Molecular Control of Sucrose Utilization in Escherichia coli W, an Efficient Sucrose-Utilizing Strain

Abstract

Sucrose is an industrially important carbon source for microbial fermentation. Sucrose utilization in Escherichia coli, however, is poorly understood, and most industrial strains cannot utilize sucrose. The roles of the chromosomally encoded sucrose catabolism (csc) genes in E. coli W were examined by knockout and overexpression experiments. At low sucrose concentrations, the csc genes are repressed and cells cannot grow. Removal of either the repressor protein (cscR) or the fructokinase (cscK) gene facilitated derepression. Furthermore, combinatorial knockout of cscR and cscK conferred an improved growth rate on low sucrose. The invertase (cscA) and sucrose transporter (cscB) genes are essential for sucrose catabolism in E. coli W, demonstrating that no other genes can provide sucrose transport or inversion activities. However, cscK is not essential for sucrose utilization. Fructose is excreted into the medium by the cscK-knockout strain in the presence of high sucrose, whereas at low sucrose (when carbon availability is limiting), fructose is utilized by the cell. Overexpression of cscA, cscAK, or cscAB could complement the WΔcscRKAB knockout mutant or confer growth on a K-12 strain which could not naturally utilize sucrose. However, phenotypic stability and relatively good growth rates were observed in the K-12 strain only when overexpressing cscAB, and full growth rate complementation in WΔcscRKAB also required cscAB. Our understanding of sucrose utilization can be used to improve E. coli W and engineer sucrose utilization in strains which do not naturally utilize sucrose, allowing substitution of sucrose for other, less desirable carbon sources in industrial fermentations.

INTRODUCTION

Carbon source is one of the major cost drivers for industrial production of bulk chemicals from microbial fermentation (1, 2). Currently, glucose (typically from corn) is the most common carbon source for industrial fermentation in Escherichia coli. Sucrose from sugarcane, however, would be preferable to corn-based glucose as a carbon substrate for E. coli-based industrial fermentation. First, it is a cheaper substrate as it can be directly fermented (either as cane juice or as molasses, or it can be easily made into pure sugar by high-temperature crystallization), whereas glucose has to be converted from starch by milling and enzymatic hydrolysis (3, 4). Second, sugarcane sucrose-based bioprocesses are more environmentally friendly and sustainable than glucose-based bioprocesses. This is because bagasse (the fibrous by-product from sugarcane mills) can be utilized to produce energy for the bioprocess, whereas corn glucose-based processes rely on fossil fuels for energy (5). Finally, as a result of these two primary factors, the associated overall bioprocess cost is decreased relative to that with glucose (5–7). In addition, sucrose is highly abundant and readily available.

The ability to metabolize sucrose as a carbon source is a highly variable feature among E. coli strains. Sucrose-fermenting strains include the enteropathogenic strains (8), B-62 (9), EC3132 and its mutants (10, 11), and W (6, 12). There are two gene clusters responsible for sucrose catabolism in E. coli: the scr regulon, encoding a sucrose phosphotransferase system (PTS) (13, 14), and the chromosomally carried sucrose catabolism (csc) regulon, encoding a sucrose non-PTS utilization system (11). Sucrose PTS genes may be found either on plasmids (14, 15), on transposons (16, 17), or on the chromosomal DNA (18–20), whereas the csc genes have been found only on the chromosome (6, 10, 11).

Escherichia coli W (ATCC 9637) grows particularly quickly on sucrose and is the only safe laboratory or industrial strain that can utilize sucrose (6). Genome sequencing indicated that sucrose is metabolized via the csc genes in this strain (6). The csc regulon was originally described in E. coli EC3132 (11) and consists of four open reading frames which encode a transcriptional repressor (CscR), a sucrose hydrolase or invertase (CscA), a sucrose permease (CscB), and a fructokinase (CscK) (10, 11). The csc catabolic genes are negatively controlled by CscR, which presumably represses transcription in the absence of sucrose and at low sucrose concentrations (<2 g/liter) (10, 11). The sucrose permease/proton symporter, CscB, transports sucrose into the cell. Intracellular sucrose is then hydrolyzed to glucose and fructose by CscA. These two sugars are then phosphorylated into glucose-6-phosphate and fructose-6-phosphate by glucokinase (Glk) and CscK, respectively. The phosphorylated sugars are then assimilated into glycolysis.

Although the genetic structure of the E. coli W csc genes is well characterized, the molecular control of the regulon and the contribution of each gene to sucrose metabolism are not. In order to investigate this in detail, expression and enzyme activity of the csc genes were examined by individual and combinatorial gene knockout (KO) approaches and by overexpression in non-sucrose-utilizing strains.

MATERIALS AND METHODS

Bacterial strains, media, and growth conditions.

Bacterial strains and plasmids used in this study are shown in Table 1. E. coli strains were grown in LB medium (23) for general cloning and maintenance and in the following media for sugar utilization experiments: M9 (23) supplemented with thiamine (1 mg/liter) and with either 2 g/liter (0.2%) sucrose (M9S2), 20 g/liter (2%) sucrose (M9S20), 20 g/liter (2%) glucose (M9G20), or 20 g/liter (2%) lactose or MacConkey agar base (Difco, BD, North Ryde, NSW, Australia) supplemented with 10 g/liter sucrose (MCS10). Sucrose utilization on MacConkey plates was scored as positive if colonies were pink and negative if colonies were yellow. Ampicillin (100 μg/ml) and/or chloramphenicol (25 μg/ml) was included in media where appropriate.

Table 1.

