Positively regulated expression of the Escherichia coli araBAD promoter in Corynebacterium glutamicum


*Corresponding author. Tel.: +33 (1) 6915-5720; Fax: +33 (1) 6915-5712; E-mail: reyes@igmors.u-psud.fr


In Corynebacterium glutamicum the promoter of the araBAD Escherichia coli gene is positively regulated by both arabinose and the araC gene product, as it is the case in the natural host. If the l-arabinose inducer and an active araC gene are present, significant amounts of araBAD promoter expression take place in the absence of the E. coli CRP protein. These results show that the C. glutamicum RNA polymerase is activated by the E. coli positive regulator of transcription AraC.


For several decades, Corynebacterium glutamicum[11] has been engineered to produce large quantities of amino acids such as glutamic acid and lysine. A desirable genetic engineering goal is the identification of ways to achieve the controlled expression of specific genes, but few such systems have been described [21]. The pMB1-derived plasmid pBAD-GFPuv contains a GFPuv reporter gene (a genetically engineered variant of the green fluorescent protein of Aequorea victoria[7] transcriptionally fused to the Escherichia coli araBAD promoter. Since the ara DNA in pBAD-GFPuv also contains the araC gene, E. coli cells transformed for this plasmid express the GFPuv protein upon induction with l-arabinose. In this work we show that this fusion also express GFPuv in C. glutamicum from the E. coli araBAD promoter (ParaBAD), under positive control by the araC gene product.

2Materials and methods


Construction of plasmids is described in Fig. 1. Plasmid pBAD-GFPuv (accession number U62637) was purchased from Clontech (catalog #6078-1). Plasmids pDG1663 (accession number U46200, [8]) and pCGL0241 (accession number X99398) have been described. Plasmid pCGL0609 (accession number AF092931) is a derivative of pCGL0243 (accession number AF091268, [15]). Plasmid pCGL1091 (K. Salim, unpublished) is a derivative of the CmR plasmid pCGL0482 (accession number AF092036) that carries the nitrilase gene of Alcaligenes faecalis (accession number D13419). The construction of pCGL1385 and pCGL3017 is described in Fig. 1.

Figure 1.

Construction of the plasmids used in this work. The fragment XhoI-ClaI from pDG1663, containing SpcR (spectinomycin resistance) was moved into a XhoI-ClaI digest of pCGL0241. This places SpcR between two XbaI sites. To construct pCGL1384, the XbaI SpecR cassette was inserted into the unique XbaI site of pBAD-GFPuv. This forms a 3495-bp NsiI-Sse8387I casette that contains araC-ParaBAD, GFPuv and SpcR. To construct pCGL1385 the above cassette was integrated into the unique PstI site of pCGL0609, a replicative shuttle vector. To construct pCGL3017, a BssHII fragment of pCGL1384 that contains the amino-terminal domain of AraC, ParaBAD, GFPuv, and SpcR, was ligated to a MluI digest of pCGL0609. The ligation mixture was transformed in DH5a, in order to conserve the pACYC184 replication origin.

2.2Bacterial strains

Corynebacterium glutamicum ATCC 14752 was obtained from the American Type Culture Collection. E. coli DH5α[22] is F, hsdR (rkmk+), D(lac-argF)U169, recA1, endA1, supE44, [f80dlacDZM15], relA1, gyrA1, rpsL150, thi-1.

2.3DNA manipulation techniques

Plasmid construction techniques have been described [2]. Plasmid DNA was transformed into, and extracted from, C. glutamicum as described previously [3].

2.4Fluorescence measurements

Cells collected from stationary phase cultures (OD570=7–10) grown in ‘Basal Medium Corynebacterium Growth’ (BMCG [9]) containing kanamycin (25 μg ml−1) or chloramphenicol (6 μg ml−1). Cells were broken in 50 mM Tris-HCl pH 8.0, as described [1], and centrifuged at 13 000×g. Fluorescence in the supernatants (F) was measured in a Biolumin 960 (Molecular Dynamics) at an excitation wavelength of 405 nm and an emission wavelength of 520 nm, and calibrated with known amounts of wild-type GFP protein (purchased at Clontech, catalog #8360-1,-2), which was assumed to be 18 times less fluorescent than wild-type GFPuv [6]. In the range of our experimental F measurements, equivalent to 1.7–110 ng of GFPuv, both variables are related by the equation {LnGFPuv}=−12.174+11.079{LnF}1−2.6941{LnF}2+0.23006{LnF}3 (r= 0.999). Background F values were determined in extracts of GFPuv-less C. glutamicum control cultures made in parallel. Protein was measured as described [14]. One specific activity unit of GFPuv is defined as 1 ng of GFPuv per mg of protein.

