PHOTOLYTIC MECHANISM OF RIBOFLAVIN

SYOICHI SHIMIZU

Department of Industrial Chemistry. Faculty of Engineering,

Kyoto University, Sakyo-ku, Kyoto.

(Received August 12, 1954)

One of the most characteristic properties of riboflavin (Ⅰ) is its extraordinary

photosensitivity. Therefore, many investigations (1-11) have been carried

out on the photolytic reaction of riboflavin and related compounds. According

to Karrer et al. (2) and Kuhn et al. (3), the type of decomposition by light is

dependent on the pH of the solution. When irradiated in neutral or acidic

solution, riboflavin is converted into lumichrome (Ⅱ), while the main photo

decomposition product of the vitamin in an alkaline medium is lumiflavin

(Ⅲ) as shown in Diagram 1.

(Ⅰ)

(Ⅱ)

(Ⅲ)

(Ⅳ)

Diagram 1.

However, little has been known of the fate of the ribityl side chain in

the photolysis. Brdicka (10) observed polarographically that nearly equiva

lent amounts of lumichrome, formaldehyde and an aldose, probably erythrose

were produced on exposure of an aqueous solution of riboflavin to light. More

recently, Halwer (11) isolated and identified acetaldehyde, formaldehyde and

formic acid as side chain products from the photolyzed solution of 9-〔2'-hydrox

yethyl〕-isoalloxazine (Ⅳ), a simple analogue of riboflavin. But the isolation

of compounds to be expected from the side chain of riboflavin itself has not

hitherto been achieved.

For the study of the exact reaction mechanism, therefore, the identifica

tion and determination of the decomposition products are of special impor

tance. The present study deals with related problems, especially the compa-

39

40 SHIMIZU 1955

rison between the acid and alkaline side in the photolysis of riboflavin.

EXPERIMENTAL

1. Photolysis in Neutral Media.

Riboflavin, dissolved in redistilled water in a concentration of 116.4γ/ml,

was exposed to sunlight in Petri dish until yellow color completely dis

appeared.

Determination of Decomposition Rate of Riboflavin. The remaining riboflavin

was determined at several stages of irradiation by measuring the optical

density at 450mμ using Beckman spectrophotometer model DU.

As shown in Table Ⅰ, the riboflavin concentration fell to 3.88% of its

original value at the end of the photolysis.

Table Ⅰ

Remaining Riboflavin in Photolysis in a Neutral Medium

* Optical density of standard riboflavin solution containing 10γ/ml was

0.378.

Paper Chromatographic Detection of Fluorescent Decomposition Products . The paper

chromatography was carried out by one-dimensional and ascending procedure

in a dark room at 25±2°, as previously reported (9 , 12).

The photolyzed solution itself and the solution concentrated to 1/10

volume were applied to Toyo filter paper No . 50 (for paper chromatographic

purpose), dried in air and then developed using n-butanol-glacial acetic acid-

water (4:1:5) (A) and 5% aqueous solution of Na2HPO4・12 H2O (B) , respec

tively, as developing solvents. Detection of spots was done by observation

under ultraviolet ray.

In case of the photolyzed solution, only one spot having RF value 0.70 (solvent A) or 0.07 (solvent B) was observed. This spot was corresponding

to that of lumichrome. With the concentrated solution two spots were obtained, one being above-mentioned lumichrome and the other undecomposed

riboflavin, the RF value of which was 0.32 (solvent A) or 0.28 (solvent B). It became obvious from this observation that the fluorescent substance pro

duced under the above condition was only lumichrome.Determination of Lumichrome . Since the fluorescence intensity of riboflavin

solution was so weak at pH 9.5 as to be negligible when compared with that

of lumichrome solution, lumichrome in photolyzed solution was determined

by measuring its fluorescence intensity after adjusting to pH 9.5.

The obtained value was 61.50γ/ml.

