INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Synthesis of the three branched-chain amino acids (BCAA), leucine, isoleucine, and valine, has been studied in detail in organisms as diverse as bacteria, fungi, and plants (for reviews, see references 6, 22, and 47). A particular feature of BCAA synthesis is that several steps of this pathway are carried out by the same enzymes. Genes encoding the enzymes responsible for BCAA synthesis are often clustered, e.g., the ilvBNC-leuACBD operon of Bacillus subtilis, the ilvBNC operon of Corynebacterium glutamicum, and the ilvGMEDA operon of Escherichia coli. In Lactococcus lactis subsp. lactis, the structural genes for BCAA synthesis are present in a single operon that also contains three additional genes (Fig. 1A) (16, 19). One of these genes, aldB, encodes an -acetolactate (AL) decarboxylase (ALDC), an enzyme usually involved in the catabolic degradation of AL to acetoin in the 2,3-butanediol pathway.

View larger version (31K): [in this window] [in a new window]   FIG. 1.   Organization of the L. lactis leu-ilv-ald cluster and schematic representation of the constructions used in this work. (A) Locations of the fragments (horizontal bars) fused with the luciferase genes. The vectors used are indicated in parentheses on the right, and the designations of the final constructions are given on the left. The stars indicate the mutation inactivating P3, the vertical bar indicates the start codon of aldB (pJIM1730 to -1740), and the white arrow indicates the 5' end of the hisC gene fused or not with ilvA (pJIM1740 and pJIM1739, respectively). (B) Schematic representation of the final constructs after integration in the chromosome for JIM5470, -5152, and -5192 or carried on replicative plasmids. P1, P2, and P3 refer to promoters (represented by arrows). Pres is a constitutive promoter from the vector present downstream of repE, encoding the protein allowing plasmid replication. tetM and erm are the resistance markers from Tn916 and pAM1. luxA, luxB, and luxAB are the native lux genes from Vibrio harveyi and the fused version. Lollipops show the terminators. (C) Sequence of the intergenic ilvA-aldB region. The arrows indicate the repeats that could form potential secondary structure of the messengers initiated at P1 or P2 by the pairing of 1-2, 3-4, 5-6, and a-b, as shown in Fig. 4. The sequence in boldface is the 10 extended box ("10 ext.") and the transcriptional start of P3. The ORFs translated below the sequence encode IlvA, AldB, and (in italics) the ORF translated in pJIM1739 (OrfX). The vertical lines show the ends of the cloned fragments in the indicated plasmids.

AL is a central metabolite involved in both anabolism (synthesis of leucine and valine) and catabolism (production of acetoin). The control of its partition between the two pathways is thus of particular importance. Previous work has shown that in L. lactis, the product of aldB was responsible for (i) the degradation of AL produced during the catabolism of sugars (18) and (ii) the regulation of the pool of AL in the cell during BCAA metabolism (19). This is important since IlvBN, the first enzyme of the pathway, is not subject to feedback control in L. lactis (3). To exert a role in the metabolic regulation of BCAA biosynthesis, ALDC should be expressed during BCAA starvation. However, to avoid the dissipation of the pool of AL during the biosynthesis, the activity of this enzyme should be tightly controlled. A first level of control, at the enzymatic level, was previously described. L. lactis ALDC is activated allosterically in vitro by leucine (36). The activation of ALDC in vivo by addition of leucine in the medium induces valine starvation (19).

Despite the low activity of ALDC in the absence of leucine, its synthesis at high levels during BCAA synthesis could have a negative effect on BCAA synthesis in the cell. This consideration raises the question of a possible regulation of ALDC synthesis. Indeed, it may be advantageous for the cell to adjust the amount of ALDC in response to different environmental conditions. For instance, ALDC may be required at high levels during carbon limitation or aerobiosis, conditions that induce a shift from lactic homofermentation to mixed acid fermentation, often concomitant with high fluxes to acetoin. In contrast, ALDC may be required at low levels to regulate the AL pool during BCAA biosynthesis (19).

