Sorbitol is claimed to have important health-promoting effects and Lactobacillus casei is a lactic acid bacterium relevant as probiotic and used as a cheese starter culture. A sorbitol-producing L. casei strain might therefore be of considerable interest in the food industry. A recombinant strain of L. casei was constructed by the integration of a d-sorbitol-6-phosphate dehydrogenase-encoding gene (gutF) in the chromosomal lactose operon (strain BL232). gutF expression in this strain followed the same regulation as that of the lac genes, that is, it was repressed by glucose and induced by lactose. 13C-nuclear magnetic resonance analysis of supernatants of BL232 resting cells demonstrated that, when pre-grown on lactose, cells were able to synthesize sorbitol from glucose. Inactivation of the l-lactate dehydrogenase gene in BL232 led to an increase in sorbitol production, suggesting that the engineered route provides an alternative pathway for NAD+ regeneration.
Lactic acid bacteria (LAB) are fermentative microorganisms that use two main sugar catabolic pathways. The Embden–Meyerhof–Parnas pathway results in almost exclusively lactic acid as end-product (homolactic fermentation). The NAD+ consumed during glycolysis is regenerated during pyruvate reduction by the L-lactate dehydrogenase (Ldh), thus maintaining the cellular redox balance. The 6-phosphogluconate/phosphoketolase pathway results in significant amounts of other end-products such as ethanol, acetate, and carbon dioxide in addition to lactic acid (heterolactic fermentation). Under certain conditions, LAB can also use alternative electron acceptors to regenerate NAD+ through suitable dehydrogenases. For instance, some heterofermentative LAB can reduce directly fructose to mannitol, by means of a mannitol dehydrogenase . In homofermentative LAB, similar reactions can take place and fructose-6-phosphate can be reduced to mannitol-1-phosphate and sorbitol-6-phosphate (Fig. 1), which are catalized by mannitol-1-phosphate dehydrogenase and sorbitol-6-phosphate dehydrogenase (S6PDH), respectively. However, their genes are clustered with those of the corresponding hexitol transporter of the phosphoenolpyruvate-dependent phosphotransferase system (PTS) family, indicating that their regular physiological role should be mediating hexitol assimilation [2,3]. This could be confirmed by the fact that hexitol accumulation has only been demonstrated in mutants lacking Ldh, where NAD+ regeneration was seriously compromised [4–7].
Lactobacillus casei is a facultative heterofermentative LAB associated with dairy products and some strains have health-promoting properties [8–12]. In L. casei, two S6PDH-encoding genes have been found [3,13]. However, sorbitol production is not likely in this bacterium during its growth on glucose or lactose, since both S6PDH genes are subject to catabolite repression and substrate induction [3,13]. The aim of this work has been the construction of sorbitol-producing L. casei strains. The gutF gene, encoding a S6PDH, was introduced in the lactose operon of L. casei and expression of the S6PDH was subject to the same regulation as lactose genes [14–16]. Production of sorbitol and other metabolites in this recombinant strain and in an L-Ldh deficient variant were determined by 13C-nuclear magnetic resonance (NMR) analysis.
2Materials and methods
2.1Bacterial strains, plasmids, primers and growth conditions
The strains, plasmids and primers used in this work are listed in Table 1. L. casei strains were grown in MRS medium (Oxoid LTD., Hampshire, England) or MRS fermentation medium (Scharlau Chemie S.A., Barcelona, Spain) supplemented with 0.5% (w/v) of different carbohydrates at 37°C under static conditions. Escherichia coli DH5α was used in all cloning experiments and was grown in Luria–Bertani (LB) medium at 37°C under agitation. When required, erythromycin was added to the L. casei growth medium at 5 μg ml−1 final concentration. Ampicillin was used at 100 μg ml−1 for E. coli.
Table 1. Strains, plasmids and oligonucleotides used in this study
2.2DNA techniques, transformation and construction of plasmids
In order to amplify the S6PDH-encoding gene, gutF, total DNA was isolated from L. casei as described before , and it was used as template in a PCR reaction with primers gutNdeI and gutEcoRI (Table 1). PCR was performed using the Expand High Fidelity PCR System (Roche Diagnostics GmbH, Penzberg, Germany), containing 0.2 mM of each dNTP and 20 pmol of each primer. The PCR product was isolated, digested with Nde I and Eco RI and cloned into plasmid pIlac (Table 1) digested with the same enzymes, giving plasmid pIlgutF. L. casei cells were transformed with this plasmid, pVBldh or pRVgut3 (Table 1) by electroporation  using a Genepulser apparatus (Bio-Rad Laboratories, Richmond, CA, USA). General recombinant DNA techniques and plasmid DNA isolation were performed by standard procedures . DNA sequencing was carried out by using the ABI PRISM dRhodamine Terminator Cycle Sequencing Ready Reaction kit and an automatic ABI 310 DNA sequencer (Applied Biosystems, Foster City, USA).
