ABSTRACT
The nif promoters of Klebsiella pneumoniaemust be activated by proteins bound to upstream sequences which are thought to interact with the ς54-RNA polymerase holoenzyme by DNA looping. NifA is the activator for most of the promoters, and integration host factor (IHF) mediates the DNA looping. While NtrC is the activator for the nifLA promoter, no IHF appears to be involved. There are two A tracts and one T tract between the upstream enhancer and the nifLA promoter. This DNA segment exhibits anomalous electrophoretic mobility, suggesting intrinsic sequence-induced curvature in the DNA. On the one hand, mutation of the A tracts or T tract individually or together, or deletion of the A tracts and the T tract reduces the anomaly; on the other hand, creation of two additional A tracts enhances the anomaly. Intrinsic curvature in the DNA has been confirmed by circular permutation analysis after cloning the DNA fragment in the vector pBend 2 and also by electron microscopy. Computer simulation with the DNA base sequence is also suggestive of intrinsic curvature. A transcriptional fusion with the Escherichia coli lacZ gene of the DNA fragment containing the nifLA promoter and the wild-type or the mutated upstream sequences was constructed, and in vivo transcription in K. pneumoniae and E. coliwas monitored. There was indeed very good correlation between the extent of intrinsic curvature of the DNA and transcription from the promoter, suggesting that DNA curvature due to the A tracts and the T tract was necessary for transcription in vivo from thenifLA promoter of K. pneumoniae.
Biological reduction of atmospheric dinitrogen is the exclusive preserve of a very limited number of microorganisms. The molecular mechanism of the process has now been worked out in considerable detail, and the gram-negative bacteriumKlebsiella pneumoniae has served as a model system for this purpose. Twenty-one potential genes contained in eight operons have been found in a single cluster (34). In addition to the structural genes (nifHDK) for the enzyme nitrogenase, these operons contain genes for cofactor synthesis, processing, carriers for electron transport, and control elements (nifLA) (34). The sigma factor involved in initiation of transcription from the nif promoters is ς54 or ςN. The nif promoters are different from the more common ς70 promoters (2, 5, 23, 43) and need the upstream activator NifA to initiate transcription (10, 37, 44). This is not a case of activation by recruitment of the RNA polymerase holoenzyme to the promoter DNA (41). NifA is counteracted by NifL in the presence of molecular oxygen and fixed nitrogen (24, 42). The upstream bound NifA interacts with the promoter-bound ς54-RNA polymerase holoenzyme by DNA looping mediated by the integration host factor (IHF) bound in between the NifA binding site and the promoter (25, 46). It has been observed in some in vitro experiments that the histone-like protein HU can substitute for IHF at least partially (12, 39). On the other hand, phosphorylated NtrC binds to two adjacent sites more than 100 bp upstream of the nifLA promoter and serves as the activator of transcription (18, 21) by interacting with the ς54-RNA polymerase holoenzyme bound to the promoter (36). The actual activation process, of course, may be dependent on oligomerization of NtrC involving protein-protein interactions in addition to protein-DNA interactions (48). It is, however, important to note that IHF does not bind to the intervening sequence between the NtrC binding sites and thenifLA promoter (45). The transcriptional activation by NtrC is, nevertheless, face-of-the-helix dependent (36), suggesting the involvement of DNA looping. The mechanism for looping by which the activator and the RNA polymerase holoenzyme would interact in this case thus appears to be different. This assumes greater importance because the nifLA operon is the master regulatory operon for all other nif operons.
The presence of intrinsic curvature in DNA because of the presence of specific base sequences has been noticed, but most of these curvatures are small compared to the marked effect produced by stretches of AT tracts, each tract being about half a helical turn long (29). Often a CA-TG doublet junction enhances the curvature (4, 37).
Sequence-induced curved DNA is present in the replication origin of bacteriophage lambda (49) and an autonomously replicating sequence of yeast (47). Curved DNA regions, inferred from anomalous electrophoretic mobility, have also been found upstream of the plasmid promoter PCIII, and their presence has been found to increase in vivo and in vitro transcription, apparently independent of any activator protein (40). On the other hand, a significant percentage of Escherichia coli promoters have A tracts in the immediate upstream region of the −35 hexamers of ς70 promoters, which have been predicted to confer a sequence-induced curvature that is involved in transcription (20). Interestingly, several in vitro studies using such promoters have revealed that addition of RNA polymerase ς70 holoenzyme alone was sufficient for transcriptional activity and that no upstream activator was necessary (3, 26, 31), giving rise to the view that such bends could facilitate and stabilize the initial binding of the RNA polymerase holoenzyme (39). Sequence induced curvature in DNA has also been inferred from computer analysis (6) of base sequence upstream of the gln Ap2 promoter (11), but no experimental data has been cited.