A previous study suggested that the transcriptional level of the specific aldB promoter is almost constitutive, regardless of the presence of BCAA (19). Moreover, aldB is transcribed from two promoters, one situated upstream of the leu-ilv operon and the other within it, since there is no terminator upstream of aldB. This feature has not been found in other microorganisms, where aldB is under the control of a promoter independent from BCAA biosynthesis genes (32, 41). The transcriptional pattern of aldB in L. lactis is thus in apparent contradiction with the previous assumption that the production of ALDC should be limited during BCAA starvation, when the leu-ilv operon is strongly transcribed.

The analysis of the sequence of the intergenic region between ilvA and aldB indicated that the mRNA could fold in a complex structure entrapping the aldB ribosome binding site (RBS) (40). It was proposed that such folding could reduce the translation of aldB produced from the upstream transcripts. Here we show that the translation of aldB from the messengers initiated upstream of its specific promoter is indeed inhibited. In addition we show that the transcription from the specific aldB promoter is tightly regulated by amino acid starvation. This suggests that the conjunction of the two mechanisms regulates ALDC synthesis to adjust optimally its level to the cell requirement in L. lactis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and media. The bacterial strains and plasmids used in this study are presented in Table 1 and Fig. 1B and C. L. lactis subsp. lactis NCDO2118 was grown at 30°C in M17 medium (46) or in chemically defined medium (CDM) (referred to as MCD in Fig. 2 and 3) (37), in which the sugar and amino acid compositions were modified as described in Results. The two amino acid analogs arginine hydroxamate and serine hydroxamate were added at 250 µg/ml in CDM without arginine and serine, respectively. E. coli and B. subtilis were grown in Luria-Bertani medium at 37°C. Transformation of L. lactis, E. coli, and B. subtilis was performed by standard procedures (1, 21, 43). When necessary, erythromycin (100 µg/ml for E. coli, 5 µg/ml for L. lactis, and 0.5 µg/ml for B. subtilis), ampicillin (50 µg/ml for E. coli), or tetracycline (10 µg/ml) was added to the medium.

DNA and RNA manipulations. Plasmids and chromosomal DNA were prepared as previously described (27, 43, 44). Southern blotting, DNA hybridization, and other molecular techniques were performed as described previously (43). DNA probes were labeled by nick translation using [-³²P]dCTP (Amersham) according to the recommendations of the supplier (Boehringer Mannheim). The sequences of all constructions were verified using a Taq Dye Primer Cycle sequencing kit and 370A sequencer (Applied Biosystems). RNA preparation and Northern hybridizations were performed as described previously (19).

Plasmids construction and mutagenesis. The relevant steps for construction of plasmids and strains are summarized in Table 1. PCR fragments were usually cloned in pBS and sequenced before their insertion in the final vector. The sites adjacent to the cloned PCR fragments in pBS were often used to facilitate the constructions. Fusion of the pBS derivatives with the appropriate vectors followed by the deletion of the pBS backbone in L. lactis was also frequently used.

Analysis of ALDC synthesis by Western blotting. The cells were recovered from the medium by centrifugation and suspended in lysis buffer (Tris [50 mM, pH 8], NaCl [0.1 M], saccharose [10%], lysozyme [1 mg/ml]) for 30 min at 4°C. The suspension was adjusted to an optical density (OD) at 600 nm of 10. Before the sample was loaded on the gel, loading buffer was added and the samples were boiled for 2 minutes. Proteins were subjected to electrophoresis in sodium dodecyl sulfate-polyacrylamide gels (24) and blotted on polyvinylidene difluoride membrane (Millipore). ALDC was revealed by immunochemiluminescence, using anti-ALDC polyclonal antibodies and ECL detection (Amersham). The antibodies were produced in rabbit after three injections of 20 µg of ALDC (36) and included in liposomes.