Strains and plasmids used in this study

Strain or plasmid	Description	Source or reference
E. coli strains
DH5α	fhuA2 Δ(argF-lacZ)U169 phoA glnV44 ϕ80dlacZΔM15 gyrA96 recA1 relA1 endA1 thi-1 hsdR17	Invitrogen
W	Wild type	NCIMB 8666a
MG1655	F− λ− ilvG mutant rfb-50 rph-1	Coli Genetic Stock Center
WΔcscR	WΔcscR::FRT	This study
WΔcscA	WΔcscA::FRT	This study
WΔcscB	WΔcscB::FRT	This study
WΔcscK-cat	WΔcscK::FRT-cat-FRT	This study
WΔcscK	WΔcscK::FRT	This study
WΔcscRK	WΔcscR::FRT-cat-FRT ΔcscK::FRT	This study
WΔcscRAKB	WΔcscRAKB::FRT	This study
WΔcscK Δmak	WΔcscK::FRT Δmak::FRT-cat-FRT	This study
Plasmids
pKD46	γ Red recombinase expression plasmid	21
pKD3	Template for amplifying the cat gene	21
pCP20	Flp recombinase expression plasmid	21, 22
pCR2.1	Cloning and expression	Invitrogen
pCSCA	pCR2.1 harboring cscA under lac promoter	This study
pCSCAB	pCR2.1 harboring cscA and cscB under csc promoter	This study
pCSCAK	pCR2.1 harboring cscA and cscK under csc promoter	This study

^aNational Collection of Industrial Bacteria, Aberdeen, Scotland. This strain is also archived as ATCC 9637 in the American Type Culture Collection.

DNA manipulations and PCR.

General recombinant DNA techniques were performed according to standard protocols (23). PCR products were purified using a MinElute PCR purification kit (Qiagen, Doncaster, VIC, Australia). Sequencing of PCR and plasmid DNA was performed by the Australian Genome Research Facility (The University of Queensland, St. Lucia, QLD, Australia) using the Sanger method.

Knockout of target genes by homologous recombination.

Chromosomal gene knockout (KO) was performed using one-step homologous recombination, as described previously (21), using minor modifications as described by Bruschi et al. (24), with the exception that cultures for electrocompetent cell preparation were grown in 100 ml SOB medium (25) and induced with arabinose for 15 min rather than 1 h. Primers used to target specific genes for homologous recombination are listed in Table 2; PCR was performed using Platinum Taq polymerase (Invitrogen, VIC, Australia) according to the manufacturer's instructions. The WΔcscR strain has been described previously (26). The cscA, cscK, and cscB genes were deleted in the wild-type (WT) strain using KOcscA_F and KOcscA_R2, KOcscK_F and KOcscK_R, and KOcscB_F and KOcscB_R primer pairs, respectively. The WΔcscRK strain was made by removal of cscR in the WΔcscK strain background using KOcscR_F and KOcscR_R2 primers. The chloramphenicol acetyltransferase (CAT) selection cassette (replacing the cscR gene) is still present in this strain. The WΔcscRAKB strain was made by removal of cscAKB from the WΔcscR strain as follows: a knockout cassette was generated using KOcscK_F and KOcscB_R primers targeted at cscK and cscB recombined through the FLP recombination target (FRT) site which remained after removal of cscR, thereby resulting in deletion of cscA, cscK, and cscB. The WΔcscK Δmak strain was created in the WΔcscK strain background using the KOmak_F and KOmak_R primers.

Table 2.

Primers used in this study^a

Primer	Sequence (5′–3′)	Specific use
cscA_R2	TTAACCCAGTAGCCAGAGTGC	To amplify cscA, cscA + cscK, and cscA + cscB
cscA_F2	ATGACGCAATCTCGATTGCATG	To amplify cscA
cscK_R2	GATAAGAGCGACTTCGCCGTT	To amplify cscA + cscK and cscA + cscB
cscB_F2	ATGGCACTGAATATTCCATTCAGA	To amplify cscB
cscB_R3	GTTTACGTCTATATTGCTGAAGGTAC	To amplify cscB
cscB_R2	CTATATTGCTGAAGGTACAGGCGT	To amplify cscA + cscB
cscA_qrtF1	GTCCGGACATTCCCACATATAG	QRT-PCR for cscA
cscA_qrtR1	AGGCAACACGGGGCAGATCCTG	QRT-PCR for cscA
cscB_F1	ATCCGTCTTCAAATACAGCGTGG	QRT-PCR for cscB
cscB_R1	CAGCACAATCCCAAGCGAACTGG	QRT-PCR for cscB
cscK_F1	GCCGGGTTACTCACAGGTCTG	QRT-PCR for cscK
cscK_R1	TTCGCCGTTACTGCAAGCGCT	QRT-PCR for cscK
cscR_F1	GTAACGATCGCGCAGCCTTTGTGG	QRT-PCR for cscR
cscR_R1	GCTGAATTGTGGTCAGCGGCGGTA	QRT-PCR for cscR
fruA_F1	AGAGTATGGAGGCGCTGAAA	QRT-PCR for fruA
fruA_R1	CGCCCATGTCAGTACACATC	QRT-PCR for fruA
dld_F1	AGCACCCTGCGTCTCGACAAGC	QRT-PCR for dld
dld_R1	CACGACGATCCAATCACCGAGTGC	QRT-PCR for dld
KOcscR_F	CTCGCCAGTGACGTCTGTTTCTGCTACAGTGCCCGTTTTACGGCAAACGGCTTGGGTGTGTAGGCTGGAGCTGCTTC	To amplify cscR KO cassette
KOcscR_R2	CCCACGGAGTGGCTGTGCTGCAACATGGAGCACTCTGGCTACTGGGTTAACATAATACATATGAATATCCTCCTTAG	To amplify cscR KO cassette
KOcscA_F	CAATTCACCAAATTTGCTTAACCAGGATGATTAAAATGACGCAATCTCGATTGCATGGTGTAGGCTGGAGCTGCTTC	To amplify cscA KO cassette
KOcscA_R2	TATGTTAACCCAGTAGCCAGAGTGCTCCATGTTGCAGCACAGCCACTCCGTGGGACATATGAATATCCTCCTTAG	To amplify cscA KO cassette
KOcscK_F	CGATTTTGCGAAAAAGAGGTTTATCACTATGCGTAACTCAGATGAATTTAAGGGAGTGTAGGCTGGAGCTGCTTC	To amplify cscK KO cassette
KOcscK_R	CGTTAAAAAATTCACGTCCTATTTAGAGATAAGAGCGACTTCGCCGTTTACTTCTCATATGAATATCCTCCTTAG	To amplify cscK KO cassette
KOcscB_F	GAATTTTTTAACGACAGGCAGGTAATTATGGCACTGAATATTCCATTCAGAAATGGTGTAGGCTGGAGCTGCTTC	To amplify cscB KO cassette
KOcscB_R	CCGGTTGAGGGATATAGAGCTATCGACAACAACCGGAAAAAGTTTACGTCTATATCATATGAATATCCTCCTTAG	To amplify cscB KO cassette
KOmak_F	GTGCGTATAGGTATCGATTTAGGCGGCACCAAAACTGAAGTGATTGCACTGGGCGGTGTAGGCTGGAGCTGCTTC	To amplify mak KO cassette
KOmak_R	TACGCCGCTGGAATCACCGTGCTTCGCCTTACGCACCGGCGTTTCACATTCGCCGCATATGAATATCCTCCTTAG	To amplify mak KO cassette

