CarD uses a minor groove wedge mechanism to stabilize the RNA polymerase open promoter complex

Abstract

A key point to regulate gene expression is at transcription initiation, and activators play a major role. CarD, an essential activator in Mycobacterium tuberculosis, is found in many bacteria, including Thermus species, but absent in Escherichia coli. To delineate the molecular mechanism of CarD, we determined crystal structures of Thermus transcription initiation complexes containing CarD. The structures show CarD interacts with the unique DNA topology presented by the upstream double-stranded/single-stranded DNA junction of the transcription bubble. We confirm that our structures correspond to functional activation complexes, and extend our understanding of the role of a conserved CarD Trp residue that serves as a minor groove wedge, preventing collapse of the transcription bubble to stabilize the transcription initiation complex. Unlike E. coli RNAP, many bacterial RNAPs form unstable promoter complexes, explaining the need for CarD.

Keywords: CarD; E. coli; RNA polymerase; Thermus aquaticus; X-ray crystallography; biochemistry; biophysics; structural biology; transcription factors.

Figure 1.. Structure of the Thermus CarD/RPo complex.

(A) Synthetic oligonucleotides used for CarD/RPo crystallization. The numbers above denote the DNA position with respect to the transcription start site (+1). The DNA sequence is derived from the full con promoter (Gaal et al., 2001). The −35 and −10 (Pribnow box) elements are shaded yellow, the extended −10 (Keilty and Rosenberg, 1987) and discriminator (Feklistov et al., 2006; Haugen et al., 2006) elements purple. The nt-strand DNA (top strand) is colored dark grey; the t-strand DNA (bottom strand), light grey; the RNA transcript, red. The colored blocks denote protein/nucleic acid interactions: σ^A, orange; β, cyan; β′, pink; CarD, green. CarD interacts exclusively at the upstream junction of the transcription bubble. (B) Overall structure of CarD/RPo—two orthogonal views. The nucleic acids are shown as CPK atoms and color-coded as above. Proteins are shown as molecular surfaces. The RNA polymerase (RNAP) holoenzyme is color coded as follows: αI, αII, ω, grey; β′, light pink; Δ1.1σ^A, light orange; β is light cyan except the β1-lobe (interacting with the CarD-RID, corresponding to RNAP β subunit residues 18–138 and 333–392) is light blue. The CarD-RID is magenta, CarD-CTD green. In the right view, the boxed region is magnified in (C). (C) Magnified view illustrating the CarD-RID/β1-lobe protein/protein interaction and CarD-CTD (α3 and α5)/DNA interactions at the upstream ds(−12)/ss(−11) junction of the transcription bubble. (D) CarD does not alter the transcription bubble. KMnO₄ footprints (t-strand) of Thermus RNAP holoenzyme on the Mtb AP3 promoter. (Top) Sequence of the AP3 promoter (Hartmann et al., 1987). T-strand thymidines rendered KmnO₄ reactive by RNAP are denoted (red arrows). (Bottom) KMnO₄ footprints. Lane 1, no protein added; lanes 2–3, RNAP holoenzyme − or + CarD (respectively); lanes 4–7, the effect of incubating with a competitor promoter trap for the indicated amounts of time. DOI: http://dx.doi.org/10.7554/eLife.08505.004

Figure 1—figure supplement 1.. Sequences of Mtb rrnAAP3 (Gonzalez-y-Merchand et al., 1996) and Tth 23S ribosomal RNA (rRNA) (Hartmann et al., 1987), promoters used in in vitro assays, and full con (Gaal et al., 2001) used for structural studies.

DOI: http://dx.doi.org/10.7554/eLife.08505.005

Figure 1—figure supplement 2.. Crystal packing interactions in CarD/RPo P4₃2₁2 crystals.

One asymmetric unit of the crystals contains two CarD/RPo complexes, complex A [RNAP(A), cyan; CarD(A), blue] and complex B [RNAP(B), pink; CarD(B), red]. One central asymmetric unit is shown (proteins as molecular surfaces), with neighboring symmetry-related complexes shown as ribbons; only symmetry-related complexes that make crystal packing contacts with the central asymmetric unit are shown. CarD(A) makes a crystal packing contact with a symmetry-related CarD(A) (circled in red), but CarD(B) is not involved in any crystal packing interactions. Nevertheless, the protein/protein and protein/DNA contacts in complex(A) and complex(B) are essentially identical. DOI: http://dx.doi.org/10.7554/eLife.08505.006

Figure 1—figure supplement 3.. CarD/β1-lobe structure.

(Top) View of the CarD/RPo structure, similar to Figure 1B (Right) except the RNAP β1-lobe and CarD are shown as backbone ribbons without surfaces. (Bottom) The CarD/β1-lobe structure (2.4 Å-resolution, Table 1) shown in the orientation corresponding to the top view. DOI: http://dx.doi.org/10.7554/eLife.08505.007

Figure 1—figure supplement 4.. Slight movement of CarD-CTD towards DNA when DNA is present.