Luciferase assay. One milliliter of culture was mixed with 5 µl of nonaldehyde, and the light emission was measured immediately in a Bertold luminometer. Lux activity was monitored during the entire length of the growth to obtain an accurate measure of the transcriptional levels at the different growth stages. Values given in this work are those read on a plot of lux versus OD at an OD of 0.5.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Characterization of P3 elements by directed mutagenesis. The P3 promoter is specific for ald genes. Its transcriptional start was previously determined (Fig. 1C) (19). An extended 10 promoter box (TGNTATAAT) without a 35 box is present at an appropriate distance from the 5' end of this transcript. A deletion analysis of the region containing P3 was performed in order to characterize the cis elements necessary for its full activity. First, we subcloned a 0.56-kb BamHI-PstI fragment containing P3 in the promoter probe vector pJIM2367 in front of the lux genes. The plasmid obtained, pJIM2808 (Fig. 1A), produced about 2,000 klx/OD unit when introduced in L. lactis NCDO2118. This strong luciferase activity indicates that the fragment carries the elements necessary for the full activity of P3. The deletion of the region downstream of DraI in pJIM2559 and pJIM2560 (Fig. 1C) led to a loss of lux activity, as expected for the removal of P3. Deletion of the sequence upstream of HpaI in pJIM2562 did not affect the activity of P3. However, the 148-bp AseI fragment that contains the start point of transcription (Fig. 1C, pJIM2566) gave a 50-fold-reduced activity. Finally, the addition of two more bases after the AseI site restored full P3 activity (pJIM2576). This indicates that the P3 promoter overlaps the AseI site.

Regulation of P3 by global response to starvation. To study the activity of P3 in a chromosomal context without the interference of the two upstream promoters, we introduced the P3::lux fusion in the tetM gene of Tn916 in JIM5192, as shown in Fig. 2. When JIM5192 was cultivated in CDM with BCAA, the luciferase activity increased constantly during growth. In contrast, in the absence of BCAA, the activity fluctuated at low levels. The fluctuations were correlated with the growth rate. Indeed, the growth curve can be divided in four phases. During the first, upon inoculation, growth was fast, possibly due to BCAA remaining in the inoculum. During the second, growth was slow, possibly because of the latency necessary to start the BCAA synthesis. The growth resumed during the third phase and stopped when the culture reached the fourth, stationary phase. The luciferase activity had a tendency to increase during rapid growth phases (1 and 3) and to decrease during the slow growth and stationary phases (2 and 4). This suggests that P3 transcriptional activity is growth rate dependant or is low in cells starved for BCAA (during latency or stationary phase).

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Growth curves (A) and luciferase expression from a P3::lux fusion carried by Tn916 in JIM5192 (B). The cells were grown in CDM with (triangles) or without (squares) BCAA. The arrows mark the transitions between the different phases of growth. In CDM without isoleucine, leucine, and valine (ILV) the phases were as follows: 1, growth after inoculation; 2, adaptation to ILV starvation; 3, exponential growth; and 4, stationary phase. Dashed lines indicate the slopes of the curves in the different phases. Results of a representative experiment is shown, out of three that were carried out.

ALDC synthesis correlates with P3, but not P1 and P2, activity. The production of ALDC was monitored by Western blotting with polyclonal antibodies directed against the L. lactis ALDC (Fig. 3). In the presence of BCAA in the medium, the amount of ALDC was similar during the growth and stationary phases. However, in the absence of BCAA, this amount varied during growth. ALDC was detectable after inoculation and then disappeared during the lag required for adaptation to BCAA starvation. When the growth resumed, ALDC was produced again, and it then decreased to a nondetectable level after the end of the exponential growth phase. The production of ALDC thus seems to occur mainly during a phase in which the activity of P3, measured by luciferase fusion in JIM5192, increased (Fig. 2).

View larger version (17K): [in this window] [in a new window]   FIG. 3.   Western blotting with ALDC antibodies, carried out on total cell extracts from JIM5192 grown in CDM with and without BCAA, as indicated. The samples were taken during the experiment presented in Fig. 2, at the times indicated above the lanes.