Detection of Formaldehyde. One hundred ml of photolyzed solution was

steam-distilled and the 50ml of the distillate was collected.

Vol. 2 RIBOFLAVIN PHOTOLYSIS 41

The presence of formaldehyde in this distillate was demonstrated by

phenylhydrazine-nitroprusside test, phenylhydrazine-ferricyanide test and phenylhydrazine-ferric chloride test (13). The final identification was done

by dimedone method as described below.Although the detection of acetaldehyde by Rimini test, Windisch test and

Leys test (14), respectively, was attempted, positive reaction did not occur.

Therefore, there must be very few, if any, acetaldehyde.Determination of Formaldehyde.

(A) From the Weight of Dimethone (15). One liter of photolyzed solution

was steam-distilled. Five hundred ml of distillate collected was adjusted to

pH 8.0 and 25ml of 1% dimedone solution in 50% ethyl alcohol was added

to it. White needles crystallized after storing in a refrigerator for several

days were collected on the glass-filter, vacuum-dried, and weighed. Total

amount of the crystals was 69.5mg, having mp 189°. After recrystallization

from ethanol mp rose to 191° and not depressed on admixture with authentic

formaldehyde-dimethone.

Dimethone, 69.5mg, corresponds to formaldehyde, 7.14mg.

(B) By Potassium Cyanide Method (16). Two hundred and fifty ml of distillate was collected from 1l of photolyzed solution. To an aliquot of 50ml,

10ml of 0.1N silver nitrate, 1ml of 30% hydrochloric acid, 10ml of 0.5%

potassium cyanide were added and the whole was shaken vigorously. After filling with distilled water to 100ml it was filtered. Using 50ml of filtrate,

the unconsumed silver nitrate was titrated with 0.01N ammonium rhodanide by Volhard method. The difference between blank and a sample was 1.95

ml, corresponding to 6.54mg of formaldehyde.Detection of Formic Acid. According to the method proposed by Fincke (17),

200ml of photolyzed solution acidified with sulfuric acid was steam-distilled under reduced pressure. To catch the volatile acids, a suspension of barium

carbonate in water was used. The filtrate of this suspension was concentrated to 20ml. After reducing an aliquot of this concentrate with hydrochloric acid and magnesium, the identification test for formaldehyde was

performed with positive results. Also when solutions of sodium acetate, hydrochloric acid, sodium chloride and mercuric chloride were added to an

aliquot of this concentrate, heated and then cooled down, a white precipitate was produced. From the results mentioned above, the existence of formic

acid in photolyzed solution was demonstrated. The absence of acetic acid in the distillate described above was demonstrated by lanthane nitrate reaction

(18).Determination of Formic Acid.

(A) By Iodometric Method (19). To 100ml of solution to be analyzed (corresponding to 1l of the photolyzed solution), 10ml of 50% sodium acetate, 2ml of 10% hydrochloric acid and 10ml of 10% mercuric chloride-15% sodium

chloride were added, the whole was boiled for 2 hours and filtered. The precipitate was dissolved by adding 10ml of 10% hydrochloric acid, 20ml

of 5% potassium iodide and 2.0ml of 0.1N iodine and an excess iodine was titrated with 0.01N sodium thiosulfate.

42 SHIMIZU 1955

The consumed amount of 0.1N iodine solution was found to be 0.5976ml, corresponding to 1.38mg of formic acid.

(B) By Alkalimetry. The determination of formic acid was likewise done by alkalimetric method. A sample solution was steam-distilled and the distillate was caught by 10ml of 0.01N sodium hydroxide. The excess

sodium hydroxide was back-titrated with 0.01N sulfuric acid. The volume of sulfuric acid equivalent to alkali consumed was found to be 2.80ml,

corresponding to 1.21mg of formic acid.Detection of Alcohols. The qualitative identification of alcohol in the distil

late of photolyzed solution was attempted, but any kinds of volatile alcohols were not detectable.

