Although it is generally accepted that Escherichia coli glycogen genes are organized in two tandemly arranged, differentially regulated glgBX and glgCAP operons, RT (reverse transcriptase)–PCR analyses carried out in the present study showed that E. coli cells possess transcripts comprising the five glgBXCAP genes. glg::lacZY expression analyses in cells lacking the region immediately upstream of the glgB gene revealed an almost total abolishment of glgB, glgX and glgC expression, but only a 50–60% reduction of the wild-type glgA and glgP expression levels. Furthermore, similar analyses showed that glgA and glgP expression was almost totally abolished in cells lacking glgA upstream sequences, including glgC, glgB and the asd–glgB intergenic region upstream of glgB. These results indicate that E. coli glgBXCAP genes are organized in a single transcriptional unit controlled by promoter sequences occurring upstream of glgB, and that an alternative suboperonic promoter is located within glgC, driving expression of the glgA and glgP genes. Computer searches for consensus promoters, and analyses of glgB::lacZY and glgA::lacZY expression in cells containing deletions of glgB and glgA upstream sequences identified regions directing glgBXCAP and glgAP expression. 5′ RACE (rapid amplification of cDNA ends) analyses located a glgBXCAP transcription start site 155 bp upstream of the glgB initiation codon, and a glgAP transcription start site 359 bp upstream of the glgA initiation codon. Finally, glg::lacZY expression analyses on cells lacking the relA or phoP regulatory genes indicated that both the glgBXCAP operon and the suboperonic promoter driving glgAP expression form part of both the RelA and PhoP-PhoQ regulons.
- Escherichia coli
- gene regulation
- glg operon
- glycogen metabolism
- PhoP-PhoQ regulatory system
- suboperonic promoter
Glycogen is a branched homopolysaccharide of α-1,4-linked glucose subunits with α-1,6-linkages at the branching points. The exact role of this polyglucan in bacteria is not as clear-cut as in animal and yeast cells [1–3], but several works have linked glycogen metabolism to environmental survival, symbiotic performance, and colonization and virulence [4–9]. Synthesized by glycogen synthase (GlgA) using ADPG (ADP–glucose) as the glucosyl moiety donor, glycogen accumulation in Escherichia coli occurs when cellular carbon sources are in excess and other nutrients are lacking [10–14]. Although it has been widely accepted that ADPG is exclusively synthesized by ADPG pyrophosphorylase (GlgC) , other studies have provided strong evidence about the occurrence of important sources, other than via GlgC, of ADPG that are linked to glycogen biosynthesis in E. coli [15,16]. In this context, our previous experimental data revealed the existence of enzyme(s) capable of catalysing the conversion of glucose 1-phosphate, synthesized by phosphoglucomutase, into ADPG .
One of the salient features of bacterial gene expression is that genes of related functions are often clustered into a single transcriptional unit or operon. Unlike other bacterial species possessing a single glycogen operon [4,5,17,18], various experimental evidence has led to the hypothesis that genes involved in E. coli and Salmonella enterica glycogen metabolism are clustered into two tandemly arranged operons: glgBX [encompassing the genes coding for the glycogen branching (GlgB) and debranching (GlgX) enzymes], and glgCAP [encoding the GlgC and GlgA anabolic enzymes, as well as the catabolic glycogen phosphorylase (GlgP)] (reviewed in ). The main data supporting this hypothesis can be summarized as follows. First, S1 nuclease protection assays identified up to four different transcripts initiating within a 0.5 kbp region upstream of glgC , suggesting that glgCAP promoter sequences are located within glgX, although no E. coli consensus promoter sequences could be found in this region . Secondly, in vitro assays employing S-30 E. coli extracts and the glg locus cloned into a plasmid vector showed that, unlike glgB, glgC- and glgA-coupled transcription–translation is enhanced by ppGpp (guanosine 5′-diphosphate 3′-diphosphate) . These observations, however, conflict with transcriptomic analyses showing that expression of the five glg genes is regulated by ppGpp . Thirdly, expression of glgC and glgA, but not that of glgB, is stimulated by exogenously added cAMP in maxicells derived from bacteria bearing the complete E. coli glg locus cloned into plasmids , which is consistent with mobility-shift assays that identified a putative cAMP/CRP (cAMP receptor protein)-binding site within the glgX gene . Fourthly, in vitro assays employing S-30 E. coli extracts and the glg locus cloned into a plasmid vector showed that, unlike glgB, glgC- and glgA-coupled transcription–translation is enhanced by the cAMP/CRP complex . However, although the latter two observations pointed to the involvement of cAMP/CRP in regulating E. coli glgCAP expression, further transcriptomic analyses conducted in two different E. coli K-12 genetic backgrounds failed to identify glg genes as members of the CRP regulon [22–24]. It is also worth noting that earlier experimental evidence obtained by other authors using chromosomally-located glgA::lacZ gene fusions and different E. coli mutant backgrounds argued against a significant in vivo role of cAMP and relA in regulating E. coli glgCAP expression at the transcriptional level .