Inhibition of aldB translation. The fact that ALDC is not produced when P1 and P2 messengers are abundant suggests that the production of this protein is inhibited at a posttranscriptional level. The possible messenger folding in the region of the aldB translational start was previously described and was proposed to modulate the translation of ALDC (40). Figure 4 shows a modification of the previously proposed folding, in order to take into account the sequence variability found in other L. lactis strains. Indeed, in closely related strains the important regulatory features, such as DNA motifs and secondary structures, are usually conserved, whereas variations occur at higher frequency in the other noncoding regions (9). In the present case, the previously proposed cloverleaf-like structure is not conserved, since a 10-bp insertion and several mutations that destabilize it were found in L. lactis subsp. cremoris strains. However, this region can produce a secondary structure named stem I in all cases. The end of the coding frame of ilvA, in which the 12 last codons code for four leucine residues and one valine residue, terminates in the top bulge of stem I (Fig. 1C and 4). A second structure (stem II), with a free energy of 12.4 kcal/mol, entraps the RBS and is almost invariable in all L. lactis strains, since only three changes were found in the paired region. None of these changes significantly modifies the stability of the structure, since two are compensatory changes (G-C to A-U) and the third results in a G-U pairing instead of G-C (Fig. 4).

View larger version (25K): [in this window] [in a new window]   FIG. 4.   Potential secondary structure present upstream of aldB in L. lactis subsp. lactis NCDO2118 and L. lactis subsp. cremoris MG1363. The structure presented is that of NCDO2118, and the changes found for MG1363 are indicated by arrows.  indicates deleted bases; arrows directed toward the sequence indicate insertions. The bulge of stem I (boxed by a dashed line) and the spacer between stems I and II are larger in MG1363. Two additional potential pairings at the bottom of stem II for MG1363 are indicated by dashed lines. The two base changes introduced to inactivate P3 are indicated in italics and are placed in parentheses (on the left of stem II). The base corresponding to the transcriptional start of P3 is marked with an asterisk at the top of stem II. The sequences belonging to the ilvA and aldB ORFs are shaded. The sequence of the aldB RBS is in boldface.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The dual role of ALDC in L. lactis, both in the catabolism of pyruvate and in the biosynthesis of BCAA, implies a tight control of its synthesis in order to avoid an excess of its activity during BCAA biosynthesis but ensure sufficient activity for acetoin production (19). The present study identifies different elements involved in the regulation of ALDC synthesis.

Translational control of aldB transcribed from leu-ilv promoters. The transcripts initiated at P1 and P2 in the absence of isoleucine and leucine terminate downstream of the ald genes. However, no detectable ALDC is produced from these transcripts, as shown by Western blot analysis. This observation suggests that ALDC synthesis is subject to a posttranscriptional control. The activity of a luciferase reporter gene under the control of the aldB RBS decreases up to more than 2,500-fold in the presence of upstream sequences. Messengers containing these sequences might fold into secondary structures that entrap the RBS and thus inhibit the translation of aldB (Fig. 5). Folding of messengers in the RBS region is already known to inhibit translation, for example, in the bacteriophage MS2 coat genes and several other E. coli genes (10, 11). The rate of synthesis of these proteins might be related to the fraction of mRNA molecules in which the RBS is unfolded and would thus depend on the free energy of the local secondary structures. Unexpectedly, the 100-fold inhibition due to stem II is much lower than the 30,000-fold inhibition calculated with the model proposed by de Smit and van Duin (10, 11). The difference between the calculated and the experimental values might be due to (i) a difference in the effect of secondary structures on RBS efficiency particular to L. lactis or (ii) factors interfering with the folding of stem II.

