The potential of bifidobacteria as a source of natural folate


Thomas Andlid, Department of Chemical and Biological Engineering/Food Science, Chalmers University of Technology, SE - 412 96 Gothenburg, Sweden. E-mail:


Aims:  To screen 19 strains of bifidobacteria for main folate forms composition in synthetic folate-free and complex folate-containing media.

Methods and Results:  HPLC was used to analyse deconjugated folates extracted from bacterial biomass. Most strains had a total folate content above 4000 μg per 100 g dry matter (DM). The highest value of 9295 μg per 100 g DM was found in Bifidobacterium catenulatum ATCC 27539 and the lowest in Bifidobacterium animalis ssp. animalis ATCC 25527 containing 220 μg per 100 g DM. Ten strains grew in a synthetic folate-free medium (FFM), showing folate autotrophy and suggesting folate auxotrophy of the remaining nine. In the autotrophic strains, a consistently higher folate level was found in FFM as compared to a more complex folate-containing medium, suggesting reduced requirements for folates in the presence of growth factors otherwise requiring folates for synthesis. The contents of total folate, 5-CH3-H4folate and H4folate were strain dependent. 5-CH3-H4folate dominated in most strains.

Conclusions:  Our results show that bifidobacteria folate content and composition is dynamic, is strain specific and depends on the medium. Suitable selection of the growth conditions can result in high levels of folate per cell unit biomass.

Significance and Impact of the Study:  This suggests that certain bifidobacteria may contribute to the folate intake, either directly in foods, such as fermented dairy products, or in the intestine as folate-trophic probiotics or part of the natural microbiota.


Folate (folic acid; vitamin B9) is a water-soluble vitamin essential for methylation and synthesis of nucleic acids, certain amino acids and proteins necessary for replication and growth (Jacob 2000; Lucock 2000). Folate deficiency is associated with increased risk for malformations in early embryonic brain and spinal cord development (neural tube defects, NTDs) (Botto et al. 1999) and megaloblastic anaemia (Wickramasinghe 2006). Data also suggest increased risk for some forms of cancer (e.g. colon, breast and bladder) (Choi and Mason 2000), Alzheimer’s disease (Clarke et al. 1998) and cardiovascular disease as a result of insufficient folate status (Institute of Medicine, 1998).

In contrast to many micro-organisms and plants, humans are not able to synthesize folate de novo and have to rely on exogenous sources of the vitamin (Sarma et al. 1995). The recommended daily intake of folate, by for instance the Nordic Nutrition Recommendations (Becker et al. 2004), is 300 μg for men, 400 μg for women and 500 μg for lactating or pregnant women (McPartlin et al. 1999). These levels, however, are not easy to reach. Foods rich in folate are, for example, green leafy vegetables like spinach and broccoli, beans and fruits such as oranges, yeast and liver (Eitenmiller and Landen 1999). Naturally, the fraction of such products in the diet depends on availability, economy and eating habits. Moreover, folate in nature is present in different forms and matrixes that vary significantly in stability and bioavailability (Gregory 1996). The overall picture is that folate shortage is a common phenomenon, especially in developing countries, poor population segments and women at childbearing age (Antony 2001).

Recommendations by health organizations have led to fortification programmes in many countries (Wright et al. 2001). For instance, USA (US Food and Drug Administration, 1996), Canada (Health Canada, 1997) and Chile (Freire et al. 2000) have mandatory fortification of flour and uncooked cereal-grain products, whereas voluntary folic acid fortification of specified foods is implemented in Australia (Metz et al. 2002). Other countries, however, have chosen not to fortify because of the potential link between high doses of synthetic folic acid and the development and progression of certain cancer forms (Mason et al. 2007; Hirsch et al. 2009), as well as masking of vitamin B12 deficiency and thereby the risk of neuropathy (Asrar and O’Connor 2005). Natural folates, in contrast to synthetic folic acid, do not mask vitamin B12 deficiency (Kim et al. 2004) and are probably of lesser risk with respect to overdosing and cancer (Gutstein et al. 1973; Mason 2002; Kim et al. 2004). Therefore, biofortification with natural folates produced by selected micro-organisms may be an alternative to fortification with synthetic folic acid.

