Because it is a precursor of coenzymes

that are required for the enzymatic oxidation of carbohydrates,

riboflavin is essential to basic metabolism.
In higher animals, insufficient riboflavin can cause loss of
hair, inflammation of the skin, vision deterioration, and growth
failure. Riboflavin is therefore widely used as an additive to
foodstuffs and feedstuffs.

Riboflavin is recovered from the broth by centrifugation after

inactivation of the microorganisms by heat. Pasteurization of the
broth ensures that no viable cells of the production organism are
present in the final product. Differential centrifugation leads
to separation of cells and riboflavin crystals because of
 differences in size and sedimentation behaviour.
Mineral acid treatment (about 2% hydrochloric or sulfuric acid)
yields 96% riboflavin, which, after drying and packaging, is
either directly sold for application in feed or used for further
processing into a formulation. The acid treatment also ensures
destruction (via depurination) of DNA from the processing
organism. Recrystallization of the 96% pure material from
concentrated hydrochloric acid (> 25%) results in a food-quality
material containing 98% riboflavin. This last purification step
is identical to that currently used for purification of
riboflavin derived from chemical synthesis.

The concentration of 8alpha-hydroxymethylriboflavin, which

is a known degradation product after acid treatment of riboflavin, was
slightly increased in the fermentation-derived product, but only in
that of 98% purity. Since this product is derived from the 96% pure
material by crystallization from acidic media, this is to be expected;
however lumiflavin, which is a known degradation product after basic
treatment, was present in the synthesized riboflavin but not in the
fermentation-derived material.
Table 1. Composition of riboflavin products derived by fermentation and by synthesis
(average % weighed samples)

Component 96% ex 98% ex 96% ex synthesis

fermentation fermentation
HPLC method HPLC method HPLC method

I II I II I II

Riboflavin 98.4 98.3 100.1 99.7 98.8 98.6

Ribityl-oxo-chinoxalic acid Trace ND ND ND Trace Trace
8alpha-Hydroxy riboflavin Trace Trace 0.45 0.2 Trace Trace
Formylmethyl-flavin acetal Trace ND 0.11 Trace 0.52 ND
Lumichrome 0.14 0.11 ND Trace 0.18 Trace
Lumiflavin ND ND ND ND 0.56 0.5
Amino acids 0.9 0.9 0.06 0.06 - -
DNAa ND ND ND ND - -

Total 99.44 99.31 100.72 99.96 100.1 99.3

2.1.3 Genotoxicity

Studies on the genotoxic potential of riboflavin derived by

fermentation from B. subtilis and of 8alpha-hydroxyriboflavin, an
impurity of the fermentation-derived material, are summarized in Table
3.

3. METHODS OF GENETIC MODIFICATION

3.1 Molecular genetic methods used to clone and express the riboflavin
(rib) operon from Bacillus subtilis

The riboflavin operon of B. subtilis containing the genes

ribT, ribH, ribA, and ribG was cloned into pUC19 as a 6.4-kb
NcoI/XbaI DNA fragment. In order to enhance expression of the rib
genes, a strong constitutive B. subtilis bacteriophage SPO1 promoter
was included. Two versions were made: (i) the SPO1-15 promoter was
inserted in front of the endogenous ribP2 promoter in a construct
later designated as pRF93 and (ii) the endogenous ribP1 promoter was
replaced by SPO1-15 in a construct later designated as pRF69. In order
to make the constructs selectable after integration into the B.
subtilis genome (the ß-lactamase gene of pUC19 is not expressed in
B. subtilis), the two constructs were provided with different
selectable genes, i.e. the tetracycline-resistant (tet) gene derived
from B. cereus was inserted into pRF93, and the chloramphenicol-
resistant (cat) gene derived from pUC194, origin S. aureus/
B. subtilis, into pRF69. Both constructs were integrated into the
genome of recipient B. subtilis strain RB50. The integration sites
were mapped for pRF69 at 209° and for pRF93 at 139° in the B.
subtilis genome.

RB50, a strain of B. subtilis that overproduces riboflavin, was

obtained by classical mutagenesis and selection procedures. With
various unknown mutations, RB50 contains a mutation that maps at the
ribC locus, which is present at 147° of the B. subtilis genome.