CarD/us-fork and CarD/RPo structures (four copies, two crystallographically independent copies from each structure) are shown superimposed by the Cα positions in the β1-lobe. In all these structures in the presence of promoter DNA, the β1-lobe is colored cyan, CarD is colored dark red, and CarD-W86 is shown in CPK format. The CarD/β1-lobe structure is also superimposed by the Cα positions of the β1-lobe (slate blue), with the CarD-RID magenta and the CarD-CTD green. Viewing the structures superimposed this way reveals a rotation of the CarD-CTD of ∼11° towards the DNA (when promoter DNA is present). DOI: http://dx.doi.org/10.7554/eLife.08505.008

Figure 1—figure supplement 5.. Data and model quality.

Plots relating data quality with model quality using the Pearson correlation coefficient (CC) analysis described by Karplus and Diederichs (2012). CC1/2 (red squares) was determined from the unmerged diffraction data randomly divided in half. Since CC1/2 underestimates the information content of the data (since it's calculated by dividing the dataset in half), CC* was calculated from an analytical relation to estimate the information content of the full data (Karplus and Diederichs, 2012). CC* provides a statistic that assesses data quality as well and also allows direct comparison of crystallographic model quality and data quality on the same scale through CC_work and CC_free, the standard and cross-validated correlations of the experimental intensities with the intensities calculated from the refined model. A CC_work/CC_free smaller than CC* indicates that the model does not account for all of the signal in the data, meaning it is not overfit. Plotted also are the standard <I>/σI for the diffraction data, as well as the R_work/R_free for the refined models. (Left) Data for Tth CarD/Taq EΔ1.1σ^A/us-fork (−12 bp) at 4.4 Å-resolution. (Right) Data for Tth CarD/Taq EΔ1.1σ^A RPo (with 4-nt RNA primer) at 4.3 Å-resolution. DOI: http://dx.doi.org/10.7554/eLife.08505.009

Figure 1—figure supplement 6.. CarD does not alter the structure of the transcription bubble.

Superimposition of the nucleic acids from the CarD/RPo (colored as in Figure 1A) and RPo (magenta) (Bae et al., 2015) structures. The only significant differences occur in the single-stranded t-strand from −11 to −7; this part of the DNA is relatively unconstrained by protein/DNA interactions and has very high B-factors. DOI: http://dx.doi.org/10.7554/eLife.08505.010

Figure 2.. CarD-CTD/promoter DNA interactions.

(A) Stereo view of the refined, B-factor sharpened (−80 Ų) 2F_o − F_c map (grey mesh, contoured at 1σ), with superimposed DNA and CarD. Density for the close approach of the CarD peptide backbone to the −14(t) DNA phosphate backbone and for CarD-W86 are clearly resolved. (B) Close up view showing interactions between the N-terminal ends of α3 and α5 of the CarD-CTD with promoter DNA at the upstream ds(−12)/ss(−11) junction of the transcription bubble. Grey dashed lines indicate potential polar interactions between the peptide backbone nitrogen of L124 and the −14(t) phosphate oxygen, and W86 Nε and O2 of T₋₁₂(nt). (C) Same view as Figure 2B. Superimposed is the simulated annealing omit map (dark green mesh, F_o − F_c, contoured at 3σ), calculated from a model where CarD-W86 was mutated to Ala. The unbiased difference Fourier density shows that the side chain position is specified in the data. DOI: http://dx.doi.org/10.7554/eLife.08505.012

Figure 2—figure supplement 1.. Alignment of CarD homologs found in bacteria from 11 diverse phyla/groups.

The CarD sequences shown are from the following organisms chosen to represent the preceding phylum/group: Deinococcus–Thermus-Tth HB8, Actinobacteria–Mtb, Acidobacteria–Candidatus Solibacter usitatus, α-Proteobacteria–Rickettsia belli, Aquificae–Desulfurobacterium thermolithotrophum, Chlamydae–Chlamydae trachomatis, Cyanobacteria–Mastigocoleus testarum, δ-Proteobacteria–Desulfobulbus propionicus, Firmicutes–Bacillus cereus, Spirochaetes–Treponema pallidum and Thermodesulfobacteria–Thermodesulfatator atlanticus. Alignments were performed using the ClustalW algorithm in MegAlign (DNASTAR). Groups of residues considered homologous (DE), (HKR), (ALMIV), (NQ), (ST), (C), (G) and (P) are shaded blue when occurring in greater than 9/11 sequences. Identical residues occurring in all 11 sequences are shaded black. Histograms above the alignment graphically illustrate residues that are absolutely conserved within each of the 11 sequences and W86 is asterisked. The 100% identical residue is listed immediately below the histogram. A larger alignment of 831 CarD sequences is included (Source code 1). DOI: http://dx.doi.org/10.7554/eLife.08505.013

Figure 3.. Function of CarD-W86.

