Structural basis of bacterial transcription activation

See allHide authors and affiliations

Science  17 Nov 2017:
Vol. 358, Issue 6365, pp. 947-951
DOI: 10.1126/science.aao1923

Structural basis for transcription activation

Bacteria can initiate transcription through two independent classes of recruitment mechanisms. Liu et al. determined the cryo-electron microscopy structure of an intact class I transcription activation complex. The positions and orientations of all the components and the detailed protein-protein and protein-nucleic acid interactions reveal how an activator interacts with the promoter DNA and recruits RNA polymerase through the class I mechanism. Together with a recently reported class II transcription activation complex, the findings complete our structural understanding of bacterial transcription activation.

Science, this issue p. 947


In bacteria, the activation of gene transcription at many promoters is simple and only involves a single activator. The cyclic adenosine 3′,5′-monophosphate receptor protein (CAP), a classic activator, is able to activate transcription independently through two different mechanisms. Understanding the class I mechanism requires an intact transcription activation complex (TAC) structure at a high resolution. Here we report a high-resolution cryo–electron microscopy structure of an intact Escherichia coli class I TAC containing a CAP dimer, a σ70–RNA polymerase (RNAP) holoenzyme, a complete class I CAP-dependent promoter DNA, and a de novo synthesized RNA oligonucleotide. The structure shows how CAP wraps the upstream DNA and how the interactions recruit RNAP. Our study provides a structural basis for understanding how activators activate transcription through the class I recruitment mechanism.

In bacteria, all genes are transcribed by an RNA polymerase (RNAP) holoenzyme consisting of a five-subunit core enzyme (α2ββ′ω) and a sigma factor (13). The polymerase recognizes certain promoter sequences to initiate transcription, a process requiring the assistance of activators (47). The Escherichia coli cyclic adenosine 3′,5′-monophosphate (cAMP) receptor protein (CRP or CAP) is one classic “simple activator.” CAP is a dimeric global transcription factor activated by its allosteric effector cAMP (8). It binds to a 22–base pair (bp) inverted-repeat DNA site and activates transcription at more than 100 different promoters in the E. coli genome (6, 9). At CAP-dependent promoters, CAP activates transcription by direct protein-protein interactions with RNAP, mainly through the RNAP α subunit carboxyl-terminal domain (αCTD) (6). The activation mechanisms are generally categorized into two classes according to their different interaction modes (5, 10). In the case of class I promoters, CAP binds at the region centered at the –61.5 site and makes contact with αCTD using its activation region 1 (AR1) to activate transcription through a recruitment mechanism (6, 11, 12). In the case of class II promoters, CAP binds at the –41.5 site, which overlaps the –35 element, and interacts with multiple regions of RNAP through its AR1, AR2, and AR3 to activate transcription through both recruitment and postrecruitment mechanisms (6, 13, 14).

The crystal structure of a CAP-DNA complex showed an overall bend of ~90° in the DNA (8) and also initiated studies to understand the structural mechanisms of transcription activation by CAP (1517). A later structure of a CAP-αCTD-DNA complex revealed the interactions among the components in further detail (18). However, high-resolution structures of intact transcription activation complexes (TACs) are necessary for more accurate determination of the interactions among all the components and a better understanding of the transcription activation mechanisms. A recent 4.4-Å-resolution crystal structure of the intact class II TAC provided a structural basis for understanding the class II mechanism (19). The structure of an intact class I TAC is believed to be more challenging to obtain because CAP binds DNA further upstream in class I TACs (–61.5) than it does in class II TACs (–41.5). A previous 20-Å-resolution negative-stain electron microscopy (EM) structure of the class I TAC first illuminated the overall architecture of the complex (20). However, the limited resolution made it hard to reveal detailed insights into the mechanism. How CAP wraps the promoter DNA, interacts with RNAP, and activates transcription still need a clearer structural explanation.

