Deciphering the Regulation of the Mannitol Operon Paves the Way for Efficient Production of Mannitol in Lactococcus lactis

ABSTRACT

Lactococcus lactis has great potential for high-yield production of mannitol, which has not yet been fully realized. In this study, we characterize how the mannitol genes in L. lactis are organized and regulated and use this information to establish efficient mannitol production. Although the organization of the mannitol genes in L. lactis was similar to that in other Gram-positive bacteria, mtlF and mtlD, encoding the enzyme IIA component (EIIA^mtl) of the mannitol phosphotransferase system (PTS) and the mannitol-1-phosphate dehydrogenase, respectively, were separated by a transcriptional terminator, and the mannitol genes were found to be organized in two transcriptional units: an operon comprising mtlA, encoding the enzyme IIBC component (EIIBC^mtl) of the mannitol PTS, mtlR, encoding a transcriptional activator, and mtlF, as well as a separately expressed mtlD gene. The promoters driving expression of the two transcriptional units were somewhat similar, and both contained predicted catabolite responsive element (cre) genes. The presence of carbon catabolite repression was demonstrated and was shown to be relieved in stationary-phase cells. The transcriptional activator MtlR (mtlR), in some Gram-positive bacteria, is repressed by phosphorylation by EIIA^mtl, and when we knocked out mtlF, we indeed observed enhanced expression from the two promoters, which indicated that this mechanism was in place. Finally, by overexpressing the mtlD gene and using stationary-phase cells as biocatalysts, we attained 10.1 g/liter mannitol with a 55% yield, which, to the best of our knowledge, is the highest titer ever reported for L. lactis. Summing up, the results of our study should be useful for improving the mannitol-producing capacity of this important industrial organism.

IMPORTANCE Lactococcus lactis is the most studied species of the lactic acid bacteria, and it is widely used in various food fermentations. To date, there have been several attempts to persuade L. lactis to produce mannitol, a sugar alcohol with important therapeutic and food applications. Until now, to achieve mannitol production in L. lactis with significant titer and yield, it has been necessary to introduce and express foreign genes, which precludes the use of such strains in foods, due to their recombinant status. In this study, we systematically characterize how the mannitol genes in L. lactis are regulated and demonstrate how this impacts mannitol production capability. We harnessed this information and managed to establish efficient mannitol production without introducing foreign genes.

INTRODUCTION

Lactococcus lactis is a lactic acid bacterium (LAB) with a long history of safe use in the dairy and food industries. Due to its safe status, its well-known metabolism, and the many genetic tools that have been developed for it, L. lactis is regarded as an excellent production platform for food ingredients, as well as for compounds with sensitive applications, such as therapeutics (1). Mannitol, a sugar alcohol, is an example of a compound that can be produced in many LAB, including L. lactis (2, 3). Different from heterofermentative LAB, which produce mannitol from fructose only, L. lactis and other homofermentative LAB (4–6) produce mannitol from fructose-6-phosphate (F6P), which is a common intermediate in the glycolytic pathway. This implies that L. lactis in principle can produce mannitol from all carbon sources that are metabolized via F6P, which would further broaden its application as a mannitol producer. Although mannitol production in L. lactis has been studied to some extent during the last 2 decades, many aspects of mannitol production are still unclear (7, 8).

Mannitol 1-phosphate dehydrogenase (M1PDH) is responsible for reducing F6P into mannitol 1-phosphate (M1P), which subsequently is dephosphorylated to mannitol by an uncharacterized phosphatase. The normal function of M1PDH is not to facilitate mannitol production, but rather growth on mannitol, and its encoding gene, mtlD, is located together with genes coding for the mannitol-specific phosphoenolpyruvate phosphotransferase (PEP PTS) components (mtlA and mtlF) and the gene encoding the regulator MtlR (mtlR). Although L. lactis has the metabolic capacity for generating mannitol, the relevant genes are under tight control and appear not to be expressed on glucose.

In most Gram-positive bacteria, the transcription activator MtlR controls the expression of the genes encoding M1PDH and the d-mannitol-specific PTS components (9). MtlR proteins from different Gram-positive bacteria have been studied, and they all share domains that have different regulatory functions, depending on the organism. In general, MtlR contains an N-terminal helix-turn-helix motif followed by an Mga-like domain, two PTS regulation domains (PRDs), an EIIB^Gat-like domain, and an EIIA^Mtl-like domain (Fig. 1). There are exceptions, however: for instance, MtlR in Lactobacillus casei lacks the PRD1 and EIIB^Gat-like domains. There are several regulatory sites in MtlR that can be phosphorylated by PTS sugar-specific enzyme IIs (EIIs) and HPr, which can activate or inhibit the activity of MtlR. In the presence of an efficiently metabolized PTS substrate (e.g., glucose), HPr is barely phosphorylated (10), and this prevents activation of MtlR. Another regulatory mechanism, which turns off MtlR when mannitol is absent, is the mannitol-specific PTS component EIIA, which can phosphorylate the EIIB^Gat-like domain of MtlR, leading to its inactivation (11). In addition to these forms of regulation, classical carbon catabolite repression (CCR) is also involved in the regulation of the mannitol operon (12), and together these three regulatory mechanisms probably explain why the mannitol genes normally are repressed during growth on other sugars, and why mannitol is not produced.

FIG 1.

Protein structure of the characterized mtlR genes from B. subtilis, G. stearothermophilus, and L. casei. The phosphorylation sites are indicated with the amino acid and their position.

