Mycobacterium tuberculosis rrn Promoters: Differential Usage and Growth Rate-Dependent Control

Abstract

Mycobacterium tuberculosis is a slow-growing pathogen and is characterized by a low content of RNA per unit of DNA. rRNAs represent a major proportion of the total RNA pool, and the entire requirement for rRNA is met by transcription from a single rrn operon that is driven by two promoters, P1 and P3. This study attempted to analyze the specific role of the rrn promoter in determining the characteristically low levels of RNA in M. tuberculosis. For this purpose, the activity of the M. tuberculosis rrn promoter as a function of the growth rate was studied by rrn-lacZ promoter fusion, hybridization, and primer extension analysis in M. smegmatis. rrn promoter signals were faithfully recognized in M. smegmatis cultures harboring the rrn-lacZ promoter construct. In M. smegmatis cultures that displayed doubling times varying between 3.06 and 6.5 h, β-galactosidase activity increased ∼sixfold in proportion to the growth rate (μ). There was a corresponding increase in the amount of lacZ-specific mRNA, while the plasmid copy number remained essentially unchanged. For any given μ, the P3 promoter was ∼twofold more efficiently utilized than the P1 promoter. Since both promoters of the M. tuberculosis rrn operon are regulatable as a function of growth rate in M. smegmatis cultures, it is implied that the inherent structure or sequence of the rrn promoter per se is not primarily responsible for the observed lack of modulation of RNA synthesis in M. tuberculosis.

Mycobacterium tuberculosis is the causative agent of tuberculosis and is characterized by slow growth. The study of ribosome regulation is extremely relevant to understanding the molecular basis of the slow growth of this organism, since protein synthesis, so critical to growth, is dependent on ribosomes. M. tuberculosis cultures contain small amounts of RNA per unit content of DNA; the total RNA content varies only twofold between stationary-phase cultures and actively growing cells (38). The reasons underlying this lack of responsiveness can be addressed by directly analyzing rRNA transcriptional activity since rRNA comprises the majority (∼80%) of the total RNA pool of a mycobacterial cell. The production of rRNA is determined by the number of rrn operons, the number of promoters, the nature of the promoter elements, and the efficiency with which the operons are transcribed. Since rRNAs represent a relatively stable population, breakdown is less likely to constitute a major regulatory mechanism and the regulation of rRNA synthesis is expected to occur at the level of RNA chain initiation. In earlier studies from our laboratory, we demonstrated that fast-growing (M. smegmatis) and slow-growing (M. tuberculosis) mycobacteria follow a similar pattern of bacterial growth comprising the lag, logarithmic, and stationary phases, with maximum rRNA levels found during the logarithmic phase of growth (7). The M. tuberculosis rrn operon is driven by twin promoters designated P1 and P3 (21); transcription start points (tsp), an RNase III-processing site, and the +1 of mature 16S rRNA were mapped in our laboratory (37). In Escherichia coli, rRNA synthesis is a rate-limiting step in ribosome production since r-protein expression is regulated by feedback mechanisms sensitive to the rRNA concentration (31). In several bacteria including E. coli, the number of ribosomes varies linearly with the growth rate, μ, over a range of conditions. This phenomenon has been termed growth rate-dependent control (GRDC) of ribosome synthesis, and it serves to maintain the cellular pool of ribosomes at a level commensurate with the requirement of the cell for protein synthesis at all times (18). GRDC of rRNA biosynthesis has been most extensively studied in E. coli and has been shown to occur at the level of rrn expression (2, 12, 16, 17, 24, 30, 34). The exact mechanism by which GRDC is attained is still under active investigation, although a large body of evidence implies a role for ppGpp and/or some translation-linked event (12). Since the rrn operon of M. tuberculosis is expressed from dual promoters, as in other eubacteria including E. coli (12) and Bacillus subtilis (35), we asked whether regulatory mechanisms operating in E. coli and B. subtilis, such as GRDC, may be applicable to M. tuberculosis rrn promoters. This study was designed to analyze the ability of the M. tuberculosis rrn promoters to respond to variations in the growth rate. To determine if the rRNA promoter sequence and structure per se impose any constraints on their usage, their activity was analyzed in M. smegmatis, a mycobacterial species often used as a surrogate host to study M. tuberculosis gene expression. We report (i) fidelity in usage and in the differential activity of the M. tuberculosis rrn P1 and P3 promoters and (ii) GRDC of the M. tuberculosis rrn promoters in cultures of M. smegmatis bearing rrn-lacZ constructs. The minimal role of the M. tuberculosis rrn promoter per se in determining the slow growth of M. tuberculosis is discussed.