In order to further increase riboflavin production, the

RB50::[pRF69]: :[pRF93] was subjected to increasing concentrations of
the antibiotics tetracycline and chloramphenicol, resulting in
amplification of pRF69 and pRF93. In general, pRF69 was multiplied
20-30 times whereas pRF93 was multiplied 8-10 times, resulting in the
production strain RB50::[pRF69]n: :[pRF93]m. Although neither
antibiotic was present during the production stage, there was no
detectable loss of either pRF (Maruo & Yoshikawa, 1989; Kreneva &
Perumov, 1990; Mironov et al., 1994; Schurter, 1994). All of the
procedures were extensively and well described.

3.2 Genetic stability of the production strain

The production strain obtained after amplification was

RB50::[pRF69]n: :[pRF93]m. Samples were taken from the fermentor
during the process, and the numbers of copies of pRF69 and pRF93 were
determined by Southern blotting. There was no apparent loss of the
foreign DNA present in RB50: :[pRF69]n::[pRF93]m during the

process, which took 48 h, with samples taken at 6-h intervals

(Hümbelin & Hermann, 1995). As all of the experiments were well
described and appeared to be sound and convincing, it can be concluded
that the production strain is genetically stable during the
fermentation process.

3.3 Foreign DNA present in the end-product

One of the purification steps involved treatment of the product

with diluted hydrochloric acid for 60 min at 90°C. Under these
conditions, DNA is heavily depurinated and degraded (Hermann &
Schurter, 1994). A sensitive PCR method was used to detect any
residual transforming DNA (monitored as a 557-bp pRF69 PCR cat
fragment), with a detection limit of 0.5 ppb DNA per mg riboflavin. No
foreign DNA was detected. Furthermore, no transforming activity was
detected after only 5 min of hot acid treatment of crude riboflavin
(Schurter & Hermann, 1995). As all of the experiments were described
in detail, and the results are convincing, it can be concluded that
the riboflavin samples were free of any detectable transforming DNA.

4. COMMENTS

The strain used in riboflavin production has been genetically

modified to overproduce riboflavin by amplification of the chromosomal
region of the riboflavin operon containing suitable promoters and
flanked by pUC19 and antibiotic-resistance marker genes. The lack of
pathogenicity and toxicity of the strain of microorganism
(B. subtilis) from which the genetic information encoding riboflavin
was cloned is well documented, and the methods for genetic
modification have been well described. The production strain of
B. subtilis was shown to be genetically stable during fermentation.
It was convincingly established that the final product is riboflavin
of a purity comparable to, or greater than, that produced by
conventional methods and is free of DNA from the production organism.
Thus, antibiotic resistance marker genes are not present.

In a 90-day study of toxicity, rats were fed diets containing

food-grade riboflavin (98% pure) produced by fermentation from
genetically modified B. subtilis or riboflavin derived by chemical
synthesis, also of 98% purity. The doses tested were 0, 20, 50, or 200
mg/kg bw per day. Some animals at the high dose had reduced weight
gain, but generally by less than 10%, and food conversion efficiency
was not affected. In the group fed the high dose of
fermentation-derived riboflavin of 98% purity, some minor changes in
red blood cell parameters were observed, but they were not considered
relevant. The NOEL for 98% pure riboflavin produced by fermentation
was 200 mg/kg bw per day, which was the same as that for chemically
synthesized riboflavin.

Fermentation-derived riboflavin and the degradation product

8alpha-hydroxyriboflavin did not induce gene mutations in bacteria
in vitro in either the absence or the presence of metabolic
activation.

5. EVALUATION

The Committee concluded that the recombinant DNA techniques used

to derive the production strain of B. subtilis were well
characterized, providing assurance that no DNA is present in the
end-product. On the basis of molecular biological data and chemical
analytical research, it can be concluded that fermentation-derived
riboflavin from genetically modified B. subtilis is substantially
equivalent to synthetic riboflavin.

For 98% pure fermentation-derived riboflavin for use in food, the

NOEL in the 90-day study of toxicity in rats was 200 mg/kg bw per day,
the highest dose tested. Fermentation-derived riboflavin was evaluated
on the basis of its substantial equivalence to synthetic riboflavin.
Therefore, the Committee included riboflavin derived from a production
strain of genetically modified B. subtilis in the previously
established group ADI of 0-0.5 mg/kg bw for synthetic riboflavin and
riboflavin-5'-phosphate.