There have been many efforts to achieve mannitol production in L. lactis. Gaspar et al. (13) reported that an L. lactis strain deficient in the lactate dehydrogenase (LDH) gene (ldh) formed small amounts of mannitol (0.7 mM) when growing on glucose; however, after glucose depletion, mannitol was metabolized. The same researchers also found that nongrowing cells resuspended in buffer were able to convert 33% of the glucose metabolized into mannitol. Wisselink et al. demonstrated that M1PDH had an impact on mannitol production, but the effect was only seen clearly in a lactate dehydrogenase-deficient strain, even under nongrowing conditions (14). By heterologous expression of the mannitol-1-phosphatase from Eimeria tenella and a M1PDH from Lactobacillus plantarum in an LDH-deficient strain, efficient mannitol production (9.0 g/liter, 50% yield) was accomplished in growing L. lactis cells for the first time (15). All reported cases of mannitol production using growing cells, involve introduction of foreign genes in L. lactis, in combination with additional modifications to metabolism. Since such strains are genetically engineered, their use in food fermentations is precluded. To overcome this problem, we recently attempted a different strategy, namely, adaptive evolution on mannitol, where the aim was to debottleneck mannitol catabolism, in the hope that this could result in more efficient production of mannitol from glucose. It was indeed possible to enhance growth on mannitol, and the mutants obtained contained a mutation upstream of mtlA, which greatly enhanced expression of the mannitol genes. By further eliminating competing NADH-consuming pathways, it was possible to achieve mannitol production from glucose with high yield (60%) and fairly high titer (6.1 g/liter) using a two-stage fermentation setup (16).

In this study, we further characterize the effects of those mutations and study the roles that MtlR and CcpA have in regulating mannitol production from glucose. Based on the knowledge accumulated, we managed to establish efficient mannitol production using stationary-phase cells and achieved, to the best of our knowledge, the highest titer and yield ever reported when relying solely on native metabolism.

(This research was conducted by H. Xiao in partial fulfillment of the requirements for a Ph.D. from the Technical University of Denmark.)

RESULTS

Organization of the mannitol genes in L. lactis.

As mentioned above, we recently reported that adaptive laboratory evolution combined with systematic elimination of NADH-consuming reactions can lead to efficient mannitol production from glucose by L. lactis. The strains obtained after adaptive evolution contained two different mutations, C-39T and G-46T, both in the mtlA promoter region, which we found were critical for mannitol production (16).

To better understand the effect of these mutations, we first examined the genetic arrangement of the mannitol genes in L. lactis and compared our findings to those previously reported for other Gram-positive bacteria. As shown in Fig. 2a, the mannitol genes appear to be organized in a similar manner in all the bacteria investigated, except for Bacillus subtilis, where the gene encoding the regulator MtlR is distantly located from the other mannitol genes. Nevertheless, we observed two distinct features for the L. lactis genes. The first one was that the spaces between mtlR and mtlF and between mtlF and mtlD appear longer in L. lactis than in other organisms. Another difference is that there is a predicted rho-independent terminator between mtlF and mtlD, suggesting that mtlD is transcribed from its own promoter. Interestingly, the gene just after mtlD, encoding a characterized 6S RNA (17), may also be affected by the mtlD upstream promoters, since no transcriptional terminator can be identified between the 6S RNA gene and mtlD.

FIG 2.

Genetic organization of the mannitol genes in L. lactis and other selected firmicutes, as well as putative promoters upstream of mtlA and mtlD in L. lactis. Arrows represent characterized or predicted promoters. The numbers beneath the genes in panel a indicate either gene length or the length of the spacers between indicated genes. 6S, 6S RNA. In panel b, putative transcriptional elements are indicated. The transcriptional start sites are indicated with green, cre sites are indicated with red, the inverse repeat sequence (i.e., the mtlR binding site) is indicated with arrows, and the −10 and −35 elements are underlined. The mutations C-39T and G-46T (the numbers indicate the distance from start codon ATG) in the mtlA promoter region gained during mannitol adaptation are indicated with red capital boldface letters.

The mtlA and mtlD upstream promoter regions in L. lactis were also analyzed, and interestingly, we found that they share more than 47% identity. Both regions have putative MtlR binding sequences and cre sites that are well conserved. Furthermore, as shown in Fig. 2b, the two mutations C-39T and G-46T were both found in the predicted putative cre site, where C-39T also overlapped with the putative −10 element region, suggesting that the mutations affect the expression of the mannitol operon.

Adaptive laboratory evolution on mannitol leads to mutations that enhance the mtlA promoter.

To assess the effect of these mutations on expression from the mtlA promoter, transcriptional fusions between the mtlA promoter and the reporter gene gusA (encoding β-glucuronidase) were constructed based on the reporter vector pTD6. As shown in Table 1, on glucose, a low promoter activity, no more than 0.1 Miller unit (MU), was observed for the wild-type promoter. In contrast, the promoters containing the C-39T or G-46T mutations were significantly stronger, in particular the promoter with the C-39T mutation, which was 40-fold stronger than the promoter containing the G-46T mutation. The β-glucuronidase activity measured depended on the host: e.g., the activity of the C-39T promoter in the mannitol-adapted strain MG1363M, having the C-39T mutation in its genome, was 52.16 MU, whereas the activity in the wild-type strain MG1363 only reached 15.6 MU. As MtlR is expressed together with mtlA and the mannitol PTS genes (one transcriptional unit), the C-39T mutation in MG1363M would result in higher expression of MtlR, which could in turn explain the higher activity of the mtlA promoter in the MG1363M background. On mannitol, as expected, we found a higher activity, and whereas the wild-type promoter reached 449 MU, the C-39T promoter activity resulted in more than 4,000 MU.