MATERIALS AND METHODS

Strains and plasmids.

Jack Crawford, Centers for Disease Control and Prevention, Atlanta, Ga., provided M. smegmatis LR222. A. K. Tyagi, University of Delhi South Campus, New Delhi, India, generously provided promoter selection vectors pSD7 and pSD5B.

Construction of rrn-cat and rrn-lacZ fusions.

For the construction of the rrn-lacZ promoter fusion (Fig. 1), the rrn upstream sequence (816 bp) was amplified by inverse PCR (37) and a ∼650-bp fragment was cloned into a XbaI site located upstream of the lacZ gene in pSD5B (26). The ligation mixture was electroporated into E. coli, and the transformants were plated on Luria-Bertani agar containing kanamycin (25 μg/ml) and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal). Cloning of the rrn promoter in the right orientation (pSD5B.16SR) with respect to lacZ produced blue colonies, and cloning in the opposite orientation (pSD5B.16SW) produced white colonies. Plasmid DNA was isolated from the transformants and electroporated into M. smegmatis to generate strains containing promoter-fusion constructs with the rrn promoter cloned in both orientations.

FIG. 1.

(A) Schematic representation of the M. tuberculosis rrn promoter. The top line shows the genomic organization of the rrn operon in M. tuberculosis (not to scale). The middle line shows the PstI fragment spanning the rrn promoter region. The region encompassed by the PstI and Sau3AI sites (mapping downstream of +1 of 16S rRNA) was amplified by inverse PCR as detailed previously (37) and cloned in pGEM3Zf (+) to generate pAV16S.2. tsp a and c represent the start points and direction of transcription from the P1 and P3 rrn promoters, respectively, and are denoted by thin and thick arrows, respectively; the nomenclature of P1 promoter for tsp a and P3 promoter for tsp c corresponds to that by Gonzalez-y-Merchand et al. (21). d and +1 represent the experimentally determined RNase III-processing site and the start of the mature transcript, respectively (37). The lower part of the figure represents the restriction fragments cloned in promoter fusion vectors pSD5B (26) in the right (pSD5B.16SR) and wrong (pSD5B.16SW) orientations. (B) Nucleotide sequence of the rrn promoter of M. tuberculosis. The numbers on the right are according to Cole et al. (11) for the M. tuberculosis genome. The sequence of the PCR-amplified fragment including the promoter region, +1 mature 16S rRNA start site, and 78 bp of coding region is shown. tsp a and c and their respective promoters are indicated by arrowheads and highlighted boxes, respectively. The experimentally determined RNase III recognition sequence is highlighted and the cleavage site ‘d’ is indicated (37). The location and orientation of the primer-annealing sites are indicated by arrows. The Sau3AI sites mark the fragment that was cloned in plasmid pSD7. (C) Comparison of the primer-annealing regions in M. tuberculosis (Mtb) and M. smegmatis rrnA (Msm A) and rrnB (Msm B) operons. The sequence of the RNA-like strand is shown. Identical residues are marked by vertical lines.

Culture conditions.

M. smegmatis was maintained on Lowenstein-Jensen medium. A loopful was inoculated into 10 ml of Youmans and Karlsons (YK) liquid medium containing kanamycin (25 μg/ml) and Tween 80 (0.2%) and supplemented with 0.5% glycerol (YKKTG) in a 50-ml conical flask at 37°C with vigorous shaking. This primary inoculum was subcultured once more in YKKTG to obtain vigorously growing cells. Cells from the second subculture were used to initiate the cultures used to evaluate the growth rate dependence of the rrn promoter. YKKT (100-ml flasks in triplicate) containing either 1, 0.5, 0.1, 0.05, or 0.01% glycerol were inoculated with twice-subcultured M. smegmatis harboring either pSD5B.16SR or pSD5B.16SW to an initial optical density at 600 nm (OD₆₀₀) of 0.04 to 0.06 and incubated with shaking at 37°C. To establish the growth curve and growth rates, aliquots were taken every 4 h and the OD₆₀₀ was measured over a period of 40 h. The cultures were appropriately diluted for OD measurement within the linear range. The growth rate, μ, was calculated from the equation μ = 1/g, where g is the time in which the initial mass, M₀, of a bacterium doubles as a consequence of cell division (9). For a typical growth curve, see Fig. 2.