Some micro-organisms important for fermented food and/or gut flora, such as yeast and bifidobacteria, can synthesize folate de novo (Klipstein and Samloff 1966; Deguchi et al. 1985; Camilo et al. 1996; Hjortmo et al. 2005). This means that live micro-organisms in food as well as in the intestinal biota may contribute to the human folate intake. It has recently been demonstrated that a portion of folate may be absorbed across the large intestine of both humans (Strozzi and Mogna 2008; Aufreiter et al. 2009) and animals (Kim et al. 2004). In the large intestine, the folate absorption rate is slower than in the small intestine (Wright et al. 2003, 2005); however, the transit time is longer and the rich microbiota probably contributes to a continuous production and hence a more stable folate level than in the small intestine (Aufreiter et al. 2009).

Bifidobacteria represent one major group of intestinal bacteria in humans and are often screened for probiotic properties and added to different kinds of dairy and pharmaceutical products (Scardovi 1986). Many strains have been found to produce folate, and information about accumulated cellular and secreted levels is presented in a few studies (Lin and Young 2000; Crittenden et al. 2003; Pompei et al. 2007a,b; Strozzi and Mogna 2008). However, further studies on bifidobacteria folates are motivated for a number of reasons. First, the impact of medium richness (growth factors) has been found crucial for the folate level in yeast. This has not been much studied in bifidobacteria. Second, the relative composition of main folate forms is not known. From yeast studies, it is evident that presence or absence of a certain growth factors as well as specific growth rate affects the level of a specific folate form, which may guide in bioprocessing for higher levels. Knowledge about main forms is also relevant for stability and availability reasons (folate forms vary in stability and bioavailability). Third, previous studies have shown that strains differ much with respect to folates. Therefore, studies on more strains may select potentially useful strains and, for instance, suggest trends for species. Fourth, as different methods for folate extraction and analysis may yield differences in folate data, it is important to compare methods. This far, the microbiological assay has mainly been used for folates in bifidobacteria; here, we use HPLC.

The aim of the present work was to investigate the folate levels and main folate form composition in bifidobacteria cultured in vitro. The impact of medium composition on biomass-specific cellular folate content and composition was addressed. We screened different species and strains of bifidobacteria originating from humans, different animals and a fermented product, with a validated HPLC method.