TABLE 1.

Effect of mutations in the mannitol operon promoter on expression level^a

Sugar	Host	Promoter activity (Miller units)
		PmtlA, WT	PmtlA, C-39T	PmtlA, G-46T
Glucose	MG1363	0.01 ± 0.01	15.6 ± 0.9	0.43 ± 0.01
	MG1363M	0.06 ± 0.01	52.16 ± 2.0	1.17 ± 0.11
Mannitol	MG1363M	449 ± 5	4,415 ± 56	4,183 ± 34

^aP_{mtlA, WT}, wild-type mtlA promoter; P_{mtlA, C-39T}, mtlA promoter containing the C-39T mutation; P_{mtlA, G-46T}, mtlA promoter containing the G-46T mutation. Data were collected at the exponential phase. All treatments were duplicated, and standard deviations are indicated.

Based on this information, the promoter mutations in the adapted strains appear to enhance the expression of mtlA, mtlR, and mtlF, as well as possibly mtlD. We previously found that deletion of mtlF in L. lactis had a beneficial effect on mannitol production (16), while others have found that strains lacking either mtlA or mtlF perform similarly in terms of mannitol production (13).

The mtlA and mtlD upstream promoters (P_mtlA and P_mtlD) are both activated by MtlR and drive expression of separate transcriptional units.

To investigate the mtlD upstream promoter, we inserted it into the promoter probe vector pTD6 and introduced the plasmid into AceR, the mtlR-depleted strain derived from Ace001M (Ace001 adapted on mannitol, containing the C-39T mutation in the mtlA promoter). We also introduced the previously constructed mtlA promoter-gusA fusion in the same strain. As shown in Fig. 3, on glucose in the AceR background, we found that both promoters were less active than in the parental strain, Ace001M. In Ace001M, the mannitol operon is expressed to a higher level, leading to more of the activator MtlR. These findings clearly demonstrated that MtlR was an activator of the mtlD upstream promoter. Interestingly, the promoter activities were much higher for cells in the stationary phase than in the exponential phase, and for both promoters, a large induction was observed on mannitol.

FIG 3.

Testing the effect of mtlR on expression of mannitol genes. (a) Effect on mtlD promoter expression. (b) Effect on mtlA promoter expression. WT, wild-type mtlA promoter; Mut, mtlA promoter with C-39T mutation. The carbon source is indicated by G (1% glucose) or M (1% mannitol). The host is indicated by Ace001, AceR, and Ace001M. Cells harvested in the exponential phase are indicated with red, while cells harvested in the stationary phase are indicated with blue. Experiments were carried out in duplicate, and standard deviations are indicated by error bars.

To investigate if the expression of mtlD could be affected by the mtlA upstream promoter (P_mtlA), we fused the mutated mtlA promoter (C-39T) with the mtlD promoter, including the transcriptional terminator upstream of mtlD, and introduced the construct into the reporter vector pTD6. In the Ace001M background, a low expression level was detected, only 1.82 ± 0.13 MU, which is similar to that found for the mtlD promoter alone. This demonstrates that mtlD is not part of the mtlA-mtlR-mtlF operon.

Deletion of ccpA increases expression of the mannitol genes.

The mannitol genes are subject to carbon catabolite repression, and this might have a negative effect on mannitol production from glucose. To investigate this, we deleted ccpA in Ace001, thereby generating Ace001C. As shown in Fig. 4, by deleting ccpA, the mtlA and mtlD promoters were both derepressed, and activities of 588 MU and 405 MU, respectively, were measured in the exponential phase (Table 2). Furthermore, we found that the wild-type mtlA promoter and mtlA promoter from the mannitol-adapted strains had similar activities in the ΔccpA background, demonstrating that the mutation had affected the functionality of the cre site (Fig. 4). Moreover, 2- to 3-fold higher promoter activities were detected in the stationary phase.

FIG 4.

Activities of mtlA and mtlD promoters in Ace001C. WT, wild-type mtlA promoter; Mut, mtlA promoter with C-39T mutation. Cells were grown in 1% glucose and harvested at the exponential phase (red) and stationary phase (blue). Experiments were carried out in duplicate, and standard deviations are indicated by error bars.

TABLE 2.

Overview of mtlA and mtlD promoter strength under different regulatory conditions

Sugar	Strain	Promoter type	Promoter activity (Miller units)
			Exponential	Stationary
Glucose	Ace001	mtlD	0.00 ± 0.00	0.00 ± 0.00
	AceR	mtlD	0.01 ± 0.01	0.01 ± 0.00
	Ace001M	mtlD	2.38 ± 0.10	122 ± 5
Mannitol	Ace001M	mtlD	871 ± 46	1699 ± 137
Glucose	Ace001	mtlAWT	0.03 ± 0.00	0.05 ± 0.01
	Ace001	mtlAMut	15.6 ± 1.1	40.0 ± 0.4
	AceR	mtlAMut	14.5 ± 0.9	41 ± 2
	Ace001M	mtlAMut	44.3 ± 2.2	2,273 ± 164
Mannitol	Ace001M	mtlAMut	4,078 ± 230	7,539 ± 687
Glucose	Ace001C	mtlAWT	588 ± 15.7	1,880 ± 86
	Ace001C	mtlAMut	523 ± 13.1	1,842 ± 48
	Ace001C	mtlD	405 ± 6.6	894 ± 12
	AceCF	mtlAWT	3,787 ± 232	6,055 ± 351
	AceCF	mtlAMut	3,413 ± 70	6,375 ± 308
	AceCF	mtlD	1,955 ± 65	3,664 ± 88
	AceCR	mtlAMut	39.0 ± 3.5	52.2 ± 1.5
	AceCR	mtlD	0.00 ± 0.00	0.06 ± 0.01

MtlR is needed for high-level expression of the mannitol genes, and expression is enhanced by deleting mtlF.