FIG. 2.

Growth curve of M. smegmatis. YKKT (100-ml flasks in triplicate) containing either 1, 0.5, 0.1, 0.05, or 0.01% glycerol was inoculated with twice-subcultured M. smegmatis harboring pSD5B.16SR to an initial OD₆₀₀ of 0.04 to 0.06 and incubated with shaking at 37°C. To establish the growth curve, aliquots were taken every 4 h and the OD₆₀₀ was measured over a period of 40 h. Similar growth curves were derived for M. smegmatis harboring pSD5B.16SW and pSD5B.

Isolation of RNA from M. smegmatis.

M. smegmatis strains containing rrn-lacZ promoter fusion constructs were cultured as described above in YK medium supporting a range of growth rates. The growth of all cultures was arrested at 20 h by adding sodium azide (to a final concentration of 10 mM), and cells were harvested at 4°C. RNA was isolated as described previously (3). The quality and quantity of the RNA were evaluated by electrophoresis on formaldehyde-agarose gels, by measuring absorbance at 260 and 280 nm, and by PCR to test for DNA contamination (28).

β-Galactosidase assays.

In parallel with RNA isolation, M. smegmatis culture aliquots were sonicated and assayed for β-galactosidase activity as described previously (29). Briefly, harvested cell pellets derived from 1-ml culture aliquots were resuspended in 0.25 to 0.5 ml of 0.25 M Tris-HCl (pH 7.4) and sonicated (total sonication time of 3 to 4 min with a pulse of 30 s alternating with 30 s of rest). The amount of protein in clarified cell lysates was estimated (8). Triplicate aliquots of cell extracts were used for measurement of β-galactosidase activity, which was calculated and expressed in nanomoles per minute per milligram of protein. Total RNA was isolated from the cultures in parallel, and β-galactosidase-specific mRNA levels were also determined.

Estimation of lacZ mRNA.

The purity of the RNA used in the hybridizations was assessed by PCR. Any residual DNA contamination was removed by DNase treatment. The samples were considered to be free from DNA when no amplification of plasmid DNA was obtained in the absence of reverse transcriptase (28). Serial dilutions (500 ng to 500 pg) of total RNA isolated from M. smegmatis cells harvested at different growth rates (from three independent cultures) were dot blotted onto nylon membranes in duplicate as described. Briefly, 1 μg of RNA was taken and serially diluted with sterile water to the required concentrations. To each RNA sample, denaturing mixture containing 50% formamide, 7% formaldehyde, and 1× SSC (0.15 M NaCl, 0.015 M sodium citrate) was added, and the RNA was denatured at 68°C for 15 min, chilled on ice, and neutralized with 2 volumes of 20× SSC. RNA dot blots on nylon membranes (Schleicher & Schuell, Dassel, Germany) were prepared with a 96-well vacuum manifold. After the slots were rinsed with 10× SSC, the blots were air dried and fixed by UV irradiation. The blots were individually hybridized to ³²P-labeled lacZ- and kanamycin-specific probes in 5× SSC–50% formamide–5× Denhardt’s solution–50 mM sodium phosphate (pH 6.8)–200 μg of salmon sperm DNA per ml–0.1% sodium dodecyl sulfate (SDS). After 16 h at 42°C, the blots were subjected to series of washings, with a final wash at 65°C in 0.2× SSC–0.2% SDS, and subjected to autoradiography. To quantitate the extent of RNA hybridization, the dots were cut and counted individually in scintillation fluid. The individual mean counts of triplicate filters obtained with 100, 50, and 25 ng of RNA dotted for pSD5B.16SR and pSD5B.16SW were determined and plotted against the growth rate.

Determination of plasmid copy number.