^aAbbreviations: KO, knockout; QRT-PCR, quantitative real-time PCR.

Three biological replicates (colonies growing on original transformation plates) were selected for each recombination experiment and sequenced to confirm knockout. The cat resistance marker was removed by Flp recombinase as described previously (21, 22). Loss of the resistance gene was confirmed by sequencing. Sucrose utilization phenotypes of each strain were scored on M9S20 and M9S2 agar plates.

Overexpression of csc genes and complementation of knockout strains.

Individual csc genes were cloned for overexpression studies. Genes were amplified using the primers shown in Table 2 and Platinum Taq polymerase (Invitrogen, VIC, Australia) according to the manufacturer's instructions. Amplicons were cloned into pCR2.1 (Invitrogen, VIC, Australia) according to the manufacturer's instructions. Ligations were transformed into E. coli DH5α by electroporation. Three clones with the insert oriented correctly relative to the lac promoter were selected for each plasmid. Sucrose utilization phenotypes were scored on MCS10 and M9S20 media. Plasmids were purified using Qiaprep Spin Miniprep (Qiagen, Doncaster, VIC, Australia). The full sequence of each insert was confirmed by sequencing. Each construct which conferred a suc⁺ phenotype in DH5α was also transferred into E. coli WΔcscRAKB by electroporation; three clones for each construct were selected, and sucrose utilization phenotypes were confirmed. Each knockout strain made in the W genetic background was also complemented using the appropriate overexpression plasmid.

Growth rate analysis in MTPs.

Growth rate analysis of the recombinant E. coli strains (with plasmids bearing the csc genes) was carried out in 96-well microtiter plates (MTPs) using a FLUOStar Omega multidetection microplate reader (BMG Labtech, Offenburg, Germany) as described previously (24).

Growth in shake flasks.

E. coli W and the knockout strains were grown overnight on LB plates at 37°C. Three single colonies were transferred into 250-ml baffled Erlenmeyer shake flasks containing 25 ml of M9 (23) with trace metal solutions (mg/liter) as described in the work of Bruinenberg et al. (27) except that the concentration of Na₂MoO₄·2H₂O was 10-fold higher: EDTA, 15; ZnSO₄·7H₂O, 4.5; CoCl₂·6H₂O, 0.3; MnCl₂·4H₂O, 1.0; CuSO₄·5H₂O, 0.3; CaCl₂·2H₂O, 4.5; FeSO₄·7H₂O, 3; Na₂MoO₄·2H₂O, 0.4; H₃BO₃, 1.0; and KI, 0.1, containing sucrose (20 g/liter or 2 g/liter) or glucose (20 g/liter). Flasks were incubated at 37°C and 250 rpm. When the optical density at 600 nm (OD₆₀₀) was 1, cells were centrifuged at 10,000 × g for 2 min at room temperature (RT), washed, and resuspended in 25 ml of the growth medium. The cells were then centrifuged again and resuspended in 5 ml of the growth medium. Cells were then transferred into three 250-ml baffled Erlenmeyer shake flasks containing 25 ml of fresh medium to a final OD₆₀₀ of 0.1. Flasks were incubated as described above. To determine growth rates, growth was monitored using a spectrophotometer at a wavelength of 600 nm. Samples were taken every 40 to 60 min until cells reached stationary phase.

Sugar analyses.

Extracellular samples were collected from shake flask fermentations by centrifugation, and the supernatant was stored at −20°C for subsequent chromatographic analyses. Concentrations of sugars (sucrose, glucose, and fructose) were analyzed by high-pressure liquid chromatography (HPLC) using an Agilent 1200 Series chromatograph (Agilent, Forest Hill, Victoria, Australia) as described previously (26).

Enzyme assays.

From glycerol stocks, cultures were streaked out on M9S20 agar. After overnight growth, cells from a single colony were inoculated into 5 ml of M9S20 medium and incubated overnight at 37°C and 200 rpm. The overnight culture was transferred to 50 ml M9S20 in a 200-ml baffled shake flask and grown at 37°C and 200 rpm until mid-log phase (OD₆₀₀, 0.5 to 1). Culture (20 ml) was centrifuged (5,000 × g, 10 min, 4°C), and the pellet was resuspended in 2 ml of 100 mM phosphate buffer (pH 6.5). Crude extracts of E. coli were prepared using a Mini Beadbeater (Biospec, Bartlesville, OK) set at 4,800 rpm for three 1-min agitations. Cellular debris was removed by centrifugation (14,000 × g, 10 min, 4°C). Supernatants were transferred to fresh tubes and either used immediately (fructokinase assays) or stored at −20°C for later use.

Invertase activities were quantified by measuring the amount of glucose released after sucrose hydrolysis (28). Cell extracts (in 100 mM phosphate buffer, pH 6.5) were incubated with 50 mM sucrose at 37°C for 1 h. The reaction was stopped by heating at 95°C for 2 min. Glucose concentrations were measured using HPLC. Invertase activities (U) are expressed as nanomoles of glucose per minute per milligram of protein.