Isolation and Identification of Lumichrome. Each of the distillation residues of

photolyzed solutions used in the experiment described above was evaporated

in vacuo to a volume less than 1/10 and the resultant concentrate was cooled

in a refrigerator overnight.

The crystals deposited were collected on a glass-filter, vacuum-dried and

weighed. The average value was found to be 39.62mg, being about 65%

of the fluorometrically determined value.

The crystals did not melt below 350° and paper chromatogram of their

solutions in 0.01N sodium hydroxide showed one spot corresponding to lumi

chrome.

The filtrates, separated from lumichrome crystal were filled to 100ml

and the whole was examined in the next experiment. This solution will

be called, for convenience, the distillation residue.

Analysis of Distillation Residue.

(A) Determination of Aldehyde Group by Ponndorf's Method (20). To 100ml of

distillation residue, obtained from 1l of photolyzed solution, 10ml of 0.01N

silver oxide solution was added and the unconsumed silver was measured

volumetrically. It was found to be 3.850ml. Since 1ml of consumed

silver solution is equivalent to 0.005mM of aldehyde group, 1l of photolyzed

solution contained 0.103mM of aldehyde group.

(B) Determination of Carboxyl Group. One hundred ml of distillation residue

was boiled in the presence of lead peroxide in nitrogen atmosphere. The

generated carbon dioxide gas was absorbed in 10ml of 0.1N barium hydrox

ide solution and the remaining barium hydroxide was back-titrated with

0.1N succinic acid solution.

The difference between blank and a sample was found to be 0.74ml,

corresponding to 0.075mM carboxyl group.

(C) Determination of α-Glycol Group (21). Ten ml of distillation residue, 2

ml of 0.1M potassium periodide, 1ml of 2N sulfuric acid were mixed.

After 24 hours, 5ml of 20% potassium iodide solution was added and the

liberating iodine was titrated with 0.1N sodium thiosulfate. The consumed

periodic acid was found, on an average, to be 0.745mM. Then, 50ml of

distillation residue was oxidized with periodic acid and the resultant solution

was steam-distilled. Formaldehyde in the distillate was determined by

dimedone method.

Vol. 2 RIBOFLAVIN PHOTOLYSIS 43

Formaldehyde formed was 0.247mM. Thus, the ratio of periodic acid

consumption to formaldehyde formation was about 3:1. This result shows that about one mole of tetrose exists in distillation residue.

2. Photolysis in Alkaline Media.Riboflavin was dissolved in 0.1N sodium hydroxide solution and illumi

nated to sunlight.Paper Chromatography of Photolyzed Solution. Besides lumiflavin, having RF value

0.46 (solvent A) or 0.16 (solvent B), which is the main photolyzed product in an alkaline medium, lumichrome was likewise detectable.

Determination of the Remaining Riboflavin. To remove the photodecomposition product, the photolyzed solution was adjusted to pH 4.5, shaken vigorously

with chloroform, and then determined colorimetrically.Table Ⅱ represents the results obtained.

Table Ⅱ

Remaining Riboflavin in Photolysis in an Alkaline Medium

Thus, 97% of original riboflavin was shown to have been destroyed .

Determination of Lumiflavin. A fluorometric method was adopted . At the

end of the photolysis, lumiflavin concentration was found to be 181γ/ml.

Determination of Lumichrome. Determination of lumichrome concentration

was done by measuring the fluorescence intensity after adjusting to pH 9.5

as described in part 1.

An average value was found to be 23.7γ/ml.

Qualitative Detection of Volatile Aldehyde and Acids. The same procedure as

described in part 1 was adopted. Formaldehyde and formic acid were found

in the distillate. But tests for acetaldehyde and acetic acid showed negative

results.

Quantitative Estimation of Formaldehyde and Formic Acid. The determination of formaldehyde was made by dimedone method. The amount of formaldehyde-

dimethone was found to be 27.5mg, corresponding to 2.82mg of formaldehyde in 1l of the photolyzed solution.

