Isolation of roseoflavin resistant Leuconostoc mesenteroides
It has previously been shown in B. subtilis and L. lactis that spontaneous resistance to the toxic riboflavin analogue, roseoflavin, frequently coincides with a riboflavin-overproducing phenotype [24,27]. In order to determine whether this method was also applicable to Lc. mesenteroides, a wildtype strain of this species was plated at mid exponential growth phase on CDM containing 100 mg/L roseoflavin. In this way a total eleven mutants were isolated. These isolates were grown in CDM and as the cells entered the stationary phase the cell free supernatant was analysed for riboflavin content. All eleven roseoflavin-resistant mutants were found to be riboflavin overproducers (Fig. 1), leading to accumulation of the vitamin into the medium although at clearly different levels ranging from 120 to 500 μg/L.
Identification of mutations resulting in riboflavin overproduction
B. subtilis roseoflavin-resistant mutants have been shown to carry mutations either in the regulatory region of the rib operon or in ribC, encoding the flavokinase/FAD synthetase responsible for the conversion of riboflavin to its active derivatives FMN and FAD [28-30]. Roseoflavin resistance in L. lactis seems to be due to mutations that reside exclusively in the regulatory region upstream of the rib operon . It has been found that this region acts as a riboswitch causing premature transcription termination of the rib operon in the presence of FMN . The RFN element is a highly conserved domain found within this 5'-untranslated region consisting of 5 stem-loop structures . A predicted RFN element was identified upstream of the Lc. mesenteroides rib operon using RFAM (indicated in Fig. 2). To investigate whether mutations were present in the homologous DNA regions in the roseoflavin-resistant Lc. mesenteroides mutants chromosomal DNA was isolated from each mutant after which the relevant regions were amplified by PCR and subjected to sequence analysis. No mutations were identified in the ribC gene, whereas in each of the isolated mutants the rib leader region [Genbank: DQ645591] was shown to contain a mutation (Fig. 2). In only one case (CB203) the mutation was represented by a deletion, while the other ten strains were shown to possess point mutations at one of four different locations within the rib leader region. At two of these locations different mutations occurred in different mutants (For example, the rib leader region of mutant CB201 contains a G to A substitution at position 77, while strain CB207 contains a G to C substitution at the corresponding position).
Isolation of riboflavin-overproducing Lb. plantarum and identification of mutations resulting in riboflavin overproduction
Bioinformatic analysis had previously shown that the sequenced strain of Lb. plantarum WCFS1 contains an incomplete rib operon, in which the entire ribG and part of ribB are absent from the genome [24,33]. As expected, this strain is unable to grow in the absence of riboflavin (Fig. 3(A)), in contrast to Lb. plantarum NCDO1752, which is capable of growth in the absence of the vitamin, thus indicating that this strain encodes a complete and functional riboflavin biosynthetic capability. The latter strain was exposed to 100 mg/L roseoflavin resulting in the isolation of six resistant mutants, which were also shown to be riboflavin overproducers (Fig. 4). In order to analyse the possible mutations in these mutants it was necessary to obtain the sequence of the region upstream of ribG, which is not present in the sequenced strain. PCR and sequence analysis of this region from NCDO1752 suggests that the WCFS1 genome harbours a deletion of approximately 1350 base pairs corresponding to the entire ribG gene, and part of ribB (Figs. 3(B) and 3(C)). The sequence of the regulatory region in NCDO1752 was determined [Genbank: DQ645592], allowing the identification of the putative RFN element using RFAM. This region was then amplified by PCR from the riboflavin-overproducing mutants and their sequence was determined. Only two types of mutations were found (Fig. 5). One strain (CB300) contained a G to T substitution, while the remaining strains (CB301 to 305) contained a 9 base pair deletion within the RFN element.
Isolation of roseoflavin-resistant Propionibacterium freudenreichii
Two different strains of P. freudenreichii were selected in order to investigate whether the roseoflavin-resistance strategy for the isolation of riboflavin overproducers could also be applied to a representative species of a high-GC gram-positive bacterium. The method was indeed successful as shown by the isolation of twelve riboflavin over-producing mutants of this species, each varying in the amount of riboflavin produced (Fig. 6). Since the genome sequence of P. freudenreichii is not publicly available and the amplification of the rib leader region of this species using degenerate primers based on the rib leader sequence of P. acnes was unsuccessful, it was not possible to determine the nature of the mutation(s) resulting in the altered phenotype.
Stability study of riboflavin overproducing phenotype in P. freudenreichii
In order for any of the strains isolated in this study to have potential industrial usefulness it is necessary that the riboflavin-overproducing phenotype is stably maintained. To determine if this is the case, P. freudenreichii NIZO B374 and two of its riboflavin-overproducing derivatives were subcultured for sixty generations in the absence of the riboflavin analogue roseoflavin and extracellular riboflavin was determined throughout. It was found that the riboflavin-overproducing phenotype is stably maintained in the absence of the selection pressure for at least 60 generations (Fig. 7).
Yoghurt study using P. freudenreichii B374 and its riboflavin-overproducing derivative B2336
In order to determine if the cultures isolated in this study could have industrial potential, a yoghurt production trial was set up, which compared the use of P. freudenreichii NIZO B374 and its riboflavin-overproducing derivative P. freudenreichii NIZO B2336. To optimize the effect of propionibacteria in a yoghurt fermentation, we compared two different fermentation processes. The effect of co-inoculation of the propionibacteria with the yoghurt starter culture was compared with pre-fermentation by the propionibacteria prior to the addition of the yoghurt starter culture. The viable counts of propionibacteria, streptococci and lactobacilli were unaffected by the presence of the wildtype or riboflavin-overproducing propionibacterial strain. Also sequential inoculation or co-inoculation was did not affect viable counts of the different strains in the starter culture (data not shown). The final pH after fermentation is an important parameter in the yoghurt production process. Figure 8(A) shows the final pH value of the yoghurt at various inoculum levels using both the wildtype and riboflavin-overproducing strains in both sequential and simultaneous inoculation. Both strains have the same effect on acidification of the yoghurt. However, a marked difference is visible between the types of inoculation. Preculturing with propionibacteria counteracts acidification, while no effect was observed when the propionibacteria were added simultaneously with the starter culture. Figure 8(B) shows the final riboflavin concentration of the yoghurt at various inoculum levels of P. freudenreichii NIZO B374 and its riboflavin-overproducing derivative P. freudenreichii NIZO B2336. Co-inoculation of the propionibacteria with the starter culture has little impact on final riboflavin levels. However, addition of P. freudenreichii NIZO B2336 in the sequential fermentation process shows a doubling of the concentration of riboflavin in the fermented end-product.