To obtain a complete TAC, we constructed a synthetic DNA scaffold representing the complete class I CAP-dependent promoter (from –74 to +14, 88-bp in total), which contains a CAP binding region as observed in the lac promoter; an αCTD binding site; a promoter consensus –35 element; an extended –10 element; a consensus –10 element; a discriminator element; a 14–nucleotide (nt) transcription bubble; and an 11-bp downstream portion (Fig. 1A). To test the ability of the class I TAC to incorporate nucleotides and obtain the structure of the complex trapped in a functional state, we incubated the complex in the presence of the nucleotide mixture [adenosine 5′-triphosphate (ATP), guanosine 5′-triphosphate (GTP), and 3′-dideoxythymidine 5′-triphosphate (ddTTP)].

Fig. 1 Cryo-EM reconstruction of the class I CAP-TAC.

(A) Schematic representation of the synthetic promoter DNA (88 bp) and a de novo synthesized RNA transcript (3 nt) in the CAP-TAC. RNA synthesis starts from the –1 position with a GTP residue, as observed previously (22). T, template; NT, nontemplate. (B) Overview of the E. coli class I CAP-TAC cryo-EM reconstruction at 3.9-Å resolution (contoured at 5.5 σ). The ω subunit and the nonconserved region (NCR) of σ70 are shown using a contour of 1 σ. The CAP-αCTD binding region was reconstructed by using a charge density distribution map (~5.5 Å, contoured at 4.4 σ) generated from the 10-Å-focused cryo-EM map where other portions were excluded during the 3D classification.

The assembled complex was subjected to negative stain– and cryo–EM analyses. A data set of 2382 movies was collected on an aberration-corrected 300-kV Titan Krios cryo-EM with a postenergy-filter Gatan K2 Summit camera, and ~835,000 particles were extracted for further processing. The cryo-EM single-particle reconstruction of the intact class I TAC was obtained at 3.9-Å resolution (Fig. 1B; figs. S1 and S2; and table S1), which enabled the unambiguous docking of all the components into the map. The densities for the nonconserved region of σ70 and the ω subunit are clear at a lower contour level compared to other portions in the density map, suggesting their low occupancies. The density for the CAP-αCTD portion is also visible at a lower contour level, but poorly defined in the initial map, suggesting its low occupancy and flexibility. Further focused three-dimensional (3D) classification in this region generated a 10-Å-resolution map with clear density (fig. S3A). On the basis of the 10-Å cryo-EM map, we then generated a charge density distribution map using Laplacian (21), which greatly improved the quality of the density and created a ~5.5-Å density map (fig. S3B), allowing confident placement of the CAP-αCTD region.

In the structure of the class I CAP-dependent TAC (CAP-TAC) (Fig. 2A), the RNAP adopts a similar overall conformation as in the E. coli σs–transcription initiation complex (σs-TIC) (22), E. coli σ70-TIC (23), and Thermus thermophilus (Tth) σA-TIC (24), suggesting that the binding of the CAP dimer does not induce substantial conformation changes of the RNAP (fig. S4). The corresponding portion of promoter DNA is also similar to those in the TICs except that the orientation of the DNA bubble region is different from that in Tth σA-TIC, probably a result of the difference in the sequence, which suggests that sequence-dependent protein–nucleic acid interactions are present in the bubble region. However, although the same promoter sequence as those in E. coli TICs was used in the assembly of CAP-TAC, the structure displays an open active site and a well-ordered nascent RNA-DNA hybrid in the posttranslocation register, which represents state II in the nucleotide addition cycle (NAC), rather than a closed active site and pretranslocation state (Fig. 2B). This observed difference at the active site could be due to the binding of CAP. Additionally, comparisons of this structure with those E. coli TICs also show that there are obvious conformation differences in the periphery of the RNAP secondary channel (fig. S5). These different conformations represent different states in the NAC, providing insights into the structural mechanism of NAC.

Fig. 2 Cryo-EM structure of the class I CAP-TAC.