To investigate the extent to which MtlR affects expression of mannitol genes, we deleted ccpA in AceR, which lacks mtlR, thus generating AceCR. As shown in Fig. 5, compared with AceR (Fig. 3), deletion of ccpA (AceCR) had only a small positive effect on the mtlA promoter, and the mtlD promoter had no activity. These results clearly demonstrate that the mannitol genes in L. lactis are tightly regulated by MtlR. To investigate whether eliminating CcpA could be beneficial for mannitol production, we deleted ccpA in strain AceF, which harbors the C-39T promoter mutation and in which mtlF has been deleted: the outcome was strain AceCF. As observed earlier, without mtlF, the activities of the mtlA and mtlD promoters were dramatically increased to 3,787 and 1,955 MU, respectively, in the exponential phase and increased to 6,055 and 3,664 MU in the stationary phase (Fig. 5 and Table 2).

FIG 5.

Activities of mtlA and mtlD promoters in AceCF and AceCR. WT, wild-type mtlA promoter; Mut, mtlA promoter with C-39T mutation. Cells were grown in 1% glucose and harvested at the exponential phase (red) and stationary phase (blue). Experiments were carried out in duplicate, and standard deviations are indicated by error bars.

Inactivation of CcpA enables mannitol production in non-mannitol-adapted strains.

To quantify the effect of MtlR on the expression level of the mannitol genes, mannitol catabolism, and mannitol production from glucose, we overexpressed mtlR from a promoter of intermediate strength in the non-mannitol producer Ace001 (MG1363 Δldh ΔldhB ΔldhX Δpta ΔadhE ΔbutBA). As shown in Table 3, this enabled mannitol production from glucose. Deletion of mtlR in the mannitol producer Ace001M abolished both growth on mannitol and mannitol production from glucose, which indicates that MtlR has a key role in both mannitol catabolism and production. AceCF and AceF were both able to produce mannitol, but unable to grow on mannitol, where the latter was expected, since mtlF encodes a critical mannitol PTS component.

TABLE 3.

Effect of overexpression/deletion of mtlR in different L. lactis backgrounds

Strain	Genotype	Growth on mannitola	Mannitol productionb
Ace001	MG1363 Δldh ΔadhE Δpta ΔbutBA	−	−
Ace001M	Ace001 adapted on mannitol	+	+
AceF	Ace001M ΔmtlF	−	+
Ace001-mtlR	mtlR overexpressed in Ace001	−	+
AceR	Ace001M ΔmtlR	−	−
Ace001C	Ace001 ΔccpA	+	+
AceCF	AceF ΔccpA	−	+
AceCR	AceR ΔccpA	−	−

^aTo assess if growth is possible, we use the following criterium: the cell density attained in M17 containing mannitol should be above an OD₆₀₀ value of >0.9 after 24 h of incubation (M17 without added sugar/mannitol supports growth to approximately an OD₆₀₀ of around 0.8).

^bTo assess mannitol production, if more than 0.001 g/liter mannitol is formed, we interpret this as an ability to produce mannitol from glucose.

MtlR has no influence on mannitol-1-phosphatase activity.

In L. lactis, two activities are needed for mannitol production to occur (Fig. 6): the first is mannitol-1-phosphate dehydrogenase (mtlD), and the second is a phosphatase activity that can convert mannitol-1-phosphate into mannitol, where the latter has not been identified. The impact that MtlR has on the expression of the mannitol genes is clear; however, it is possible that MtlR could serve as an M1Pase or be able to increase the M1Pase activity in cells, as MtlR contains several phosphoryl transfer domains. To test this, we measured the M1pase activity in Ace001-mtlR and AceR. We finally detected similar M1Pase activity in Ace001-mtlR, AceR, and Ace001 (0.48, 0.49, and 0.49 mM/h, respectively). Based on these results, we concluded that MtlR cannot be responsible for the M1Pase activity in L. lactis.

FIG 6.

Scheme of mannitol production in L. lactis. PEP-PTS, phosphoenolpyruvate (PEP)-dependent glucose phosphotransferase system (PTS); F6P, fructose-6-phosphate; M1PDH, mannitol-1-phosphate 5-dehydrogenase; M1P, mannitol-1-phosphate.

Mannitol production using two-step fermentation.

As deletion of ccpA leads to increased expression of the mannitol genes, we decided to test if this could also be beneficial for mannitol production. We therefore tested AceCF and AceF in the two-step fermentation setup used previously (16). As shown in Table 4, AceCF, the strain lacking ccpA, had similar performance to its parental strain, AceF, both in terms of titer and yield for mannitol. One possible explanation is that glucose metabolism is hampered to such an extent that the beneficial effect coming from overexpressing the mannitol genes is lost. For AceR-mtlD, a strain that overexpresses mtlD and lacks MtlR, surprisingly a titer of 10.1 g/liter with a 55% yield was attained, which is, to the best of our knowledge, the highest ever reported for L. lactis. This clearly demonstrated the key role of the mannitol 1-phosphate dehydrogenase in mannitol production. Furthermore, we also tested if the strains AceCF and AceR-mtlD were able to produce mannitol under semianaerobic growth conditions, where slow diffusion of oxygen into the culture medium was allowed (no active aeration), and indeed, slow growth and mannitol production were observed.

