ABSTRACT
The gal operon of Escherichia coli has 4 cistrons, galE, galT, galK, and galM. In our previous report (H. J. Lee, H. J. Jeon, S. C. Ji, S. H. Yun, H. M. Lim, J. Mol. Biol. 378:318–327, 2008), we identified 6 different mRNA species, mE1, mE2, mT1, mK1, mK2, and mM1, in the gal operon and mapped these mRNAs. The mRNA map suggests a gradient of gene expression known as natural polarity. In this study, we investigated how the mRNAs are generated to understand the cause of natural polarity. Results indicated that mE1, mT1, mK1, and mM1, whose 3′ ends are located at the end of each cistron, are generated by transcription termination. Since each transcription termination is operating with a certain frequency and those 4 mRNAs have 5′ ends at the transcription initiation site(s), these transcription terminations are the basic cause of natural polarity. Transcription terminations at galE-galT and galT-galK junctions, making mE1 and mT1, are Rho dependent. However, the terminations to make mK1 and mM1 are partially Rho dependent. The 5′ ends of mK2 are generated by an endonucleolytic cleavage of a pre-mK2 by RNase P, and the 3′ ends are generated by Rho termination 260 nucleotides before the end of the operon. The 5′ portion of pre-mK2 is likely to become mE2. These results also suggested that galK expression could be regulated through mK2 production independent from natural polarity.
INTRODUCTION
Polycistronic operons in bacteria show a differential expression of the constituent cistrons (1). A Northern blot analysis showed that there are 6 different species of mRNA specific to the galactose operon in wild-type E. coli cells grown exponentially in the presence of galactose (2). Five of the 6 mRNA species, mE1, mE2, mT1, mK1, and mM1, have their 5′ ends at the transcription initiation region, and their 3′ ends at 5 different locations within the operon, four of which (all but mE2) are at the ends of the galE, galT, galK, and galM cistrons, respectively (Fig. 1A). There is one distinct mRNA species, designated mK2, that has 5′ ends not at the promoter region but at the middle of galT. The existence of these mRNA species automatically establishes a gradient of gene expression, higher in the promoter-proximal region and lower in the promoter-distal region, which has been referred to as “natural polarity” (3). Natural polarity is intrinsically different from what has been known as polarity that is caused by a mutation (4), because it can be observed in cells harboring the wild-type operon (2, 5–9). The term “polarity” refers to the phenomenon in which a mutation in one gene of an operon decreases the expression of the subsequent genes of the operon. The cause for polarity is well established. The cessation of translation by a nonsense mutation uncouples transcription from translation, allowing the transcription termination factor, Rho, to bind to the nascent RNA and terminate transcription at the next available termination signal. This Rho-mediated transcription termination leaves the rest of the operon untranscribed, creating polarity (3, 4, 10, 11).
gal-specific mRNAs and two different types of mCONGRAD. (A) A schematic representation of the galactose operon in E. coli. +1 indicates the transcription initiation site from the P1 promoter (P1). Numbers indicate the positions of the stop codons for the cistrons relative to +1. The five gal-specific mRNAs that differ only in their 3′ ends are shown. Double arrows indicate the location of the primers used for qRT-PCR to measure mCONGRAD. The sixth gal mRNA, mK2, encodes galactokinase and has a 5′ end different from that of other gal mRNAs. The stem-and-loop structure at the end of the operon is shown. (B) The type 1 mCONGRAD measured from WT MG1655 cells grown exponentially in LB with 0.5% galactose using the qRT-PCR primers shown in panel A. (C) The type 2 mCONGRAD measured from MG1655Δcya cells.
In our previous report (2), we could not determine the specific cause of natural polarity in the gal operon. In this study, we asked whether the natural polarity in gal is caused by (i) mRNA processing of a long transcript by endoribonucleases, (ii) differential decay rate of the different gal mRNAs, (iii) intrinsic DNA sequence of the gal operon, or (iv) intraoperonic transcription termination. Results from a quantitative measurement of the ends of transcripts from a series of 3′-end deletion constructs of the gal operon indicated that transcription termination at the end of each cistron is the primary cause of natural polarity.
Expression of the galK gene in the gal operon is regulated by Spot 42, a small RNA (sRNA) that belongs to a group of noncoding RNA of a predominant size range of 50 to 250 nucleotides. The sRNA controls gene expression through sequence-specific binding to the target mRNA, causing target RNA degradation or translation inhibition (12). Spot 42 decreases GalK production by binding to the ribosome binding site of the galK transcript (13) and overproduction of Spot 42 from a plasmid decreases galK transcript (14). Among the 6 gal-specific mRNAs, mK1, mK2, and mM1 have the binding sites for Spot 42. We hypothesized that the target mRNA for Spot 42 is mK2, because only mK2 has the Spot 42 binding site near the 5′ end. Spot 42 may not be able to bind mK1 and mM1 due to the translating ribosomes. During the experiments designed to answer whether the cause of natural polarity resides at mRNA processing, we found that the 5′ end of mK2 is produced by a RNase P-mediated cleavage of a long gal transcript. In this study, we also asked (i) how mK2 is produced and (ii) how the production of mK2 is related to the establishment of natural polarity.