The plasmid copy numbers in triplicate cultures of M. smegmatis growing at different rates (μ = 0.150 to 0.296 for cultures containing pSD5B.16SR and μ = 0.159 to 0.298 for cultures containing pSD5B.16SW) were determined by the hybridization method (1). Briefly, M. smegmatis sonicates containing various amounts of protein (5, 10, 15, 20, and 25 μg) were spotted in duplicate onto nitrocellulose membranes as described previously (1). Sonicated pSD5B plasmid DNA (ranging from 0.1 to 819.2 ng) was spotted alongside in duplicate. Plasmid DNA (6.4 ng) spiked with M. smegmatis sonicates (0.1 to 51.2 μg of protein equivalent) was also spotted in duplicate. Probe pSD5B was dephosphorylated, end labelled with [γ-³²P]ATP, and hybridized to the immobilized RNA in 50% formamide–5× SSC–5× Denhardt’s solution–50 mM Tris-HCl (pH 7.5)–200 μg of denatured salmon sperm DNA per ml. The filters were washed in 2× SSC–0.1% SDS at room temperature and 0.2× SSC–0.2% SDS at 65°C and subjected to autoradiography. Numbers of plasmid copies are expressed as nanograms of plasmid DNA per microgram of total protein in M. smegmatis sonicates.

Primer extension.

Total RNA (10 μg) isolated from M. smegmatis cultures growing at different rates and containing pSD5B.16SR, pSD5B.16SW, or pSD5B (vector control) was used in primer extension experiments with primers T4In3 and T4In5. Briefly, the RNA sample and 10⁶ cpm of γ-³²P-labelled primer were mixed, resuspended in 20 μl of piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES) hybridization mixture, transferred to siliconized glass capillaries, denatured at 95°C for 5 min, and incubated at 55°C (primer T4In5) or 58°C (primer T4In3) in a water bath for 8 h. Despite perfect sequence homology of T4In3 primer at its 3′ end to sequences of the rrnA operon of M. smegmatis (Fig. 1C), good specificity in primer extension was retained on account of the stringent temperature (58°C) used for annealing primer to RNA and the sequence divergence at the 5′ end of the primer. Reaction mixtures were subsequently transferred to sterile Eppendorf tubes, cleaned, and subjected to primer extension with 200 U of Superscript II (Gibco BRL, Grand Island, N.Y.) at 42°C for 90 min. After RNase treatment, phenol-chloroform extraction, and ethanol precipitation, the products were run on a 6% polyacrylamide–7 M urea gel alongside a sequencing ladder generated with the same primers and corresponding plasmid constructs. Primer extension experiments were performed with two individual RNA preparations for each growth rate. For a representative figure, see Fig. 4.

FIG. 4.

Usage of M. tuberculosis rrn promoters as a function of growth rate. (A) Schematic map of the relevant portion of pSD5B.16SR containing the rrn promoter region of M. tuberculosis. The annealing regions of primers T4In5 and T4In3 are indicated by black boxes, and their primer-extended products are indicated by arrows. Numbers above the arrows denote the length (in bases) of the primer-extended products. The brick box represents the rrn promoter region, and the striped box represents lacZ sequences. OriM and Ter3 are part of vector pSD5B. e, start site of mature 16S rRNA. (B) Primer extension products obtained with primer T4In5. Lanes T, C, G, and A represent sequencing reactions performed with T4In5 and rrn promoter containing plasmid pAV16S.2. Lanes: 1 to 3, RNA from M. smegmatis cultures harboring pSD5B.16SR harvested at μ = 0.274, 0.202, and 0.153 respectively; 4, RNA from M. smegmatis cultures harboring pSD5B.16SW harvested at μ = 0.282; 5, RNA from M. smegmatis cultures harboring pSD5B vector harvested at μ = 0.288; 6, RNA from M. tuberculosis H37Ra logarithmic-phase culture; 7, no RNA. (C) Primer extension products obtained with primer T4In3. Lanes: 1 to 4, RNA from M. smegmatis logarithmic-phase cultures harboring pSD5B.16SR harvested at μ = 0.274, 0.234, 0.202, and 0.153, respectively; 5, RNA from M. smegmatis cultures harboring pSD5B.16SW harvested at μ = 0.186; 6 and 7, RNA from M. smegmatis cultures harboring pSD5B vector harvested at μ = 0.288 and 0.177, respectively; 8 and 9, RNA from M. tuberculosis H37Rv and H37Ra logarithmic-phase cultures, respectively; 10, no RNA; T, C, G, and A, sequencing reactions performed with T4In3 and rrn promoter containing plasmid pAV16S.2. ∗ and ∗∗ indicate probable processing intermediates of the P1- and P3-derived transcripts, respectively. (D) Comparison of M. tuberculosis rrn promoter usage in M. smegmatis at various growth rates. tsp a and c in the primer extension experiments of panels B and C were scanned by using the UVP gel documentation system and analyzed with Gelblot software. The relative area units (mean of two experiments) for each signal are plotted against the growth rate. Primers T4In5 and T4In3 and their homology to the rrn promoter sequence of M. smegmatis are indicated in Fig. 1B.