The evidence presented in this paper suggests that AT-tract-mediated intrinsic curvature in native DNA is instrumental in ensuring the interaction between the upstream activator and the promoter-bound RNA polymerase-ς54 holoenzyme, resulting in transcription from the promoter.
MATERIALS AND METHODS
Cloning of the nifLA promoter and the upstream regulatory region of K. pneumoniae, and construction of the deletion mutant.The EcoRI-BglII (∼1-kb) fragment from pGPD44 (16), containing the K. pneumoniae nifLA promoter and the upstream activator site for NtrC binding (bases 17681 to 18648 as described by Arnold et al. [1]), was cloned at theEcoRI-BamHI sites of pUC19. This was then digested with HincII and SmaI, and the 553-bpHincII-SmaI fragment (bases 17983 to 18536 as described by Arnold et al. [1]) was cloned at theHincII site of pUC19 in an orientation such that theBamHI site in the pUC19 polylinker was upstream of thenifLA promoter and the HindIII site was downstream (see Fig. 1). This construct was named pDJ22. TheBamHI-HindIII fragment from this construct would be 583 bp long. To obtain the deletion derivative, theHindIII-SalI fragment (571 bp) was removed from pDJ22 and digested completely with Sau3A and the largest fragment (275 bp), which hadHindIII-Sau3A ends, was collected after electrophoresis in 2% agarose. The sameHindIII-SalI fragment was also partially digested with Sau3A, and the 213-bpSau3A-SalI fragment was isolated after similar electrophoresis. These two fragments were then ligated and cloned in the HindIII-SalI sites of pUC19. This construct was named pDJ225.
Construction of other mutants.TheBamHI-HindIII fragment from pDJ22 was cloned in the phage vector M13mp19, and oligonucleotide-directed mutagenesis was carried out as described by Kunkel (30) with the Mutagene M13 kit (Bio-Rad). The oligonucleotides were synthesized in the laboratory by the phosphoramidite method of solid-phase synthesis (19), using a 391 DNA synthesizer (Applied Biosystems).
Acrylamide gel electrophoresis.The DNA samples (∼1 μg) were electrophoresed in a 7.5% acrylamide gel with 25% glycerol and 1× TBE (88 mM Tris, 88 mM boric acid, 2.5 mM EDTA [pH 8.3]) at 4°C for 36 to 40 h at 7 V/cm by using a vertical gel electrophoresis apparatus (Hoefer).
Electron microscopy of DNA.The plasmid containing the DNA fragment to be analyzed was digested with restriction endonucleases and electrophoresed on a 0.8% low-melting-temperature agarose (Sigma Chemical Co.) gel. The gel piece(s) containing the DNA fragment to be eluted was cut under long-wavelength UV light and kept in a microcentrifuge tube. The tube was incubated at 65°C for 5 min, and 0.1 volume of 5 M NaCl was added to the molten agarose. It was mixed well and incubated again at 65°C for another 5 min. The agarose was removed by extraction with an equal volume of phenol, (saturated with 100 mM Tris [pH 8.0]), and centrifugation in a microcentrifuge at room temperature for 5 min. After the centrifugation, the upper, aqueous layer was transferred to a fresh microcentrifuge tube and extracted with 2 volumes of diethyl ether. The two layers were allowed to separate, and the upper layer was aspirated. Two volumes of ethanol was then added to the aqueous layer and mixed well, and the DNA was precipitated at −70°C for 30 min. The DNA was sedimented by centrifugation in a microcentrifuge at 12,000 rpm (Eppendorf Plol rotor) at 4°C for 15 min. The pellet was washed with chilled 70% ethanol, dried in a vacuum desiccator, and dissolved in a buffer containing 10 mM Tris and 1 mM EDTA (pH 8.0). DNA samples at approximately 1 μg/ml were incubated with 10 mM ZnCl2 or ZnSO4 (8, 22) at 55°C for 5 min and incubated at 37°C for 5 min. The mixture was then brought to room temperature before being prepared for electron microscopy.