Fructokinase activities were assayed spectrophotometrically at 340 nm by coupling the formation of fructose-6-phosphate and glucose-6-phosphate, respectively, to the oxidation of NADH as previously described by Sprenger and Lengeler (29). The fructokinase assay mixture contained the following in 0.2 M Tris-HCl, pH 7.8: 5 mM ATP, 10 mM MgCl, 5 mM NADP, 5 mM fructose, phosphorglucose isomerase (3 U ml⁻¹), glucose-6-phosphate dehydrogenase (2.5 U ml⁻¹), and cell extracts. The assay was performed at 25°C. Under these conditions, the amount of phosphorylated products formed in the assay is proportional to the amount of NADP reduced as measured by the change in absorbance at 340 nm per min per mg protein. Protein concentrations were determined by the Bradford method using a Bio-Rad protein assay reagent (Bio-Rad Laboratories, Richmond, CA) with bovine serum albumin as a protein standard (25).

RNA extraction.

E. coli W and the knockout strains were grown at 37°C in M9 medium with 20 g/liter sucrose, 2 g/liter sucrose, 20 g/liter glucose, or 20 g/liter lactose to an OD₆₀₀ of 1. The cell suspension (1 ml) was inoculated into 25 ml of fresh medium and incubated at 37°C. When the culture reached an OD₆₀₀ of 0.8 (early exponential phase), cells were collected by centrifugation. RNAprotect bacterial reagent (Qiagen, Doncaster, VIC, Australia) was added to stabilize RNA according to the manufacturer's recommendations. The cell pellets were stored at −20°C. RNA was purified using an RNeasy RNA isolation kit (Qiagen, Doncaster, VIC, Australia), and the samples were treated with RQ1 RNase-free DNase (Promega), as recommended by the manufacturer. DNase-treated RNA was repurified using a second RNeasy column. Purified RNA was eluted with 30 μl of RNase-free water and stored at −80°C. RNA integrity was determined using an Agilent 2100 Bioanalyzer (Agilent, CA).

RT-PCR.

Real-time PCRs (RT-PCRs) were performed using Superscript III Plus RNase H⁻ reverse transcriptase (Invitrogen, Mulgrave, Australia) with random hexamer primers (Invitrogen, Mulgrave, Australia) according to the manufacturer's instructions. cDNA samples were stored at −20°C. Amplification, detection, and analysis of mRNA were performed using an ABI7900 sequence detection system (Applied Biosystems). An Eppendorf epMotion 5075 robotics system was used to set up the reaction mixtures. Reactions were performed in 10-μl volumes with 0.2 μM concentrations of the appropriate forward and reverse PCR primers (Table 2) and 1 μl of a 20× dilution of cDNA using Platinum SYBR Green qPCR Supermix UDG (Invitrogen, Mulgrave, Australia) according to the manufacturer's instructions. Annealing and extension were performed at 65°C for 30 s. For each set of primers, a standard amplification curve was plotted (cycle threshold [C_T] against log of concentration) to determine the amplification efficiency of the primer pair/gene combination. C_T values were standardized using both amplification efficiency and an internal reference gene. To select an appropriate internal reference gene, two different d-lactate dehydrogenase genes (dld and ldhA) and the 16S rRNA gene (rrsA) were tested. The dld gene was selected since transcript levels of rrsA were much higher than those of the genes being measured for the experiment, and the amplification efficiency of ldhA was found to vary between different strains. In order to perform the inference analysis, logR values are used: logR = −logA · C_T + logA_ref · C_T,ref, where A is the amplification efficiency for the target gene, A_ref is the amplification efficiency for the internal control, C_T is the cycle threshold time for each target gene, and C_T,ref is the cycle threshold time for the internal control. Experiments were performed with three technical replications for each gene/condition and three biological replications for each strain.

Statistical analysis.

The enzyme activity, growth rate, and quantitative reverse transcription-PCR (QRT-PCR) data set were analyzed using multiway analysis of variance (ANOVA). Before performing ANOVA, it was confirmed that the data satisfied the appropriate criteria, i.e., that the residuals could be assumed to follow a normal distribution with constant variance (Bartlett's test for homogeneity of variance). All statistical analyses were performed using R software.

RESULTS AND DISCUSSION

E. coli W grows particularly quickly on sucrose, a carbon source of industrial importance. Sucrose is metabolized in E. coli W via the csc genes; however, relatively little is known about the molecular control of this regulon. The csc regulon from E. coli W is almost identical to the previously characterized csc regulon from EC3132 and to that of other strains containing the csc regulon; changes at the nucleotide level are mostly silent or conservative at the protein level. As shown previously for EC3132 (11), E. coli W could grow on high concentrations (2%) of glucose or sucrose (Fig. 1) but could not grow on low (0.2%) sucrose (Table 3). The high growth rate of ∼1 h⁻¹ on 2% sucrose or 2% glucose was similar to that reported previously for E. coli W on minimal medium (26). Also as shown for EC3132, the csc genes were transcriptionally repressed in the absence of sucrose and derepressed in the presence of 2% sucrose (Fig. 2). The repressor CscR is itself constitutively expressed at low levels under all conditions (Fig. 2A and B). Consistent with cscR mutants generated in previous studies (10, 11, 26), knockout of cscR enabled growth on low sucrose (Table 3; Fig. 1); repressor gene expression could not be detected (Fig. 2B), and the remaining csc genes (cscA and cscKB) are derepressed (Fig. 2C, D, and E). Invertase and fructokinase activities were also detected in this strain at low sucrose (Fig. 3). The growth rate was not affected by the cscR knockout at 2% sucrose; the growth rate in 0.2% sucrose was about 55% of the growth rate in 2% sucrose (Fig. 1). The similarities in sequence and behavior between the csc regulon in E. coli W and that in E. coli EC3132 suggest that information derived from E. coli W analysis will be broadly applicable to csc-controlled sucrose utilization.

