Implementation of the agmatine-controlled expression system for inducible gene expression in Lactococcus lactis

Background Lactococcus lactis has been safely consumed in fermented foods for millennia. This Gram-positive bacterium has now become of industrial importance as an expression host for the overproduction of lipopolysaccharide-free recombinant proteins used as food ingredients, therapeutic proteins and biotechnological enzymes. Results This paper reports an agmatine-controlled expression (ACE) system for L. lactis, comprising the lactococcal agmatine-sensor/transcriptional activator AguR and its target promoter PaguB. The usefulness and efficiency of this system was checked via the reporter gene gfp and by producing PEP (Myxococcus xanthus prolyl-endopeptidase), an enzyme of biomedical interest able to degrade the immunotoxic peptides produced during the gastrointestinal breakdown of gluten. Conclusion The ACE system developed in this work was suitable for the efficient expression of the functional recombinant proteins GFP and PEP. The expression system was tightly regulated by the agmatine concentration and allowed high protein production without leakiness.


Background
Heterologous protein production is a multi-billion dollar market of particular importance to manufacturers of biopharmaceuticals and enzymes for industrial use. Microbial production systems are often the best option for making such products given their ease of handling and high synthesis rates [1]. At present, Escherichia coli remains the first choice of host system given its high overexpression yields, ease of genetic handling, and the wealth of information available on this microorganism [2]. However, it is not without its drawbacks, such as the formation of inclusion bodies, the presence of an outer membrane that hampers secretion, its relatively complicated aerobic fermentation system, and the formation of endotoxins such as cell wall lipopolysaccharides [3]. The presence of bacterial endotoxins is one of the major concerns of regulatory agencies, and the need to add downstream steps to ensure their removal can make otherwise simple processes quite costly [4].
The Gram-positive bacterium Lactococcus lactis has emerged as an attractive alternative for the overproduction of recombinant proteins. Due to its long, safe history of use in dairy fermentations, this bacterium has been classified as a food grade microorganism 'Generally processes), and a very simple metabolism that allows for rapid growth without aeration-all properties that facilitate scaling-up [4]. Moreover, L. lactis is likely to provide a good membrane environment for the overproduction of eukaryotic proteins [6]. Indeed, a number of eukaryotic membrane transporters, yeast mitochondrial proteins and human proteins have been heterologously expressed in this host [7]. Further, L. lactis is an efficient cellular factory able to turn out recombinant viral antigens, interleukins, allergens, virulence factors, bacteriocins and enzymes [8,9]. It can even be genetically engineered to produce proteins from pathogenic species on its cell surface, and thus be used as a vector for the production and delivery of oral vaccines against HIV, cholera, malaria, human papillomavirus, stomach ulcers, tetanus and brucellosis [10][11][12][13][14][15][16][17][18].
For most of these applications, the availability of vectors that allow the cloning and expression of foreign genes is of paramount importance. Although the genetic accessibility and ease of handling of L. lactis lags far behind that of E. coli, the molecular biology techniques and genetic tools available for use with this bacterium have increased over recent years [9,19]. So far, a number of inducible expression systems regulated by environmental factors have been documented, including the chloride-inducible expression cassette [20], the zincinducible expression system [21], the lactate-inducible P170 system [22], the heat shock-inducible system [23], systems based on sugar or peptide concentration-regulated promoters, and bacteriophage-derived promoters [24,25]. However, some of these systems are less useful since they are controllable only to a limited extent, show low efficiency, or are associated with some degree of basal expression [21,25,26]. These limitations may be due to (1) the corresponding inducer being an essential nutrient or metabolite, the concentration of which in the culture cannot be fully controlled, (2) by being strongly sensitive to catabolite repression (i.e., certain sugar-inducible systems) [21,25,26], or (3) the promoter showing leaky activity. To date, the most widely used and potent gene expression system in L. lactis has been the nisin-inducible controlled expression (NICE) system. When added to the medium as an inducer, nisin binds to the membrane receptor NisK, which subsequently activates NisR by phosphorylation, and the activated NisR induces the nisin A promoter [27][28][29][30].
It has been shown that the agmatine deiminase (AGDI) cluster of L. lactis subsp. cremoris CECT 8666 (formerly GE2-14) encodes the enzymatic activities responsible for the catabolism of agmatine to putrescine [31]. Briefly, aguD codes for the agmatine/putrescine antiporter, aguA encodes agmatine deiminase, aguB encodes putrescine transcarbamylase, and aguC encodes a specific carbamate kinase [32,33]. Transcriptional analyses of these genes has shown the expression of aguB, aguD, aguA and aguC to be driven by the aguB promoter, and has confirmed the four genes to be co-transcribed as a single polycistronic mRNA [32,33]. A cre site exists in the promoter of aguB which is transcriptionally regulated by carbon catabolite repression (CCR) mediated by the catabolite control protein CcpA [32,33]. Also included in the AGDI cluster, upstream of the aguBDAC genes, is the aguR gene, which is transcribed constitutively under its own promoter (P aguR ) (Fig. 1). We previously showed that aguR is a regulatory gene encoding a transmembrane protein (AguR) that acts as a one-component signal transduction system able to sense the extracellular agmatine concentration and trigger transcriptional activation of the aguB promoter (which drives expression of the aguBDAC operon) [34].
The present work reports the adaptation of the ACE system-an inducible gene expression system that involves aguR and the P aguB promoter (the latter with its natural ribosome binding site)-to L. lactis. This was successfully tested via the production of green fluorescent protein (GFP) and Myxococcus xanthus prolyl-endopeptidase (PEP) [35,36].