MATERIALS AND METHODS
Bacterial strains and growth conditions.Strain MG1655 was used as the wild-type (WT) strain in this study. Chromosomal deletion strains of the entire gal operon were generated by deleting the corresponding gene(s) from MG1655 using λ red-mediated recombination (15). The primers used are listed in Table S1 at http://cnu.ac.kr/~hmlim/. The MG1655Δcya strain was described previously. The RNase E temperature-sensitive mutant GW20 (W3110 zce-726::Tn10 rne-1), RNase G-defective strain GW11 (W3110 zce-726::Tn10 rng::cat), and strain GW10 (W3110 zce-726::Tn10), from which GW11 and GW20 were derived, were provided by H. Aiba (Nagoya University, Japan). An RNase P temperature-sensitive mutant, NHY322 [Δ(proBlac) ara gyrA thi zic-501::Tn10 rnpA49], and strain NHY312 [Δ(proBlac) ara gyrA thi zic-501::Tn10 rnpA+], from which NHY322 was derived, RNase III mutant strain SDF205 (W3110 rnc105 TD1-17::Tn10), and strain SDF204 (W3110 TD1-17::Tn10), from which SDF205 was derived, were provided by Y. H. Lee (KAIST, South Korea). HME60 (W3110 rho::bla; nine amino acids of rho at the C terminus were replaced by bla) was provided by D. Court (NIH, USA) (16, 17). Except for GW20, GW10, NHY322, and NHY312, which were cultured at 30°C overnight, cells were grown at 37°C in LB medium (10 g tryptone, 5 g yeast extract, and 10 g NaCl per liter of water) supplemented with 0.5% (wt/vol) galactose. After a 1/100 dilution in fresh medium, cells were grown at 30°C for 1 h and then incubated at a nonpermissive temperature (44°C) for an additional hour before harvest and further processing. Nine bases exist between the stop codon of galE and the start codon of galT, while three bases exist between galT and galK. The stop codon of galK overlaps the start codon of galM.
RNA preparation.Equal numbers of cells (about 2 × 108) were harvested at an A600 of 0.6 and resuspended in 50 μl protoplasting buffer (15 mM Tris-HCl, pH 8.0, 0.45 M sucrose, and 8 mM EDTA). Five microliters of lysozyme (50 mg ml−1) was added, and then the sample was incubated for 5 min at 25°C. A phenolic detergent (1 ml TRI Reagent; Molecular Research Center, USA) was added, and the mixture was vortexed for 10 s before incubation for 5 min at 25°C. Chloroform (200 μl; Sigma-Aldrich, USA) was added to the mixture, which was vortexed vigorously for 20 s and then incubated for 10 min at 25°C. The resulting mixture was centrifuged at 10,000 × g for 15 min at 4°C. The aqueous phase (500 μl) was transferred to a new tube, mixed with 500 μl of isopropanol (Sigma), and then incubated for 10 min at 25°C. RNA was collected by centrifugation at 10,000 × g for 15 min at 4°C and washed with 1 ml of 75% cold ethanol. The precipitated RNA was dissolved in 50 μl of RNA storage buffer (Ambion, USA). RNA concentration was determined by measuring the absorbance at 260 nm using a NanoDrop spectrophotometer (Thermo Fisher Scientific, USA).
qRT-PCR.For quantitative real-time RT-PCR (qRT-PCR), genomic or plasmid DNA in the reaction mixture was removed with Turbo DNAfree (Ambion) according to the manufacturer's recommendations. Reverse transcription was performed with the following method: 2.0 μg of total RNA was incubated at 37°C for 2 h in a 20-μl reaction volume containing 4 U of Omniscript reverse transcriptase (Qiagen, Germany), 0.5 mM each deoxynucleoside triphosphate (dNTP), 10 μM random hexamer primer (TaKaRa, Japan), and 10 U of rRNasin (Promega, USA). PCR primer sets used in qRT-PCR are listed in Table S1 at http://cnu.ac.kr/~hmlim/. qRT-PCR was performed in 10-μl reaction mixtures containing 5 μl of iQ SYBR green Supermix (Bio-Rad, USA), 3 μl of nuclease-free water, 0.5 μl each of forward primer (10 mM) and reverse primer (10 mM), and 1 μl of the cDNA template under the following conditions: an initial denaturation step at 94°C for 3 min, and then 40 cycles of 15 s of denaturation at 94°C, 20 s of hybridization at 60°C, and 15 s of elongation at 72°C (CFX96; Bio-Rad). The results from each strain were normalized against those for the rrsB gene coding for 16S rRNA.