(A) Cryo-EM density maps for the overall structure of the class I CAP-TAC (ribbon diagram). The maps are shown in transparent surface representation to allow visualization of all the components of the complex. The contours and color schemes are same as those in Fig. 1. (B) The complete class I CAP-dependent promoter DNA (88 bp) is shown as a ribbon diagram. The inset is the close-up view of the DNA-RNA hybrid region with the mesh cryo-EM map (contoured at ~5.1 σ). A and G indicate adenine and guanine, respectively.

The CAP dimer in the complex binds in the major grooves at the –61.5 site of the promoter DNA, mainly through helix-turn-helix motifs as suggested previously (6, 8), and shows a similar conformation to those observed in the CAP-DNA and CAP-αCTD-DNA complexes (8, 1518). Analysis of the upstream DNA indicates that the CAP dimer binds the promoter DNA and bends it at three regions (Fig. 3A). The upstream DNA first bends ~21° at the –35 element, then bends ~55° and ~33° at the normal sites, as observed in the CAP-DNA complex. These bend angles are slightly different from those (~51° and ~43°) observed in the CAP-DNA complex (8). In addition, when these structures are superimposed on one CAP (CAP I), a comparison of the structures shows a ~2- to 3-Å shift in the orientation of the second CAP (CAP II) and its corresponding bound DNA (fig. S6). These changes may be due to the CAP dimer functioning as a component of the activation complex rather than as an independent complex.

Fig. 3 Structural mechanism of the class I transcription activation by CAP.

(A) Overall structure of the class I CAP-TAC shown in ribbon representation. The characteristic elements on the promoter—the –10 element, extended –10 element, –35 element, αCTD binding site, and CAP-binding region—are colored in red. cAMPs are shown in red stick representations. The other color schemes are same as in Fig. 1. Solid blue lines show the reference lines for determining the bend angles. FTH, β flap-tip helix. (B) A close-up view of the CAP-αCTD region with the interaction residues: AR1 (orange), 261 determinant (plum), 265 determinant (pink), 287 determinant (magenta), and 596 determinant (salmon) shown in sphere representations.

In the structure of the class I CAP-TAC, the CAP dimer makes contact with the 287 determinant of αCTD mainly through AR1 (Fig. 3B and fig. S7). The αCTD binds at a minor groove in the –40 to –45 region with its 265 determinant and interacts with the 596 determinant of σ4 mainly with its 261 determinant (Fig. 3B and fig. S7). The σ4 domain of σ70 binds at the major groove in the –35 element through a helix-turn-helix motif (Fig. 3B). These interactions support the major properties of the previous structures of the partial complexes and biochemical data (5, 6, 18, 25, 26). Structural comparisons with those structures of the partial complexes show similar interactions between CAP I and αCTD, but a ~3- to 4-Å shift in the orientation of the σ4 domain, and its corresponding bound DNA was observed (fig. S6E). The displayed interacting surface between αCTD and σ4 is slightly different in orientation upon CAP binding than that of an earlier model proposed primarily from mutant analysis (27) and is quite different from one that suggested that there is substantial remodeling of RNAP-promoter interactions (25).

Comparison of this structure with the 20-Å-resolution negative-stain EM structure of the class I TAC (20) shows considerable differences (Fig. 4, A and B). First, the CAP dimer binding region shows an obviously different orientation in the 20-Å-resolution structure (Fig. 4B), which results in the much shorter distances between CAP II and the RNAP α subunit amino-terminal domains (αNTDs) (~13 Å) and between the farthest end of the upstream DNA and αNTDs (~18 Å). In this structure, those distances are ~22 and ~52 Å, respectively. Second, the upstream DNA bends at five sites in the 20-Å-resolution structure (Fig. 4C): the spacer region between the –35 and –10 elements (30°); the –35 element (26°); the linking site between αCTD and CAP binding regions (22°); and two CAP binding sites (51° and 35°). Note that there are differences in the lengths of the spacer and upstream DNA between the present (17- and 74-bp) and previous (18- and 78-bp) EM structures. The difference in the conformation of the spacer region between the two structures could be due to the different length of spacer. The structures containing a 17- or 18-bp spacer may represent two subclasses of class I promoter: a consensus CAP-independent promoter and one as in the E. coli lac promoter. Third, the interactions between the CAP I protein and αCTD and between αCTD and σ4 are quite different in the 20-Å-resolution structure from those in this 3.9-Å-resolution structure (Fig. 4C). The interaction of the β flap-tip helix and αCTD was not observed in this structure, although such an interaction was suggested by the 20-Å-resolution model. Structural comparison with the 4.4-Å-resolution crystal structure of the class II TAC (19) shows that the TAP dimer (a CAP homolog found in Tth) displays a different position from that in the class I TAC and binds at the –41.5 site to bend the promoter DNA only at two regions (54° and 20°). It also shows that αCTD displays a totally different position and different interactions with other components in the class II TAC from those in the class I TAC and that there is no bending in the spacer region in the class II TAC, which is same as observed in this structure (Fig. 4D).