3Results and discussion

3.1Regulated expression of GFPuv in C. glutamicum

We subcloned the araC-ParaBAD-GFPuv fragment on a plasmid replicative in C. glutamicum (pCGL1385 Fig. 1). Then, a derivative of pCGL1385 carrying a deletion of 65% of the carboxy-terminal end of the araC gene (pCGL3017) was constructed. This domain is involved in AraC binding to the regulatory araCBAD sites [12]. A related plasmid lacking both the ara and GFPuv genes (pCGL1091) was used as autofluorescence control. C. glutamicum ATCC 14752 carriers of plasmids pCGL1385, and pCGL3017 were examined for the production of the reporter gene GFPuv, both by fluorescence and immunological assays.

The fluorescence measurements (Fig. 2) indicate that GFPuv is expressed only in C. glutamicum pCGL1385 transformed cells, and that its appearance is dependent of the addition of l-arabinose to the culture. GFPuv is undetectable in l-arabinose induced C. glutamicum cells transformed with the ΔaraC plasmid pCGL3017. This was true for the three carbon sources studied, namely glucose, lactate and maltose.

Figure 2.

l-Arabinose and AraC dependence of the expression of the E. coli araBAD promoter in C. glutamicum ATCC 14752. Horizontal axis, logaritm of the external l-arabinose concentration (g per 100 ml of solution); vertical axis, GFPuv specific activity (ng of fluorescent GFPuv per mg of protein). Glucose and lactate as carbon source. White and black symbols represent independent experiments. Squares, pCGL1385 (AraC+) transformant; triangles, pCGL3017 (AraC) transformant. The carbon source concentration was 180 mM. Maltose as carbon source. White squares and crosses, pCGL1385 transformant at 180 mM maltose; white circles, pCGL1091 GFPuv-less strain at 180 mM maltose. The graph is cut to show the activity at 0% external l-arabinose. Black squares, pCGL1385 transformant at 90 mM maltose; black triangles, pCGL3017 transformant at 90 mM maltose.

Similar results (not shown) were obtained in Western blot measurements of GFP performed in glucose and lactate cultures [19]. The plasmid pCGL3017 was recovered from the C. glutamicum transformant and retransformed to the AraC+ strain E. coli DH5α. This new strain expresses normal amounts of GFPuv (data not shown), which indicates that its ParaBAD-GFPuv fusion is intact. Thus we conclude that l-arabinose-induced expression of ParaBAD in C. glutamicum depends of the presence of the araC gene.

At the same inducer (5%) and optimal carbon source concentration (180 mM, 1.5 h−1 generation time for glucose and lactate; 3 h−1 generation time for maltose), the GFPuv activities observed in lactate and maltose are respectively 2.6-fold, and 7.8-fold higher than that observed in glucose (167 units). At 90 mM maltose, the GFPuv activity is only 2.3-fold higher than in glucose. This is to be compared with the performance of an E. coli (DH5α) transformant for pCGL1385, which produces 8587 GFPuv units in conditions of maximal induction (2% glycerol, and 0.2%l-arabinose).

3.2Positive regulation in C. glutamicum

Some examples of negative regulation of E. coli promoters by its cognate repressor protein have been reported in C. glutamicum. Such is the case for the trpR, lacPZ, and l PL,R promoters [20]. This is not surprising, given the similarity of the −35 and −10 DNA consensus regions reported for the RNA polymerases of E. coli and C. glutamicum[13]. Thus, occupancy of the E. coli operators by its specific repressors could prevent C. glutamicum RNA polymerase action by mechanisms similar to those proposed for E. coli[6].