3′ RACE and 5′ RACE assay.Total RNA was extracted as described above. For the 3′ RACE assay, RNA ligation was performed at 37°C for 3 h in a 15-μl reaction volume containing 2.5 μg of total RNA, 2 nM synthetic RNA oligomer possessing a 5′-phosphate and 3′-inverted deoxythymidine (27-mer; Dharmacon, USA), 5 U of T4 RNA ligase, and 10 U of rRNasin. The RNA ligation reaction was applied to a G-50 column. One microgram of RNA (eluted from the G-50 column) was reverse transcribed at 37°C for 2 h in a 20-μl reaction volume containing 4 U of Omniscript reverse transcriptase, 0.5 mM each dNTP, 0.4 μM 3RP primer complementary to the RNA oligomer (see Table S1 at http://cnu.ac.kr/~hmlim/), and 10 U of rRNasin. A 2-μl sample of this reaction was used as the template for PCR amplification of the gal cDNA with gene-specific primers and the 3RP primer (see Table S1 at http://cnu.ac.kr/~hmlim/) using HotStar Taq DNA polymerase (Qiagen, Germany). To assay the 3′ ends of the gal mRNAs, the amplified cDNA was purified and used as a template for a primer extension reaction performed in a volume of 20 μl with a 32P-labeled primer (complementary to different regions of the gal operon mRNA; see Fig. 4) and 1 U of Taq polymerase (Qiagen, Germany). The reaction products were resolved on an 8% polyacrylamide urea sequencing gel, and the radioactive bands were visualized after exposure to X-ray film. The 3′ ends of mE2 were not detected in the 3′ rapid amplification of cDNA ends (RACE) assay. We believe that the 3′ ends that appear to be the result of RNA processing did not ligate to the RNA aptamer during the ligation reaction in the 3′ RACE assay.
The procedure for 5′ RACE was almost the same as that for the 3′RACE assay, with a little modification. The 5S rRNA was used as an RNA aptamer for ligation instead of synthetic RNA oligomer. For the reverse transcription, a final concentration of 10 μM random primer (hexamer; TaKaRa, Japan) was used. PCR amplification of the gal cDNA was performed with a forward primer complementary to the 3′ end of 5S rRNA and a reverse primer specific to the galK region (see Table S1 at http://cnu.ac.kr/~hmlim/).
Measurement of gal mRNA decay kinetics.MG1655 cells were cultured until early-log-phase growth was achieved (optical density [OD] of 0.6). Rifampin was added to the culture to a final concentration of 500 μg ml−1, and cells were harvested at 0, 2, 4, 6, and 8 min. Harvested cells (2 × 108 at each time point) were mixed immediately with 10% buffer-saturated phenol in ethanol (1/10 volume) and chilled rapidly on ice. RNA preparation, cDNA synthesis, and real-time PCR were performed as described above.
In vitro transcription assay.The pHL1277 plasmid was used as a DNA template for in vitro transcription. pHL1277 was obtained by cloning the entire galactose operon into the pCC1BAC plasmid. The in vitro transcription reaction was performed using Escherichia coli Eδ70 (Epicentre) according to the manufacturer's instructions. Briefly, DNA template (2 nM) was incubated at 37°C for 5 min in reaction buffer (20 mM Tris-acetate, pH 7.8, 10 mM magnesium acetate, 200 mM potassium glutamate, 1 mM ATP, and 1 mM dithiothreitol [DTT] [18]) containing 2 U Eδ70 and 40 U rRNasin (Promega) in a 47.5-μl reaction. The reaction was initiated by adding 2.5 μl NTP mix (final concentration, 0.1 mM each NTP) to the mixture. After 30 min, the reaction was terminated by phenol-chloroform extraction, and then 30 μl of supernatant was purified using a G-50 column. Purified RNA was then used for qRT-PCR or 3′ RACE assays as described above.
Plasmid construction.The pHL1277 plasmid was constructed by inserting the galactose operon (from −75 to +4333) between the EcoRI and BamHI sites of pCC1BAC (Epicentre Biotechnologies, USA). PCR primers used to amplify the galactose operon from genomic DNA are listed in Table S1 at http://cnu.ac.kr/~hmlim/. Sequences corresponding to Brevibacterium albidum tRNAarg from pCAT-tRNAArg (19) were amplified by PCR using primer sequences shown in Table S1 at http://cnu.ac.kr/~hmlim/, digested with HindIII, and then inserted immediately upstream of the cat gene in pKK232-8 (GE Healthcare, USA) to create the pHL1141 plasmid. DNA fragments containing different portions of the gal operon, from nucleotide position −73 to various positions downstream, were obtained via PCR amplification of the genomic DNA with the corresponding primer pairs (see Table S1 at http://cnu.ac.kr/~hmlim/). The resulting PCR fragments were digested with BamHI and SalI and then ligated to pHL1141 to generate pHL1142, pHL1143, pHL1144, pHL1145, and pHL1146.
Northern blot analysis.Total RNA was isolated as described above. A specific amount (depending on the experiment; see below) of total RNA (with 1 mg/ml ethidium bromide) was resolved by 1.2% (wt/vol) formaldehyde-agarose gel electrophoresis at 5 V/cm for 4 h. After electrophoresis, RNA integrity was assessed under UV light and the RNA was transferred overnight to a positively charged nylon membrane (Ambion, USA) using a downward transfer system (TurboBlotter; Whatman, United Kingdom) and then fixed to a nylon membrane by baking at 80°C for 1 h. The Northern probe was prepared as follows. First, a 500-bp DNA fragment in the galK region (from +2103 to +2603) was prepared by PCR using primers indicated in Table S1 at http://cnu.ac.kr/~hmlim/. The RNA probe (generated from the 500-bp DNA fragment by in vitro transcription) then was labeled with digoxin according to the manufacturer's protocol (DIG Northern starter kit; Roche, Switzerland). The transferred RNA was hybridized with the RNA probe (100 ng/ml) at 68°C for 8 h and detected immunologically using the anti-digoxin antibody by chemiluminescence.