Fig 1.

Growth rates of knockout strains grown in a shake flask using minimal medium supplemented with 2% glucose, 2% sucrose, or 0.2% sucrose. Errors are standard deviations (n = 3).

Table 3.

Map of the csc genes in E. coli strains and the strain phenotype on M9 agar plates supplemented with 2% or 0.2% sucrose^a

^aA plus sign indicates growth, while a minus sign indicates no growth.

Fig 2.

Quantitative real-time PCR analysis. LogR values are presented for expression levels of each gene/condition relative to the internal control, dld. n = 3 biological replications; error bars are standard deviations; **, P < 0.01. (A) Relative expression levels of csc genes for the control conditions (grown on glucose or lactose). (B to E) Relative expression levels of cscR (B), cscA (C), cscK (D), and cscB (E) in strains grown in 2% sucrose (open bars) and 0.2% sucrose (shaded bars).

Fig 3.

Enzyme activities in chromosomal knockout mutants. Invertase (A) and fructokinase (B) are expressed as nanomoles of product per minute per milligram of protein. Error bars are standard deviations (n = 2).

cscA and cscB are essential for sucrose utilization in the wild-type strain.

In order to characterize the contributions of the different catabolic csc genes to sucrose utilization, we performed a detailed chromosomal knockout analysis. The growth of each knockout mutant was examined on minimal agar plates containing either 2% or 0.2% sucrose as a sole carbon source (Table 3). When the entire csc regulon was removed, the mutant could not grow on sucrose. Previously, we sequenced the E. coli W genome and observed that no other known sucrose uptake and utilization systems were present (6). These data confirm that no other sugar transport/utilization system can compensate for the loss of the csc regulon.

Strains with cscA or cscB knockouts could not grow on sucrose (Table 3), demonstrating that these two genes are essential for sucrose utilization in the wild type. It is well known that disaccharide transporters display plasticity in substrate recognition (30); the essentiality of cscB demonstrates that no other transporter in E. coli W is capable of transporting sucrose. This result also implies that inversion of sucrose occurs intracellularly in the wild-type E. coli W, in contrast to the yeast sucrose utilization system, where sucrose can be hydrolyzed outside the cell by extracellular invertase (31, 32). This is consistent with the presence of the cscB transporter gene in the csc regulon. In addition, the essentiality of cscA demonstrates that no other invertase activity is available in E. coli W.

cscK is not essential for sucrose utilization.

WΔcscK could grow on 2% sucrose, demonstrating that the cscK gene is not essential for sucrose utilization (Table 3). Furthermore, the growth rate at 2% sucrose was not compromised by the loss of cscK (Fig. 1). Fructokinase activities were in fact relatively low in all strains compared to invertase activities (Fig. 3); the low activity of the fructokinase is curious considering that the sucrose utilization model indicates that fructose must be phosphorylated prior to entry into the glycolytic or pentose phosphate pathways (10, 11, 33). Furthermore, fructokinase activity could be detected in the K-12 and W strains growing on glucose and in the cscK-knockout strains growing on high sucrose (Fig. 2), despite transcription of cscK being strongly downregulated (when growing on glucose) or completely absent (in the case of the knockout strains and K-12) under these conditions (Fig. 2A). Together, these data suggested the presence of an alternate pathway for fructose utilization (this will be discussed further below).

Knockout of cscK relieves repression at low sucrose levels.

Surprisingly, knockout of cscK (WΔcscK) enabled growth on low sucrose (Table 3; Fig. 1). Initially, we speculated that derepression happened as a result of interference with repressor binding due to the change in sequence context downstream of the binding site in the knockout. This might allow expression of cscB in the cscK knockout. However, when WΔcscK is grown in the presence of noninducing sugars (glucose or lactose), the transcription of the csc catabolic genes is identical to that observed in the wild type under the same conditions (with the exception of the loss of the cscK transcript) (Fig. 2A). Together, these data indicate that repressor activity is not affected in the cscK knockout.

Since the CscR repressor system is clearly intact and functioning in WΔcscK, derepression in this strain must have been effected by an increase in the concentration of the unknown sensor metabolite. This may be associated with an increase in the intracellular concentration of unphosphorylated fructose as a result of deletion of the fructokinase. Free intracellular fructose (and to a lesser extent, fructose-1-phosphate) is a molecular inducer for the Scr sucrose PTS regulon (sucrose-6-phosphate, sucrose, and glucose-6-phosphate are not inducers) (34). This may also be the case for the csc regulon. Alternatively, the increased concentration of the sensor metabolite could be a result of increased transport activity by cscB. A previous study showed that in addition to truncation or knockout of cscR, a gain-of-activity mutation in the cscB gene can confer growth on low sucrose concentrations in EC3132 (10, 11). It is known that fewer full-length RNAs are transcribed from genes near the end of an operon than from those near the beginning of an operon (35), and removal of cscK places the cscB gene in much closer proximity to the promoter than it was in the wild-type strain. Transcript analysis indicated that cscB was transcribed at similar levels in WΔcscK and the wild-type (WT) strain at 2% sucrose (Fig. 2E); however, cscB transcript levels were significantly higher on 0.2% sucrose than on 2% sucrose (P = 0.006). This might explain improved sucrose utilization at 0.2% sucrose; however, if the improvement in cscB transcription was solely related to its location relative to the transcription start site, transcription levels would be expected to be similar at both low and high sucrose. These data imply that sucrose concentration may have a direct effect on moderation of transcription levels which is independent of repressor activity, with expression being inversely proportional to sucrose concentration. In support of this, invertase (cscA) transcript levels were also higher at 0.2% than at 2% sucrose for all strains (Fig. 2C); this will be further discussed below.

Although differences in transcript levels at exponential growth might give some clues as to how sucrose is metabolized in these strains, ultimately it is not possible to infer from these measurements whether reduced conversion or increased transport caused increased accumulation of the sensor metabolite and derepression in the early stage of cultivation. Once derepression has been effected, sufficient transcription of catabolic genes will ensure that the available sucrose is utilized.