The AraC protein of E. coli regulates, among others, the promoter of the araBAD operon (ParaBAD), that encodes for enzymes of the arabinose catabolism. AraC can be described as a ParaBAD repressor in equilibrium with a ParaBAD transcriptional activator. The ligand l-arabinose shift the equilibrium from the repressor to the activator form of AraC [16], [17]. AraC thus belongs to a group of E. coli proteins termed positive regulators, that are known to exert transcription activation by acting at specific DNA sites [6], [17]. Since AraC is both a repressor and a positive activator, the question arises whether the observed GFPuv activity observed results from the unaided action of the C. glutamicum RNA polymerase on ParaBAD, once the AraC repressor activity is lifted by arabinose. This is not the case, since in the absence of AraC no significant ParaBAD expression is observed at all l-arabinose concentrations studied. This indicates that in AraC-less strains, ParaBAD is inactive towards the C. glutamicum RNA polymerase. However, in the presence of AraC, C. glutamicum expresses ParaBAD at high l-arabinose concentrations. This observation indicates that the heterologous E. coli AraC-C. glutamicum RNA polymerase interaction at ParaBAD determines positive regulation.

The observed AraC activation at ParaBAD takes place in the absence of the E. coli cyclic AMP receptor CRP protein. Though optimal expression of ParaBAD in E. coli requires both AraC and CRP, the CRP requirement is not absolute, since in its absence still a 1/20 of the optimal transcription levels is still observed. An excess of AraC alleviates the CRP requirement. This is also seen at other AraC-activated promoters [16, 17]. Specific interactions between the CRP activator and the E. coli RNA polymerase may intervene in AraC positive regulation [17, 23].

On the other hand, the C. glutamicum RNA polymerase may not require CRP in order to transcribe CRP-dependant E. coli promoters, since some lacPO region mutations that permit the expression of the E. coli lac promoter in C. glutamicum are located outside the CRP binding site [4].

The following observations suggest that AraC availability is not the limiting factor for ParaBAD activation in C. glutamicum.

First, our best result for ParaBAD expression in C. glutamicum is still 6.5-fold under that observed in E. coli. However, this is an appreciable level of expression, taking into consideration that in E. coli, ParaBAD should show 20-fold less activity in the absence of CRP [16, 17]. Thus, the ParaBAD activity observed in C. glutamicum is at least equal to that expected in the absence of CRP. This suggests that the araC gene is expressed at physiological levels in C. glutamicum.

Second, though full ParaBAD induction in E. coli is reached at 0.03%l-arabinose [10], the maximal C. glutamicum ParaBAD activity level was not reached even at 3%l-arabinose. This indicates that the C. glutamicum ParaBAD expression is limited by the inducer availability, and not by the disposability of AraC.

The preceding observations suggest that C. glutamicum permeability toward l-arabinose is low. This is consistent with the fact that C. glutamicum does not use l-arabinose as a unique carbon source. Though d-arabinogalactans are a major component of the cell envelope [5], no pathway involving l-arabinosides has been described in C. glutamicum. Some l-arabinose may enter C. glutamicum through a low-affinity uptake system, or diffuse through aqueous channels [4], [20]. The proportionality between the l-arabinose dose and ParaBAD response may be explained by l-arabinose permeability variations within the C. glutamicum population [18].

In E. coli, cultivation in glucose abolishes ParaBAD expression [10], due to catabolic repression. However, C. glutamicum shows significant albeit reduced levels of ParaBAD expression when cultured in glucose. We don't know the cause of the variation of ParaBAD efficiency in different carbon sources, though at least in the case of glucose and maltose we can exclude a large variation in plasmid copy number (data not shown). The possibility of modulate the levels of expression of ParaBAD in C. glutamicum through the nature of the carbon source and its concentration make this promoter an useful research tool.

Our work raises the question of whether the C. glutamicum RNA polymerase will respond to other E. coli positive regulators, for example CRP. Besides its interest as a research tool, the examination of the behavior of Gram-positive RNA polymerases toward well characterized E. coli transcription activators provides an interesting approach to the characterization of their fine functional relationships with the E. coli RNA polymerase. Such approach could take advantage of the large collections of E. coli transcription activator mutants available.