Site-directed mutagenesis of the cre site of P aguB
In previous work [34] we showed that P aguB of L. lactis CECT 8666 drives expression of the aguBDAC operon in response to agmatine supplementation. Moreover, aguB-DAC expression is regulated by CCR, mediated by the catabolite control protein A (CcpA) [33]. To develop an efficient gene expression system, CCR had to be inactivated so that glucose could be used as a carbon source. This allowed high bacterial cell densities to be obtained, and in turn high recombinant protein yields. For this we introduced arbitrarily directed mutations into the cre site (sequence 5′-TGAAATCGTTCCCA-3′; the nucleotides aguR aguB aguA P 1 kb cre aguD aguC P Fig. 1 Genetic organization of the AGDI cluster of L. lactis CECT 8666. Physical map showing the cluster to be composed of five genes: aguR which encodes a transcription regulator, followed by aguB, aguD, aguA and aguC, which encode the proteins involved in the putrescine biosynthesis pathway (GenBank: AZSI00000000.1). The P aguR and P aguB promoters are shaded in grey, and the terminators indicated. The predicted transcripts are indicated below (dotted arrows). The cre site is shaded in black matching the cre consensus sequence are underlined) within P aguB (see "Methods"). The pAGDI plasmid [33], containing the P aguR -aguR-P aguB cassette fused to the reporter gene gfp (encoding green fluorescent protein [GFP]), was used to construct new plasmids to assess the effect of cre mutation on P aguB activity. Using this plasmid as a template, and the primers indicated in Table 1, three new plasmids were generated which contained specific mutations in the cre site: pAGDIcre1 (containing 10 nucleotide mutations from positions 5-14 of the cre site), pAGDIcre2 (containing 3 nucleotide mutations from positions 6-8 of the cre site) and pAGDIcre3 (containing 1 nucleotide mutation at position 5) (Fig. 2a). The 10 and 3 nucleotide mutations (plasmids pAGDIcre1 and pAG-DIcre2 respectively) both had a dramatic effect on P aguB : no activity was detected with either construct at any glucose concentration ( i.e., expression was not repressed at 120 mM glucose (Fig. 2e). This mutated promoter was therefore selected for the construction of the expression vector.