Crude extracts were prepared as described before  and S6PDH activity was determined with 2 mM sorbitol-6-phosphate and 2 mM NAD+ in 10 mM Tris–HCl, pH 8.5, as previously reported . The reduction of fructose-6-phosphate was assayed in 10 mM Tris–HCl, pH 7.5, with 0.1 mM NADH and 2 mM fructose-6-phosphate. The rate of NAD+ reduction or NADH oxidation was followed spectrophotometrically by measuring the rate of absorbance variation at 340 nm.
2.413C-nuclear magnetic resonance analysis
L. casei strains were grown in MRS fermentation medium supplemented with 0.5% lactose, cells were collected by centrifugation and suspended in 150 mM potassium phosphate buffer pH 5.5 to 10 mg (dry mass) of cells per ml. [1-13C]glucose (Sigma–Aldrich, St. Louis, USA) was added to a final concentration of 10 mM and after a 30 min incubation at 37°C the supernatants were analyzed by NMR. 13C NMR spectra were recorded at 75.47 MHz by using a Bruker Avance DPX-300 spectrometer. Acquisition parameters were: spectral width, 18 kHz; data size, 32 K; repetition delay, 1 s; 30° pulse angle (5.5 μs); number of scans, 14.000. NMR tubes with a 5 mm diameter were used and 10% (vol/vol) 2H2O was added to provide a lock signal. The 13C chemical shifts were relative to dioxane, designated at 67.19 ppm.
2.5Quantification of sorbitol and glucose
Sorbitol production and glucose consumption were measured in resting cells, which were prepared as mentioned in Section 2.4 and incubated for 2 h at 37°C with 50 mM glucose. Samples were taken, centrifuged and supernatants were stored at −20°C until they were analyzed. Sorbitol was quantified by the enzymatic method previously described  with some modifications. Sorbitol oxidation by sorbitol dehydrogenase (Roche Diagnostics GmbH, Mannheim, Germany) was determined in a reaction mixture containing 100 mM Tris–HCl, pH 9, 2 mM NAD+ and 10-fold diluted supernatant of resting cells. Glucose consumption was determined as described previously .
3.1Integration of gutF into lac operon
In order to integrate the S6PDH-encoding gene of L. casei, gutF, into the lactose operon, the integrative plasmid pIlgutF was used to transform L. casei BL155 (Fig. 2). The selection strategy for recombinant colonies was the same as the previously described for pIlac . After the second recombination event, clones that contained the gutF gene integrated between lacG and lacF could be selected as erythromycin sensitive and lactose positive colonies. One of these integrants was selected and designated BL232. The gutF gene in this strain is transcribed from the lac promoter and translated from a typical ribosomal binding site for Lactobacillus previously introduced in the vector pIlac . BL232 strain does not contain any heterologous DNA fragments and, hence, it is suitable for use in the food industry. In order to determine if the expression of gutF was following the regulatory pattern of the lac operon, S6PDH activity was measured in BL232 grown on MRS fermentation medium supplemented with ribose, lactose or glucose plus lactose. Data in Table 2 show that lactose induced S6PDH activity in BL232, but not in the wild-type strain BL23, and also that the presence of glucose repressed this activity in BL232. These results showed the functional integration of gutF within the lac operon.
Table 2. Sorbitol 6-phosphate dehydrogenase activity in L. casei strains
L. casei strain
Activity (nmoles min−1 mg protein−1)a
Stol-6P → Fru-6P
Fru-6P → Stol-6P
ND, not detected.
aResults are from at least three independent experiments.