RESULTS

In a preliminary experiment, a ∼310-bp Sau3AI fragment (Fig. 1A) mapping between −233 and +74 relative to the +1 of mature 16S rRNA and containing P1 and P3 rrn promoters was cloned into the promoter selection vector pSD7 (13) to generate pSD7.16S. This fragment supported an extremely high chloramphenicol acetyltransferase activity (16,669 nmol/min/mg of protein) in M. smegmatis (data not shown) in comparison to the relatively low activity observed for randomly cloned promoters of M. tuberculosis (5 to 2,500 nmol/min/mg of protein) in M. smegmatis, with the majority of promoters displaying activity in the range of 5 to 100 nmol/min/mg of protein (13).

Growth rate dependence of the rrn promoter(s) of M. tuberculosis.

For the analysis of the rrn promoter, a ∼650-bp rrn promoter-containing fragment was subcloned upstream of the lacZ reporter gene in shuttle plasmid vector pSD5B, a low-copy-number plasmid that is maintained at ∼3 copies per mycobacterial cell (32) (Fig. 1A). Blue and white colonies of M. smegmatis transformants were obtained on a Luria-Bertani agar plate containing X-Gal and 20 μg of kanamycin per ml. When cloned in the right orientation (pSD5B.16SR), the M. tuberculosis rrn promoter drove β-galactosidase expression, yielding blue colonies, whereas when cloned in the wrong orientation (pSD5B.16SW), white colonies were obtained, clearly indicating that the promoter was functional in one direction only.

M. smegmatis cells carrying rrn-lacZ fusions were grown as shake cultures in YKKT medium (supplemented with 0.01 to 1% glycerol) at various growth rates and were assayed for β-galactosidase activity. The growth curves followed the expected pattern and yielded a range of growth rates (μ = 0.153 to 0.327), which were required to study the growth rate-dependence of rrn promoter (Fig. 2). The generation time of M. smegmatis cultures corresponding to the growth rates of 0.153 to 0.327 was in the range of 6.49 h to 3.06 h, respectively. The rrn promoter, when cloned in the right orientation (pSD5B.16SR), gave the steep positive slope of activity with increasing growth rate which is characteristic of GRDC (Fig. 3A). M. smegmatis carrying the rrn promoter cloned in the wrong orientation (pSD5B.16SW) showed negligible activity over a range of growth rates (data not shown). The presence of the fusions used in this study did not appear to impose any metabolic load, since there were no differences in the growth rates of cultures carrying or not carrying the plasmids.

FIG. 3.

A. β-Galactosidase activity as a function of growth rate. β-Galactosidase activity was measured in M. smegmatis sonicates prepared from cells cultured at various growth rates as described in Materials and Methods. pSD5B.16SR contains the rrn promoter fragment cloned in the right orientation. (B) Amount of lacZ-specific mRNA as a function of growth rate. The amount of lacZ mRNA was determined by hybridization to filter-bound lacZ DNA as described in Materials and Methods and is expressed as a percentage of the total input radioactivity that is bound to the lacZ DNA filter. (C) Amount of plasmid DNA as a function of growth rate. Sonicated cell extracts used for β-galactosidase assays were assayed in parallel for the amount of plasmid DNA by using a dot blot hybridization procedure as described in Materials and Methods. Results are expressed as nanograms of plasmid DNA per microgram of total protein in the cell extract. Triplicate cultures were grown and processed in triplicate for each estimation. Values in the panels are expressed as a mean of nine readings with a standard deviation of <10% of the mean.

Variation of β-galactosidase transcript levels with growth rate.