TABLE 4.

Effects of deletion of CcpA on mannitol production

Conditiona	Strainb	Mannitol produced (g/liter)	Yield (%)	Biomass (OD600)
TSF	AceF	6.08	0.6
	AceCF	5.92	0.59
	AceR-mtlD	10.1	0.55
AG	AceF	2.18	0.41	1.38
	AceCF	0.62	0.34	0.65
	AceR-mtlD↑	0.71	0.34	0.47

^aTSF, two-step fermentation; AG, anaerobic growth. For anaerobic growth, 35 mM arginine was added, and biomass and yield were measured and calculated after 7 days of incubation.

^bIn strain AceR-mtlD, mtlD is overexpressed.

DISCUSSION

In L. lactis, mtlD encodes M1PDH, which is a key enzyme in mannitol metabolism. In this study, we found that MtlR is essential for maintaining the expression of mtlD, thus allowing for mannitol production in nongrowing or slowly growing cells.

The phosphatase responsible for dephosphorylating mannitol-1-phosphate remains to be found.

Although several studies have shown that L. lactis is able to convert M1P into mannitol, the enzyme responsible has not been identified (13, 15). It has been suggested that a mannitol-specific PTS component could be involved in this (e.g., EIIBC^mtl), as it is able to transfer a phosphoryl group between M1P and mannitol. However, Ramos and coworkers showed that neither EIIBC^mtl nor EIIA^mtl is responsible for the M1Pase activity in L. lactis (13). In this study, we show that the phosphoryl-transferring protein, MtlR, is also not responsible for the M1Pase activity. Inducer expulsion, which is a unique mechanism only found in few Gram-positive bacteria, including L. lactis, could be the explanation. Inducer expulsion is the process by which intracellular sugar phosphates are dephosphorylated and exported when preferred sugars, like glucose, are present (18–22). In our case, when cells are incubated with glucose, M1P accumulates, is dephosphorylated to mannitol, and is transported out of the cell. Interestingly, in L. lactis a 10-kDa sugar phosphatase having a broad substrate specificity was characterized and found to be responsible for inducer expulsion (23, 24). If this enzyme is the M1Pase, we probably could overcome the bottleneck in mannitol production by enhancing its expression, thus avoiding heterologous gene expression (15).

CCR is relieved when cells reach stationary phase.

CcpA is a global transcriptional regulator, which is responsible for CCR in Gram-positive bacteria. It functions as a complex with P-Ser-Hpr and exerts CCR by binding to the cre site upstream of various genes. In this study, we validated that CcpA serves as a strong repressor, preventing expression of the mannitol genes when glucose is present. We characterized several mutants adapted to growth on mannitol, and in one of these, a mutation was found in the mtlA promoter region (C-39T). In the C-39T mutant, the mtlA promoter had become more active, both on glucose and, especially, on mannitol. However, after inactivation of the ccpA gene, the C-39T promoter displayed a similar activity to the wild-type promoter. This indicates that the C-39T mutation had an influence on the functionality of the cre site in this promoter. In stationary-phase cells, the mannitol genes appeared to be upregulated around 50-fold compared to the exponential phase. This provided an explanation for why the two-step fermentation approach used previously resulted in high mannitol production in the presence of glucose, which promotes catabolite repression (16). This derepression could be due to a decreased level of P-Ser-Hpr in the cells, and it has been reported that Hpr mainly exists in the dephosphorylated form in stationary-phase cells (25). We speculate that a decreased level of CcpA could also contribute to the observed behavior; however, to clarify this, further studies are needed (e.g., where the transcriptome of stationary-phase cells is scrutinized).

Expression of a 6S RNA gene located downstream of mtlD is influenced by the mtlD promoter.

van der Meulen et al. characterized a 6S RNA gene located 105 bp downstream of mtlD (17). Interestingly, it was shown that the expression of this gene was upregulated by 3-fold after deletion of the ccpA gene, and by growing the cells on galactose or cellobiose, repression could also be relieved. Although there are a predicted promoter and a cre site upstream of this gene, there is no predicted transcriptional terminator between this gene and mtlD, suggesting that increased expression of the mtlD gene could positively affect the expression of this 6S RNA gene. In addition, it was also shown that several noncoding RNAs and protein-coding genes are regulated by this 6S RNA, and it was deduced to be active when CCR is relieved during the stationary phase and/or growth on alternative carbon sources (17).