Materials and methods

Bacterial strains and culturing media

A total of 19 Bifidobacterium strains were investigated for their capacity to produce folate. The strains were obtained from Bologna University Scardovi Collection of Bifidobacteria and from ATCC or DSMZ Collections (Table 1). Prior to proceed to the analysis, all the strains of Bifidobacterium spp., provided as freeze-dried cultures, were subcultured in ‘trypticase–phytone–yeast extract’ broth (TPY) (Biavati and Mattarelli 2006), for 24 h, at 37°C and anaerobically (Anaerocult A; Merck, Solna, Sweden). Two different culturing media were used for the experiments: (i) TPY, containing trypticase peptone (10 g l−1), phytone peptone (5 g l−1), glucose (15 g l−1), yeast extract (2·5 g l−1), Tween 80 (1 ml l−1), cysteine hydrochloride (0·5 g l−1), di-potassium hydrogen phosphate (2 g l−1) and magnesium chloride-hexahydrate (0·5 g l−1); (ii) synthetic medium developed as ‘folate-free medium’ (FFM) from the ‘Medium D’ (MD) described by Modesto et al. (2003) with modifications mostly consisting of the elimination of sources of folate and addition of a trace metals solution. FFM contained the following components: glucose 15 g l−1, sodium acetate 10 g l−1, ammonium sulfate 10 g l−1, di-potassium hydrogen phosphate 5 g l−1, di-hydrogen phosphate 3 g l−1, urea 2 g l−1, ascorbic acid 10 g l−1, Tween 80 1 ml l−1, minerals (MgSO4·7H2O 0·2 g l−1, FeCl2·4H2O 10 mg l−1, MnSO4·4H2O 8 mg l−1, NaCl 10 mg l−1), trace metals (EDTA 30 mg l−1, CaCl2·2H2O 9 mg l−1, ZnSO4·7H2O 9 mg l−1, FeSO4·7H2O 6 mg l−1, H3BO3 2 mg l−1, MnCl2·2H2O 1·2 mg l−1, Na2MoO4·2H2O 0·8 mg l−1, CoCl2·2H2O 0·6 mg l−1, CuSO4·5H2O 0·6 mg l−1, KI 0·2 mg l−1), vitamins (pyridoxamine pantothenate-HCl 2 mg l−1, nicotinic acid 2 mg l−1, thiamine 2 mg l−1, calcium pantothenate 1 mg l−1, riboflavin 1 mg l−1, p-amino-benzoic acid 50 μg l−1, biotin 50 μg l−1), aminoacids (cysteine-HCl 0·5 g l−1, l-alanine 0·2 g l−1, dl-arginine 0·2 g l−1, dl-asparagine 0·2 g l−1 l-aspartic acid 0·2 g l−1, glycine 0·2 g l−1, l-histidine 0·2 g l−1, l-glutamic acid 1 g l−1, l-isoleucine 0·1 g l−1, l-lysine 0·1 g l−1, l-leucine 0·2 g l−1, l-methionine 0·2 g l−1, dl-phenylalanine 0·2 g l−1, l-proline 0·2 g l−1, dl-serine 0·2 g l−1 l-threonine 0·2 g l−1, dl-tyrosine 0·2 g l−1, l-trytophan 0·2 g l−1, dl-valine 0·2 g l−1). The pH of the medium was adjusted to 6·6–6·8 with a 4 mol l−1 NaOH solution.

Table 1.   Species and source of 19 Bifidobacterium strains screened for the ability to produce folate
Collection of bifidobacteria*ATCC/DSMZ†
  1. T, type strain.

  2. *Received from Bologna University Scardovi Collection of Bifidobacteria.

  3. †Deposited in the collection.

Bifidobacterium adolescentis ATCC 15703TFaeces of adultReuter (1963), Scardovi et al. (1971)
 ATCC 15706Faeces of adultReuter (1963), Scardovi et al. (1971)
B 5005 Faeces of adultBiavati et al. (1986)
Bifidobacterium bifidumB 1760 Faeces of infantScardovi et al. (1979)
B 2009 Faeces of infantScardovi et al. (1979)
B 2531 Faeces of infantScardovi et al. (1979)
Bifidobacterium breveB 1501 Faeces of infantScardovi et al. (1979)
Bifidobacterium catenulatum ATCC 27539TFaeces of adultScardovi and Crociani (1974)
B 2130 Faeces of infantScardovi et al. (1979)
Bifidobacterium longum ssp. infantisB 1860 Faeces of infantScardovi et al. (1979)
B 1954 Faeces of infantScardovi et al. (1979)
Bifidobacterium longum ssp. longumB 1990 Faeces of infantScardovi et al. (1979)
B 2160 Faeces of infantScardovi et al. (1979)
B 2327 Faeces of infantScardovi et al. (1979)
Bifidobacterium pseudocatenulatumB 1280 Faeces of infantScardovi et al. (1979)
Bifidobacterium animalis ssp. animalis ATCC 25527TFaeces of ratScardovi and Trovatelli (1974), Masco et al. (2004)
Bifidobacterium animalis ssp. lactis DSMZ 10140TFermented milkMeile et al. (1997), Masco et al. (2004)
P 17 Faeces of chickenScardovi et al. (1979)
Ra 23 Faeces of rabbitScardovi et al. (1979)

Both media were autoclaved at 121°C for 15 min. Trace metals and glucose were autoclaved separately. Ascorbic acid and the different solutions of minerals, vitamins and amino acids were filter sterilized through a 0·22-μm-pore-size filter.

