaponifiable matters, TLC and GLC analysis was applied to the usual prepa

1 Studies on the Gas-Liquid Chromatographic Determination of Vitamin D . Part III. Part II, see Ref. 2. 2 Following abbreviations are used: pre-D and D3 , precalciferol and precholecalciferol; pyro-D3, pyrocholecalciferol; isopyro-D3, isopyrocholecalciferol; CA, cholesteryl acetate; GLC, gas-liquid chromatography; TLC, thin-layer chromatography; I. U., international unit; TMS, trimethylsilyl. 3 小林正 ,足立昌子,古田肇子 4 Vitamin D usually means the sum of vitamin D and pre -D .

215 216 T. KOBAYASHI, A. ADACHI, and K. FURUTA rations, whereas the second included phosphate-treated alumina column chromato graphy (3) in addition to the above method for special preparations containing excess amounts of vitamin E. In this paper, the determination of vitamin D3 in tuna liver and vitamin D3 resin5 oils is one of the studies in this series. The method proposed previously for usual multivitamin preparations (1) was first applied to tuna liver oils, but it was not successful because of interferences by large amounts of sterols and unknown substances. Digitonin-Celite column chromatography according to SHEPPARDet al. (4) was used for the unsaponifiable matters to eliminate sterol influence. The vitamin D3 eluate obtained by chromato graphy was applied to the TLC and then the vitamin D3 fraction was analyzed without trimethylsilylation by the GLC reported previously (1). However, peaks due to unknown substances accompanying the vitamin D3 fraction disturbed both pyro and isopyro-D3 peaks (thermal cyclized products of vitamin D3) on the gas chromatogram. Although various kinds of column chromatography and TLC were investigated, the unknown substances could not be separated from vitamin D3. However, when trimethylsilylation was used followed by GLC, the pyro-D3 TMS ether peak was successfully separated from the TMS ethers of the unknown substances. Because tuna liver oils do not contain excess amounts of vitamin E, the use of phosphate-treated alumina column chromatography is un necessary. From these results and considerations, a routine method for deter mination of vitamin D3 in tuna liver oils was established. During the investigations, samples of vitamin D3 resin oils were obtained from Dr. F. J. Mulder (associate referee of the AOAC6) of Philips-Duphar Co., the Netherlands, for the 1975 comparative test of vitamin D assay methods of the AOAC. When GLC determination of vitamin D3 in these oils was investigated, we found that it was possible to adapt the method proposed for tuna liver oils to the vitamin D3 resin oils without any modification. Therefore, we presently proposed the method described in this paper as a routine one for both tuna liver and vitamin D3 resin oils.

EXPERIMENTAL

1. Materials and reagents Celite 545. Commercial grade, John-Manville Sales Co. Digitonin. Guaranteed reagent, E. Merck Co. Pyridine. Guaranteed reagent, redistilled, Nakarai Chem. Ind. Co. (Redis tillation was an essential procedure.) TMS reagents. Hexamethyldisilazane and trimethylchlorosilane, Nakarai Chem. Ind. Co.

Vitamin D3 resin is a concentrate of photo-irradiated 7-dehydrocholesterol after elimina tion of the unreacted sterol by precipitation with methanol cooling. Therefore, the resin usually contains lumisterol, tachysterol and other photoisomers with vitamin D3. 6 AOAC : Association of Official Analytical Chemistry. GLC DETERMINATION OF VITAMIN D (III) 217

Cholesterol acetate (CA). Cholesterol, Nakarai Chem. Ind. Co., was acety

lated according to a conventional method and recrystallized from acetone : mp 116•Ž.

Internal standard solution. CA was dissolved in benzene-pyridine (95: 5)

to a concentration of 100ƒÊg/ml. Used as an internal standard solution for GLC. Vitamin D3 standard solution. Vitamin D3 and CA were dissolved in benzene

pyridine (95: 5) to make concentrations of 80 ƒÊg/ml (3,200 I. U. /ml) for vitamin D3 and 100, ƒÊg/ml for CA, respectively.

Other materials and reagents were similar to those used in a previous paper (1) .