L. lactis MtlR is different from homologs in other firmicutes.

In this study, we substantiated that MtlR is indeed a transcriptional activator, which plays an important role for the expression of mannitol genes, and we characterized a mutant mtlA promoter found in a mannitol-adapted strain that grows fast on mannitol. The mutated mtlA promoter retained an activity of around 40 MU in the absence of mannitol, which appears to be due to less carbon catabolite repression, as the mutation is located in the predicted cre site. In B. subtilis, MtlR binds to an incomplete inverse repeat sequence upstream of the mtlA promoter (11). In L. lactis, a similar inverse repeat sequence can also be found upstream of both mtlA and mtlD (Fig. 2b). Interestingly, these inverse repeat sequences are likely to form rho-independent terminators, as they contain a GC-rich harpin following by several T’s. From a protein alignment between studied MtlRs from other Gram-positive bacteria, we found that L. lactis MtlR is much more similar to the Streptococcus mutans MtlR (see Table S1 in the supplemental material). However, in one study, it was claimed that MtlR in S. mutans was not needed for expression of the mannitol genes and that this was due to a long insertion in the mtlR gene (26). Despite this, in most other studies of MtlR in firmicutes, it has been shown that MtlR indeed acts as transcriptional activator for the mannitol operon (26–31). MtlR generally consist of two parts. The first part is the helix-turn-helix Mga domain, which was first characterized in the Streptococcus pyogenes virulence gene regulator Mga, where it is responsible for binding to its DNA target. The second part consists of PRD domains and EIIA^Mtl-like domains (as well as EIIB^Gat-like domains in some firmicutes), which regulate the activity of MtlR through their phosphorylation states. In general, it has been found that the PRD1 domain has little effect on the activity of MtlR, whereas the phosphorylation status of the PRD2 domain greatly affects activity, where phosphorylation activates MtlR. Furthermore, dephosphorylation of the EIIA^Mtl-like and EIIB^Gat-like domains has positive effects on the activity of MtlR. Interestingly, the L. casei MtlR is only regulated by its EIIA^Mtl-like domain (9). By sequence alignment with other characterized MtlR proteins and domain prediction, L. lactis MtlR appears structurally similar to L. casei MtlR, which does not have PRD1 and EIIB^Gat–like domains (see Fig. S1 and S2 in the supplemental material). However, the histidine in the PRD2 domain is conserved in L. lactis MtlR. Moreover, in B. subtilis, it was shown that regulation of the mannitol operon mainly takes place via phosphorylation of MtlR and that CcpA-mediated carbon catabolite repression does not play a significant role, as expression of mannitol genes was not significantly affected in a ccpA-deficient strain (12). In contrast, in L. lactis, we have shown that CcpA is a strong repressor of the mannitol genes, while MtlR seemingly is still active in the presence of glucose. In G. stearothermophilus and L. casei, phosphorylation in the EIIA^Mtl-like domain by P∼EIIB^Mtl inhibits the activity of MtlR (9, 27). Several studies have shown that mannitol production can be enhanced by deleting mtlF, and it has been concluded that this is due to prevention of reuptake of the mannitol produced (13, 16, 32). In this study, we detected a 4-fold higher MtlR activity in an mtlF-deficient strain, where EIIB^Mtl always is in an unphosphorylated state because of the absence of its phosphoryl group donor, EIIA^Mtl (mtlF encodes EIIA^Mtl). This suggests that the EIIA^Mtl-like domain in L. lactis MtlR might be functional and that phosphorylation of the EIIA^Mtl-like domain could negatively affect the activity of L. lactis MtlR (Fig. 7). If so, deletion of the mtlF gene not only inhibits uptake of mannitol from the medium, but also enhances the M1PDH activity by activating MtlR, thus stimulating mannitol production in L. lactis.

FIG 7.

The proposed phosphoryl group transfer route in L. lactis MtlR.

Deletion of ccpA is not a good strategy for enhancing mannitol production in L. lactis.

In this study, we have shown that there is an important role for MtlR in mannitol production. However, the sole role of MtlR appears to be in activating expression of the mtlD gene. In an attempt to improve mannitol production from glucose, we deleted the ccpA gene, which resulted in high constitutive expression of the mtlD promoter. Unfortunately, we found that both titer and yield did not change when ccpA was deleted. In L. lactis, when ccpA is deleted, glycolysis is hampered (33–35). Since PTS uptake of glucose requires phosphoenolpyruvate (PEP), a glycolytic intermediate, this will have a direct effect on mannitol production, which is what we indeed observed. It appears that a more promising strategy for relieving carbon catabolite repression would be to inactivate the CcpA binding site upstream of mtlA and mtlD, without reducing the activities of these promoters, which perhaps could be accomplished by classical mutagenesis followed by screening.

In our previous study, to achieve a high mannitol titer, we relied on a two-stage production setup, where we first accumulated biomass and then subsequently used this biomass to produce mannitol (16). In this study, we found out that the success of this approach was mainly due to the relief of CCR and high expression of mtlD in stationary-phase cells. To bypass the CCR, we tested whether overexpression of mtlD, the mannitol-1-phosphat dehydrogenase, would enable anaerobic growth and mannitol production; however, this was not the case. It appears that the mannitol-1-phosphatase is a bottleneck in mannitol production, and irrespective of a high mannitol-1-phosphate dehydrogenase activity, anaerobic NAD⁺ regeneration will remain hampered as long as the phosphatase activity is limiting, and the only way to achieve a high mannitol productivity is to use dense cell suspensions. We also speculated whether the cells could be starving for ATP, since mannitol production is ATP demanding. To ameliorate the latter, we supplemented the cell suspensions with arginine, which can be metabolized via the arginine deiminase pathway, leading to ATP formation (36). This indeed had a beneficial effect: however, it was only a small one. Boosting the inherent mannitol-1-phosphatase activity (e.g., by mutagenesis) thus appears to be a first priority if a successful mannitol production in L. lactis is to be established.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Lactococcus lactis subsp. cremoris MG1363 and its derivatives were used in this study (described in Table 5). L. lactis strains were cultured in M17 broth supplemented with 1% glucose (Sigma-Aldrich) at 30°C, with shaking at 240 rpm. When needed, tetracycline was used at 5 μg/ml for selection and 2 μg/ml in physiological growth experiments. The optical density at 600 nm (OD₆₀₀) of cell suspensions was measured using a spectrophotometer (Shimadzu). Resting cells were prepared and two-step fermentation was performed as described previously (16).