Glucose, sodium acetate, ammonium sulfate and magnesium chloride-hexahydrate were purchased from Scharlau Chemie, S.A. (Barcelona, Spain). Phytone and yeast extract were purchased from Becton Dickinson and Company (Franklin Lakes, NJ, USA). All other chemicals were obtained from Sigma-Aldrich (Stockholm, Sweden).

Culture conditions

For the screening in FFM, all the strains were subcultured in broth at 37°C anaerobically (Anaerocult A; Merck). Therefore, the strains able to grow for five completed growth cycles were subjected to the analysis.

The experiments were conducted in both media, at least in duplicate. Bifidobacterial precultures were cultivated anaerobically (Anaerocult A; Merck), overnight, at 37°C in the respective broth. The steps of preparation of medium, bacterial inoculation and cultivation were performed using a Hungate technique (Hungate 1969) modified according to the methods as described by Talwalkar et al. (2001). The medium contained in 15-ml pirex tubes was insufflated with sterile nitrogen gas during boiling to create anaerobic conditions. To preserve anoxic condition, the tubes were sealed with a rubber stopper immediately after the cooling process. Resazurin (2 mg l−1; Sigma-Aldrich) was added into the medium as redox indicator dye.

Once tubes were ready, precultures were inoculated with a sterile syringe to an initial optical density (OD610) of 0·2 and incubated at 37°C. Bifidobacterial cultures were collected in the late exponential phase of growth. This stage was normally reached at OD about 1·0 (on average 6–8 h) (Lin and Young 2000). For Bifidobacterium breve B 1501 and Bifidobacterium longum ssp. longum B 2160, we assessed folates as a function of time after inoculation in FFM batch culture. From the moment bifidobacteria reached OD = 1·0, samples were collected at intervals for HPLC analysis.

The bacterial cells were collected by centrifugation (6000 g, 4°C, 15 min) and washed twice with cold 0·9% NaCl. The pellet was stored in the freezer (−80°C) and, when deeply frozen, freeze-dried for 2 days.

Folate analysis by HPLC

Concentrations of intracellular folate were determined by a validated high-performance liquid chromatography (HPLC) method (Patring et al. 2005). During folates extraction, samples were protected from light, extracted under a nitrogen atmosphere and stored on ice. The samples were analysed in duplicates. Cell extracts were prepared as follows: 0·025 g of freeze-dried cells were added to 25 ml of a freshly prepared 0·1 mol l−1 phosphate buffer (pH 6·1) containing 2% ascorbic acid and 0·1% 2,3-dimercapto-1-propanol (Patring et al. 2005). The suspended cells were boiled for 12 min in a water bath and cooled on ice. The supernatant was recovered by centrifugation (27 000 g, 15 min, 4°C) and stored in the freezer (−80°C) until deconjugation. Rat serum (Scanbur, Uppsala, Sweden) was dialysed in 0·1 mol l−1 phosphate buffer containing 0·1% 2,3-dimercapto-1-propanol, at 4°C during stirring in dialysis tube (cut off 12 000–14 000 Da) for 3 h. The buffer was changed three times. Deconjugation of folate polyglutamates to monoglutamates was performed by adding 50 μl of the dialysed rat serum to 1 ml of extracted sample in a glass tube. This solution was incubated on a shaking water bath at 37°C for 3 h. Rat serum deconjugase enzymes were inactivated by boiling the extracts in a water bath for 5 min. After cooling on ice, the samples were centrifuged (27 000 g, 10 min, 4°C) and supernatants were analysed by HPLC.

Quantification of folates in TPY was performed in 10 ml of freeze-dried medium. The methodology used for the extraction was the same as described above. The supernatant was purified with SAX-HPLC column prior to the HPLC analysis.