Electron microscopy of the DNA fragment was carried out essentially as described by Davis et al. (15). About 100 ng of DNA sample was mixed with cytochrome c (from horse heart, type III [Sigma Chemical Co.]) to a final concentration of 0.005% and formamide (EM Sciences) to a final concentration of 40% and spread on water. The cytochrome c film containing DNA was gently picked up on to a carbon-coated grid (400 mesh; Polysciences Inc.) and stained with uranyl acetate (Polysciences Inc.) for 30 s (5 mM stock in 50 mM HCl, freshly diluted 100-fold in 90% ethanol). The grid was then washed with 100% ethanol for 10 s. To increase the contrast of DNA molecules further, the grid was rotary shadowed at an angle of ∼7° with Pt-Pd (80%:20%) (Polysciences Inc.). The grids were then examined in a Philips EM 410 transmission electron microscope. Two other methods of spreading have been checked. In the method of Inman and Schnos (27) ∼10-ng DNA samples (in 10 mM Tris–1 mM EDTA [pH 8.0]) were briefly mixed with the buffer containing 65 mM Na2CO3, 10 mM EDTA, 33% HCHO, and 40 mM HCl. Cytochrome c and formamide were then added to the mixture (to final concentrations of 0.01 and 50%, respectively), which was spread, stained, and rotatory shadowed. In the method described by Chattoraj et al. (13), the carbon-coated grids were treated with alcian blue (0.2% stock in 3% acetic acid, diluted 100-fold with water just before use) and then dried. About 5 ng of DNA (in 10 mM Tris–1 mM EDTA [pH 8.0]) was applied directly onto the grid and allowed to adsorb. The grid was washed in water, stained with 1% aqueous uranyl acetate solution, dried, and rotary shadowed with tungsten.
Construction of lacZ fusion and assay of β-galactosidase activity.The low-copy-number broad-host-range transcriptional fusion plasmid vector pGD499 (17) was used as described previously (32). The DNA fragments containing the mutated nifLA promoter and upstream sequence were removed from M13mp19 as BamHI-HindIII fragments and cloned at the same sites of pGD499. E. coliS17.1 or E. coli TB1 was transformed by the method of Cohen et al. (14). The construct was mobilized from E. coli S17.1 into K. pneumoniae M5a1 by conjugation. A β-galactosidase assay (35) was performed under maximally inductive conditions in the absence of NH4+ and O2. E. coli cells were grown in M9 glucose medium (devoid of NH4Cl) supplemented with Casamino Acids (200 μg/ml). K. pneumoniae cells were grown in NFDM medium (9), supplemented with Casamino Acids (500 μg/ml), in a 10-ml volume in acetylene reduction bottles (with rubber caps) sealed with a metalic seal. These bottles were then flushed with argon for 5 min to remove traces of oxygen. The cultures were grown at 30°C until the absorbance at 600 nm reached 0.3 to 0.5.
Mapping of the locus of bending.The wild-type DNA comprising nucleotides −131 to −38 with respect to the transcription initiation site was amplified by PCR and cloned by blunt-end ligation in the SmaI site of the plasmid vector pNEB 193 (construct pAN 94). The DNA was then removed as anEcoRI-BamHI fragment and cloned in theXbaI site of the vector pBend 2 by blunt-end ligation after end filling in (construct pADH I). This construct was then digested separately with BamHI, SspI, NruI,PvuII, EcoRV, XhoI, SpeI,BglII, and MluI. The fragments released were then electrophoresed in an acrylamide gel as described above.
RESULTS
The DNA stretch between the NtrC binding sites and thenifLA promoter has anomalous electrophoretic mobility: mutation or deletion of the two A tracts or one T tract present there reduces the anomaly.Figure 1reveals the presence of two A tracts centered at positions −88 and −80 and one T tract centered at −67 with respect to the transcription initiation site. The A tract at −80 is preceded by a GC at the 5′ end.
Sequence of the 583-bpBamHI-HindIII fragment encompassing thenifLA promoter and the upstream sequence of K. pneumoniae. The bases above the filled bars represent the two NtrC binding sites (9). The bases above the open bars depict −26 and −12 regions containing the nifLA promoter (5). The two A tracts and the T tract are between the NtrC binding sites and the promoter.
The 583-bp BamHI-HindIII fragment (Fig. 1) migrated like a 690-bp fragment on acrylamide gel electrophoresis (Fig.2; Table1). Mutating an A or T tract in isolation reduced the anomaly in electrophoretic mobility to some extent (Fig. 2; Table 1). However, when two A tracts were mutated in conjunction or an A and a T tract were mutated together, reduction in the anomaly in electrophoretic mobility was substantial (Fig. 2; Table 1). When the 83-bp stretch (positions −137 to −54) containing both the A tracts and the T tract was deleted, the resultant 500-bp fragment migrated on electrophoresis with practically no anomaly. Mutation of bases resulting in two additional A tracts at a 10-bp interval upstream of and in phase with the existing A tracts accentuated the anomaly in electrophoretic mobility.