In the presence of high sucrose, the fructose moiety is not used by the cscK-knockout strain.

To examine what happens to fructose in the absence of cscK, we first analyzed extracellular sugars to determine if the fructose was excreted from the cell or was metabolized inside the cell (Fig. 4). At high sucrose concentrations, sucrose is gradually consumed with little/no glucose or fructose accumulating in fermentations using the wild-type or repressor knockout strains (Fig. 4A and B). However, in the cscK-knockout strains, fructose accumulated in the growth medium at approximately stoichiometric levels (20 mM fructose at the end of the cultures where 20 mM sucrose had been metabolized) (Fig. 4D, F, and H). This demonstrated that the fructose moiety is not utilized under these conditions. The fact that the growth rate is not compromised at high sucrose concentrations (Fig. 1) and that fructose is exported from the cell and not reimported suggests that the glucose moiety is sufficient for maximal growth under these conditions. Despite this, the absence of fructose in the medium in wild-type and cscR-knockout strain fermentations indicates that CscK is responsible for incorporation of fructose from inversion into the glycolytic pathway in these strains. However, clearly the fructose moiety is not a major determinant of growth rate.

Fig 4.

Culture growth and sugar utilization of E. coli strains grown in shake flasks. OD₆₀₀ was used to measure growth (□). Sugars were measured in the extracellular medium by HPLC: sucrose (■), glucose (▲), and fructose (●). The strains and growth conditions are as follows: W in 2% sucrose (A), WΔcscR in 2% sucrose (B), WΔcscR in 0.2% sucrose (C), WΔcscK in 2% sucrose (D), WΔcscK in 0.2% sucrose (E), WΔcscK Δmak in 2% sucrose (F), WΔcscK Δmak in 0.2% sucrose (G), WΔcscRK in 2% sucrose (H), and WΔcscRK in 0.2% sucrose (I).

At low sucrose concentrations, an alternative fructose utilization pathway is available.

At low sucrose concentrations (0.2%), no fructose accumulation was observed in the fructokinase-knockout strains (Fig. 4E, G, and I). Furthermore, the growth rate was decreased in the cscK knockout relative to the cscR knockout (Fig. 1). Together, these observations suggest that carbon availability becomes limiting for the cscK knockout at 0.2% sucrose and that other methods to utilize the available fructose moiety are engaged.

As noted previously, our fructose phosphorylation experiment indicated that an alternative fructose→fructose-6-phosphate activity was available in both W and MG1655 strains (Fig. 3B). For the sucrose PTS, inactivation of the fructokinase component (ScrK) also does not affect growth on sucrose (14). This is due to phosphorylation of intracellular fructose by manno(fructo)kinase (mak), a gene which is cryptic in wild-type K-12 but has broad substrate specificity for d-riboses and can be decryptified by mutation of various kinases (15, 36–38). A search in the E. coli W genome sequence (6) identified a mak gene. To examine whether Mak was contributing to fructose phosphorylation and sucrose utilization, we deleted the mak gene in the WΔcscK strain background (WΔcscK Δmak). Fructose still accumulated in the growth medium at stoichiometric levels at 2% sucrose (Fig. 4F). This strain could also grow on low sucrose (Fig. 1); however, fructokinase activities were undetectable (Fig. 3). Though these experiments demonstrate that the Mak gene encodes an alternative fructokinase activity, there is no evidence that Mak has any functional relevance as the growth rate relative to that of the cscK-knockout strain was not compromised by loss of mak at either 2% or 0.2% sucrose (Fig. 1). Curiously, mak-mediated fructokinase activity is much higher at 2% sucrose than at 0.2% sucrose (Fig. 3; compare in cscK- and cscRK-knockout strains); it is therefore unlikely that mak accounts for fructose utilization at low sucrose levels.

A fructose PTS (the fru regulon) is encoded on the E. coli W genome, and E. coli W can grow on fructose as a sole carbon source (6). This is an obvious alternative route for fructose utilization in the cscK-knockout strains. However, under normal conditions the fructose must be extracellular for the PTS to efficiently phosphorylate it, since phosphorylation is coupled with transport (39). This clearly does not happen in the case of the high sucrose fermentation, as sucrose accumulates in the extracellular medium (Fig. 4D, F, and H). In the case of low-sucrose conditions, there are two possible mechanisms for PTS-mediated fructose utilization. First, fructose which is excreted into the medium may be imported and phosphorylated using the normal PTS mechanism. However, there is no evidence that fructose is ever excreted at low sucrose concentrations since we could not detect fructose in the extracellular medium (Fig. 4C, E, G, and I). It is also theoretically possible that intracellular fructose can be phosphorylated by both fructose and mannose PTS fructokinase functions (reviewed in reference 39). The transport and carbohydrate phosphorylation functions of EIIs are coupled under normal physiological conditions but are mechanistically separate processes. Some EIIs can carry out facilitated diffusion at low rates when they are unphosphorylated (i.e., in the presence of high glucose; phosphorylation when glucose is low/absent significantly increases the transport rate). Unphosphorylated carbohydrates can dissociate after transport and then rebind and be phosphorylated by EII at low efficiency. This mechanism may provide an alternative route for phosphorylation of intracellular fructose at low concentrations, thus explaining the absence of fructose in the extracellular medium at low sucrose levels.