Given the above contradictory data regarding regulatory aspects of the expression of glycogen genes in E. coli, we decided to characterize in detail glg genes transcription using RT (reverse transcriptase)–PCR and glg::lacZY fusion approaches. We also identified the glgBXCAP transcriptional regulatory regions by both deletion analyses on cells bearing glg::lacZY fusions and 5′ RACE (rapid amplification of cDNA ends). In addition, we investigated the expression patterns of the five glgBXCAP genes in different transcriptional regulator mutants. Taken together, our results show that the five E. coli glgBXCAP genes are transcribed in a single transcriptional unit under the control of promoter sequences occurring upstream of glgB. Moreover, we show the occurrence of an alternative promoter located within glgC that controls the expression of glgA and glgP.
E. coli mutants and culture conditions
Strains, mutants and plasmids used in the present work are shown in Table 1. E. coli K-12 derivative BW25113 single-gene-knockout mutants were obtained from the Keio collection . Unless otherwise indicated, cells were grown at 37 °C with rapid gyratory shaking in liquid Kornberg medium (1.1% K2HPO4, 0.85% KH2PO4 and 0.6% yeast extract from Duchefa) supplemented with 250 μM MgCl2, 50 mM glucose and the appropriate selection antibiotic, or in Mops medium supplemented with 50 mM glucose and amino acids at the concentrations described in Traxler et al. , after inoculation with 1 volume of an overnight culture in 50 volumes of fresh medium.
RNA extraction and RT–PCR
Total RNA from cells entering the stationary phase was extracted using the GenElute Bacterial Total RNA Miniprep kit (Sigma–Aldrich). For the synthesis of cDNA, 1 μg of DNaseI- (TaKaRa) treated RNA was incubated for 1 h at 43 °C in 20 μl of Expand RT (Roche) assay mixture containing 5 μM oligonucleotide O1 (Supplementary Table S1 at http://www.BiochemJ.org/bj/433/bj4330107add.htm). For PCR amplification of the resulting cDNA, 2 μl of the RT assay mixture was added to 20 μl of a PCR cocktail containing Taq polymerase (Biotools) and 0.5 μM each of O2 and O3 (for PCR I), O4 and O5 (for PCR II), O6 and O7 (for PCR III), or O8 and O9 (for PCR IV) oligonucleotides (Supplementary Table S1). Using a programmable thermal cycler we performed amplification for 30 cycles, which consisted of denaturation at 94 °C for 1 min, annealing at 55 °C for 30 s, and extension for 3 min, with a final extension at 72 °C for 7 min. A total of 2 μl of reverse transcription reactions performed without RT were used as negative controls in the RT–PCR to rule out the possibility of contaminating DNA traces. See Figure 1(B) for schematic representation of the oligonucleotides annealing positions and the expected PCR products.