Agmatine-induced heterologous expression of gfp
To verify the usefulness of the ACE system in the expression of heterologous genes, the reporter gene gfp was cloned into the pACE vector under the control of P aguB , thus generating the plasmid pACE-gfp. L. lactis NZ9000-a strain without the AGDI cluster-was transformed with pACE-gfp, thus resulting in L. lactis pACEgfp. The presence of 10 mM agmatine in a culture of L. lactis pACE-gfp induced the expression of gfp, which was measured in terms of the fluorescence produced (7.87 arbitrary units) (Fig. 3). It should be noted that the expression of gfp was undetectable (<0.5 arbitrary units) in agmatine-uninduced cultures of L. lactis pACE-gfp. Fluorescence was also undetectable (<0.5 arbitrary units) in parallel agmatine-induced cultures of L. lactis NZ9000 harbouring the pACE vector (L. lactis pACE).

Sensitivity of the ACE system to the inducer: doseresponse curve
The production of GFP in cultures of L. lactis pACEgfp induced using a range of agmatine concentrations (between 0 and 60 mM) was analysed by whole-cell fluorescence. The ACE expression system showed great sensitivity to very low agmatine levels; a significant increase in fluorescence was seen after induction with agmatine at concentrations as low as 10 −5 mM (Fig. 3). Above this concentration, the induction levels increased in line with the agmatine concentration until a maximum induction level (fluorescence ~8 arbitrary units) was reached at 0.5 mM agmatine (no significant increases in induction were seen with concentrations of >0.5 mM). The absence of any leaky activity of the promoter P aguB , as verified by the absence of fluorescence in uninduced cultures (0 mM agmatine, Fig. 3), is remarkable.

Heterologous production of GFP using the ACE system
The efficiency of the ACE system in overexpressing recombinant protein was tested with the reporter protein GFP. The expressed His-tagged GFP protein was purified using immobilized metal affinity chromatography (IMAC). The eluted protein fractions were examined by SDS-PAGE (data not shown) and their GFP activity. Pure protein with GFP activity was obtained in fraction two Table 1 Oligonucleotides used in this study

GfR1
Cloning of gfp (R) CACACACACCTGCAGACTAGTTTTGTAGAGCTCATCCATGC (of the four obtained); the yield was 47 % (Table 2). Fluorescence was found in the soluble fraction only (data not shown).

Heterologous overproduction of M. xanthus prolyl-endopeptidase and comparison with the NICE system
To confirm the usefulness and efficiency of the ACE system, the pep gene of M. xanthus, which encodes a prolyl-endopeptidase of biomedical interest, was cloned into appropriate plasmids for introduction into L. lactis NZ9000. The resulting L. lactis NZ9000 pACE-pep was induced with different agmatine concentrations (0, 0.001, 0.01, 0.1, 0.5, 1, 2, 5, 10, 20, 40 and 60 mM) and the PEP activity assayed. No activity was detected in cultures without agmatine, but was observed even with the lowest agmatine concentration tested (0.001 mM). Above this concentration, the induction level increased with the agmatine concentration until 0.1 mM agmatine (21.04 mU mg −1 ) (no significant increase in PEP activity was obtained with concentrations of >0.1 mM) (Fig. 4). PEP activity was sought in the soluble and insoluble fractions, but was only seen in the former. For comparison, PEP was also produced using the NICE system at different nisin concentrations (0, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10 ng ml −1 ) (Fig. 4). The activity increased with the nisin concentration until 2.5 ng ml −1 . The highest specific activity obtained (15.2 mU mg −1 ) was lower than that obtained with the ACE system

Influence of agmatine on Lactococcus lactis growth
Since high induction concentrations of agmatine were tested in the present work, assays were performed to see whether these affected bacterial fitness. L. lactis cultures were grown in liquid GM17 supplemented with 0, 0.001, 0.01, 0.1, 0.5, 1, 2, 5, 10, 20, 40 or 60 mM agmatine. Similar growth curves were obtained (data not shown). Table 3 shows the OD 600 , μ max and pH values and lactate concentrations reached after 24 h of incubation. Growth slightly decreased with increasing concentration of agmatine. Although the 0.5 and 0.1 mM agmatine concentrations led to the highest GFP and PEP activities, no differences in OD 600 were observed compared to the uninduced cultures. Organic acids and sugars were analyzed by HPLC, and no differences observed in the presence or absence of agmatine (data not show). Agmatine had a weak effect on L. lactis cell viability after 24 h of exposure; optical density values suitable for protein production were obtained even at high agmatine concentrations.