RESULTS
The 6 mRNAs of the gal operon and their measurement.We have demonstrated in our previous report (2) with quantitative real-time RT-PCR (qRT-PCR) using primers specific to each cistron of the gal operon (Fig. 1A) that expression of galE is greater than that of galT, and that galM has the lowest level of expression of the four cistrons (Fig. 1B). The level of galK transcript, due to its more distal location from the promoter, was expected to be less than that of galT; however, galK expression actually was found to be greater than that of galT. This exception to the transcription gradient may be due to the presence of another gal-specific mRNA species, mK2 (Fig. 1A and B), which depends on cyclic AMP (cAMP). We termed this phenomenon, which provides natural polarity at the level of transcription, the mRNA concentration gradient (mCONGRAD). Thus, mCONGRAD measured in the WT shows more galK than galT, and we termed this type 1. However, when mCONGRAD is measured in a cya mutant strain, galT is greater than galK, and we termed this type 2 (Fig. 1C). Throughout this study, unless otherwise noted, mCONGRAD was measured in cells grown in LB containing 0.5% galactose. Moreover, total RNA was isolated from 2 × 108 cells during early-log-phase growth (OD600 of 0.6). The same amount of total RNA (2 μg) was used for qRT-PCR.
RNase E, III, and G are not involved in 3′-end generation of the gal mRNAs.To better understand mCONGRAD and its causes, we investigated several possible factors that could influence how the 3′ ends of the gal mRNAs are generated. First, we investigated the effects of processing full-length mRNA (mM1) into different sizes. To accomplish this, we measured mCONGRAD in E. coli strains carrying mutations in endoribonucleases known to be involved in RNA processing. WT MG1655 cells exhibited a typical mCONGRAD status (type 1) with the relative amount of galT, galK, and galM transcripts measured as 0.5, 0.65, and 0.07 of the galE transcript (Fig. 1B). Results from the endoribonuclease mutant strains showed that the mCONGRAD status at a nonpermissive temperature (44°C) for the endonuclease RNase E temperature-sensitive strain (lacking rne) was similar to that of the isogenic strain (rne+) from which the mutant strain was derived (Fig. 2). The mCONGRADs in the G-negative (rng::cat) and RNase III (rnc105) strains were also identical to those of the corresponding WT strains (rng+ and rnc+) (Fig. 2). Although the total amount of transcript in the RNase E and RNase III strains was 0.34 and 0.26, respectively, relative to the corresponding WT strain, the expression ratio of genes to galE did not change (Table 1). These data suggest that RNA processing by RNase E, III, or G is not involved in generation of the 3′ ends of the gal mRNAs.
mCONGRAD measurements of gal mRNAs in RNase mutant strains. GW10 (rne+, RNase E control) cells were grown at 30°C for 1 h and then shifted to 44°C for 1 h. Strains used were GW20 (rne mutant; RNase E temperature sensitive), GW10 (rng+; RNase G control), GW11 (rng::cat; RNase G downmutation), SDF204 (rnc+; RNase III control), SDF205 (rnc105; RNase III downmutation), NHY312 (rnpA+; RNase P control), NHY322 (rnpA49; RNase P temperature sensitive). Independent measurements were performed at least three times. The amount of transcript relative to that of galE in each control is presented.
Expression of the gal genes relative to the first gene, galE
RNase P-mediated cleavage of a pre-mK2 mRNA could yield 5′ ends of mK2 and possibly 3′ ends of mE2.Interestingly, mCONGRAD in the temperature-sensitive RNase P strain (rnpA49) at the nonpermissive temperature (44°C) appeared the same as that observed in a cAMP-deficient (MG1655 Δcya) strain lacking mK2 (Fig. 2 and Table 1). In this type 2 mCONGRAD event, the amount of galK was less than that of galT, yielding a linear gradient of mRNA concentration. This linear gradient in the absence of RNase P activity implies its potential involvement in mK2 production. Therefore, we expected that there would be no mK2 mRNA in the RNase P mutant strain. We visualized mK1, mK2, and mM1 by Northern blotting (Fig. 3) and found that there was no discrete band of mK2 in the RNase P mutant at the nonpermissive temperature.
Northern analyses of mK2, mK1, and mM1 in the RNase P mutant strain. rnpA+, RNase P control strain; rnpA49, RNase P temperature-sensitive strain. Cells were grown at 30°C for 1 h and then shifted to 44°C for 1 h.