Identification of transcription start sites by 5′-RACE
The glgBXCAP transcription start site was determined using 5′-RACE on extracted bacterial RNA after treatment or not with TAP (tobacco acid pyrophosphatase) as indicated in . Briefly, 10 μg of DNaseI-treated RNA was subjected to the action of TAP (Epicentre Technologies). After phenol/chloroform extraction and ethanol precipitation, RNA was resuspended in 15 μl of water together with 500 pmol of an RNA adapter (5′GAUAUGCGCGAAUUCCUGUAGAACGAACACUAGAAGAAA-3′), which was ligated to the bacterial RNA using T4 RNA ligase (New England Biolabs) according to the manufacturer's instructions. After phenol/chloroform extraction, the RNA was resuspended in 20 μl of diethyl-pyrocarbonate-treated water. For identification of the main glgBXCAP transcription start site, 5 μl of the RNA preparation was included in a 20 μl RT reaction using the glgB-specific O10 oligonucleotide (Supplementary Table S1). A total of 2 μl of the RT reaction was then used in a PCR using the O11 oligonucleotide annealing at glgB, and the O14 oligonucleotide annealing at the RNA adapter (Supplementary Table S1). For identification of the transcription start site at the glgBXCAP suboperonic promoter, 5 μl of the RNA preparation was included in a 20 μl RT reaction using the glgA-specific O12 oligonucleotide (Supplementary Table S1). A total of 2 μl of the RT reaction was then used in a PCR using the O13 oligonucleotide annealing in glgA, and the O14 oligonucleotide annealing at the RNA adapter (Supplementary Table S1). Amplicons obtained from TAP-treated samples were cloned into pGEM-T (Invitrogen) and transformed into XL-1 blue cells. The amplicon sequences present in different clones were then determined at the Secugen S.A. DNA sequence service (Madrid).
Disruption of genomic sequences
Δup-glgB cells (lacking the 253 bp sequence located at positions −270 to −17 immediately upstream of the glgB gene), ΔglgBX* cells (ΔglgBX cells lacking the 270 bp region upstream of glgB), ΔglgBXC* cells (ΔglgBXC cells lacking the 270 bp region upstream of glgB) and different deletions of the glgB and glgA upstream regions were produced following the method of Datsenko and Wanner  (Supplementary Figure S1 at http://www.BiochemJ.org/bj/433/bj4330107add.htm). A selectable antibiotic resistance gene was generated by PCR from a freshly isolated colony of E. coli MC4100 containing a spectinomycin-resistance cassette, using 80-nucleotide-long primer pairs that included 60 nucleotide extensions for the targeted locus and 20 nucleotides priming sequences for the resistance gene (Supplementary Table S2 at http://www.BiochemJ.org/bj/433/bj4330107add.htm).
LacZY transcriptional fusions
The KmR (kanamycin resistance) cassette of the ΔglgB, ΔglgX, ΔglgC, ΔglgA and ΔglgP cells of the Keio collection  was removed by using temperature-sensitive plasmid pCP20 carrying the FLP recombinase . The scar sequence left after removal of the resistance cassette included a 34-nucleotide FRT site , which was used to build lacZY transcriptional fusions as reported in . Briefly, ΔglgB, ΔglgX, ΔglgC, ΔglgA and ΔglgP cells of the Keio collection carrying the pCP20 plasmid were transformed with pKG137, which has functional lacZY and a KmR cassette, which integrated in the proper orientation at the FRT site by the action of the FLP recombinase, yielding lacZY transcriptional fusions where the original resistance cassette of the Keio collection was placed. Transcriptional fusions were P1-transduced  into the different mutants as needed. All fusions were verified by PCR using an oligonucleotide (5′-TTCAGGCTGCGCAACTGTTGG-3′) annealing within lacZ (+150 bp reverse orientation) and oligonucleotides specifically annealing at positions 500 bp upstream of the insertion point. lacZY fusions inserted at the expected sites yielded ~750-bp PCR amplification products.
Western blot analyses
Bacterial extracts were separated by SDS/PAGE (10% gels), transferred on to PVDF filters, and immunoblotted by using rabbit antisera raised against GlgC  and a goat anti-(rabbit IgG)–alkaline phosphatase conjugate (Sigma).
Bacterial growth was followed spectrophotometrically by measuring the absorbance at 600 nm. β-Galactosidase activity was measured and reported as described by Miller . Protein content was measured by the Bradford method using a Bio-Rad prepared reagent.
Promoters were predicted using the BPROM program (SoftBerry, http://linux1.softberry.com/berry.phtml?topic=bprom&group=programs&subgroup=gfindb).