Discussion
Lactococcus lactis has long been used in the food industry and has emerged as a cost-effective cellular factory for the production of proteins of interest [4]. At present, the genome sequences of several strains have been elucidated, and the genetic tools available for use with lactic acid bacteria (LAB) are ever increasing the number of fully lipopolysaccharide-free recombinant proteins that can be produced [3,37].
The control of gene expression is critical in achieving high protein yields. For example, it is essential for ensuring the conservation of energy for the production of biomass prior to the directed overproduction of the target protein, and in controlling the production of products that might be toxic to the host cell (i.e., membrane proteins, autolysins, lysis-related proteins from phages) [3,38]. Systems that allow for well controlled inducible expression, and that allow no basal expression, are essential in setting production rates. A number of inducible expression systems for L. lactis are available [20-22, 25, 39], however, the use of some of them, especially in large-scale fermentations, may be hampered by the lowlevel induction achieved, high background expression  before induction (leaky activity), and/or a lack of control over the inducer [21,25,26,39]. Such drawbacks are not suffered with the NICE system, probably the most commonly used regulated expression system for Grampositive bacteria [27][28][29]40]. However, when using a host strain other than L. lactis NZ9000, the NICE system requires the regulatory genes nisR and nisK to be supplied in trans. Thus, although the NICE system has been optimised to be used as a single-plasmid expression vector [41], its exploitation nearly always requires the integration of nisR and nisK, either in the chromosome of the host strain or via their cloning into a separate plasmid (the dual-plasmid strategy).
The present work describes an alternative system based on the aguR/P aguB cassette (the regulatory part of the AGDI cluster of L. lactis CECT 8666) for agmatinecontrolled gene expression in L. lactis using a vector that includes all the elements required. It should be noted that this system requires the expression of no additional genes supplied in trans. The developed vector relies on the regulatory transmembrane protein AguR, which responds to extracellular agmatine, and in so doing triggers the induction of gene expression via P aguB [34]. It was previously shown that P aguB of L. lactis CECT 8666 contains a cre site involved in CCR and that this is repressed by glucose concentrations of >30 mM [33]. Since higher concentrations of glucose may be required in culture media to obtain the densities of bacterial cells needed to provide high recombinant protein yields, this repression mechanism was eliminated in the present work through the introduction of a single A > T mutation at position 5 in the cre site. Larger mutations of either 3 or 10 nucleotides completely impaired the activity of the promoter, most likely by preventing some additional regulatory signal. Other authors have also shown CCR to be relieved when single mutations occur in the cre site of CCR-controlled genes in L. lactis. For example, a single mutation in the cre site of the celB promoter allows fully active transcription of the cryptic cel cluster involved in lactose utilization in L. lactis MG1363 [42]. Similarly, two single mutations in the cre site of the ptcC promoter do away with the glucoserepressor effect and allow cells to constitutively metabolize cellobiose [43].
An expression vector combining the one-component signal transduction system, i.e., aguR and the aguB promoter, followed by convenient cloning sites for introducing the gene of interest, was constructed. An important feature of the developed pACE vector is the possibility of fusing a His-tag to the protein of interest by cloning the encoding gene in frame into the NcoI-PstI sites. Histags have been used in other L. lactis expression vectors previously shown to perform efficiently in the immunodetection and purification of proteins [6,44]. We here confirm the functionality of the His-tag in purifying the GFP (protein yield 47 %).
The control of the ACE system was assessed via the expression of gfp, the reporter gene coding for GFP. Strong fluorescence was seen in the presence of ≥0.5 mM agmatine (8 arbitrary units compared to 0 in uninduced cells). Interestingly, the system was associated with no basal expression, indicating P aguB to have no leaky activity. Further, agmatine is not present in common culture media, thus allowing for tight control over the gene to be expressed. Neither is it found in milk nor any derivative dairy environment where L. lactis occurs [45]. The optimal moment of induction, which can change from one overexpressed gene to another, needs to be evaluated. In the ACE system, the inducer agmatine can be added when the culture medium is prepared. The AguR/ P aguB cassette, on which this expression system is based, is the regulatory part of the AGDI cluster of L. lactis [33] and is not active until the transition between the exponential and stationary phases is reached (5-6 h of culture) [46]. Thus, even when agmatine is supplied to the culture medium, the time of net expression would lie between 5 and 7 h after culturing began. This induction time is comparable to that associated with the NICE system (2-3 h after adding nisin). The addition of the inducer at the beginning of culturing avoids the problem of monitoring the culture's optical density to determine the optimal moment for induction. Moreover, the abolition of sampling and inducer addition steps may prevent contamination, which could have serious economic consequences in industrial protein production. As seen for nisin, agmatine affected cell viability and caused a 20 % reduction in bacterial yield. However, the final OD (>3.3) of agmatine-induced cultures was still optimal for industrial protein production. The price of the inducer is important in large scale fermentations. That of agmatine varies widely, depending on the supplier, but even the cheapest found (marketed as dietary supplement) worked properly as an inducer (data not shown). Certainly, it was much cheaper than nisin. An agmatine induction system combining AguR and P aguB of Enterococcus faecalis was earlier used to develop an expression vector suitable for the latter species [47]. However, the proposed lactococcal ACE system shows some differences to the E. faecalis system: (1) the aguR gene is in the same orientation as P aguB (reflecting the corresponding organization of the AGDI cluster in each system), (2) the highest expression rate is reached at ≥0.5 mM agmatine in the ACE system but at ≥60 mM in the E. faecalis system, and (3) the level of induction (as determined by fluorescence) is less than that achieved with the E. faecalis system.