(6S)-5,6,7,8-Tetrahydrofolate sodium salt (H4folate) and (6S)-5-CH3-5,6,7,8-tetrahydrofolate sodium salt (5-CH3-H4folate) were used as references for HPLC (Merck Eprova AG, Schaffhausen, Switzerland). Folic acid was added as a reference when performing quantification of folates in TPY. The purity of all standards was checked according to the procedure of Van den Berg et al. (1994) using molar extinction coefficients reported by Eitenmiller and Landen (1999). The concentration of all standard stock solutions was corrected for purity.

The HPLC system consisted of a gradient quaternary pump (Jasco PU-2089 plus; Jasco, Mölndal, Sweden), a cooled autosampler (8°C) (Jasco AS-2057 plus), a UV detector (Chrompack) and a fluorescence (Jasco FP-920). Individual forms of folate were detected by UV (290 nm) and fluorescence detector (excitation 290 nm, emission 360 nm), and the software jasco chrompass was used as for controlling the HPLC system and for processing the data. The analytical column was Aquasil C18 150 mm × 4·6 mm, 3 μm (Thermo Electron Corp., Västra Frölunda, Sweden) and the mobile phase consisted of 30 mmol l−1 phosphate buffer (pH 2·3) and acetonitrile. The acetonitrile gradient started at 6% for 5 min, thereafter increasing linearly to 25% in 20 min followed by an increase to 45% in 5 min, which was kept for another 5 min, finally back to the 6% in 1 min. The injection volume was 20 μl, and flow rate was 0·4 ml min−1.

Statistical analysis

We used GLM ancova in spss 15.5 (SPSS Inc., Chicago, IL) to test for effects of the culture medium (TPY or FFM, fixed factor) and strain (random factor) on the dependent variable folate content (total, 5-CH3-H4folate or H4folate). Growth rate was added as a covariate potentially affecting folate content. To increase statistical power, higher order interaction effects were removed from the model when not significant. All continuous variables were tested for normal distribution using Kolmogorov–Smirnov tests, and no variable deviated from expectations. Differences were considered significant at < 0·05.

Throughout the manuscript, bacterial folate content is expressed as mean ± the minimum and maximum values from two or three independent experiments ((MAX–MIN)/2).


Folate content in bifidobacteria cultured in TPY

The results of the screening in complex medium are presented in Fig. 1, and details about the including strains are described in Table 1. The data show two main things: (i) the content of folate in bifidobacteria is highly strain dependent (Table 2a, b and c) and varies extensively between strains, and (ii) the folate form 5-CH3-H4folate dominates in most strains.

Figure 1.

 Intracellular biomass-specific folate content and composition in different bifidobacteria cultured in TPY. The strains were harvested at late exponential phase (c. OD = 1). Error bars indicate the minimum and maximum values from two or three independent experiments (mean ± (MAX–MIN)/2). Samples from each experiment were independently extracted and analysed twice by HPLC. (inline image) Total; (inline image) 5-CH3-H4folate and (□) H4folate.

Table 2. ancova results showing the association of quantities of different folate forms (total, H4folate, 5-CH3-H4folate) with culturing medium, bifidobacteria strain and growth rate
  1. *df, degrees of freedom and critical value α = 0·05.

  2. F, anova test statistic.

(a) Total folateMedium1,3539·88<0·001
Growth rate1,356·470·016
(b) H4folateMedium1,3557·02<0·001
Growth rate1,350·260·617
(c) 5-CH3-H4folateMedium1,3512·550·001
Growth rate1,359·550·004

Bifidobacterium adolescentis ATCC 15703 and Bif. longum ssp. longum B 1990 contained the highest amounts of total folate: 6415 (±1225) and 6015 (±860) μg per 100 g dry matter (DM), respectively. In particular, eight strains were far below, showing a folate content at least four times lower than the highest value found. The lowest levels of folate were found in strains isolated from animals and from fermented milk: Bifidobacterium animalis ssp. lactis DSMZ 10140, P 17, Ra 23 and Bif. animalis ssp. animalis ATCC 25527. More precisely, these four strains contained total folates ranging from 220 (±20) μg per 100 g DM (Bif. animalis ssp. animalis ATCC 25527) to 385 (±60) μg per 100 g DM (Bif. animalis ssp. lactis DSMZ 10140). This means a 29-fold difference between the highest and the lowest folate content per unit dry bacterial biomass in our selected strains, when screened in TPY. The medium itself, TPY, was found to contain 265 ng of folic acid mL−1.