To confirm that the growth rate-dependent response seen for M. tuberculosis rrn promoters cloned into pSD5B reflected the transcriptional activity of the promoters, the level of lacZ mRNA at different growth rates was directly determined by dot blot hybridization of RNA isolated from M. smegmatis cultures harvested at the same time point at which β-galactosidase activity was measured. The level of lacZ mRNA in cells carrying the rrn promoter cloned in the right orientation (pSD5B.16SR) increased in a growth rate-dependent manner. In contrast, hybridization with the kanamycin probe did not increase with an increase in μ for either pSD5B.16SR or pSD5B.16SW (Fig. 3B). These findings confirmed that (i) the increase in β-galactosidase activity with an increase in growth rate was due to enhanced transcriptional activity of the M. tuberculosis rrn promoter and (ii) the presence of the rrn promoter did not modulate kan gene transcription from the same plasmid.

Association of growth rate with plasmid copy number.

Because a plasmid system was used to study rrn-lacZ transcription, we considered the possibility that fluctuations in plasmid copy number with different growth rates would present as a growth rate-dependent response. Others have found that growth conditions and strengths of inserted promoters can significantly affect the copy number (1, 36). Therefore, an estimate of the amount of plasmid DNA per unit of total cellular protein was obtained for promoter-fusion clones containing the rrn promoter as described in Materials and Methods. It is clear that the amount of plasmid DNA remained essentially unaltered as the growth rate increased; thus, it could not account for the increase seen in lacZ mRNA and β-galactosidase specific activity (Fig. 3C).

Promoter fusion studies have proven particularly useful in demonstrating that with an increase in growth rate, E. coli rrn promoter-directed transcription and translation of the reporter gene also increases (20, 27). Although reporter technology for the analysis of promoter activity is very extensively used, it is not entirely natural. We have therefore attempted to correct for potential artifacts within the system that could mimic a growth rate-dependent response by directly measuring promoter utilization, amount of lacZ mRNA, and plasmid copy number and by using a mycobacterial host rather than E. coli. The conclusions derived from the experiments confirm that the promoter fusion assay was a valid means of assessing GRDC of M. tuberculosis rrn promoters.

Faithful recognition and growth rate-dependent usage of M. tuberculosis rrn promoters in M. smegmatis.

The utilization of the M. tuberculosis rrn P1 and P3 promoters was assessed by primer extension experiments with M. smegmatis cells carrying rrn-lacZ fusion constructs (right and wrong orientations and vector alone) grown in different media to achieve a range of growth rates. With this technique, which uses an excess of end-labelled primer to generate cDNAs complementary to RNA, the strength of the signal from the extended product obtained is a reflection of the concentration of that particular RNA species within the cell at the time of harvesting. Comparison of the signal strengths enables one to evaluate the strengths of multiple promoters for a single operon such as the rrn operon of M. tuberculosis. Primers T4In5 and T4In3 were used to assess transcription from the P1 and P3 promoters. Primer T4In5 had little sequence homology to either rrn operon of M. smegmatis and was specific for the rrn-lacZ transcript generated from plasmid construct pSD5B.16SR (Fig. 1B and C). This primer enabled the detection of two major products, tsp a and tsp c, corresponding to the P1 and P3 promoters, respectively, and two minor products (Fig. 4). The major products comigrated with RNA products a and c obtained with genomic RNA of M. tuberculosis H37Rv and H37Ra (Fig. 4B). The upstream tsp a (primer extension product of ∼190 bases) was directed by the P1 promoter while the tsp c (primer extension product of ∼110 bases) was directed by the P3 promoter, as demonstrated previously (21, 37). The P3 tsp was preceded by the E. coli ς⁷⁰ −10 and −35 consensus motifs TATTAG and TTGACT, respectively (Fig. 1B); the sequence, position, and spacing between them closely resembled those of the P2 promoter of rrn operons from E. coli and B. subtilis. In contrast, the putative P1 promoter was composed of an E. coli ς⁷⁰-like −10 motif but lacked a −35 sequence (Fig. 1B) and resembled the bulk of M. tuberculosis promoters (4). In M. tuberculosis as well as in M. smegmatis, the signal from tsp a was weaker than that from tsp c, suggesting that the P3 promoter was the better utilized of the two promoters (Fig. 4). Transcript d, a product of RNase III-mediated processing, was generated only with genomic RNA of M. tuberculosis (Fig. 4B, lane 6), consistent with the requirement of a panhandle structure between sequences flanking 16S rRNA. From the above experiments, it was clear that (i) the rrn promoter is very strong and thus distinct from the majority of M. tuberculosis promoters and (ii) the rrn promoter is faithfully expressed in M. smegmatis.