Table 3 Effect of agmatine on growth, μ max , pH and production of lactic acid
In the present work, the performance of the ACE system for the heterologous production of the M. xanthus PEP was compared to that obtained using the NICE system, the most widely used and potent gene expression system in L. lactis [40]. The proposed system achieved higher PEP activity (circa 38 %) under similar laboratory conditions. The observed differences might be related to the characteristics of each promoter, or to the effect on signal transduction of the two-component NICE system compared to the one-component ACE system. More studies are required to understand how the agmatine signal is transduced.
In summary, the present results confirm the ACE system as an attractive candidate for high level recombinant protein production. The lactococcal aguR/P aguB system can effectively control the expression of genes in response to agmatine in L. lactis without any basal expression, and combines both the expression cassette and regulatory gene in one plasmid. This vector expands the genetic toolbox available for this species, and could be a powerful and straightforward alternative system for overexpressing proteins in lactococcal strains lacking nisR and nisK. It might also be used to complement the NICE system and be used in co-expression.

Conclusions
The present work describes the construction of a L. lactis agmatine-controlled expression system based on the aguR/P aguB cassette of the putrescine biosynthesis gene cluster. A single mutation of the cre site in P aguB abolished the CCR of this promoter. This system was assessed by expressing the reporter gene gfp, and fluorescence was found strictly dependent on the agmatine concentration added to the culture medium, with maximum induction occurring at 0.5 mM agmatine (7 arbitrary units compared to 0 in uninduced cells). An important potential benefit of this system is the lack of leaky activity associated with it, and the fact that gene expression can be tightly controlled via the addition of the appropriate concentration of agmatine. The pACE vector allowed the agmatine-inducible expression of the gene encoding M. xanthus PEP, an enzyme that can degrade the immunotoxic peptides of gluten breakdown. Moreover, enzymatic activity was greater than that obtained with the NICE expression system. The addition of a His-tag to the pACE vector renders the system suitable for protein purification and immunodetection purposes. Together, these findings suggest that the ACE expression system could be a very valuable addition to the L. lactis genetic toolbox, and offers a straightforward, alternative inducible gene expression system that to be used in functional studies and in the large-scale production of recombinant proteins.