E. coli cells possess transcripts comprising the five glg genes
Genes involved in glycogen metabolism are clustered in a single chromosomal locus of the E. coli genome in the order glgBXCAP (Figure 1A). In the present work we investigated the possible occurrence of transcripts comprising the five glg genes in E. coli. Towards this end, we reverse transcribed total E. coli RNA using as primer an oligonucleotide (O1) annealing to the 5′ region of glgP transcripts (for further details see the Experimental section) (Figure 1B). The resulting cDNAs were then subjected to PCR analyses using as primers oligonucleotides corresponding to sequences occurring in glgB (O2 and O3 for PCR I), glgX and glgC (O4 and O5 respectively, for PCR II), glgX and glgA (O6 and O7 respectively, for PCR III), and glgC and glgA (O8 and O9 respectively, for PCR IV) (Figure 1B). As shown in Figure 1(C), PCR I amplified a fragment whose size (~2.1 kbp) corresponds to the whole glgB gene. PCR II amplified a ~2.0 kbp fragment corresponding to the whole glgX gene and a glgC fragment. PCR III amplified a ~1.8 kbp fragment corresponding to a glgX fragment, the whole glgC gene and a glgA fragment. Finally, PCR IV amplified a ~2.7 kbp fragment corresponding to the whole glgC and glgA genes. The overall results thus show that E. coli cells possess single transcripts comprising all the five glg genes.
The five glg genes are transcribed from a single operon under the control of promoter sequence(s) located upstream of glgB
Whether expression of the five glg genes is controlled by promoter sequences occurring upstream of the glgB gene was explored by analysing the expression of chromosomal glg::lacZY transcriptional fusions on both WT (wild-type) and on Δup-glgB cells, lacking 253 bp of the asd–glgB intergenic region located upstream of glgB (nucleotide positions −270 to −17 from the glgB translation initiation codon; Supplementary Figure S1). As shown in Supplementary Figure S2 (at http://www.BiochemJ.org/bj/433/bj4330107add.htm), the five glg genes followed the same time course of expression in WT cells, with the β-galactosidase activity increasing during the exponential growth phase and reaching a plateau in the stationary phase. In contrast, Δup-glgB cells displayed sharply reduced expression of glgB, glgX and glgC fusions (≤15% that of WT cells at the onset of the stationary phase), and reduced but still very significant levels of expression of glgA and glgP (40–50% of those of WT cells) (Figure 2A and Supplementary Figure S2).
We also carried out Western blot analyses of GlgC in ΔglgB* and ΔglgX* cells (ΔglgB and ΔglgX cells lacking the KmR cassette, Supplementary Figure S1), and in Δup-glgB cells at the onset of the stationary phase. In agreement with the results obtained employing glg::lacZY fusions (Figure 2A), these analyses revealed that Δup-glgB cells accumulated drastically reduced levels of GlgC when compared with WT cells, whereas ΔglgB* and ΔglgX* cells accumulated similar to WT GlgC levels (Figure 2B). The small amount of GlgC accumulated by Δup-glgB cells (Figure 2B) is ascribed to the combination of both residual levels of glgC expression in these mutants, as judged by the analyses of β-galactosidase activity in glgC::lacZY-expressing Δup-glgB cells (Figure 2A), and low GlgC proteolytic turnover, allowing this protein to accumulate within the cell (Supplementary Figure S3 at http://www.BiochemJ.org/bj/433/bj4330107add.htm). Residual glgC expression occurring in Δup-glgB cells could be ascribed to low RNA polymerase affinity for non-specific sites [33,34], rather than to active promoter sequences occurring within glgX. This conclusion is supported by comparative analyses of β-galactosidase activity in glgC::lacZY-expressing Δup-glgB and ΔglgBX* cells (ΔglgBX cells lacking the 270 bp sequence immediately upstream of glgB, Supplementary Figure S1), which revealed marginally low and similar residual β-galactosidase activities in both strains (Supplementary Figure S3). The overall results thus provide strong indications that (i) the five glgBXCAP genes are organized in an operon mainly controlled by promoter sequence(s) located upstream of glgB and (ii) glgCAP expression is only marginally controlled by sequences occurring within glgX.