Folate content in bifidobacteria cultured in FFM

The chemical environment is likely to influence the cellular folate level in bifidobacteria. To initiate such studies, we also performed a screening in a synthetic defined medium (FFM). In addition to generally reduced complexity (e.g. less growth factors), we excluded folic acid from the formula of FFM to study autotrophy for folate in our bifidobacteria strains. The HPLC data from the second screening are shown in Fig. 2 and illustrates that in FFM, the contents of total folate were generally higher than in TPY; FFM yielded significantly higher content of both H4folate and 5-CH3-H4folate compared with same strains in TPY (Table 2b and c). Moreover, folate levels were again strain dependent and with large variations between strains.

Figure 2.

 Intracellular biomass-specific folate content and composition in different bifidobacteria cultured in folate-free medium. The strains were harvested at late exponential phase (c. OD = 1). Error bars indicate the minimum and maximum values from two or three independent experiments (mean ± (MAX–MIN)/2). Samples from each experiment were independently extracted and analysed twice by HPLC. (inline image) Total; (inline image) 5-CH3-H4folate and (□) H4folate.

However, in FFM, only ten strains, all of human origins, were able to grow, suggesting inability of the remaining strains to synthesize folate. Six strains contained in FFM a total folate level above 7000 μg per 100 g DM. In particular, Bif. adolescentis ATCC 15703 and Bifidobacterium catenulatum ATCC 27539 showed a very high cellular folate levels in FFM: 8865 (±355) μg per 100 g DM and 9295 (±750) μg per 100 g DM of total folates, respectively. Compared to the other strains, Bif. breve B 1501, Bif. longum ssp. longum B 2160 and Bif. longum ssp. infantis B 1860 contained low levels of total folate in both media.

Composition of folate forms

A validated method for extracting, deconjugating and analysing folates from microbial biomass was used (Patring et al. 2005). In addition to the main forms – H4-folate and 5-CH3-H4folate – it has also been shown that it is possible to determine the less abundant 5-HCO-H4folate from microbial biomass (Patring et al. 2006). In bifidobacteria, however, 5-HCO-H4-folate was not detectable, which means that it was below the limit of quantification by our method (LOQ: 840 μg per 100 g dry biomass for 5-HCO-H4-folate; 53 and 20 μg per 100 g for H4-folate and 5-CH3-H4folate, respectively). The relative levels of the main folate forms H4folate and 5-CH3-H4folate were found to vary largely between strains in both media (Figs 1 and 2; Table 2a, b and c). Hence, composition of folate forms as well as total folate content is a strain-specific characteristic. The predominant folate form found in the majority of the strains cultivated in TPY was 5-CH3-H4folate. In four strains, Bif. catenulatum B 2130, Bif. longum ssp. infantis B 1860, Bif. longum ssp. longum B 2160 and Bif. breve B 1501, more H4folate than 5-CH3-H4folate was found (Fig. 1). Common for those four strains was relatively low total folate content. Finally, 5-CH3-H4folate was the only, by our method, detectable form in bifidobacteria of nonhuman origin: Bif. animalis ssp. lactis DSMZ 10140, P 17, Ra 23 and Bif. animalis ssp. animalis ATCC 25527. Those bifidobacteria also contained the lowest amount of total folate (Fig. 1).