In a separate transcriptomics study, we observed that the fructose PTS (but not the mannose PTS) was upregulated during growth on sucrose relative to glucose in the wild-type strain (Y. Arifin, unpublished results). This is not surprising given the general derepression of sugar catabolic genes observed in the absence of glucose-mediated carbon catabolite repression (39). We also observed that transcript levels of fruA (encoding the fructose PTS permease) were significantly higher in the cscK-knockout strain than in the cscR-knockout strain (Fig. 5). This suggests that the fructose PTS kinase function may be available for phosphorylation of the fructose moiety and compensate for loss of cscK. Levels of fruA transcript in the cscK knockout are higher than those in the cscR knockout at both 0.2% and 2% sucrose; however, obviously at 2% sucrose the PTS is not functioning for uptake of extracellular fructose. Therefore, at low sucrose concentrations, the fructose PTS may provide an alternative route for utilization of the fructose moiety; however, this pathway is not available at high sucrose concentrations. There is no obvious mechanism by which the PTS might be inhibited at high sucrose but not at low sucrose; however, the intracellular concentration of unphosphorylated fructose is presumably higher at 2% than at 0.2%, and this may be important.

Fig 5.

Quantitative real-time PCR analysis of fruA gene in WΔcscK and WΔcscR grown in high (2%) and low (0.2%) sucrose concentrations. LogR values are presented for expression levels of the fruA gene relative to the internal control, dld. n = 3 biological replications; error bars are standard deviations; **, P < 0.01; *, P < 0.05.

Regardless of the mechanisms by which fructose is utilized at low sucrose and by which utilization is repressed at high sucrose, it is clearly unfavorable to utilize fructose at high sucrose concentrations in the absence of CscK. If carbon is not limiting at high sucrose (as we presume from the growth rate data), it would be energetically unfavorable to spend phosphoenolpyruvate (PEP) for phosphorylation of the fructose moiety, whereas at low sucrose (under carbon-limiting conditions), it becomes favorable to phosphorylate and utilize the available fructose. Whatever mechanism is at play for fructose utilization in the absence of CscK, it clearly functions only when carbon is limiting (at low sucrose). Presumably, the cell has less control over the activity of CscK and cannot prevent fructose phosphorylation (and the associated ATP cost) when it is present. Acquisition of the csc genes is thought to be a relatively recent evolutionary event in E. coli (11); possibly, such control mechanisms have yet to evolve. Alternatively, ATP-based phosphorylation by CscK is energetically preferred over substrate-level phosphorylation by the PTS, as fructose does not accumulate in the wild-type strain harboring CscK.

Low sucrose levels are associated with increased invertase transcription and enzyme activity.

On 2% sucrose, all mutant strains containing the invertase gene had levels of invertase activity similar to that of the wild-type strain (two-way ANOVA, P = 0.1169) (Fig. 3A). However, on 0.2% sucrose, the invertase activity levels were ∼60% higher than that seen on 2% sucrose (two-way ANOVA, P = 1.935 × 10⁻⁸). This increase in activity was matched by an 80 to 100% increase in cscA transcript levels (two-way ANOVA, P = 0.002) (Fig. 2C). Conversely, fructokinase transcript levels and enzyme activities were similar on both 2% and 0.2% sucrose in WΔcscR, the only strain containing cscK that grew on 0.2% sucrose (other discrepancies in fructokinase activities have been discussed above). For permease transcripts, the pattern was more complicated. No significant difference was observed between transcript levels on low and high sucrose where the repressor was knocked out; when the repressor was present (in the cscK knockout), transcription was again higher on low sucrose. These data provide further circumstantial evidence that sucrose concentration (on top of the presence/absence of sucrose) affects transcription of the csc genes, with a particularly strong effect on the cscA transcription unit.

Combinatorial KO of cscR and cscK confers improved growth rates on low sucrose.

Deletion of cscR in the cscK deletion strain (double cscRK deletion) resulted in a higher growth rate on 0.2% sucrose than those for the individual knockout strains (WΔcscR and WΔcscK) (P < 10⁻⁵ for both) (Fig. 1). An increase in invertase transcripts at both sucrose concentrations relative to the individual knockout strains and to the WT at 2% sucrose was observed (Fig. 2C). However, all three strains had similar invertase activities of ∼1,700 nmol min⁻¹ mg⁻¹ at 2% sucrose and ∼2,500 nmol min⁻¹ mg⁻¹ at 0.2% sucrose, so the increase in relative transcript levels clearly does not affect final enzyme activity for CscA. Transcript levels for cscB were not significantly different for any of the knockout strains at 0.2% sucrose (Fig. 2E). These data suggest that neither CscB nor CscA is responsible for the improved growth rates in the double knockout. Since either there is no difference in transcript (in the case of cscB) or the difference in transcript does not carry through to the protein level (in the case of cscA), it is not clear why the growth rate is improved in the double deletion strain. Indeed, there is no obvious reason why deletion of cscR should improve growth rates in an already derepressed strain. We are currently investigating this phenomenon further. Regardless of the mechanism, the WΔcscRK strain may have a further improved performance for production of industrial compounds in fed-batch fermentation given that the WΔcscR strain has an improved performance relative to the wild type (26).

Overexpression of cscAB confers reasonable plasmid-mediated sucrose utilization in E. coli K-12.

The ability to easily confer a sucrose utilization phenotype on non-sucrose-utilizing strains is desirable for industrial applications. However, we and others have previously observed plasmid instability (and, consequently, phenotypic instability) when plasmids are used to confer sucrose catabolic properties on non-sucrose-utilizing strains of E. coli (9, 24, 40). Here, individual genes, as well as gene combinations, were tested by overexpression in both K-12 (DH5α) and WΔcscRAKB to examine the effect of the genetic background of different strains on sucrose utilization. Overexpression of cscA alone has previously been shown to confer sucrose utilization in different E. coli strains, albeit with low growth rates (9, 33, 41, 42). We also observed this for K-12 (DH5α) and WΔcscRAKB (Table 4). The growth rate of WΔcscRAKB carrying the pCSCA plasmid was quite low on both 2% and 0.2% sucrose (0.26 h⁻¹ and 0.20 h⁻¹, respectively); in the case of K-12, the growth rate was extremely low, and accurate growth rate analysis could not be performed due to phenotypic instability. Sahin-Tóth et al. (41) originally suggested that overexpression of the invertase might result in leakage of the excess invertase across the cell membrane; however, this could not be shown conclusively. Lee et al. (33) later demonstrated that CscA leakage did indeed happen, as they could identify invertase activity in the extracellular medium. This provides a mechanistic explanation for cscA-mediated sucrose utilization in plasmid-overexpressing strains. This mechanism does not occur in the wild-type strain, which utilizes sucrose intracellularly exclusively (see above). Lee et al. (33) also proposed that resulting extracellular fructose was imported via the fructose PTS; our data (discussed above) suggest that this may be possible at low sucrose concentrations but not at high sucrose concentrations.