It was envisioned that the 5′ ends of mK2 are generated by RNase P-mediated endonucleolytic digestion of a pre-mK2 mRNA. We searched the 5′ ends of mK2 by using the 5′ RACE assay. Briefly, 5′ ends of total RNA were ligated to the 3′ end of the 5S rRNA, which is abundant in total RNA preparations. Reverse transcription was performed using a random primer. PCR then was performed with a primer specific to the 3′ end of the 5S rRNA and a gal-specific primer that binds to the beginning of galK (2). This PCR would amplify the cDNA whose 5′ ends are roughly between +1500 and +2100, where 5′ ends of mK2 might reside (Fig. 1A). The amplified 5′ ends were detected using a primer extension reaction with 5 different radioactive primers. We identified two 5′ ends at +1764 and +1777 that could be observed in a WT control strain of the RNase P mutant but disappeared from the RNase P temperature-sensitive mutant strain at the nonpermissive temperature (Fig. 4). Based on these results, we concluded that the 5′ ends of mK2 are generated by RNase P digestion of a pre-mK2 at +1764 and +1777. By simple logic based on the location of the 3′ ends of mE2, which are at the middle of galT and close to the 5′ ends of mK2, we suppose that the 5′ portion of the gal mRNA generated by the RNase P digestion of the pre-mK2 would become mE2.
5′ Ends of mK2 visualized by 5′ RACE assay of the rnpA+ RNase P control strain and the RNase P temperature-sensitive strain (rnpA49). Cells were grown at 30°C for 1 h and then shifted to 44°C for 1 h. Numbers indicate the positions of the 5′ ends of mK2 relative to the transcription initiation at +1.
Neither RNA decay rates nor gene sequence is involved in mCONGRAD formation.We next addressed whether different decay rates of the gal mRNAs could account for the generation of different species of the gal mRNAs. mCONGRAD was measured in WT cells at 0, 2, 4, and 8 min after addition of rifampin, an inhibitor of bacterial RNA polymerase (RNAP), to the culture. The relative amount of transcript representing each cistron decreased after rifampin treatment. When changes in the amount of galE, galT, galK, and galM transcripts were plotted over time, it became evident that transcripts representing each cistron degraded at nearly the same rates (Fig. 5). The level of transcript corresponding to each cistron is the sum of different mRNA species (e.g., the amount of galE is the sum of mE1, mE2, mT1, mK1, and mM1). Therefore, these results showing that the gal transcripts undergo similar decay rates while maintaining the same mCONGRAD suggests that the general RNA degradation process does not lead to the generation of the 3′ ends of the gal mRNAs.
Decay kinetics of the gal transcripts. Relative amounts of the gal mRNAs to that of galE at time zero are presented. The relative amount of mRNA was plotted against time to determine the decay rate.
We then investigated whether the DNA sequence itself, or a property intrinsic to the DNA sequence at the end of each cistron, could terminate transcription. We measured the mCONGRAD of transcripts derived from in vitro transcription of the entire gal operon cloned into a low-copy-number plasmid, pHL1277 (see Materials and Methods). Our results demonstrate that the amount of transcript representing the first three cistrons, namely, galE, galT, and galK, were identical, and that the level of galM transcript was almost half that of the others (Fig. 6). The data that galM transcription is half that of galK suggest that RNAP alone (without any protein factors) could terminate transcription at the end of galK and that mK1 could be generated by transcription termination at the end of galK. The data that the amount of galE, galT, and galK is the same suggest that an in vivo factor(s) exists that governs the generation of the 3′ end of mE1 and mT1. The amount of galK is greater than that of galM in all of the RNase mutants tested in Fig. 2, suggesting that the galK termination is not the result of RNA processing.
Relative amount of transcript of each gal cistron from in vitro transcription of the entire gal operon.
Transcription termination generates 3′ ends of the gal mRNAs.Because we found that the 3′ ends of mE1, mT1, mK1, and mM1 are not generated by differential endoribonuclease processing or mRNA decay rates, we investigated intraoperonic transcription termination in the gal operon. Based on the locations of the 3′ ends of mE1, mT1, mK1, and mM1 in the gal operon shown in Fig. 1A, we anticipated transcription termination at the end of each cistron. For this, we constructed a series of gal operon deletion mutants and cloned the deletions in a high-copy-number plasmid, pHL1141. As diagrammed in Fig. 7A, each plasmid has the same 5′ portion of the gal operon, including both the P1 and P2 promoters and operators, but different 3′ ends. The plasmid pHL1142 has a portion of the gal operon from −73 to +1031, pHL1143 has up to +1699, pHL1144 has up to +3087, pHL1145 has up to +3791, and pHL1146 has the entire gal operon, from −73 to +4344. The argX gene coding for tRNAarg from Brevibacterium albidum is cloned at the end of each gal deletion. If transcription termination occurs at the end of each cistron, the amount of the tRNAarg transcribed from these plasmids will be the greatest in the longest deletion plasmid, pHL1142, and smallest in the shortest deletion plasmid, pHL1146.
In vivo transcription termination assay. Schematic illustration of the series of plasmids used to measure intraoperonic transcription in vivo. Each plasmid contains a 5′ portion of the gal operon as indicated. For example, pHL1142 has a portion from −73 to +1031. pHL1146 expresses the entire gal operon from −73 to +4344, including the stem-and-loop structure at the end of galM. The 5′ portions of the gal operon were cloned in front of the tRNAarg gene of pHL1141. A chloramphenicol resistance (cat) gene (the determinant) was fused to the 3′ end of tRNAarg. (A) The strong Rho-independent rrnB transcription termination signal (31) was cloned after the cat gene and also in front of the gal promoters to prevent aberrant transcription originating from plasmid DNA. Thus, transcription initiated from the gal promoters would transcribe the 5′ end of the operon DNA, tRNAarg, and cat before termination at the rrnB unless terminated at the intercistronic transcription terminator. The relative amount of tRNAarg was measured in MG1655 (B) and in HME60 (C), the Rho-negative strain. The amount of tRNAarg was measured using qRT-PCR and is presented relative to that of pHL1142. Data represent at least three independent experiments.