Formic acid was estimated by iodometric method and found to be 3.22mg.

Detection of Volatile Alcohol. No alcohol could be detected.

Analysis of Distillation Residue.

(A) Determination of Aldehyde and Carboxyl Groups. Aldehyde group, 0.52mM,

and 0.44mM of carboxyl group were found in the distillation residue from

1l of the photolyzed solution.

(B) Determination of α-Glycol Group. Periodic acid oxidation method was

used. Periodic acid consumption was 2.7mM and formaldehyde formation

was 0.94mM. From these results it was evident that primary alcohol group

adjacent to the hydroxylated carbon, CH2OH-CHOH-, and two other α-glycol

44 SHIMIZU 1955

groups existed.

DISCUSSION

The results obtained are summarized in Table Ⅲ and Ⅳ.

Table Ⅲ

Photodecomposition in a Neutral Solution (Lumichromic Cleavage)

Table Ⅳ

Photodecomposition in an Alkaline Solution (Lumiflavinic Cleavage)

The remarkable difference between the lumichromic and lumiflavinic cleav

ages can be found in the quantity of formaldehyde formed, which is marked

only in the former case. The amount of formaldehyde formed was, however,

found to be nearly equal to that of lumichrome in both cases.

It is suggested from this fact that formaldehyde would be produced only

when the conversion of riboflavin to lumichrome occurs. This leads us to

the assumption that formaldehyde might result from the carbon atom at

position 1' of ribityl side chain and the corresponding carbon atom might

remain at position 9 in lumiflavin.

Based upon this assumption, a photodecomposition mechanism of ribofla-

Vol. 2 RIBOFLAVIN PHOTOLYSIS 45

vin is conceived as illustrated in Diagram 2.

Diagram 2.

(Ⅰ)

(Ⅱ)

(Ⅲ)

The first step in photolysis is the dehydrogenation on the second carbon atom of ribityl side chain, whereby a carbonyl group is formed. This idea

coincides with Karrer's hypothesis (2), but no evidence could be obtained regarding the actual formation of 2'-keto-riboflavin.

Owing to its lability, the hydrolytic rupture of a carbon-carbon bond instantly occurs between the position 1' and 2', so that, in an alkaline medium,

N-methylene radical combines with proton, leading to N-methyl compound known as lumiflavin, while, in an acid medium, it combines with hydroxyl

group, forming 6, 7-dimethyl-9-hydroxymethyl-isoalloxazine, which is so unstable that it converts to lumichrome by liberating formaldehyde.

The reason why a small amount of formic acid is formed may be based on the oxidation of a part of formaldehyde once produced. The production

of erythronic acid besides erythrose in lumichromic cleavage may be understood from the same reason.

Considering that lumichrome and one carbon compound, chiefly formaldehyde, were produced in photolysis in a neutral medium, while lumiflavin

which has one more carbon than lumichrome was produced in an alkaline medium, it seems natural that in both cases four carbon compounds are con

tained in the distillation residue. This was demonstrated by periodic acid oxidation. Thus, primary alcohol group in position 5' of parent riboflavin

has been left harmless in four carbon compounds. In this point the mechanism

presented here does not agree with the explanation proposed by Halwer (11).The photodecomposition of riboflavin is complicated and involves probably

more than one mechanism. Therefore, a slight variation of the irradiation condition would affect the mechanism. But the author believes that the

above mentioned mechanism represents the most essential one for the photodecomposition process in an ordinary condition.

SUMMARY

In order to clarify the photodecomposition mechanism of riboflavin, detec-

46 SHIMIZU 1955

tion and determination of photolytic products were made.

(1) In neutral solution, the photolytic products, calculated on the basis of 1M of initial riboflavin, were as follows: 0.82M of lumichrome, 0.74M

of formaldehyde, 0.09M of formic acid and about 1M of sugar, probably erythrose.