Table 4.

Growth rateS of DH5α and WΔcscRAKB::FRT harboring various different csc genes overexpressed from a high-copy-number plasmid using the lac promoter (cscA) or the native csc promoter (cscAK and cscAB)^a

Strain	Growth rate (h−1)
	2%	0.2%
DH5α/pCSCA	NM	NM
DH5α/pCSCAK	NM	NM
DH5α/pCSCAB	0.54 ± 0.09	NM
WΔcscRAKB/pCSCA	0.26 ± 0.02	0.20 ± 0.02
WΔcscRAKB/pCSCAK	0.82 ± 0.04	0.38 ± 0.02
WΔcscRAKB/pCSCAB	1.15 ± 0.1	0.76 ± 0.02

^aAll strains were able to grow on minimal agar plates with 2% or 0.2% sucrose. Growth rates were determined using MTP cultures in the same medium (liquid). NM, growth rate was not measurable in MTPs. Errors are standard deviations (n = 3).

Combinatorial overexpression of cscK or cscB improved growth rates relative to expression of cscA only (Table 4). This demonstrates that, where growth rate is limited (in the case of the cscA overexpression strain), the presence of CscK can improve sucrose utilization despite the fact that it is not essential for sucrose utilization. Consistent with our knockout data showing that CscB is more important for sucrose utilization than is CscK, co-overexpression of cscB with cscA resulted in a higher growth rate than that of co-overexpression of cscK with cscA. Previously, we have also examined the effect of overexpression of cscAKB in the WΔcscRAKB strain (24). The growth rate of this strain on 2% sucrose was the same as the growth rate that we observed for WΔcscRAKB harboring the cscAB construct. This demonstrates that the maximum growth rate on 2% sucrose in the WΔcscRAKB strain can be achieved by overexpression of just two genes, cscA and cscB. This is also consistent with our knockout data, which showed that removal of cscK did not affect the growth rate on 2% sucrose (Fig. 1). However, the growth rate of K-12 on 2% glucose is 0.7 h⁻¹ (24), significantly higher than that of the K-12 cscAB overexpression strain on 2% sucrose (0.54 h⁻¹; Table 4). Our previous experiments demonstrated that when cscAKB was overexpressed from a plasmid, the growth rate was only 0.2 h⁻¹ on sucrose, and severe phenotypic instability was observed (24). It is not immediately apparent why this is the case; clearly, however, transfer of cscAB without cscK is the best method to achieve the most efficient sucrose utilization phenotype when using a plasmid-mediated overexpression system in K-12. Chromosomal integration of cscAB may also be sufficient to confer reasonably good sucrose utilization on non-sucrose-utilizing strains; integration of smaller DNA sequences is technically less challenging, so decreasing the number of genes required to confer a good phenotype is desirable for strain construction.

In the K-12 strain, only the cscAB plasmid conferred stable growth and only on 2% sucrose (Table 4). However, the growth rate was half of that of the W-derived strain harboring the same construct. Interestingly, no phenotypic instability was observed in the WΔcscRAKB strain for any of the plasmids. Together, these observations suggest that W has an adapted advantage over K-12 for sucrose utilization. This may be due to a contribution of genes other than csc genes to sucrose utilization in W. The W genome is significantly larger than the K-12 genome and carries 271 more genes (6). Genes found only in W might be contributing to the improved phenotypic stability. Alternatively, differences in expression patterns of specific genes may be responsible. With regard to the relative growth rates, it is known that K-12 strains do not grow as fast as W, even on the preferred carbon source, glucose (24, 26). It is therefore likely that the slower growth of K-12 on sucrose is related to a general physiological difference between the two strains and is not a sucrose-specific phenomenon.

Conclusions.

The data presented here and previously (10, 11) indicate that the csc regulon shares several regulatory similarities with the lac operon. However, unlike the lac operon, the csc regulon includes two distinct transcription units. Several lines of evidence presented here suggest that sucrose affects transcription of the two units (and, consequently, sucrose catabolism) differently at high and low sucrose concentrations (at least, in the absence of CscK). It is possible that this is mediated mechanistically through the fructose moiety. This is consistent with recent speculation around the importance of fructose as a carbohydrate source for bacteria. There is strong circumstantial evidence that fructose catabolism was a driver for evolutionary development of carbohydrate metabolic pathways in bacteria (reviewed in reference 43), and it is thought that glycolysis originally evolved as a fructose catabolic pathway (44–46). The catabolite repressor/activator (Cra) protein (originally characterized as the repressor for the fructose regulon and known as FruR) is a pleiotropic global transcription factor which regulates a wide variety of genes in both positive and negative fashion to determine whether metabolism will be fermentative or oxidative/gluconeogenic (43, 47). Micromolar concentrations of fructose-1-phosphate or millimolar concentrations of fructose-1,6-diphosphate bind to Cra, displacing it from its binding site and presumably derepressing negatively regulated transcripts and preventing activation of positively regulated transcripts. Interestingly, in the csc promoter, a sequence very similar to the consensus for the Cra binding site (RSTGAAWCSNTHHW, where R represents A/G, S represents C/G, W represents A/T, H represents A/C/T, and N represents any) can be found; it overlaps the transcriptional start site and the putative CscR binding site for cscA and is 160 bp upstream of cscKB. Future work will be directed at determining if Cra is involved in regulation of the csc genes and at examining exactly how the fructose moiety affects sucrose metabolism. This may provide a mechanism explaining the sucrose concentration-dependent responses that we have observed here.