To study rrn promoter usage in cultures growing at different rates, RNA was isolated from M. smegmatis cultures, harboring the rrn-lacZ promoter construct, grown in YK medium supplemented with 0.01 to 1% glycerol as described in Materials and Methods. Primer extension experiments were performed with primer T4In5; a five- to sixfold increase in signal intensity in a growth rate-dependent manner was observed with tsps c and a, respectively, over a growth rate range of 0.153 and 0.274, which correspond to cell doubling times of 6.49 and 3.64 h respectively. Further, at any given growth rate, signal c was twice as intense as signal a, suggesting that the P3 promoter is utilized more efficiently than the upstream P1 promoter (Fig. 4D) is. A second primer, T4In3, which mapped only ∼55 bases downstream from tsp a (Fig. 4A), was also used, since somewhat faint signals were obtained with primer T4In5. Using T4In3, an 80-base product was generated that comigrated with the product generated with RNA of M. tuberculosis (Fig. 4C). Over a growth rate range of 0.153 to 0.274, a ∼12-fold increase in P1 promoter activity was noted with the T4In3 primer; the repression of P1 promoter was particularly marked below μ = 0.234 (Fig. 4C and D) and most probably represented the basal activity of this promoter. This apparent discrepancy in the degree of induction of the P1 promoter (12-fold with T4In3 versus 5- to 6-fold with T4In5) could be ascribed to variations in the efficiency of the primers to read across the rRNA sequence to yield primer extension products of significantly different lengths (191 bases for the T4In5-derived product and 80 bases for the T4In3-derived product). These primer extension studies revealed that (i) the P1 and P3 promoters of M. tuberculosis when in tandem are both under GRDC and (ii) the downstream promoter, which is conventionally considered to be a weak promoter in rrn operons in other bacteria such as E. coli, is well expressed in M. smegmatis; in fact, it is more efficiently expressed than the P1 promoter and is the major promoter.

DISCUSSION

M. tuberculosis in vivo would most probably be limited for oxygen and nutrients and therefore would display rather long generation times. However, in laboratory-grown cultures, these constraints do not exist and yet M. tuberculosis divides every ∼18 h on average and M. smegmatis divides every 3 to 5 h. It is reasonable to think that multiple properties of the tubercle bacillus contribute to its slow growth. First, the unique composition of the mycobacterial envelope is likely to present permeability barriers. Since cell wall lipids constitute a high proportion of the dry weight of mycobacteria (6) and cell wall synthesis imposes a considerable energy demand on the cell, its biosynthetic rate may also be limiting for growth. Second, RNA chain growth in M. tuberculosis was ∼10 times lower than that in E. coli, and the low transcription rate was attributed to a low rate of transcription initiation; rates for M. smegmatis were not determined (25). This was reflected in a rather low content of RNA per unit of DNA compared to that in other bacteria. Thus, in M. bovis and M. tuberculosis, the RNA/DNA ratio varied between only 1:1 and 2:1, while in M. smegmatis, it reached 5:1 in rapidly growing cultures (38). The G+C content of mycobacterial DNA, particularly that of the promoter regions, has been suggested as another constraint for the low rate of transcription; the upstream regions of mycobacterial genes have a higher G+C content than do the corresponding regions from M. smegmatis (4). rRNA gene dosage is also considered a critical factor influencing growth. Slow growers such as M. tuberculosis depend entirely for their total ribosome pool on a single rrn operon driven by two promoters, while fast growers including M. smegmatis typically have two rrn operons per genome (5) and possess multiple promoters to increase their capacity for rRNA synthesis (22).

We have demonstrated the exceptional strength of the M. tuberculosis rrn promoter in comparison with the bulk of M. tuberculosis promoters. We addressed whether the M. tuberculosis rrn promoter has some unique sequence or structure which precludes its modulation in conditions of varying nutrient supply. M. smegmatis was chosen as a surrogate host for this purpose since (i) unlike M. tuberculosis, it is equipped to regulate its RNA synthesis rates in response to nutrient supply (38) and (ii) it would provide a milieu devoid of the constraints impeding the growth of M. tuberculosis. M. smegmatis has been proposed to be a good surrogate host for the study of M. tuberculosis transcriptional activity, protein expression, and some aspects of genetics. This is because the efficiency and fidelity of transcriptional recognition, at least for vegetative promoters, is conserved in M. tuberculosis and M. smegmatis (4). In the present study, the usage of M. tuberculosis rrn promoters was determined by primer extension analysis with genome-derived RNA from logarithmic-phase cultures of M. tuberculosis and comparing the signals with those obtained with plasmid-derived transcripts from M. smegmatis cultures harboring the rrn-lacZ construct. The results indicate that fidelity of transcription initiation and usage of M. tuberculosis rrn P1 and P3 promoters was maintained in M. smegmatis. The P3 promoter was by far the stronger rrn promoter in M. tuberculosis as well as in M. smegmatis. The G+C content of the promoter region mapping from +1 to −50 relative to the tsp showed an inverse correlation with promoter strength; it was 50 and 58% for the P3 and P1 promoters, respectively (Fig. 1), substantiating the observation that the high G+C content of mycobacterial promoters may have a bearing on their lower activity (4).