2. Samples

Tuna liver oil. A commercial tuna liver oil (raw fish oil), kindly supplied by Riken Vitamin Oil Co., Tokyo, was used. The vitamin A content estimated

according to the Japan Pharmacopeia VIII method (5) was 95,300 I.U./g . Vitamin D3 resin oils (samples No. 1 to 16). Vitamin D3 resin oils kindly

supplied by Dr. F. J. Mulder as samples of the 1975 comparative test of vitamin assay methods of the AOAC are peanut oil solutions of vitamin D3 resin. Samples No. 1 to 8 contain 100,000 I. U. /g (indicated value) of vitamin D3, while samples

No. 9 to 16 contain 10,000 I. U ./g (indicated value) of vitamin D3.

3. Procedure 1) Digitonin-Celite column preparation. The column is prepared accord ing to SHEPPARD et al. (4). Ten g of Celite 545 dried at 110•Ž for 6hr is stirred in a flask with 5ml of an aqueous digitonin solution (60mg/ml). Three g of digitonin treated Celite 545 is transferred to a glass tube (1.0•~20cm) with suitable quantities of n-hexane and then allowed to settle while slight pressure is made to the top of tube. The height of the Celite reaches about 12cm. The stopcock is opened to drain n-hexane until the top of the Celite is covered n-hexane to a height of 1 cm.

2) Saponification and isolation of the unsaponifiable matters. About 1g of an oil is weighed accurately in a saponification flask. After adding 50 ml of aldehyde-free ethanol, 20 ml of 20% pyrogallol solution in ethanol and 8ml of 90% (w/v) KOH solution are added. Saponification and isolation of the un saponifiable matters are carried out according to the previous paper (1). 3) Digitonin-Celite column chromatography. An accurate amount of the benzene solution obtained above equivalent to about 8,000 I. U. of vitamin D3 is taken and the solvent is evaporated under reduced pressure (lower than 40•Ž).

After dissolving the residue in 3 ml of n-hexane, the solution is transferred onto the digitonin-Celite column with aid of 5ml of n-hexane. The elution is started with n-hexane at a flow rate of 1-2 ml/main and the first 30ml of the eluate is collected in a round bottom flask. The solvent is evaporated under reduced pressure and then the residue is dissolved in 1.0ml of acetone. 218 T. KOBAYASHI, A. ADACHI, and K. FURUTA

4) TLC. Using a micropipet, 0.2ml of the acetone solution obtained above is taken and applied as a zone onto a Kieselgel GF254 plate. The TLC is developed and zones corresponding to vitamin D3 and pre-D3 are scraped off according to the previous paper (1). After extracting the two combined zones with 30ml of acetone and then filtering, the remaining powder on the filter paper is repeatedly washed with small quantities of acetone. The washed solutions are combined with the filtrate and the solvent is evaporated to dryness under reduced pressure. 5) Trimethylsilylation. After dissolving the residue obtained above in 0.5ml of the internal standard solution, trimethylsilylation is carried out by adding

0.5 ml of hexamethyldisilazane and 0.1ml of trimethylchlorosilane. The mixed solution is allowed to stand for 10 min at room temperature and then centrifuged for 15 min to eliminate the precipitation formed by adding the TMS reagents. 6) GLC. Using a 10 ƒÊl microsyringe, 5 ƒÊl of the supernatant obtained above is taken and applied to the GLC using the following apparatus and analytical conditions: Apparatus. A Shimadzu GC-4BPFE gas chromatogramh equipped with a hydrogen flame ionization detector. Column. A glass column (0.4•~250cm) packed with 1.5% OV-17 on 80

100 mesh of Shimalite W (purchased from Wako Chem. Ind. Co.). Operating parameters. Temperature : column 225•Ž, detector 300•Ž, injec tion port 250•Ž. Flow rate of carrier gas (N2): 120 ml/min. 7) Trimethylsilylation and GLC of vitamin D3 standard solution.•@ Tri methylsilylation and GLC are similarly carried out on 0.5 ml of the vitamin D3

standard solution. 8) Calculation. The peak area ratios of pyro-D3 TMS ether to CA are estimated on the gas chromatograms obtained from the sample and vitamin D3

standard solutions, respectively. The content of vitamin D3 in a sample is calcu lated by the following formula : Content of vitamin D3 (I.U./g)= Psa: Peak area ratio of pyro-D3 TMS ether to CA on the gas chromato gram obtained from sample.Pst : Peak area ratio of pyro-D3 TMS ether to CA on the gas chromato gram obtained from vitamin D3 standard solution. s: Concentration of vitamin D3 in the vitamin D3 standard solution (I. U. /ml). This is 3,200 I. U. /ml in the above case. V: Multiple for dilution ratio. W: Weight of sample taken for the determination (g).