2.67 ± 1.06
17.56 ± 8.18
1.80 ± 0.77
Glucose + Lactose
8.57 ± 4.36
3.69 ± 3.67
29.82 ± 2.98
67.4 ± 17.27
181.59 ± 0.46
206.76 ± 70.24
Glucose + Lactose
5.58 ± 3.26
5.89 ± 0.45
2.62 ± 1.16
6.59 ± 2.03
181.77 ± 11.43
221.47 ± 18.26
Glucose + Lactose
5.97 ± 0.31
36.03 ± 4.13
67.58 ± 23.26
116.97 ± 11.42
200.89 ± 40.93
Glucose + Lactose
7.65 ± 3.64
5.41 ± 3.85
3.2Sorbitol synthesis by the recombinant BL232 strain
Lactose grown cells of L. casei BL23 (wild-type) and BL232 were used to analyse by NMR the pattern of fermentation products excreted in a resting cells system after the addition of [1-13C] glucose. Consumption of glucose by lactose-grown cells is possible because, in L. casei, glucose is constitutively transported by a glucose-specific PTS (María J. Yebra, Vicente Monedero and Gaspar Pérez-Martínez, unpublished data). The 13C NMR spectra of the metabolites formed by BL23 (Fig. 3(A)) and BL232 (Fig. 3(b)) clearly showed resonances due to lactate (at 20.7 ppm), ethanol (17.5 ppm), acetate (23.7 ppm) and mannitol (63.9 ppm). A resonance peak due to sorbitol (63.1 ppm) was only observed in the supernatant of BL232 cells, due to the expression of gutF on lactose (Table 2). This would mean that S6PDH would catalyze the conversion of fructose-6-phosphate into sorbitol-6-phosphate, which then would need to be dephosphorylated and excreted (Fig. 1). Sorbitol production and glucose consumption were also biochemically quantified in resting cells of BL23 and BL232 pre-grown on ribose, lactose or lactose plus glucose. Sorbitol was below the detection level in all the supernatants of BL23 and in the supernatants of lactose plus glucose BL232 grown cells. However, conversion rates (calculated as moles of sorbitol excreted per moles of glucose consumed) of 2.4% for BL232 cells grown on lactose and 1.5% on ribose were determined (Table 3).
Table 3. Amount of sorbitol formed per mol of consumed glucose in L. casei resting cells
mol sorbitol mol glucose−1b
nd, not determined.
aResting cells were prepared as described in Section 2 from cultures grown on MRS fermentation medium with 0.5% of sugar.
bResults are from three independent experiments.
Glucose + Lactose
0.015 ± 0.001
0.024 ± 0.005
Glucose + Lactose
0.043 ± 0.008
Glucose + Lactose
0.023 ± 0.002
Glucose + Lactose
3.3Effect of ldhL inactivation on sorbitol production by BL232
To study the effect of ldhL inactivation on sorbitol production by strain BL232, the plasmid pVBldh (Table 1), containing an internal DNA fragment of ldhL gene, was integrated in the chromosome of BL232. The ldhL knockout was confirmed by PCR and sequence analysis. Then, production of sorbitol in the new mutant strain, called BL233, was analyzed by 13C NMR as for the parental strain BL232. A broader pattern of metabolites was found, including ethanol, acetoin, acetate, pyruvate and mannitol, as a consequence of pyruvate accumulation (Fig. 3(c)). Furthermore, BL233 showed a resonance peak for sorbitol about 2.5 times higher than that of BL232. This result was confirmed by quantification of sorbitol production in lactose-grown resting cells. BL233 had higher conversion rate of glucose to sorbitol (4.3%) than BL232 (2.4%) (Table 3). S6PDH activity in BL233 was quantified and it showed to be similar to that determined in BL232 (Table 2), indicating that the production of sorbitol must depend on other factors.
3.4Inactivation of the gutF gene in the gut operon did not suppress sorbitol production by BL232
In order to prove that all the effects and activities described in this work were exclusively due to the newly introduced gene (gutF), the native chromosomal copy of gutF was inactivated in BL232, using the suicide vector pRVgut3 that contained an internal DNA fragment of gutF. All integrants obtained had a sorbitol-negative phenotype and one of them, named BL234, was chosen for further experiments. Production of sorbitol by this strain was analyzed as described above, it showed a resonance pattern (Fig. 3(d)) very similar to BL232 (Fig. 3(b)) and a conversion rate of glucose to sorbitol by lactose induced cells of 2.3% (Table 3). This demonstrated that the synthesis of this polyol in BL232 is the consequence of S6PDH activity induced from the lac operon.
Sorbitol is assumed to have several beneficial effects, as a low-calorie sweetener that is poorly absorbed in the small intestine and as an alternative to sugars, since it does not contribute to the formation of dental caries. In addition, sorbitol is an excellent humectant and texturizing agent. The utilization of sorbitol-producing lactic acid bacteria might therefore be a good strategy to synthesize sorbitol in situ during fermentation of dairy products.