Identification of a main promoter sequence in the intergenic asd–glgB region driving E. coli glgBXCAP expression
Computer searches for putative bacterial promoters in the E. coli glg locus identified sequences at −35 (TTACCG) and −10 (GGCTATTCT) in the 273-bp asd–glgB intergenic region starting at 194 and 170 bp upstream of the glgB translation initiation codon respectively (Figure 3A). The role of these sequences in directing expression of downstream glgBXCAP genes was analysed by measuring the β-galactosidase activity levels of cells expressing glgB::lacZY fusions and containing different deletions of the glgB upstream region (Figure 3B). As seen in Figure 3(C), cells containing deletions encompassing DNA regions located over 800 and 250 bp upstream of the glgB initiation codon (BI and BII) displayed similar β-galactosidase activities to cells bearing no deletions (B0). On the contrary, removal of the genomic region located between 250 and 150 bp upstream of glgB resulted in a sharp reduction in β-galactosidase levels (BIII), strongly indicating that this region contains the main promoter(s) directing expression of the E. coli glgBXCAP operon. To explore this possibility we identified transcription start site(s) upstream of glgB by 5′ RACE (for details, see the Experimental section). Amplicons of ~350 bp were generated in TAP-treated (but not in non-treated) RNA samples, which were cloned and sequenced. Sequence analysis of different clones identified (in all of them) a single transcription start site located 155 bp upstream of the glgB translation initiation codon (Figure 3A), a result which fully agrees with both computer searches and deletion analyses for the location of glgBXCAP promoter sequences (see above).
Identification of an alternative suboperonic promoter region within glgC driving glgA and glgP expression
A salient feature of prokaryotic gene expression is the phenomenon of operon polarity (i.e. the reduced expression of promoter distal genes with respect to the proximal genes). This rule may sometimes be over-ridden at the suboperonic level by the presence of an internal promoter [35,36]. During the course of our investigations we realized that expression levels of glgA and glgP are higher than that of the upstream glgC gene (Figure 2A). This and the fact that the removal of the glgB upstream region only resulted in a 50–60% reduction of glgA::lacZY and glgP::lacZY fusions expression (Figure 2A) suggested that glgA and glgP may be under the control of an alternative suboperonic promoter located within glgC. To test this possibility we compared the expression of chromosomal glgA::lacZY and glgP::lacZY transcriptional fusions in ΔglgBX* and ΔglgBXC* cells (ΔglgBXC cells lacking the 270 bp sequence immediately upstream of glgB, Supplementary Figure S1) with WT cells. As illustrated in Figure 4, these analyses showed that, whereas glgA and glgP expression levels in ΔglgBX* cells were still approx. 40–50% that of WT cells, glgA and glgP expression levels in ΔglgBXC* cells reduced further to values ≤15% of those of WT cells. The overall results thus strongly indicate that expression of E. coli glgA and glgP is not solely controlled by a promoter located upstream of glgB, but also by an alternative promoter located within glgC.
To identify promoter sequences located within glgC directing expression of glgAP, a deletion approach similar to that used above (Figure 3) was employed using glgA::lacZY fusions, except that the genomic region located between asd and part of the glgC gene in these cells was removed to eliminate the main glgBXCAP promoter regions and any other potential promoter sequences located upstream of the glgC gene. Putative −35 (TCGCAATT) and −10 (ACCTACAAT) consensus promoter sequences were identified starting at 394 and 369 bp upstream of the glgA translation initiation codon respectively (Figure 5A). As seen in Figures 5(B) and 5(C), similar levels of β-galactosidase activity were detected in cells expressing glgA::lacZY fusions under the control of regions located over 450 bp upstream of the glgA initiation codon (deletions AI and AII). In contrast, removal of the genomic region between 450 and 300 bp upstream of the glgA initiation (deletion AIII) resulted in a drastic reduction in β-galactosidase activity, strongly suggesting that suboperonic promoter sequences directing glgAP expression are located in this region. To explore this possibility we identified by 5′ RACE the transcription start site upstream of glgA (for details, see the Experimental section). Amplicons of ~450 bp were generated in TAP-treated (but not in non-treated) RNA samples, which were cloned and sequenced. Sequence analyses of different clones identified (in all of them) a single transcription start site located 359 bp upstream of the glgA translation initiation codon (Figure 5A), a result which fully agrees with both computer searches and deletion analyses for the location of alternative suboperonic sequences driving glgAP expression (see above).
