Structural Insights into the Mechanism of the Allosteric Transitions of Mycobacterium tuberculosis cAMP Receptor Protein

Abstract

The cAMP receptor protein (CRP) from Mycobacterium tuberculosis is a cAMP-responsive global transcriptional regulator, responsible for the regulation of a multitude of diverse proteins. We have determined the crystal structures of the CRP·cAMP and CRP·N⁶-cAMP derivative-bound forms of the enzyme to 2.2- and 2.3 Å-resolution, respectively, to investigate cAMP-mediated conformational and structural changes. The allosteric switch from the open, inactive conformation to the closed, active conformation begins with a number of changes in the ligand-binding cavity upon cAMP binding. These subtle structural changes and numerous non-bonding interactions between cAMP, the N-domain residues, and the C-domain helices demonstrate that the N-domain hairpin loop acts as a structural mediator of the allosteric switch. Based on the CRP·N⁶-cAMP crystal structure, binding of N⁶-cAMP with a bulkier methylphenylethyl extension from the N6 atom stabilizes the cAMP-binding domain, N-domain hairpin, and C-terminal domain in a similar manner as that of the CRP·cAMP structure, maintaining structural integrity within the subunits. However, the bulkier N6 extension of N⁶-cAMP (in R conformation) is accommodated only in subunit A with minor changes, whereas in subunit B, the N6 extension is in the S conformation hindering the hinge region of the central helix. As a result, the entire N-domain and the C-domain of subunit B integrated by the cAMP portion of this ligand, together tilt away (∼7° tilt) from central helix C, positioning the helix-turn-helix motif in an unfavorable position for the DNA substrate, asymmetrically. Together, these crystal structures demonstrate the mechanism of action of the cAMP molecule and its role in integrating the active CRP structure.

Introduction

Mycobacterium tuberculosis (Mtb)³ is thought to enter a latent phase (1) to survive hostile environments like starvation and hypoxia (2). During times of duress, transcriptional regulators belonging to the CRP/fumarate and nitrate reduction regulator family respond to a broad spectrum of intracellular and exogenous signals associated with low oxygen stress and starvation, modulating the expression of various metabolic genes in many facultative or strictly anaerobic bacteria (3, 4). In Escherichia coli, CRP controls the expression of over 100 genes in response to changes in the intracellular concentration of cAMP (5). CRP is activated by the binding of cAMP and subsequently becomes an active transcriptional regulator by means of allosteric structural changes (6–10). Several essential genes are up-regulated during intracellular growth in macrophages (11), suggesting that cAMP signaling may be important to M. tuberculosis during its interaction with the host. Indeed, Rv3676 (Mtb-CRP), the CRP homolog in M. tuberculosis H37Rv, is a cAMP-responsive transcription factor (12–14). Deletion of Rv3676 results in impaired growth in laboratory medium, in bone marrow-derived macrophages, and in tubercle bacillus grown in a mouse model system (13). These observations suggest that Mtb-CRP plays a critical role in Mtb survival, especially in macrophages.

Several models have been offered as possible explanations for how the sensory module can transmit the signal to the distant DNA-binding domain (9, 15–17). Perhaps the most intriguing question about the CRP family is how ligand binding within the effector domain can influence the properties of the DNA-binding domain. Answers to this question remain elusive, in part because the only CRP crystal structure available for Mtb is that of the cAMP-free form (18) The Mtb-CRP apo structure displayed asymmetry between the subunits as compared with the E. coli CRP·cAMP crystal structure, which shares only a 32% sequence identity with Mtb-CRP. Although the asymmetry observed in the apoMtb-CRP crystal (18) has been interpreted as an important factor in its mechanism, NMR studies in solution could not find any evidence of structural asymmetry in either the E. coli apoCRP or the cAMP·CRP (19–21). In contrast, recent NMR studies (22) reported cAMP binding resulting in a coil to helix transition that extends the coil-coil dimerization and functions as a regulatory switch. Nevertheless, the allosteric mechanism by which ligand binding induces a conformational transition remains elusive from the same organism and atomic level structural comparisons are needed to fully understand the transition.

Although there is mounting evidence to suggest that the E. coli CRP undergoes allosteric conformational changes upon cAMP binding from a variety of spectroscopic and biochemical techniques (23–26), little is known about the nature of the conformational change to the allosterically activated conformation in Mtb-CRP. Wild-type CRP is activated by cAMP only, although it can bind other cyclic nucleotides with comparable affinity (27). Ligands such as cGMP and inosine 3′,5′-monophosphate, which are structurally similar to cAMP, fail to activate transcription (28). In E. coli chemical modifications of cAMP that retain function have provided important clues concerning the allosteric mechanism of action. Experiments with cAMP analogs have shown that the 2′OH-, 3′O-, and 5′O-groups, the overall negative charge, and the N6 amino group cannot be modified without losing biological activity in vivo, whereas the N1 and N7 groups of adenine are not essential (29). Ebright et al. (30) reported analogs of cAMP with bulky additions at C2 and N6, which bind CRP and induce a conformational change but do not induce transcription. Moreover, it has also been suggested that the binding of the CRP·cAMP complex to DNA introduces distinct alterations in the CRP structure in proximity to the N6 moiety of the bound cAMP molecule. To further investigate this phenomenon, one of the N6-modified cAMP analogs (N⁶-(1-methyl-2-phenylethyl)adenosine-3,5-cyclic monophosphate, hereafter N⁶-cAMP) reported by Ebright et al. (30) was employed as a non-functional analogue of cAMP to investigate the differential effects of allosteric transitions. Although N6 modification of the cAMP ligand has lead to information regarding CRP activation in E. coli, no atomic level structures of these substituted ligands are available.