RESULTS

1. GLC of vitamin D and steryl acetate Suitable quantities of vitamin D2, D3, their TMS ethers and steryl acetates GLC DETERMINATION OF VITAMIN D (III) 219 were applied to the GLC described in the procedure. Their retention times and relative retention times are shown in Table 1. Stigmasteryl acetate was used as an internal standard in the previous reports (1, 2), but CA was used for this purpose in this report because of the following reasons: (a) The retention times of vitamin D3 related compounds are usually shorter than those of the corresponding vitamin D2 related compounds and moreover these of TMS ethers are shorter than those of the corresponding unmodified compounds; (b) since the retention time of CA is shorter than that of stigmasteryl acetate, the procedure time can be saved; (c) even though CA showing shorter retention time was used instead of stigmasteryl acetate, no interference to the CA peak was observed in vitamin D3 determination.

Table 1. Retention times and relative retention times of vitamins D and steryl acetate.

Note: Vitamin D gives twin peaks due to pyro and isopyro-D.

2. Condition of trimethylsilylation when vitamin D2 in irradiated ergosterol mixtures was determined as de

scribed in the foregoing papers (6, 7), trimethylsilylation was carried out in pyridine

solution. Since rather large amounts of vitamin D2 (mg order) were determined in the foregoing cases and therefore high sensitivity was unrequired, no inter

ference due to the tailing of pyridine was observed. On the other hand, since rather high sensitivity (ƒÊg order) was required in this report, solvent tailing due

to pyridine disturbed the peak of pyro-D3 TMS ether. After investigating various conditions, we found that trimethylsilylation was successful in a mixed solvent of

benzene-pyridine (95: 5) and solvent tailing could be avoided.

3. Calibration curves After dissolving 2,000 to 10,000 I. U. of vitamin D3 with 100 ƒÊg of CA in

benzene-pyridine (95: 5), 0.5ml aliquots of the solution were trimethylsilylated and then applied to the GLC according to the procedure in order to obtain the calibration curves. As shown in Fig. 1, linear correlations between the weight

ratios (vitamin D3/CA) and peak area ratios (pyro or isopyro-D3 TMS ether/CA) were obtained. These results showed that trimethylsilylation by modified con

ditions and thermal cyclization of vitamin D3 into pyro and isopyro-D3 were

quantitatively carried out and therefore both peaks were available for the deter mination of vitamin D3. 220 T. KOBAYASHI, A. ADACHI, and K. FURUTA

Fig. 1. Calibration curves of vitamin D3 TMS ether by the GLC.

4. Recovery test of vitamin D3 by the digitonin-Celite column chromatography After applying 4,000 I. U. of vitamin D3 to the dlgltonln-Celite column chro matography described in the procedure, the first 30ml of the n-hexane eluate was collected and the solvent evaporated under reduced pressure. The residue was dissolved in 1.0ml of the internal standard solution and then 0.5ml of this solution was applied to the GLC according to the procedure. When the recovery was calculated by comparing with the results without application of the column chro matography, satisfactory results were obtained as shown in Table 2.

5. Determination of vitamin D3 in a tuna liver oil Figure 2 shows the gas chromatograms of a commercial tuna liver oil after various treatments. When the unsaponifiable matters was directly applied to the GLC, large peaks due to sterols disturbed the peaks of both pyro and isopyro-D3 as shown in Chart I. After applying the unsaponifiable matters to the digitonin Celite column chromatography followed by the TLC, most of sterol peaks could be eliminated but peaks due to unknown substances still disturbed the peaks of both pyro and isopyro-D3 as shown in Chart II. Although various kinds of column chromatography, TLC, further precipitation method by digitonide forma