(A) The effect of CarD-W86 substitutions on activation of abortive initiation (UpG dinucleotide + α-³²P-CTP) on the Tth rrnA-23S promoter (normalized with respect to no CarD). Error bars denote the standard error from a minimum of three experiments. (B) The effect of promoter −12 base pair substitutions on activation of abortive initiation (GpU dinucleotide + α-³²P-UTP) by CarD on the Mtb rrnA-AP3 promoter. Error bars denote standard errors. DOI: http://dx.doi.org/10.7554/eLife.08505.014

Figure 3—figure supplement 1.. Complete gels for the abortive initiation assays shown in (A) Figure 3A and (B) Figure 3B.

DOI: http://dx.doi.org/10.7554/eLife.08505.015

Figure 4.. Inter-domain crosslinking confirms the functional conformation of CarD.

(A) View of the Thermus CarD/RPo complex. RNAP holoenzyme and nucleic acids are shown as in Figure 1B; Tth CarD is shown as an α-carbon ribbon (Tth CarD-RID, magenta; Tth CarD-CTD, green) but with W86 shown in CPK format and colored dark green. Also shown is Mtb CarD from the Mtb CarD/β1-β2-lobe structure (4KBM; Mtb CarD-RID, dark red; Mtb CarD-CTD, brown, but with W85 colored dark brown), superimposed by alignment of 145 Cα atoms from the β1-lobe (1.39 Å rmsd). The boxed region is magnified in (B). (B) (Left) Magnified view showing the modeled Mtb CarD in the context of RPo. The α-carbons of CarD-RID-P12 and CarD-CTD-G99, shown as red spheres, are ∼24 Å apart (red dashed line). A disulfide bond between these two positions in Mtb CarD2C (P12C/G99C substitutions) would disallow this conformation of CarD. (Right) Magnified view of the Thermus CarD/RPo complex. CarD-RID-P13 and CarD-CTD-G100 are ∼5.2 Å apart (red dashed line). A disulfide bond between the corresponding two positions in Mtb CarD2C would lock this DNA-interacting conformation of CarD. (C) Purification of disulfide crosslinked (lanes 5, 6) and reduced (lanes 7, 8) CarD2C. Non-reducing SDS-PAGE illustrates that CarD2C is oxidized (crosslinked) in the absence of reducing agent dithiothreitol (DTT) and is reduced (uncrosslinked) in the presence of DTT. Samples were excised from gels and LC-MS was used to confirm oxidation states. (D) Effect of oxidation state on Mtb CarD2C activation of abortive transcription on the Mtb AP3 promoter (GpU dinucleotide + α-³²P-UTP). Conformationally locked (no DTT) Mtb CarD2C exhibits wild type activation activity. DOI: http://dx.doi.org/10.7554/eLife.08505.016

Figure 4—figure supplement 1.. Complete gel for the abortive initiation assay shown in Figure 4D.

DOI: http://dx.doi.org/10.7554/eLife.08505.017

Figure 5.. CarD increases the lifetime of Thermus RPo.

(A) Sequences of Mtb rrnAAP3 (Gonzalez-y-Merchand et al., 1996) and Tth 23S rRNA (Hartmann et al., 1987) promoters used in in vitro assays. (B, C) Lifetimes of promoter complexes measured by abortive transcription. At the top of each panel, [³²P]-labeled abortive transcript production at times after addition of a large excess of competitor promoter DNA trap was monitored by polyacrylamide gel electrophoresis and autoradiography. On the bottom, transcript production was quantified by phosphorimagery and plotted. The lines indicate single-exponential decay curves fit to the data points. The calculated decay half-lives (t_1/2) are shown to the right of the gel images. Assays were performed on the following templates: (B) Tth rrnA-23S promoter (UpG dinucleotide + α-³²P-CTP). (C) Mtb rrnA-AP3 promoter (GpU dinucleotide + α-³²P-UTP). DOI: http://dx.doi.org/10.7554/eLife.08505.018

Figure 5—figure supplement 1.. Complete gels for the abortive initiation assays shown in Figure 5B.

DOI: http://dx.doi.org/10.7554/eLife.08505.019

Figure 5—figure supplement 2.. Complete gels for the abortive initiation assays shown in Figure 5C.

DOI: http://dx.doi.org/10.7554/eLife.08505.020

Figure 6.. CarD increases the lifetime of Thermus RPo by preventing transcription bubble collapse.

(A) Synthetic duplex (23S_DS) and artificial bubble (23S_Bub) promoters used in in vitro assays. (B) Lifetimes of promoter complexes formed on synthetic templates measured by abortive transcription (UpG dinucleotide + α-³²P-UTP). (Left) [³²P]-labeled abortive transcript production at times after addition of a large excess of competitor promoter DNA trap was monitored by polyacrylamide gel electrophoresis and autoradiography. (Right) transcript production was quantified by phosphorimagery and plotted. The lines indicate single-exponential decay curves fit to the data points. The calculated decay half-lives (t_1/2) are shown to the right of the gel images. Assays were performed on the synthetic double-stranded (23S_DS) and bubble (23S_Bub) templates. DOI: http://dx.doi.org/10.7554/eLife.08505.021

Figure 6—figure supplement 1.. Complete gels for the abortive initiation assays shown in Figure 6B.

DOI: http://dx.doi.org/10.7554/eLife.08505.022