We performed qRT-PCR to measure the amount of tRNAarg expressed in MG1655 cells harboring the plasmids described above. Our results demonstrated the formation of a gradient of tRNAarg expression, with the longest deletion construct (pHL1142) showing the greatest tRNAarg expression and the shortest deletion construct (pHL1146) exhibiting the least (Fig. 7B). There was a gradual decrease in tRNAarg levels in cells harboring the intermittent constructs. These data, along with the location of the 3′ ends of mE1, mT1, mK1, and mM1 in the gal operon, suggest the likelihood of transcription termination at the end of each cistron. Using the equation termination frequency = 1 − readthrough/upstream transcripts (20), we were able to calculate the transcription termination frequency occurring at different sites in the operon (Table 2). These data suggested that these terminations are stochastic. Each termination occurs with a finite efficiency of less than 100%. The half-life of the tRNAarg transcripts from the various constructs was measured and found to be similar in all cells, with an average of 1.3 min (see Fig. S2 at http://cnu.ac.kr/~hmlim/). These data indicate that intraoperonic transcription termination generates the 3′ ends of mE1, mT1, mK1, and mM1.
Transcription termination efficiency at the end of each cistron
The transcription terminations are Rho dependent.We next tested whether the transcription termination at the end of each cistron is mediated by the Rho factor. We measured the amount of tRNAarg from the same series of plasmids used above in the Rho-impaired E. coli strain, HME60. The mutant rho gene rho-15 (16, 17) in the HME60 strain has been selected as a nonfunctional transcription terminator, and the corresponding mutant Rho protein differs from that of the WT by nine amino acids at its C terminus (D. Court, NIH, USA, personal communication). Our results showed that, contrary to WT cells, there was little variation in tRNAarg levels in HME60 cells harboring pHL1142, pHL1143, and pHL1144 (Fig. 7C), suggesting that transcription termination did not occur at the end of galE and galT in the rho-15 mutant. Therefore, transcription termination at the end of galE and galT that generates 3′ ends of mE1 and mT1, respectively, is Rho dependent. Transcription termination efficiency at the end of galK drops from 65% in the WT to 8% in HME60, while the corresponding values at the end of galM are 71% in the WT and 30% in HME60, suggesting that those two terminations are partially Rho dependent (Table 2). Considering that the galM transcript level was half that of galK after in vitro transcription of the gal operon (Fig. 6), these results indicate that the DNA sequence itself at the end of galK contains a transcription stop signal (see below). Similarly, transcription termination still occurs at the end of galM even in the absence Rho, because the termination efficiency in HME60 cells was 30% (Table 2). A stem-and-loop structure on mRNA was found 6 bp downstream of the galM translation stop codon, followed by 3 continuous thymine residues. We interpreted this sequence as a conventional Rho-independent terminator. This Rho-independent terminator may explain why 30% termination still occurs in the HME60 strain.
The specific location of the 3′ ends of mK2.The 3′ ends of mE1, mE2, mT1, mK1, and mM1 have been identified and mapped (2), but those of mK2 have not been identified. To investigate the exact location of the 3′ ends of mK2, we performed the 3′ RACE assay to visualize all 3′ ends of the mRNAs generated from the entire gal operon. Briefly, 3′ ends of total RNA were ligated to a synthetic RNA aptamer composed of 27 nucleotides, and reverse transcription was performed using a DNA primer complementary to the RNA aptamer (Fig. 8A). PCR was then performed with the reverse transcription primer and a gal-specific primer, and the amplified 3′ ends were detected using a primer extension reaction with a radioactive primer. With the 5 PCR primers and 21 primer extension primers nested throughout the operon, we were able to visualize most of the 3′ ends of the mRNAs transcribed from the gal operon (Fig. 8A).
Specific location of the 3′ ends of mK2. (A) The nested 3′ RACE assay was performed, and the five PCR primers (thicker arrows) and 21 extension primers (arrows) are shown. The nested 3′ RACE assay was performed in 2 different strains, the WT and rho mutant (rho::bla) strains. Numbers by each band indicate the positions of the 3′ ends relative to +1, the transcription initiation site of the P1 promoter. (B) Only the 3′ ends of mK2 are shown. (C) What is shown as mK1 3′ ends are the results from 3′ RACE assay performed on RNAs transcribed from the gal operon in vitro.
The locations of most 3′ ends of gal mRNAs concurred with the data of our previous report (2), except for mE2 and mK1 (see below). The 3′ ends of mE1 are located 82 to 160 nucleotides downstream from the stop codon of galE. Those of mT1 and mM1 are located 25 to 70 and 30 to 51 nucleotides downstream from the stop codons of galT and galM, respectively (data now shown). Several bands that appeared to be the 3′ ends of mK2 existed at nucleotides +4025 to +4227 toward the end of galM (Fig. 8B). Based on the size of mK2, about 2.2 kb (2), and the location of its 5′ ends (+1764 and +1777) (Fig. 4), we concluded that these are indeed 3′ ends of mK2. When the same 3′ RACE assay was performed in the rho mutant strain, HME60, we found that the bands that appeared as the 3′ ends of mK2 disappeared (Fig. 8B, second lane). These data clearly demonstrated that the 3′ ends of mK2 are generated by Rho termination at the specific locations shown in Fig. 8B.