Table 2. Recovery of vitamin D3 by digitonin-Celite column chromatography. GLC DETERMINATION OF VITAMIN D (III) 221

Fig. 2. Gas chromatograms of tuna liver oil. 1 and 1•L, pyro-D3 and its TMS ether; 2 and

2•L, isopyro-D3 and its TMS ether; U and U•L, unknown substance and its TMS ether; CA, cholesteryl acetate (internal standard). Chart I: the unsaponifiable matters. Chart

II: the unsaponifiable matters followed by treatment with digitonin-Celite column and TLC. Chart III: the unsaponifiable matters followed by treatment with digitonin

Celite column, TLC and trimethylsilylation. Chart IV: vitamin D3 (dotted line) or its TMS ether (solid line).

tion and malefic anhydride reactions were tried out, the unknown substances could not be separated from vitamin D3. However, when trimethylsilylation was ap plied, the peak of pyro-D3 TMS ether could be successfully separated from those of TMS ethers of unknown substances as shown in Chart III. According to the proposed method, vitamin D3 in the tuna liver oil was determined with or without addition of a known amount of crystalline vitamin D3. The recovery and results were quitely satisfied as shown in Table 3.

6. Determination of vitamin D3 in vitamin D3 resin oils Figure 3 shows the gas chromatograms of vitamin D3 resin oils after various treatments. Since large amounts of phytosterols derived from peanut oil coexisted in the oily samples, many large peaks due to the sterols were observed on the gas chromatogram of the unsaponifiable matters as shown in Chart I. Since the samples were thought to contain large amounts of lumisterol and tachysterol as the photoisomers of irradiated 7-dehydrocholesterol, trimethylsilylation was ap plied since the initial stage because it has been known in the foregoing investiga tions (6, 7) that lumisterol could not be separated from pyro-D by GLC without application of trimethylsilylation. The removal of phytosterols seemed to be almost completely achieved by digitonin-Celite column chromatography as shown 222 T. KOBAYASHI, A. ADACHI, and K. FURUTA

Table 3. Determination and recovery test of vitamin D3 in a tuna liver oil.

Fig. 3. Gas chromatograms of vitamin D3 resin oil. 1, 2, 3, 4 and U are the TMS ethers of gyro-D3, isopyro-D3, lumisterol, tachysterol and unknown substance, respectively. CA is cholesteryl acetate (internal standard). Chart I: the unsaponifiable matters fol lowed by trimethylsilylation. Chart II: the unsaponifiable matters followed by treatment with digitonin-Celite column and trimethylsilylation. Chart III: the unsaponifiable matters followed by treatment with digitonin-Celite column, TLC and trimethylsilylation. Chart IV: vitamin D3 followed by trimethylsilylation. in Chart II, whereas the peaks due to lumisterol and tachysterol were still observed in the gas chromatogram because they could not form digitonides with digitonin reagents. After applying the TLC, clean up was almost completely achieved as shown in Chart III. Determination of vitamin D3 in the samples was carried out according to the proposed method and the results are shown in Table 4. The exact contents of vitamin D3 in the samples were unknown to us, but the results must be reliable because clean up was almost completely achieved as shown in Fig. 3. GLC DETERMINATION OF VITAMIN D (III) 223

Table 4. Determination of vitamin D3 in vitamin D3 resin oils.

DISCUSSIONS

Determination of vitamin D3 in tuna liver and vitamin D3 resin oils was in vestigated. These oily samples usually contain large amounts of sterols as inter fering substances. Since the elimination of their interferences could not be per formed by the method proposed previously (1, 2), a procedure for their removal must be added. The precipitation methods by either methanol cooling (lower than l0•Ž) or the formation of digitonide have been widely used for this purpose for some time, but the procedures are rather complicated and time-consuming.

KATZ and KEENEY (8) reported on the digitonin-Celite column chromatography in order to eliminate the influences of sterols on the determination of vitamin E, while SIIEPPARD et al. (4) modified this method. We applied SHEPPARD's modified method to vitamin D assay. The recovery of vitamin D3 was almost 100% as shown in Table 2 and the removal of sterols from both tuna liver and vitamin D3 resin oils were satisfactorily achieved as shown in Figs. 2 and 3.