A surprising finding was that both the P1 and P3 promoters of the M. tuberculosis rrn operon were under GRDC. In contrast, in E. coli and B. subtilis, only one of the two promoters of intact rrn operons is under GRDC (12, 15), although promoter dissection experiments have demonstrated that the downstream P2 promoter also is regulatable by growth rate in E. coli, albeit to a lower extent than P1 (17, 20). Thus, the features in the DNA sequence that govern GRDC of the M. tuberculosis rrn promoter appear to be present around both promoters. AT-rich and upstream activating elements, Fis-binding sites, and GC discriminator sequences are characteristic features of E. coli rrn operons (12, 27, 30, 31, 33) and are involved in their regulation. Since these sequences and the Fis-encoding gene were not present in the M. tuberculosis rrn promoter or genome (11), some unique mechanism(s) is suggested for regulation. The upstream regions of rrn operons in M. smegmatis and M. tuberculosis are predicted to form similar secondary structures to generate potential binding sites for putative trans-acting proteins (21). It is possible that these putative trans-acting factors that participate in recognition of the structure and sequence of the rrn promoter regions are present in M. smegmatis but not in M. tuberculosis. A noteworthy observation was that the two promoters exhibited differential usage; the P3 promoter was ca. twice as active as the P1 promoter at all growth rates. The purpose of a weak upstream P1 promoter that is poorly expressed at all growth rates remains a puzzle. It is functionally equivalent to the downstream P2 promoter of E. coli rrn operons in relation to low level constitutive expression. Since M. tuberculosis possesses only one rrn operon, a possible advantage of having a stronger downstream promoter than upstream promoter is that promoter occlusion effects would be minimized and rRNA transcription would be maximized. On the other hand, in E. coli, which possesses seven rrn operons, the downstream P2 promoter is subject to occlusion by transcription from the P1 promoter (17).

In the context of differential usage, it may be noted that the experiments described in this report were performed with M. smegmatis cultures grown at various growth rates. The M. smegmatis cultures grown at μ ≥0.234 were in logarithmic phase, while those cultured in media containing limiting amounts of glycerol, i.e., 0.05 and 0.01% glycerol (μ ≤0.20), reached stationary phase at 20 h of incubation, the time point at which the experiments described in this study were performed (Fig. 2). While our manuscript was in review, Gonzalez-y-Merchand et al. reported that the rate of rRNA transcription initiation from either the P1 or the P3 promoter varied little regardless of whether M. tuberculosis cultures were in the logarithmic or stationary phase of growth (23). These findings confirmed the observation made nearly three decades ago that RNA-DNA ratios of M. tuberculosis cultures altered only marginally as a function of growth rate (38). A recent report stated that in stationary-phase M. tuberculosis cells, the P1 promoter rather than the P3 promoter assumes charge of rrn transcription in a SigF-dependent manner, suggesting that sigma factor-dependent regulation of the rrn operon occurs in M. tuberculosis (10).

Despite the caveats in studying the regulation of mycobacterial genes from slow growers in rapid growers, these cross-species experiments have clearly shown that the P1 and P3 promoters of the rrn operon of M. tuberculosis are both regulatable over a range of growth rates in the environment provided by M. smegmatis. In conclusion, the present study clearly indicates that rrn promoter sequence and structure do not play a significant role in determining the low levels of RNA in M. tuberculosis. Other factors such as replication, cell division, cell wall biosynthesis and/or permeability, dosage of rRNA genes, and absence of trans-acting proteins, probably serve as primary factors in determining the inability of the tubercle bacillus to respond to changes in nutrient supply.