The specific location of the 3′ ends of mK1.The Northern blot of the gal operon clearly showed a discrete band of mE2 and mK1 (2). However, the 3′ RACE assay performed to generate Fig. 8 did not detect any bands specific to the 3′ end of mE2 and mK1 (data not shown). We considered the 3′ ends detected with the 14, 15, 16, 17, and 18 primer extension primers (Fig. 8A) that are designed to detect the end of galK to be non-gal-specific signals, because these same bands were detected in the MG1655Δgal strain, which lacks the entire gal operon (see Fig. S1 at http://cnu.ac.kr/~hmlim/). The 3′ ends of mK1 reported in our previous publication (Fig. 7 in reference 2) must have been non-gal specific. One possible reason for this is that the 3′ ends of mE2 and mK1 are engaged in further processing or degradation, and because of that the 3′ ends are not ligation proficient to the RNA aptamer during the 3′ RACE assay. We reasoned that we would be able to see the 3′ ends of mK1 if we perform 3′ RACE on RNAs generated by in vitro transcription of the entire gal operon, because no RNA degradation or processing is possible in in vitro transcription. With the 14, 15, 16, 17, and 18 primer extension primers, we found 3′ ends from +3218 to +3287 that appear to be the putative 3′ ends of mK1 (Fig. 8C). We could not detect any DNA sequence upstream of the 3′ ends that could form a stem-and-loop structure on mRNA that appears as a signal for intrinsic transcription termination (21). Taking these findings and the interpretation from Fig. 6 that there is in vitro transcription termination at the end of galK, it is likely that the 3′ ends of mK1 can be generated without any protein factors by nonconventional transcription termination.
mCONGRAD formation immediately after induction of the gal operon.So far, this study has demonstrated that a transcription termination event at the end of each cistron, galE, galT, galK, and galM, generates mE1, mT1, mK1, and mM1, as shown in Fig. 1A. The stochastic nature in each of the transcription terminations suggests that a certain proportion of RNAP molecules that left the gal promoters would dissociate from DNA template at the end of each cistron. However, mK2 is produced by endoribonuclease digestion of the pre-mK2. We questioned how fast and what type of mCONGRAD would be formed as soon as the operon becomes active in transcription. We tested this by measuring the mCONGRAD after inducing derepression of the gal operon. For this, WT cells were grown in M9 minimal media with 0.5% glucose. When the growth reached an OD600 of 0.6, galactose was added to a final concentration of 0.5%.
Cells were taken at 1, 2, 3, and 4 min, and mCONGRAD was measured. The results showed that after 1 min, only galE is induced (Fig. 9). After 2 min, the level of total gal transcripts was 4.5 times that at time zero, suggesting that the operon can be active in transcription after addition of galactose even in the presence of glucose. Nevertheless, after 2 min, type 2 mCONGRAD had already been established. Assuming that the average speed of transcription by the sigma70 RNA polymerase is 42 nucleotides/s (22), it would take 1.8 min to transcribe the entire 4.2-kb gal operon. Thus, the formation of the type 2 mCONGRAD in 2 min after the induction of transcription suggests that a single round of transcription is enough to establish mCONGRAD. These data are consistent with our hypothesis that stochastic transcription termination is the primary cause of mCONGRAD formation. After 3 min, galK started to exceed galT, and type 1 mCONGRAD was established in 4 min (Fig. 9), suggesting that mK2 is being produced while mCONGRAD is formed. Since type 2 is established first after one round of transcription and type 1 follows, it can be suggested that mK2 is produced posttranscriptionally.
Kinetics of gal mRNA production after induction. MG1655 cells were cultured in M63 glucose medium to an OD600 of 0.6, and transcription from the gal operon was induced by addition of galactose to a final concentration of 0.5%. Cells were harvested at 0, 1, 2, 3, and 4 min after induction. The gal-specific mRNA was measured using qRT-PCR, and the values relative to that of the galE at time zero are shown. The means ± standard deviations from at least three independent experiments are shown.
However, it is not clear whether RNase P digestion occurs only on pre-mK2, whose transcription termination has occurred at the 3′ ends of mK2, or if any transcript that is cleaved by RNase P is terminated at the 3′ ends of mK2.
DISCUSSION
The efficiency of transcription termination at each cistron junction is less than 100%, and it is different from one cistron junction to another. This is probably caused by a stochastic behavior of individual RNA polymerase and/or transcription factors on different DNA sequences involved in transcription termination. It is the stochastic nature of transcription termination that establishes the gradient in gene expression that is highest in the first gene (thus, most proximal to the promoter), which is galE. Nature has maintained a gradient in gene expression that is likely to have the first gene product (GalE) in the largest amounts, and that is probably what the cell needs most for the catabolism of d-galactose in E. coli. One way for the E. coli cells to change the gene expression of each cistron in the established natural polarity is to regulate the stochastic nature of transcription termination. These considerations raise the possibility that any factor(s) that could affect the stochastic nature of transcription termination would change gene expression established in natural polarity. It was suggested that Rho termination frequency can be modulated by altering elongation kinetics of RNAP (23). A very precise but subtle change in the expression of each cistron (natural polarity) could be imagined if proteins or other factors regulate the elongation kinetics of RNAP according to the external or internal signals. In addition to the well-known protein factors, such as NusG and NusA, another group of proteins that can be oligomerized on DNA to form a scaffold, such as H-NS, Cnu, and Hha, was recently shown to participate in modulation of Rho termination efficiency (24).