Vitamin A and sterols were thought to be main interfering substances on vitamin D3 assay in tuna liver oils, but it was found that some unknown substances besides them also interfered the assay. The unknown substances were accom panied with vitamin D3 after the digitonin-Celite column chromatography, TLC and GLC without trimethylsilylation as shown in Charts I and II of Fig. 2. How ever, the problem was solved by applying the vitamin D3 fraction of TLC to GLC after trimethylsilylation as shown in Chart III of Fig. 2. The chemical structures of unknown substances will be investigated in the future. The method proposed here was applied to a commercial tuna liver oil and the vitamin D3 content was estimated to be 10,800•}150 I. U. /g (mean•}S. D.), and the I. U. ratio of vitamin

A to vitamin D3 was 8.8. The former was within the range of yellowfin tuna liver oils (10,000-45,000 I. U. /g) mentioned by YAMAKAWA and KINUMAKI (9) while the latter was higher (2-3.5). It is well known that tuna liver oils contain greater amounts of vitamin D3 since BROCKMANN (10) initially isolated it in pure form from the liver oils. There 224 T. KOBAYASHI, A. ADACHI, and K. FURUTA fore, the determination of vitamin D3 in the liver oils might be one of the easier examples. Because many other fish liver oils (e.g., cod, pollack and dogfish liver oils) contain very small amounts of vitamin D3 with large amounts of interfering substances, the determination in such oils might be very difficult. Indeed, we applied the proposed method to a commercial pollack oil which was said to con tain 20 I. U. fg of vitamin D3 (9), but the interferences by unknown substances were too large to determine vitamin D3. The determination of vitamin D3 in such sam ples will be investigated in the future. Vitamin D3 resin oils are usually prepared by mixing the resin with vegetable oils (peanut oils were used for samples No. l to 16) to make suitable concentrations. Therefore, the main interfering substances for vitamin D3 assay might be lumi sterol and tachysterol as the photoisomers of 7-dehydrocholesterol from the resin and then phytosterols from vegetable oils. The gas chromatogram as shown in Chart I of Fig. 3 revealed the coexistences of large amounts of phytosterols. After removing most of the sterols by passing the unsaponifiable matters through a digitonin-Celite column, the coexistences of lumisterol and tachysterol were recognized as shown in Chart II of Fig. 3. Since there were not so large differences between the retention times of pyro-D3 and lumisterol TMS ethers (Chart II of Fig. 3), the coexistences of large amounts of lumisterol might interfere with the vitamin D3 assay. Therefore, we decided to add the clean up procedure by TLC before trimethylsilylation. From these results and considerations, the method proposed for vitamin D3 resin oils was identical with that for tuna liver oils.

The authorswish to thank Dr. F. J. Mulderof Philips-DupharCo. for his gift of vitaminD3 resin oils and Riken VitaminOil Co. for gift of a commercialtuna liver oil.

REFERENCES

1) KOBAYASHI,T. and ADACHI,A., J. Nutr. Sci. Vitaminol., 22, 41 (1976). 2) KOBAYASHI,T. and ADACHI,A., J. Nutr. Sci. Vitaminol., 22, 209 (1976). 3) MULDER, F. J. and DE VRIES,E. J., J. Assoc. Offic. Anal. Chem., 57,1349 (1974). 4) SHEPPARD,A. J., PROSSER,A. R., and HUBBARD,W. D., Method Enzymol., 18, C, 356 (1971). 5) Japan Pharmacopeia VIIIth ed., Commentary, Hirokawa Shoten, Tokyo, p. 194B (1971). 6) KOBAYASHI,T. and YASUMURA,M., J. Vitaminol.,18, 78 (1972). 7) KOBAYASHI,T. and YASUMURA,M., J. Nutr. Sci. Vitaminol.,19, 123 (1973). 8) KATZ, I. and KEENEY,M., J. Dairy Sci., 50, 1764 (1967). 9) YAMAKAWA,T. and KINUMAKI,T., Vitamins (in Japanese), 46, 167 (1972). 10) BROCKMANN,H. and BUSSE, A., Z. Physiol. Chem., 249, 176 (1937); Naturwissenschaften, 26, 122 (1938).