DNA manipulation
The procedures used for DNA manipulation and recombination were essentially those described by Sambrook et al. [49]. Table 1 lists the sequences of primers used. Genetic constructs were achieved in L. lactis NZ9000. The isolation of L. lactis plasmids and total DNA, and the transformation procedures followed, were as previously described [50]. All plasmid constructs were verified by nucleotide sequencing at Macrogen Inc. (Seoul, Republic of Korea). All enzymes for DNA technology were used according to the manufacturer's specifications.

Construction of plasmids
The core of the lactococcal vector pNZ8048 [29]-which includes the replication cassette and the chloramphenicol resistance marker-was used as a starting point for the construction of the pACE vector. First, a fragment of the AGDI cluster from L. lactis CECT 8666 (Table 4), including the aguR promoter (P aguR ), the aguR gene, and the aguB promoter (P aguB ) carrying the mutation in the cre site, was PCR-amplified (using pAGDIcre3 as a template) and cloned into the BglII-NcoI sites of pNZ8048. Subsequently, a fragment including the multicloning site and a histidine tag encoding 10 consecutives histidines (His-tag) was amplified from plasmid pNZErmC [6] and cloned into the NcoI-XbaI sites of the previous construct, thus yielding vector pACE (Fig. 5). This vector offers the option to fuse in-frame the gene encoding the protein of interest to a C-terminal His-Tag by cloning the target gene into the NcoI-PstI sites. Thus, this vector could be used for immunodetection or purification of the proteins encoded by overexpressed genes. The target gene could also be cloned without the His-tag for use in functional studies.
For the heterologous expression of GFP, the gfp gene (amplified from pAG2 [34]) was inserted into the NcoI-PstI sites of pACE, thus generating the plasmid pACE-gfp. To produce PEP using the ACE system, the pep gene was PCR-amplified from pNZ8048-pep and cloned into the NcoI-XbaI sites in the pACE vector, resulting in the pACE-pep plasmid. To produce PEP using the NICE system, the pep gene was released from plasmid pUC57-pep [36] as a NcoI-XbaI fragment and cloned into the same sites in pNZ8048 under the control of the nisA promoter [29], thus generating plasmid pNZ8048-pep.

Directed mutagenesis of the cre site of P aguB
Modification of the cre site was achieved by in vitro sitedirected mutagenesis. Mutation(s) were introduced by PCR using two divergent primers (Table 1) spanning the cre site of P aguB and containing the desired mutation(s). Each primer was complementary to the opposite strand of the pAGDI vector, which was used as template to generate plasmids pAGDIcre1, pAGDIcre2 and pAGDIcre3 containing the specific mutations in the cre site. The pAGDI plasmid was first methylated with Dam methylase and S-adenosyl methionine following the manufacturer's instructions (New England Biolabs, Hertfordshire, UK). Phusion High-Fidelity DNA Polymerase (New England Biolabs) was used to amplify both plasmid strands with high fidelity. The PCR thermoycling conditions were as follows: initial denaturation (98 °C for 30 s), 32 cycles of amplification (98 °C for 10 s; 55 °C for 30 s; and 72 °C for 2.5 min) plus a final extension step (72 °C for 7 min). An EcoRI target site was included in the primers so that the obtained amplicons could be digested with EcoRI and self-ligated. Before transformation in L. lactis NZ9000, the ligation mixture was treated with DpnI order to digest the original, Dam methylated pAGDI plasmid used as a template.

Strain/plasmid Characteristics References
Strains L. lactis CECT8666 (formerly GE2-24) Isolated from an artisanal cheese, containing AGDI cluster [31] L. lactis NZ9000 Expression host for NICE system, lacking AGDI cluster [29] Plasmids pNZ8048 Lactococcal plasmid containing the nisA promoter and the Cm r [29] pNZErmC pNZ8048 derivative containing the His-tag and the Cm r [6] pUC57-pep pUC57 derivative containing the pep gene from M. Xanthus and the Ap r [36] pAGDI pNZ8048 derivative bearing the P aguR -aguR-P aguB -gfp fusion and the Cm r [34] pAG2 pNZ8048 derivative bearing the P aguB -gfp fusion and the Cm r [33]