S6PDH is required for sorbose and sorbitol metabolism in L. casei. [3,13]. This enzyme catalyzes the oxidation of sorbitol-6-phosphate and also the reverse reaction, that is, the reduction of fructose-6-phosphate with NAD+ regeneration. In order to test this function in vivo and to determine if L. casei is able to synthesize sorbitol, we have constructed a new recombinant strain of L. casei (BL232) by the integration of a S6PDH-encoding gene (gutF) in the chromosomal lactose operon. S6PDH expression by BL232 followed the same regulation pattern to that of the lac genes. In L. casei, these genes are tightly repressed on glucose and induced on lactose, with some regulatory escape in non-PTS sugars, such as ribose [14,16,22].
When gutF was expressed in BL232, BL233 and BL234, resting cells of these strains produced sorbitol as an end product of glucose metabolism. Thus, S6PDH would catalyze the conversion of fructose-6-phosphate into sorbitol-6-phosphate, and this sugar-phosphate would be converted into sorbitol by a yet unknown enzyme. To our knowledge sorbitol-6-phosphatase activity has not been studied in bacteria. In L. plantarum, dephosphorylation of other hexitols-phosphate, such as mannitol-1-phosphate, has been proposed to occur through the mannitol-specific PTS component, EIIMtl. However, mannitol-1-phosphatase activity was determined in Lactococcus lactis and it was concluded that enzymes other than EIIMtl could be responsible for dephosphorylation of mannitol-1-phosphate [7,23].
Among the metabolites detected by NMR (Fig. 3), lactate appeared as the main product of glucose catabolism of L. casei. However, under the experimental conditions used, mannitol and ethanol were also clearly detected by NMR, which suggested that growth on lactose prompted an increase in reducing equivalents, when compared to glucose grown cells . In homofermentative LAB, mannitol could be produced through the activity of a mannitol-1-phosphate dehydrogenase that reduced fructose-6-phosphate to mannitol-1-phosphate linked to the regeneration of NAD+ (Fig. 1). In these micro-organisms, the main pathway for ethanol formation from pyruvate would involve pyruvate formate lyase and acetaldehyde/alcohol dehydrogenase, to produce ethanol with consumption of NADH excess (Fig. 1). Production of both metabolites has been well reported in lactic acid bacteria and it is due to the tight regulation of carbon fluxes by the cellular NADH/NAD+ ratio [5,25–28].
The little difference in sorbitol production between BL232 cells grown on lactose (2.4%) and on ribose (1.5%) did not match with the difference in S6PDH activities found (Table 2). These results emphasize that appropriate intracellular conditions, perhaps availability of fructose-6-phosphate precursor, NADH or other factors, are restricting sorbitol accumulation. Intriguingly, sorbitol could not be detected either in supernatants of BL232 cells actively growing in the presence of lactose or in BL232 resting cells with lactose. This may be due to a limiting pool of fructose-6-phosphate, an effect already discussed for the accumulation of mannitol in L. lactis ldh mutants .
Increasing the NADH pool by knocking out the Ldh activity is a strategy used by different groups to re-direct the glycolytic flux towards the production of mannitol in L. lactis[7,23]. We have previously described that inactivation of the ldhL gene in L. casei produced a shift from an almost homolactic conversion of glucose to a mixed-acid fermentation, probably as a way to accomplish the redox balance . In this work, we analysed the effect of the ldhL inactivation in the sorbitol synthesis by L. casei. Larger amounts of sorbitol were produced by the gutF-expressing Ldh-deficient strain (BL233) than by BL232 strain. Since S6PDH activity in BL233 was similar to that determined in BL232, it seems that ldhL inactivation could cause a higher NADH/NAD+ ratio in the cells which could be used by the S6PDH pathway.
The engineering strategy presented here led to the construction of L. casei strains with the S6PDH-encoding gene integrated in the lactose operon and subjected to its regulation. Lactose induction of gutF expression conducted to sorbitol synthesis, which might be of considerable interest in the food industry due to the claimed physiological values of this polyol. Based on this work, new metabolic strategies could be designed to enhance sorbitol production.
We are grateful to Dr. V. Monedero for critical reading of the manuscript. This work was financed by funds of the Spanish Ministry for Science and Technology (Projects with References BIO2001-01616 and AGL2004–03886). L.N. was the recipient of a fellowship from the Agrarian Faculty of Alma Mater Studiorum, Universitá di Bologna.