The glgBXCAP operon belongs to the RelA and PhoP-PhoQ regulons
During nutrient starvation E. coli elicits the so-called stringent response that switches the cell from a growth-related mode to a maintenance/survival/biosynthesis mode . The hallmark of this pleiotropic response is the accumulation of the alarmone (p)ppGpp, which is synthesized by the relA product . This nucleotide binds prokaryotic RNA polymerase to increase transcription of amino acid biosynthesis genes during nutritional starvation, and to down-regulate the production of ribosomes, nucleic acids and proteins . Therefore, when ppGpp accumulates, growth is arrested and an important pool of ATP diverts towards glycogen biosynthesis under conditions of carbon source excess. Consistent with this view, experimental evidences provided by different authors have shown that (i) relA positively affects glycogen accumulation [13,14,40,41] and (ii) glgCAP expression is under the control of the RelA product ppGpp [14,20,21]. Whether the glgBXCAP operon is under relA control was investigated by comparing the expression levels of chromosomal glg::lacZY transcriptional fusions in WT cells and ΔrelA cells cultured in either Kornberg or Mops medium, supplemented with glucose (Figure 6 and Supplementary Figure S4 at http://www.BiochemJ.org/bj/433/bj4330107add.htm). As shown in Figures 6(A) and 6(B), these analyses clearly revealed that the expression of the five glg genes was largely abolished in ΔrelA cells. The overall results thus not only reinforce the notion that glgBXCAP composes a single transcriptional unit, but also indicate that it forms part of the E. coli RelA regulon.
PhoP-PhoQ is a two-component regulatory system occurring in E. coli and Salmonella spp. that monitors the availability of extracellular Mg2+, and transcriptionally controls the expression of many genes [42,43]. We have shown previously that PhoP-PhoQ highly regulates glycogen metabolism at submillimolar environmental Mg2+ concentrations since (i) cells impaired in the PhoP and PhoQ functions accumulate low glycogen [13,14], and (ii) expression of glgC::lacZY was positively regulated by the PhoP-PhoQ regulatory system at low environmental Mg2+ . Whether the glgBXCAP operon is under PhoP-PhoQ control was investigated by comparing the expression levels of chromosomal glg::lacZY transcriptional fusions in WT and ΔphoP cells cultured in liquid Kornberg medium supplemented with glucose under limiting environmental Mg2+ concentrations. As shown in Figure 6(C), these analyses clearly revealed that the expression of the five glg genes was lower in ΔphoP cells than in WT cells, reinforcing further the idea that the five glg genes belong to a single glgBXCAP operon that is under positive control of the PhoP-PhoQ regulatory system under limiting environmental Mg2+ concentrations.
The suboperonic promoter driving glgAP expression belongs to the RelA and PhoP-PhoQ regulons
Whether the alternative suboperonic promoter region driving glgAP expression is also under control of the relA and phoP/phoQ gene products was investigated by comparing the expression levels of the chromosomal glgA::lacZY transcriptional fusion in ΔglgBX* and ΔglgBX*ΔrelA cells cultured in either Kornberg or Mops medium supplemented with glucose. Furthermore, we also compared the expression levels of chromosomal glgA::lacZY transcriptional fusion on ΔglgBX* and ΔglgBX*ΔphoP cells cultured in liquid Kornberg medium supplemented with glucose under limiting environmental Mg2+ concentrations. As shown in Figure 7, these analyses clearly showed that glgA expression was lower in ΔglgBX*ΔrelA and ΔglgBX*ΔphoP cells than in ΔglgBX* cells, the overall results showing that the suboperonic promoter driving glgAP expression belongs to both the RelA and PhoP-PhoQ regulons.
Expression of E. coli glgBXCAP genes is not altered in a Δcya background
cAMP produced by the product of the cya gene is an important cellular mediator of glucose effects in E. coli. The cAMP/CRP complex is a positive transcriptional regulator of a number of catabolic operons, and as such, plays a role in catabolite repression, whereby secondary carbon sources are not catabolized in the presence of glucose. Both cAMP and CRP are required for expression of glgS and PTS (phosphotransferase system)-related genes required for normal glycogen production [22,44]. On the basis of in vitro and in vivo experimental data, Preiss and co-workers have proposed that expression of E. coli glgCAP genes is positively regulated at the transcriptional level by the cAMP/CRP complex [19,21,45] (reviewed in ). However, a role of cAMP in regulating E. coli glgCAP expression has been questioned by other authors on the basis of results obtained using cells bearing chromosomally-located glgA::lacZ gene fusions [25,44]. Moreover, further transcriptome analysis conducted in different E. coli strains also failed to include glg genes in the CRP regulon [22,23]. In addition, our own prior analysis using glgC::lacZY fusions also indicated that glgC expression was not significantly affected in an E. coli BW25113 Δcya mutant background, which lacks the ability to produce cAMP . To investigate whether glgBXCAP may be part of the cAMP regulon, we compared the expression levels of chromosomal glg::lacZY transcriptional fusions in WT cells and Δcya cells cultured in liquid Kornberg medium supplemented with 5 mM glucose. Furthermore, to investigate whether the alternative suboperonic promoter driving glgAP expression is regulated by cAMP we compared the expression levels of chromosomal glgA::lacZY transcriptional fusion on ΔglgBX* cells and ΔglgBX*Δcya cells cultured in liquid Kornberg medium supplemented with 5 mM glucose. As shown in Supplementary Figures S5(A) and S5(B) (at http://www.BiochemJ.org/bj/433/bj4330107add.htm), these analyses revealed that the expression of glgBXCAP in WT cells, and of glgA in ΔglgBX* cells was not affected by the lack of the cya function.