In the synthetic medium FFM, 5-CH3-H4folate was again the predominant folate form in the majority of the tested strains (all from human origin) (Fig. 2). As for TPY, strains containing a low level of total folate, such as Bif. breve B 1501, had relatively more H4folate than 5-CH3-H4folate. This observation – correlation between low total folate content and relatively higher H4folate – was true for both media. At this stage, it was not clear whether this was a result of slow growth rate at the time for harvest or reflected inherent differences in folate metabolism. To more precisely monitor the relative levels of main forms, we assessed folates at intervals in the low-folate strains Bif. breve B 1501 and Bif. longum ssp. longum B 2160 in FFM as a function of time after inoculation. The results show for both strains no change in 5-CH3-H4folate during the progress of the batch cultivations (Fig. 3). The levels of H4folate however varied considerably throughout the growth period; a continuous decline from the highest level in the early exponential phase to the last sample approaching stationary phase was found. It is clear that the dominating as well as dynamic pool of folates in Bif. breve B 1501 and Bif. longum ssp. longum B 2160 is H4folate, whereas 5-CH3-H4folate was low, stable and independent of growth stage. This is in contrast to the vast majority of strains in which 5-CH3-H4folate markedly dominated in all repetitions (irrespective of slight variations in cell state at harvest).

Figure 3.

 Intracellular folate content and growth in folate-free medium of Bifidobacterium longum ssp. longum B 2160 (a) and Bifidobacterium breve B 1501 (b) during controlled batch cultivation. (inline image) H4folate; (inline image) 5-CH3-H4folate; (inline image) Total and (inline image) OD.


Bifidobacteria are much used in foods such as fermented milk products and constitute a main part of the intestinal flora of humans and many animals (Biavati and Mattarelli 2006). Potentially, selected bifidobacteria may be used to either raise the level of folate in certain foods or as folate-trophic probiotics, that is, in vivo folate synthesis in the intestine, as shown possible in rats (Pompei et al. 2007b) and humans (Strozzi and Mogna 2008). However, success in any such application requires fundamental knowledge about folates in different bifidobacteria as well as understanding of the impact by the environment on cellular folates.

Our selection criteria for the screening were as follows: (i) strains representative for the human bifidobacteria biota, that is, prevalence in the gut of human individuals (Biavati et al. 1986; Gueimonde et al. 2004), (ii) animal strains with strong capacity to colonize the bowel of the host, shown in previous animal and human feeding trials (Sanders 2006; Modesto et al. 2009) and (iii) a strain typical for fermented milk products. The screening results show that bifidobacteria folate content can be high. The majority of strains had intracellular folate levels above 4000 μg per 100 g DM in both media, and the highest value of 9295 μg per 100 g DM was found in Bif. catenulatum ATCC 27539. This is fairly equal to the levels found in different strains of Saccharomyces cerevisiae cultured in a synthetic folate free medium (Hjortmo et al. 2005), ranging from 4000 to 14 500 μg per 100 g dry biomass. Yeasts are well known to contribute to the folate content in foods such as bread (Hjortmo et al. 2005) and kefir (Patring et al. 2006). Our data suggest that certain bifidobacteria may contribute to the folate intake at a level similar to yeast.

Sequenced strains of Bif. animalis ssp. lactis have previously been found to lack all the genes for the biosynthesis of 6-hydroxymethyl-7,8-dihydropterin pyrophosphate (DHPPP), a precursor for folate synthesis and thus should be auxotrophic for folates or 7,8-dihydropteroate (DHP), even in the presence of pABA (KEEG, 2011; for a review see Rossi et al. 2011). In our study, strains of Bif. animalis ssp. lactis as well as ssp. animalis were found to contain a very low-folate level in TPY; Bif. animalis ssp. lactis DSMZ 10140, for instance, had a level of total folate of 385 (±65) μg per 100 g DM, which is roughly 17 times lower than the maximum value in TPY found in Bif. adolescentis ATCC 15703 (6415 μg per 100 g DM). These low-folate strains did not grow slower than strains with high folate content and are therefore most likely not limited by folate or other compounds in TPY (found to contain 265 ng folic acid ml−1). In FFM on the other hand, the strains of Bif. animalis did not grow at all, suggesting absence of functioning folate biosynthesis. Our data further suggest that folate auxotrophic bifidobacteria, when given folates from the medium, keeps a lower folate level than the folate autotrophic bifidobacteria able to synthesize folates de novo. A practical consequence is that Bif. animalis ssp. lactis DSMZ 10140 was found to be not suitable for folate biofortification.