TABLE 5.

Strains and plasmids used in this work

Strain or plasmid	Relevant genotypea	Source or reference
Strains
MG1363	Wild-type L. lactis subsp. cremoris	18
MG363-M	MG1363 adapted on mannitol	16
Ace001	MG1363 Δ3ldh Δpta ΔadhE ΔbutBA	40
Ace001-M	Ace001 adapted on mannitol	16
AceF	Ace001-M ΔmtlF (previously designated AceM)	16
AceR	Ace001-M ΔmtlR	This work
Ace001C	Ace001 ΔccpA	This work
AceCF	Ace001-M ΔmtlF ΔccpA	This work
AceCR	Ace001-M ΔmtlR ΔccpA	This work
Ace001-mtlR	mtlR overexpressed in Ace001, 20% TPI promoter	This work
Ace001-mtlD	mtlD overexpressed in Ace001, 20% TPI promoter	This work
MC1000	E. coli cloning host	41
DH10B	High-efficiency competent E. coli host
Plasmids
pTD6	Derivative of pAK80 containing gusA reporter gene	42
pPmtlAWT	pTD6 with wild-type mtlA promoter preceding gusA reporter gene	This work
pPmtlAmut	pTD6 with C-39T mutant mtlA promoter preceding gusA reporter gene	This work
pPmtlAmutII	pTD6 with G-46T mutant mtlA promoter preceding gusA reporter gene	This work
pPmtlD	pTD6 with mtlD promoter preceding gusA reporter gene	This work
pPmtlAD	pTD6 with fused mtlA and mtlD promoter preceding gusA reporter gene
pJET1.2	High-efficiency cloning plasmid for blunt-ended ligation	Thermo Fisher Scientific
pG+host8	E. coli/L. lactis shuttle vector, Tetr, thermosensitive replicon	43
pKmtlR	Used to knock out mtlR	This work
pKCcpA	Used to knock out ccpA	This work

^aΔ³ldh represents ΔldhX ΔldhB Δldh.

Methods and tools for bioinformatics analysis.

The information about mannitol genes in different bacteria was obtained from NCBI under the following accession numbers: NC_009004 (L. lactis MG1363), NC_000964 (Bacillus subtilis 168), U18943 (Geobacillus stearothermophilus ATCC 7954), U53868 (Clostridium acetobutylicum ATCC 824), and AE014133 (Streptococcus mutans UA159). The bioinformatics analysis was performed using different webtools: BDGP (37) and SAPPHIRE (https://sapphire.biw.kuleuven.be/) for promoter prediction, De Novo DNA (https://www.denovodna.com) for prediction of the ribosome binding site, and Arnold (http://rssf.i2bc.paris-saclay.fr/toolbox/arnold) for prediction of transcriptional terminators.

Molecular techniques.

Electrocompetent cells were prepared by the protocol of Holo and Nes (38), using GM17 medium containing either 0.5% (for strains AceCF and AceCR) or 2% glycine and 0.25 M sucrose. The medium was filtered. Electroporation was performed as previously described by Holo and Nes, using a MicroPulser electroporator (Bio-Rad) (38). Phusion DNA polymerase (Thermo Fisher Scientific) was used for PCR amplifications. The Zyppy plasmid miniprep kit (Zymo Research) was used for extraction of the plasmids from L. lactis after a pretreatment with 20 mg/ml lysozyme (7,000 U/mg; Fluka) at 37°C for 2 h. FastDigest restriction enzymes SalI, SacI, and BamHI were obtained from Thermo Fisher Scientific.

Deletion of mtlR in strain Ace001M.

To delete mtlR in strain Ace001M, pG⁺host8, a plasmid with a thermosensitive replicon, was used. Two synthetic DNA fragments, representing the upstream and downstream regions of mtlR and able to overlap, were ordered from Integrated DNA Technologies: a 985-bp fragment upstream of mtlR into which a SalI restriction site had been introduced in the mtlR distal end, and a 993-bp fragment downstream of mtlR, with a BamHI restriction site added in the mtlR distal region. After the two fragments were fused, they were inserted into pJET1.2, using the CloneJET PCR cloning kit, by blunt ligation as described in the kit’s manual and transformed into Escherichia coli DH10B. The insert was excised as a SalI/BamHI fragment and introduced into the corresponding sites in pG⁺host8 using E. coli MC1000 as the host. The resulting plasmid, designated pKmtlR (pG⁺host8/mtlR_up-mtlR_dwn), was subsequently used to delete mtlR in Ace001-M in the same manner as described previously (16). The Ace001-M derivative with deleted mtlR was designated AceR.

Construction of strains with deleted ccpA.

The same approach used to delete mtlR was taken for deletion of ccpA. In this case, a 978-bp fragment upstream of ccpA equipped with a SalI restriction site and a 984-bp fragment downstream of ccpA with a BamHI site were used. The subsequent cloning steps were identical to the ones mentioned above. To obtain the strains with ccpA deleted, we relied on the slow-growth phenotype of ccpA mutants (33), and for this reason, small colonies were selected and investigated. Deletion of ccpA was done for strains AceF, Ace001, and AceR, and the derivatives were designated AceCF, Ace001C, and AceCR, respectively.