Lumiflavin, acetaldehyde, acetic acid and volatile alcohols could not be detected.

(2) From the solution irradiated in an alkaline medium, 0.65M of lumiflavin, 0.09M of lumichrome, 0.1M of formaldehyde and 0.06M formic acid

were detected. Besides, four carbon compounds, probably mixtures of erythrose and erythronic acid, were produced. Acetaldehyde, acetic acid and

volatile alcohols were absent.From these results, it may be considered that formaldehyde is produced

accompanying lumichrome. Accordingly, the following photodecomposition mechanism of riboflavin is conceived:

At first, hydroxylated carbon at position 2' is oxidized to carbonyl group. Then, hydrolytic rupture of carbon-carbon bond occurs between 1' and 2'. At

that time, hydroxyl group is bound to flavin-methylene radical, forming 9-hydroxymethyl flavin which converts instantly to lumichrome by liberating

formaldehyde in lumichromic cleavage. On the other hand, in lumiflavinic cleavage, flavin-methylene group combines with proton, forming flavin-methyl

compound known as lumiflavin. Existence of formic acid may be resulted by a further oxidation of a part of formaldehyde once produced.

ACKNOWLEGEMENT

I wish to express my sincere thanks to Prof. Ryohei Takata for his en

couragement and valuable suggestions offered during the course of this work.

This investigations were aided by a grant from Fundamental Scientific

Research of Education Ministry.

REFERENCES

1) Warburg, O., and Christian, W., Naturwiss. 20, 980 (1932)2) Karrer, P., Salomon, H., Schopp, K., Schlitter, E., and Frizche, H., Helv. Chim. Acta 17, 1010 (1934), Karrer, P., Salomon, H., Schopp, K., and Schlitter, E., ibid. 17, 1165 (1934), Karrer, P., Schlitter, E., Pfaehler, K., and Benz, F., ibid. 17, 1516 (1934), Karrer, P., and Meerwein, H. F., ibid. 18, 480, 1126 (1935), Karrer, P., Kobner, T.,

and Zehender, F., ibid. 19, 261 (1936), Karrer, P., and Neef, R., ibid. 19, 1029 (1936)3) Kuhn, R., Rudy, H., and Wagner-Jauregg, T., Ber. 66, 1950 (1933), Kuhn, R., and Bar,

F., ibid. 67, 898 (1934)4) Theorell, H., Biochem. Z. 279, 156 (1935)5) Koschara, W., Z. physiol. Chem. 229, 103 (1934)6) Sakurai, Y., and Hukai, T., Vitamins 5, 413 (1952)7) Hotta, K., Vitamins 5, 606 (1952)8) Watanabe, A., and Asahi, T., Vitamins 4, 146 (1951)9) Shimizu, S., J. Ferment. Tech. 29, 10, 15 (1951)10) Brdicka, R., Coll. Czechoslov. Chem. Comm. 14, 130 (1949)11) Halwer, M., J. Amer. Chem. Soc. 73, 4870 (1951)

Vol. 2 RIBOFLAVIN PHOTOLYSIS 47

12) Shimizu, S., J. Ferment. Tech. 30, 13 (1952)13) Association of Official Agricultural Chemists, Official and Tentative Methods of

Analysis. 4th ed., p. 436 (1935)

14) Klein, G., "Handbuch der Pflanzenanalyse." II, p. 267 (1932)15) Weinberger, W., Ind. Eng. Chem. Anal. Ed. 3, 365 (1931)16) Romijn, G., Z. anal. Chem. 36, 21 (1897)17) Fincke, H., Z. Nahr. 21, 1 (1911), 25, 387 (1913)18) Kruger, D., Ber. 63, 826 (1930)19) Riesser, O., Z. physiol. Chem. 96, 357 (1916)20) Ponndorf, W., Ber. 64, 1913 (1931)21) Organic Reaction II, 341 (1947)