The hexameric Rho protein requires an emerging mRNA region known as a Rho utilization site (rut) for binding and function as a transcription terminator. Typically, the rut site lacks certain secondary structures and is composed of an ∼80 nucleotide stretch called the CG bubble that is rich in cytidine but poor in guanine residues (25). Using a computer algorithm (26), we identified a CG bubble at each Rho-dependent termination site that generates 3′ ends of mE1, mT1, mK2, and mM1 (Fig. 10). Each CG bubble spans about 122, 90, 182, and 114 nucleotides, respectively, and appears upstream of the termination site. These data concur with the current model of Rho termination that Rho binds to mRNA before acting upon RNAP to terminate transcription (27, 28). An interesting observation is that there is an additional larger CG bubble, spanning 191 nucleotides, in the middle of the galE gene from +281 to +472 (Fig. 10, box). Since no transcription termination activity after this CG bubble in the middle of galE has been observed in WT cells and it is located in the middle of the first gene of the operon, where 4 Rho terminations are expected to occur downstream, we propose that this is where Rho is loaded onto a paused RNAP for the terminations downstream. We actually found a transcription pause in this CG bubble (data not shown). Thus, as transcription on the gal operon DNA progresses, RNAP that deploys the associated Rho through the CG bubble identified at the cistron junctions (circled in Fig. 10) would terminate transcription. These considerations raised the possibility that the stochastic nature of Rho termination came at the deployment of Rho from RNAP.
CG bubbles within the gal operon. Cytosine (black) and guanine (gray) frequencies of the nontemplate strand of the entire gal operon are shown below a map of the operon. The high-C, low-G region (CG bubble), known to be the binding site for Rho (rut), is also indicated. The CG bubbles that lead to Rho-mediated termination are marked with ovals, and the first CG bubble at the start of the operon that does not lead to Rho-mediated termination is marked with a rectangle. The upward-pointing arrows indicate the location of the 3′ ends of gal-specific mRNAs. The Emboss Freak program (http://emboss.bioinformatics.nl/cgi-bin/emboss/freak), with a stepping value of 1 and an averaging window of 78, was used to calculate the frequencies and outputs, which were exported to Microsoft Excel for graphical representation.
Expression of galK can be controlled independently from natural polarity.The mCONGRAD status in the mid-log growth phase of the WT strain is type 1, where galK expression is greater than that of galT, forming a nonsmooth gradient of expression. The mCONGRAD in a cya strain is type 2, where galT expression is greater than that of galK, forming a smooth gradient of expression; thus, the order of expression is galE > galT > galK > galM, and there is almost no mK2 in the cya strain. These data suggest that the difference between type 1 and type 2 mCONGRAD is caused by mK2 production: more mK2 is produced in type 1 than in type 2. As mentioned above, the stochastic nature of transcription termination at the end of each cistron establishes the type 2 mCONGRAD. Thus, E. coli cells could regulate galK expression over the already established mCONGRAD by regulating mK2 production. Since it is the RNase P cleavage on a gal transcript that initiates mK2 production, one of the ways to produce more mK2 so that cells can switch to type 1, where galK expression becomes greater than that of galT, would be either to increase the synthesis rate or to decrease the decay rate of mK2. Note that since the RNase P digestion of a gal transcript would create a new 5′ end in front of the galK gene and Spot 42 can bind to the 5′ end of galK and decrease galK transcript levels (13, 14), it is likely that E. coli cells regulate degradation of mK2.
Perhaps Spot 42 and Hfq, bound to a newly generated 5′ end of mK2, block translation and further engage in degradation of mK2 by recruiting RNase E (29). The mK2 degradation by Spot 42 is currently being investigated. Since cAMP-CRP is a negative transcription factor for Spot 42 production (30), the data from this study suggest that any external or internal changes that evoke cAMP increase would not only promote the well-established notion of gal transcription initiation but also exert effects on expression of galK only through mK2 production independently from transcription initiation and from natural polarity.
ACKNOWLEDGMENTS
This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education, Science, and Technology (2013007271). This research was also supported by a grant from Chungnam National University (2012-1689).
FOOTNOTES
- Received 15 February 2014.
- Accepted 26 April 2014.
- Accepted manuscript posted online 2 May 2014.
- Address correspondence to Heon M. Lim, hmlim{at}cnu.ac.kr.
↵* Present address: Sang Chun Ji, Department of Clinical Pharmacology and Therapeutics, Seoul National University College of Medicine and Hospital, Seoul, South Korea; Sang Hoon Yun, Alteogen Inc., Daejeon, Republic of Korea.
REFERENCES
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