Electrophoretic mobility of the DNA fragments of the wild-type and mutant nifLA promoters and the upstream sequences obtained by digestion of the plasmids withBamHI-HindIII (lanes c to j). Lanes: a and l,HinfI-digested φ×174 replicative-form DNA. b and k,HaeIII-digested pBR322 DNA; c, fragment with a deletion from −135 to −53 with respect to the transcription initiation site; d, wild-type fragment; e, AAAA at positions −89 to −86 mutated to ACAC; f, AAAAA at positions −82 to −78 mutated to ACACA; g, AAAA at positions −89 to −86 mutated to ACAC and AAAAA at positions −82 to −78 mutated to ACACA; h, TTTT at positions −69 to −66 mutated to TGTC; i, AAAA at positions −89 to −86 mutated to ACAC and TTTT at positions −69 to −66 mutated to TGTC; j, CGGG at positions −99 to −96 mutated to AAAA and GCGG at positions −109 to −106 mutated to AAAA.
The apparent and relative sizes of the DNA fragments of the wild type and various mutants with mutations in the upstream region of the nifLA promoter as estimated from their respective electrophoretic mobilitiesa
Anomalous electrophoretic mobility on acrylamide gels has been accepted as a manifestation of DNA fragments containing an intrinsic sequence-induced bend (4, 33). We have therefore inferred that the two A tracts and the one T tract contribute in a cumulative manner to the curvature of the DNA stretch between the NtrC binding sites and the nifLA promoter.
Predictions based on computer simulations of the DNA base sequence are suggestive of the presence of intrinsically curved DNA.Images of DNA sequences of the wild-type and mutated DNA generated by using CURVATURE (6) and NUVIEW (37) software are presented in Fig. 3. The estimate of cumulative bending angle for the wild-type DNA was 79°, which was reduced to 65° on mutation of the two A tracts. Deletion of the 83-bp region containing the two A tracts and the one T tract resulted in a cumulative bending angle of 50°, while introduction of two A tracts in addition to the preexisting ones enhanced the cumulative bending angle to 85°.
Electron microscopy of the DNA fragments confirms the presence of curvature in the DNA due to the A tracts and T tract.The 583-bp wild-type DNA fragment and the mutated versions were examined by electron microscopy. Figure 4 shows photographs of representative fields. The wild-type DNA and the DNA in which two additional A tracts were introduced revealed many molecules with hairpin bends and some looped molecules. These curved structures very closely resemble the curved structures observed by Hoover et al. (25) when they allowed IHF to bind upstream of thenifH promoter of K. pneumoniae. The fragment in which the two A tracts were mutated and also the one in which both the A tracts and the T tract were deleted showed predominantly linear molecules and no molecule with a hairpin bend or looped molecule at all. We have screened for molecules which exhibited bend angles of >90°, and the results are presented in Table2.
Representative electron micrographs of the fragments containing the nifLA promoter and the upstream sequence. (A) Wild type (583 bp). (B) Two existing A tracts have been mutated (583 bp). (C) An 83-bp stretch containing the two A tracts and one T tract has been deleted (500 bp).
Data showing the population of bent moleculesa as determined from the electron micrographs of various DNA samplesb
The frequency of bent molecules has been found to be similar irrespective of the method of spreading used.
Experimental mapping of the locus of bending.The wild-type DNA comprising the region from −131 to −38 with respect to the transcription initiation site was cloned in the vector pBend 2 (28), and the electrophoretic mobilities of DNA fragments generated from the duplicated circularly permuted restriction sites were determined (Fig. 5). An analysis of the mobilities is presented in Table 3. A plot of the relative sizes of the fragments against the base position enabled us to map the locus of bending to be around the G residue at position −95 with respect to the transcription start site (Fig.6).
Electrophoretic mobility of DNA fragments generated by digestion of pADH.1 (a construct containing a 94-bp stretch comprising the nifLA promoter and the upstream region cloned in the plasmid vector pBend 2). Lanes: a and m, HaeIII-digested φ×174 RF DNA; b and l, HinfI-digested φ×174 DNA; c to k, the fragment released on digestion of pADH 1 with BamHI (lane c) with SspI (lane d) (there is an additionalSspI site in pBend 2 besides the two in the polylinker; the upper band in lane d is a result of this); with NruI (lane e), with PvuII (lane f), with EcoRV (lane g), with XhoI (lane h), with SpeI (lane i), withBglII (lane j), and with MluI (lane k).