We also compared the glgB::lacZY expression levels on WT cells cultured in glucose and in other carbon sources such as maltose, fructose and galactose, conditions that are known to induce cAMP synthesis. As shown in Supplementary Figure S5(C), these analyses revealed that glgB expression in WT cells was not affected by the nature of the carbon source, the overall results reinforcing the idea that cAMP does not control E. coli glycogen metabolism at the level of transcription of glg genes.
In the present work we have shown that the five E. coli glgBXCAP genes are transcribed in a single transcriptional unit under the control of promoter sequences located upstream of glgB. Moreover, we have also shown the occurrence of an alternative promoter located within glgC that controls the expression of downstream glgA and glgP genes. Finally, we have provided evidence that the glgBXCAP operon forms part of both the RelA and PhoP-PhoQ regulons. Previous studies, by other authors, using S1 nuclease protection assays indicated the existence in E. coli of glgCAP transcripts initiating within the glgX coding region [19,46]. However, computer searches failed to identify putative consensus bacterial promoter sequences within glgX (, also confirmed in the present study). Moreover, 5′ RACE analyses using glgC-specific oligonucleotides failed, in our hands, to detect any putative transcription start site within glgX (results not shown). Finally, comparative analyses of β-galactosidase activity in glgC::lacZY-expressing Δup-glgB and ΔglgBX* cells revealed marginally lower and similar residual activities in both strains (Supplementary Figure S3). It is thus likely that the results obtained by other authors [19,46] could be ascribed to low basal levels of non-specific transcription initiation occurring in regions immediately upstream of glgC, which is a situation reported for many E. coli genes . Another possible explanation is the nucleolytic processing of longer transcripts. In this context, it is worth noting that longer transcripts attributed to read-through transcription from the upstream glgBX genes were also detected in S1 protection assays .
The existence of glg genes clustered in a single chromosomal locus forming part of a single operon, and showing suboperonic promoters has been previously reported in α-proteobacterial species, such as Agrobacterium tumefaciens, Mesorhizobium loti and Rhizobium tropici [4,5,18]. In some of these cases, however, the order of glg genes and the position of the suboperonic promoters in the corresponding operons are different to those of E. coli in the present study, possibly because they respond to different metabolic needs. Internal promoters in large bacterial operons reinforce expression of downstream genes, which would otherwise be transcribed in lower amounts due to the effects of operon polarity. Therefore the suboperonic promoter within glgC may have evolved to guarantee adequate glgAP expression levels. Alternatively, internal promoters may also allow the temporal expression of certain genes of the operon under conditions of reduced expression of other members of the same operon, with their functions redundantly fulfilled by other loci [35,36]. The relevance of glgA expression when glgC is only expressed at low levels is well exemplified by glgA-expressing glgC− mutants, which accumulate substantial glycogen content  due to the occurrence of important sources, i.e. other than via GlgC, of ADPG . Since glycogen plays relevant roles in the survival of E. coli and S. enterica in sporadic periods of famine, and/or in colonization and virulence, and because glycogen metabolism is highly interconnected with multiple and important cellular processes [13,14], it is tempting to speculate that redundancy of ADPG sources and regulatory mechanism(s) guaranteeing glgA expression have been selected during evolution to warrant glycogen production in these particular species.