Fig. 4 Structural comparisons with other class I and class II TACs.

Superimposition of this complex with the previous E. coli class I CAP-TAC and Tth class II TAP-TAC onto the RNAP. (A) Overall structure of the class I CAP-TAC in this study shown in ribbon representation. The β, β′, and ω subunits are colored in light gray. Other color schemes are the same as in Fig. 1. (B) Overall architecture of the 20-Å-resolution negative-stain EM structure of a class I CAP-TAC [Protein Data Bank (PDB) 3IYD] (20). (C) A view of the superimposed promoter DNAs (left) and a close-up view of the CAP-αCTD-σ4-DNA regions (right) between the two class I TACs. Superimposition of the CAP-αCTD-σ4-DNA regions was performed via the αCTD. The T strand, NT strand, CAP I, αCTD, FTH, and σ4 of the 20-Å-resolution structure are colored in dark gray, light gray, white, blue, olive, and white, respectively. Magenta lines show the reference lines for determining the bend angles. (D) Overall architecture of the 4.4-Å-resolution crystal structure of a class II TAP-TAC (PDB 5I2D) (19). A view of the superimposed promoter DNAs between the class I and class II TACs is shown on the right side. The T strand, NT strand, and RNA transcript in the 4.4-Å-resolution structure are shown in dark gray, light gray, and pink, respectively.

The structure of this intact class I CAP-TAC clearly reveals the position and orientation information of all the components and the detailed protein-protein and protein–nucleic acid interactions. Although these observations support the major concepts in the previous structural and biochemical data, there exist many notable differences in detailed interactions, such as how the CAP dimer wraps and bends the promoter DNA and how the αCTD interacts with CAP and the σ4 domain. The structure also confirms the previous proposed recruitment mechanism on the class I transcription activation (5).

Supplementary Materials

Figs. S1 to S7

Table S1

References (2843)

References and Notes

Acknowledgements: We thank Y. Wang for constructing E. coli CAP and σ70 plasmids, J. Wang for advice on converting the cryo-EM map to a charge density distribution map, and Y. Yang for his careful and critical reading of the manuscript. We also thank the staff of the Center for Structural Biology Facility at Yale University for computational support. This work was supported by grant GM22778 to T.A.S. from the NIH. T.A.S. is an investigator of the Howard Hughes Medical Institute. B.L. performed protein-sample preparations and complex assembly used in the structure determination, negative-stain transmission EM analysis, all model building, refinement, and structural analysis experiments. C.H. and Z.Y. designed the cryo-EM experiment. C.H. performed cryo-EM grid preparation, screening and optimization, high-throughput data collection, image processing, and map generation with input from R.K.H. and Z.Y. B.L. and T.A.S. designed the experiments. B.L. and T.A.S. principally wrote the manuscript with input from all. The 3D cryo-EM density map and the coordinates for the structure of the class I CAP-TAC have been deposited in the Electron Microscopy Data Bank and Protein Data Bank under the accession codes EMD-7059, EMD-7060, and 6B6H, respectively. The authors declare no competing financial interests.
View Abstract

Stay Connected to Science

Navigate This Article