View larger version (22K): [in this window] [in a new window]   FIG. 5.   Model for the regulation of translation of aldB. The translation of the messengers initiated at P1 and P2 is inhibited by the formation of stem II. The strength of translation inhibition is enhanced by the presence of stem I and the translation of ilvA (upper panel). Messenger initiated at P3 lacks the RBS-entrapping structure, and the ribosome can always load (lower panel). The RBS is represented by an open box, and its complementary sequence in stem II is represented by a filled box; the gray lines represent the ilvA and aldB ORFs on the messengers. In brackets are presented two hypotheses to explain the enhanced inhibition due to stem I and the translation of ilvA. Under BCAA limitation, the ribosome could stall at the leucine and valine codons located at the end of ilvA (diamonds). This would allow the stabilization of stem II, which entraps the aldB RBS, possibly by the formation of a pseudoknot. In the presence of BCAA, a possible activation of aldB translation could result from the destabilization of the structure made by stem II, either by the formation of stem I or by the binding of a factor facilitating the loading of a ribosome. From our experimental data, ilvA translation would increase the stability of stem II more than 20-fold. In the presence of BCAA, the destabilization of stem II would be incomplete and would still lead to a significant decrease of translation efficiency of aldB compared to a messenger initiated at P3.

ALDC synthesis from its specific promoter, P3. The aldB information is carried not only on mRNA initiated at P1 and P2 but also on shorter messengers that start at a position corresponding to the loop of stem II (19). We confirm here the earlier hypothesis proposing that the short messengers are produced from P3, a specific promoter, rather than by the processing of the longer messenger, since (i) a 150-bp fragment containing the potential promoter and the start of transcription allowed full transcriptional activity, (ii) directed mutagenesis of P3 promoter 10 extended consensus bases inactivates the transcriptional activity of this fragment, and (iii) P1 and P2 transcripts are repressed in the presence of BCAA, whereas the intensity of the messengers is maximal in rich medium. P3 does not have a 35 consensus box but instead contains an extended TGN consensus upstream of the 10 box. A single substitution, TGATAATAT to TATATAAT, inactivates P3 to the same extent as the replacement of the first T of the conventional 10 box. The TG motif is recognized by a region of the RNA polymerase ⁷⁰ subunit conserved in all bacteria (2). Interestingly, the effect of the G-to-A mutation is much stronger in L. lactis than in E. coli, since this change decreases the activity of P3 200-fold, versus 20- and 50-fold for P_RE and galP1, respectively (23). This might reflect the greater importance of this motif, which is very frequent in gram-positive bacteria such as B. subtilis (20), Streptococcus (42), and Lactococcus (12). The complete absence of a 35 box does not impede P3 activity, the level of transcription of which is comparable to those of promoters from several highly expressed glycolytic genes (E. Jamet and P. Renault, data not shown).

Conclusion. In this work, we show that ALDC synthesis is inhibited at the translational level when it is transcribed from P1 and P2, the two promoters induced during BCAA starvation. However, P3, a third promoter, allows the synthesis of ALDC independent of this translational control. Surprisingly, the transcription of aldB from this promoter is not strongly activated when pyruvate catabolism is shifted to the acetoin-butanediol pathway as in other bacteria, such as B. subtilis, Klebsiella terrigena, and Oenococcus oeni (5, 14, 41). The transcription from P3 is reduced during lag phases, probably under the control of the stringent response and is active almost constitutively during exponential growth phases. Lastly, ALDC activity is further regulated at a posttranslational level: (i) its enzymatic activity requires the presence of leucine in the cell (19, 36), and (ii) ALDC is rapidly degraded when cells are shifted to BCAA starvation. The multiple controls for the synthesis of ALDC should allow a rapid adjustment of the amount of this enzyme, which is required at different levels in response to the environmental conditions. Indeed, during BCAA synthesis, the pool of AL could be controlled right away by ALDC, to counterbalance the absence of IlvBN retroinhibition by the products of the BCAA pathway (3, 19). During a shift to heterofermentation, the acetoin pathway will be activated without delay, since the expression of AlsS, the catabolic acetolactate synthase and the first enzyme of this pathway, is constitutive (31).