Measurement of green fluorescence
For whole-cell fluorescence measurements, overnight cultures of L. lactis NZ9000 harbouring either pACE or pACE-gfp were transferred (1 %) to fresh medium (GM17) supplemented with different agmatine concentrations (0-60 mM) and grown for 7 h. Equal amounts of cells were harvested, washed, and then resuspended in 50 mM potassium phosphate buffer, pH 7.2, as previously described [6]. GFP emission was measured in a volume of 200 μl of cells, using a Cary Eclipse fluorescence spectrophotometer (Varian Inc., Palo Alto, CA, USA) at an excitation wavelength of 485 nm and an emission wavelength 530 nm. To facilitate direct comparisons, the bacterial cultures used for GFP fluorescence measurements contained similar amounts of cells (estimation was made based on OD 600 ). Background fluorescence levels were assessed using nonfluorescent control cells (lacking the gfp gene), and these values subtracted from the experimental results.

Purification His-tagged protein
Purification of the His-tagged GFP protein was performed by immobilized metal ion affinity chromatography (IMAC). Cells (200 ml) induced with 10 mM agmatine were harvested by centrifugation at 8000g, at 4 °C for 10 min after 7 h of growth. The supernatant was discarded and cells washed twice and resuspended in 4 ml phosphate buffer (50 mM pH 7.5). They were then disrupted using a French Press operating at 2.3 kbar [Constant Cell Disruption Systems (Low March, Daventry, Norttants, UK)]. Cell debris was removed by centrifugation (10,000g for 30 min at 4 °C). Imidazole was then added to a concentration of 10 mM, and the resulting samples employed in protein purification using the QIAexpressionist kit (Quiagen, Madrid, Spain) following the manufacturer's instructions. Aliquots of collected fractions were analyzed by SDS-PAGE using 12 % acrylamide gels to determine the purity of the His 6 -taggeted proteins. Their GFP activities and protein concentrations were determined using the protocols mentioned above to determine the protein yield.

Soluble and insoluble protein fractions
Soluble and insoluble proteins fractions were prepared following the protocol described by Cano-Garrido et al. [51]. Samples (10 ml) of bacterial cultures grown at different agmatine concentrations were pelleted by centrifugation at 8000g at 4 °C for 10 min and the sediment resuspended in 1 ml of the appropriate buffer depending on the protein expressed (GFP-phosphate buffer 50 mM, pH 7.5; PEP-phosphate buffer 50 mM, pH 7.5, 0.2 M NaCl). The samples were then disrupted with 200 mg glass beads (<106 µm) (Sigma-Aldrich) in a Fast-Prep FP120 Instrument (Thermo Savant-BIO101/Q-Biogen, CA, USA) for 6 × 30 s at a power setting of 4.5 (with intermittent cooling). Total cell extracts were centrifuged at 15,000g at 4 °C for 15 min. Finally, the insoluble fractions were resuspended in 1 ml of the appropriate buffer and fluorescence and PEP activity monitored in both the soluble and insoluble-resuspended fractions.

Determination of organic acids and sugars by HPLC
Sugar and organic acid concentrations were determined using a chromatographic system composed of an Alliance 2690 module injector, a Photodiode Array PDA 996 detector, and a 410 Differential Refractometer detector, all controlled with Millennium 32 software (Waters, Milford, MA, USA). Supernatants (50 μl) were isocratically separated in a 300 × 7.8 mm HPX-87H Aminex ion-exchange column (Hewlett Packard, Palo Alto, CA, USA) protected by a cation H+ Microguard cartridge (BioRad, Laboratories, Richmond, CA, USA), at a flow rate of 0.7 ml/min and a temperature of 65 °C. Sulphuric acid (3 mM) was used as the mobile phase. A PDA 996 detector at 210 nm was used to identify and quantify the organic acids detected, whereas the sugars were analyzed with a 410 Refractometer. Solutions of lactic and acetic acids, glucose, galactose, lactose, and sucrose were used as standards in the identification and quantification procedure.

Statistical analysis
The Student t test was used to examine differences between groups. Significance was set at p < 0.05.