Regulation of glg gene expression involves a still not well-defined assemblage of factors that are adjusted to the physiological and nutritional status of the cell. Results from the present study showing that the five E. coli glgBXCAP genes can be transcribed in a single mRNA unit opens new clues on the mechanisms regulating glycogen metabolism in enterobacterial species. Depending on both external and internal conditions, complex operons may be regulated by the presence and/or combination of various factors, including differential transcriptional or translational efficiency, differential localized instability in the polycistronic mRNA, presence of internal promoter(s), regulatory structures at the 5′ UTRs (untranslated regions) of the transcripts, etc. [34,47,48]. Notably, the positions of the two promoter regions identified in the present work, regulating expression of glg genes with respect to the initiation codons (compare Figure 3A and Figure 5A), point to the occurrence of relatively long UTRs on each of the corresponding transcripts. 5′ UTRs serve as targets of regulatory factors including RNA-binding proteins, RNase E and regulatory small RNAs, and can form riboswitch structures interacting with low-molecular-mass effectors. All these elements can control the fate and stability of transcripts . It is therefore tempting to speculate that these regions may play important roles in regulating translation capability and/or stability of glgBXCAP and glgAP transcripts. In this context, we have shown recently that glycogen accumulation in E. coli is severely compromised in a Δhfq genetic background . Hfq is a chaperone that stabilizes many regulatory small RNAs, and facilitates the base pairing between small RNAs and their target mRNAs thus affecting both transcript translation and degradation [47,49]. Thus it is highly conceivable that the glycogen-deficient phenotype of Δhfq cells may result from small RNAs- and/or Hfq-mediated regulation of glgBXCAP expression. In fact, recent analyses of mRNA targets of Hfq in Salmonella have shown that glgB, glgX, glgC and glgA transcript levels are similarly reduced in a Δhfq background . The global regulator CsrA is a RNA-binding protein that controls the expression of genes involved in carbohydrate metabolism. glgB::lacZ and glgC::lacZ translational fusion expression analyses have revealed that CsrA exerts a negative effect on glycogen accumulation through its negative control on glgB and glgC expression . Although more recent studies focused their attention on the effect of CsrA in glgC expression, and showed that CsrA binds sequences immediately upstream of glgC to promote mRNA decay and prevent glgC translation , further studies will be necessary to elucidate whether CsrA can also bind the 5′ UTR of glgBXCAP transcripts to prevent glycogen biosynthesis by regulating their nucleolytic processing. Needless to say, further investigations are necessary to understand the many regulatory aspects of bacterial glycogen metabolism at both transcriptional and post-transcriptional levels.
Manuel Montero, Goizeder Almagro, Gustavo Eydallin, Francisco Muñoz, Abdellatif Bahaji, Jun Li, Mehdi Rahimpour and Edurne Baroja-Fernández were responsible for performing the experiments. Manuel Montero, Goizeder Almagro, Alejandro Viale and Javier Pozueta-Romero were responsible for planning the experiments and analysing results. Manuel Montero, Goizeder Almagro, Alejandro Viale and Javier Pozueta-Romero were responsible for writing the manuscript.
We thank Cristina Cirauqui, Maite Sesma and Maite Hidalgo (Institute of Agrobiotechnology) for technical support.
This research was partially supported by the Comisión Interministerial de Ciencia y Tecnología and Fondo Europeo de Desarrollo Regional (Spain) [grant number BIO2007-63915] and by Iden Biotechnology S.L. M.M. acknowledges a post-doctoral contract from I3P program of Consejo Superior de Investigaciones Científicas. M.R. acknowledges a pre-doctoral JAE fellowship from the Consejo Superior de Investigaciones Científicas. A.M.V. expresses his gratitude to the Ministerio de Educación y Cultura, the Consejo Superior de Investigaciones Científicas and the Public University of Nafarroa for their support. G.A. acknowledges a pre-doctoral fellowship from the Public University of Nafarroa.
Abbreviations: ADPG, ADP–glucose; CRP, cAMP receptor protein; KmR, kanamycin resistance; ppGpp, guanosine 5′-diphosphate 3′-diphosphate; RACE, rapid amplification of cDNA ends; RT, reverse transcriptase; TAP, tobacco acid pyrophosphatase; UTR, untranslated region; WT, wild-type
- © The Authors Journal compilation © 2011 Biochemical Society