Apparent and the relative sizes of the DNA fragmentsa as estimated from their respective electrophoretic mobilitiesb
Permutation analysis of the 258-bp fragment generated by digestion of pADH 1 with different restriction endonucleases containing the putative bent locus in the upstream region of the nifLApromoter of K. pneumoniae.
Mutations that modulate DNA bending also modulate transcription from the nifLA promoter in vivo.Transcriptional fusions with lacZ were constructed by using the low-copy-number plasmid pGD 499 (17), and β-galactosidase activity was determined in K. pneumoniae M5a1 and E. coli TB1 (Table 4). Mutation of the individual A tracts or T tract caused a small but significant reduction in transcription from the nifLA promoter, but mutation of both the A tracts or one A tract and the T tract or deletion of 83 bp (positions −137 to −54) containing both the A tracts and the T tract indeed resulted in substantial reduction. Mutation of bases at 10-bp intervals upstream of but in phase with the existing A tracts, creating two additional A tracts, led to considerable stimulation in transcription from the nifLA promoter. Mutation of the 10 bases at positions −131 to −122, keeping the same purine-pyrimidine bias, did not have any deleterious effect on transcription.
β-Galactosidase activities of lac fusions of the region containing the nifLA promoter and upstream sequencea
DISCUSSION
The presence of intrinsic curvature in DNA due to the presence of two A tracts and one T tract between the NtrC binding sites and thenifLA promoter of K. pneumoniae appears certain on the basis of anomalous electrophoretic mobility and electron microscopy (Fig. 2, 4, and 6; Table 1). Mutation in an individual A tract or T tract or two tracts simultaneously or deletion of all the three tracts reduces both curvature in DNA and in vivo transcription from the promoter, and the magnitude of reduction of both phenomena is in the same order, as mentioned above (Tables 1 and 4). This is the first report on the parallelism between A- and T-tract-mediated bending of native DNA and transcription in vivo. However, synthetic curved DNA sequences have been shown to be capable of effectively replacing DNA fragments containing a CRP binding site upstream of the galpromoter of E. coli (7). We have assumed that the role of the A- and T-tract-mediated bent DNA is to facilitate the interaction of RNA polymerase-ς54 holoenzyme with the upstream activator NtrC, which results in transcription from thenifLA promoter of K. pneumoniae. This is analogous to the situation with the NifA-dependent promoters, although in the latter cases IHF is instrumental in promoting the interaction (25, 46).
It has been reported that open-complex formation in vitro at theglnAp2 promoter of E. coli present on linear DNA was dependent on the native sequence of nucleotides situated in the intervening region between the enhancer and the promoter (12). Replacement of the native sequence of nucleotides by random sequence affected open-complex formation. Computer simulation based on the Trifonov algorithm (6) led Carmona et al. to infer the presence of an intrinsic bend in the DNA between the enhancer and the glnAp2 promoter. No experimental data has been presented to substantiate bending.
However, the interesting observation was that the replacement of the native sequence of nucleotides, which possibly induced the bend in the DNA, did not affect open-complex formation in vitro when the promoter was present on supercoiled DNA. Plasmid DNA in vivo is more likely to be supercoiled than linear or open circular. Therefore, if the in vitro experiments can be extrapolated to the in vivo situation, the putative bent DNA upstream of the glnAp2 promoter is inconsequential in vivo.
Carmona et al. (12) have also inferred the presence of bent DNA upstream of the K. pneumoniae nifLA promoter by computer simulation, but again no physical evidence has been provided by actual experimentation. Our experiments have established that bent DNA is essential for optimal expression in vivo, at least for thenifLA promoter.
ACKNOWLEDGMENTS
We are grateful to E. N. Trifonov and M. Bansal for making available computer softwares for analysis of curvature in DNA and to S. Adhya for providing pBend 2. Long and useful discussions with M. Bansal are also gratefully acknowledged. D. J. Sengupta and A. Manna deserve thanks for helping with some of the experiments.
The work was supported by the Department of Biotechnology, Government of India.
FOOTNOTES
- Received 15 March 1999.
- Accepted 22 June 1999.
- Copyright © 1999 American Society for Microbiology