The level of folates and composition were strain dependent in both media, rather than species specific. The exception is Bif. animalis spp., which generally showed low-folate levels. Of the nine species analysed, there was no indication that a specific species was superior in folate content per unit biomass. The defined medium resulted in higher folate content, as shown also for yeasts (Hjortmo et al. 2008), which indicates a reduced requirement for folates when compounds normally needing folate for biosynthesis are supplied in the medium as in TPY. Bifidobacterium adolescentis ATCC 15703, Bif. longum ssp. longum B 1990, Bif. catenulatum ATCC 27539, Bifidobacterium pseudocatenulatum B 1280 contained high levels of folate in both media. Interestingly, all strains of Bifidobacterium bifidum (B 2009, B 1760, B 2531) were able to grow in FFM, whereas for the other species some strains failed to grow without supplementing synthetic folate. This suggests that the folate biosynthetic pathway may have been lost during the evolution of some strains.

The pronounced difference in folate levels between the two media shows a dynamic response to the environment. As TPY contains for instance peptone and yeast extract, many compounds may have reduced the folate levels. Peptone dramatically reduces the level 5-CH3-H4folate in yeast cells when added to a synthetic medium (Hjortmo et al. 2008), suggesting reduced folate requirement in the presence of amino acids. On the other hand, glycine and serine (both closely involved in the 1C metabolism in which folates are central) both raised the level of cellular folates in yeast when added to a synthetic medium. Moreover, Pompei et al. (2007a) found that supplementation of folate (50 ng ml−1) reduced the net folate production in some strains of Bif. adolescentis. If understood, specific medium components can be used to increase cellular folate content and even specific folate forms.

All strains showing growth in FFM were of human origin, whereas the animal strains and the strain from fermented dairy products were found auxotrophic for folate. The reason for this is unclear and currently under investigation using, for instance, different animals with different feeding patterns.

Bifidobacterium longum ssp. infantis B 1860, Bif. longum ssp. longum B 2160 and Bif. breve B 1501 differed from the trend by showing a lower level of 5-CH3-H4folate compared to H4folate. The contents of H4folate were higher in the early exponential phase and declined successively towards the stationary phase, whereas 5-CH3-H4-folate remained constant. A practical consequence of the decline in folates as a culture ages is that this must be considered when producing a folate-containing food. Interestingly, these strains generally grew slower compared to the other strains and reached a relatively higher final OD value (data not shown). The relation between growth rate, folate content and composition in bifidobacteria is not understood. It is clear however that physiological state impacts on the specific folate content of a cell, which potentially can be used.

More research is needed to fully understand bioavailability and stability of specific microbial folate forms. In addition to form, the degree of glutamate conjugation may affect bioavailability. There are however conflicting data on this (McKillop et al. 2006), and conjugation has not been studied in bifidobacteria. Folates are to various degree sensitive to heat, light and oxygen, which may lead to degradation and interconversion between forms. Of the two forms present in bifidobacteria, the most unstable is H4folate, suggesting that high 5-CH3-H4folate is advantageous (O’Broin et al. 1975). Furthermore, an intact bacterial cell probably protects the intracellular folates from degradation. However, a probiotic colon strain must also release some of the intracellular folates, to enable uptake by the colonocytes. This requires lysis of a fraction of the intestinal population. Spontaneous lysis of bacteria has been shown to take place in the colon of mice and is likely to happen also in humans (Kotzéet al. 2011).

In conclusion, this study shows that bifidobacteria folate content and composition is dynamic and depends on medium components, physiological state and specific strain. If these factors are properly selected, high levels of folate per unit biomass can be reached and hence expected to make a difference either in foods, such as dairy products, or in the intestine as folate-trophic probiotics or natural microbiota.


This publication is part of a research work at Chalmers University of Technology, thanks to a Swedish Institute scholarship to M.R.D’A. We further acknowledge Lundgrenska Stiftelserna for support in project costs.