Overexpression of mtlD and mtlR.

An expression cassette consisting of a promoter, a ribosome binding sequence, and mtlD was generated using a nested PCR approach. First, a fragment containing the entire mtlD gene was amplified using primers 5′-ATTTCGGGAGACACATCTGGC-3′ and 5′-GTGTTCTCGCTTCGCATCAG-3′. For the second PCR, a forward primer containing the 20% TPI promoter sequence (39) and a ribosome binding sequence was used, 5′-CACGCGTCGACATAGATTAGTTTATTCTTGACACTACAAGCTAAATGTGGTATAATCCCATAGATATACTAGGTAAGTAATAAAATATTCGGAGGAATTTTGAAATGAAAAAAGCAGTACATTTTGGTGCAGGAAAT-3′, together with reverse primer 5′-ATCCGGAGCTCATATTCTCTGTCTACTTGCTGTCAT-3′ (the restriction enzyme recognition sites are underlined). The same approach was used to generate an expression cassette for mtlR. First, a fragment containing the entire mtlR gene was amplified using primers 5′-AAGAAGAATTCACGGCAATA-3′ and 5′-ATCGCTGAAACATAATTTGAG-3′. For the second PCR, the following primers were used: 5′-CACGCGTCGACATAGATTAGTTTATTCTTGACACTACAAGCTAAATGTGGTATAATCCCATAGATATACTAGGTAAGTAATAAAATATTCGGAGGAATTTTGAAATGTTTTTAACAAGTCGTGAG-3′ and 5′-ATCCGCTCGAGCTAATCACCATACTGTTTAACAGC-3′ (the restriction enzyme recognition sites are underlined). All primers were ordered from IDT. The amplified fragments were digested with restriction enzymes and inserted into pTD6, and then the constructs were introduced into Ace001 and verified by Sanger sequencing (Macrogen).

Quantification of mannitol production by high-performance liquid chromatography.

For quantification of glucose and mannitol, we used a high-performance liquid chromatograph equipped with a Bio-Rad Aminex HPX-87H column (Bio-Rad) and a RI-101 detector (Shodex). The mobile phase consisted of 5 mM H₂SO₄ at a flow rate of 0.5 ml/min. The column oven temperature was set to 60°C. The samples for high-performance liquid chromatography (HPLC) analysis were filtered using 0.22-μm-pore filters (Labsolute) and loaded onto the high-performance liquid chromatograph immediately after sampling at the appropriate cell density.

M1Pase activity assay.

To detect M1Pase activity, exponentially growing culture samples were quenched in wet ice. The cells were washed with ice-cold MES (morpholineethanesulfonic acid) buffer (50 mM MES, 10 mM MgCl₂, pH 7.0), and resuspended to an OD₆₀₀ of 50.0. To permeabilize the cells, 12.5 μl 0.1% SDS and 25 μl CHCl₃ were added. After a vigorous vortexing for 10 s and equilibrium at 30°C for 5 min, an appropriate volume of M1P (Sigma-Aldrich) was added to a final concentration of 5 mM to initiate the reaction. Meanwhile, for negative controls, the same volume of MES buffer was added. After 1 h of incubation at 30°C, cells were centrifuged and the supernatant was filtered with 0.22-μm-pore filters (Labsolute). Finally, the filtered supernatant was immediately analyzed by HPLC to determine formation of mannitol. M1Pase activity was descried as the amount of mannitol (mM) formed per hour.

β-Glucuronidase assays.

Strains were cultured in M17 broth supplemented with 1% of the indicated sugar and 2 μg/ml tetracycline. Cells were harvested after reaching an OD₆₀₀ of 1.0 or in the stationary phase (OD₆₀₀ of 4.5; for strains containing plasmid, the OD₆₀₀ was 4.0) and quenched on wet ice followed by centrifugation at 4°C. Harvested cells were washed in ice-cold phosphate-buffered saline (PBS) and resuspended in ice-cold Z buffer (0.06 M Na₂HPO₄, 0.04 M NaH₂PO₄, 0.01 M KCl, 0.001 M MgSO₄, 0.05 M β-mercaptoethanol, pH 7). Cells were appropriately diluted, and the cell density was measured at OD₆₀₀. For permeabilization, 12.5 μl 0.1% SDS and 25 μl CHCl₃ were added, and the suspension was vortexed vigorously. After equilibration at 30°C for 5 min, 100 μl of 4 mg/ml p-nitrophenyl-β-d-glucoside (PNPG) was added, and the time (minutes) needed for development of a yellow color was recorded. To stop the reaction, 800 μl Na₂CO₃ was added. Absorbances at 420 nm and 550 nm were measured. Promoter strength was calculated as 1,000 × [(OD₄₂₀ − 1.75 × OD₅₅₀)/(time × OD₆₀₀ × ml sample)] and expressed in Miller units.

Two-step fermentation procedure for production of mannitol.

For production of mannitol, a two-step fermentation setup was used. First, cells were cultivated in M17 supplemented with 1% glucose at 30°C at 240 rpm until the cell density reached an OD₆₀₀ of 4.5 (for strains containing plasmid, the OD₆₀₀ was 4.0). Then the cells were shifted to shaking at 20 rpm, while at the same time, glucose and arginine were added to final concentrations of 2% and 35 mM, respectively. Samples were taken at appropriate times and loaded to the high-performance liquid chromatograph immediately to determine glucose consumption and mannitol production. Mannitol yield was calculated by the mannitol conversion ratio from glucose during the stationary period.