We present here two crystallographic structures: that of Mtb-CRP bound to cAMP and that of Mtb-CRP bound to an N⁶-cAMP ligand. Determination of both the CRP·cAMP and CRP·N⁶-cAMP structures has enabled us to explore various allosteric models, giving more detailed insight into the allosteric behavior of this transcriptional regulator. Our structures of the cAMP- and N⁶-cAMP-bound CRP allow for the first comparison of apoCRP to cAMP and inhibitor-bound CRP within the same species. Crystallographic analysis of these structures has lead to a model in which the Mtb-CRP exists as a symmetric dimer, in which both subunits exist in an open form. Furthermore, our structures implicate a model for the allosteric switch from the inactive apo form to the active cAMP-bound form. Our model implies that binding of cAMP triggers alterations in the cAMP-contacting residues and shifts the N-terminal domain, consequently allowing the DNA-binding domain to accept DNA. The structural basis of our CRP inhibitor is based upon abrogation of dimer interactions and a concomitant helix unwinding at the interface of one of the subunits. Ultimately, these results will aid in understanding the regulation mechanisms controlling the expression of the corresponding CRP-regulated genes at a level that enables guided inhibition and antibiotic design.

EXPERIMENTAL PROCEDURES

The cAMP was obtained from Sigma, and N⁶-cAMP (N⁶-(1-methyl-2-phenylethyl)adenosine-3,5-cyclic monophosphate) from BIOLOG Life Science Institute, Germany. A Hi-Trap Ni²⁺ chelating column from Amersham Biosciences was used for protein purification. Crystal screen solutions from Hampton Research and Emerald Biosciences were used for obtaining CRP crystals.

Mtb-CRP Cloning, Expression, and Purification

A 675-bp DNA fragment containing the CRP gene (Rv3676c) was amplified by PCR from M. tuberculosis H37Rv genomic DNA as the template, using the following oligonucleotides as the forward and reverse primers, respectively: 5′-AGATGAAGCCATATGGACGAGATCCTGGCCAGGGCAGGA-3′ and 5′-AGA GTA AGC TTA CCT CGC TCG GCG GGC CAG TCT-3′. The amplified DNA fragment was digested with NdeI and HindIII and cloned into the corresponding restriction sites in the pET28b vector (Novagen) to generate a recombinant vector containing a 5′ sequence encoding a 20-amino acid N-terminal His tag and a tobacco etch virus-cleavage site. The CRP-pET28b vector was transformed into E. coli BL21(DE3) cells by heat shock transformation. The transformed cells were grown to mid log exponential phase at 37 °C in Luria Broth (LB) media containing 50 μg/ml kanamycin. Expression of CRP was induced with 1 mm isopropyl-1-thio-β-d-galactopyranoside, and cells were harvested 12 h after induction at 25 °C.

The harvested cell pellet was resuspended in Buffer C (20 mm Tris-HCl, pH 8.0, 10 mm imidazole, 0.5 m NaCl, and 10% glycerol) with 1 mm phenylmethanesulfonyl fluoride and Complete^TM EDTA-free protease inhibitor mixture (Roche Applied Science). The cells were disrupted by two passages through a cooled French pressure cell. The resulting cell extract was centrifuged at 15,000 rpm at 25 °C for 1 h (the protein precipitates at 4 °C). The cleared supernatant was loaded onto a Hi-Trap Ni²⁺ chelating column (Pharmacia Biosciences) and washed with 300 ml of Buffer A containing 20 mm Tris-HCl, pH 7.5, 75 mm imidazole, 0.5 m NaCl, and 5% glycerol. His-tagged CRP was eluted with a 150-ml linear gradient of 100–500 mm imidazole in 20 mm Tris-HCl, pH 7.5, and 0.5 m NaCl.

The TEV tag was then removed by incubation with tobacco etch virus protease (1:50 ratio) at room temperature while dialyzing to remove imidazole. The TEV and tag were then separated from Mtb-CRP by passing the dialyzed sample through another nickel affinity chromatography cartridge. Purified Mtb-CRP was dialyzed for the second time in the presence of 20 mm Tris-HCl (pH 7.5), 50 mm NaCl, 5% glycerol, and 1 mm dithiothreitol at room temperature. After purification to near homogeneity by size-exclusion chromatography on a Superdex S-200 column (Amersham Biosciences), CRP was dialyzed against 20 mm Tris-HCl (pH 7.5), 50 mm NaCl, 5% glycerol, and 1 mm dithiothreitol at room temperature. Purified recombinant protein was concentrated to 20 mg/ml. The protein was >95% pure as observed by SDS-PAGE.

The pET28b-CRP plasmid was transformed into E. coli B834 (DE3) (Novagen) Met auxotroph strain. Cells were grown in LB media to an optical density of 0.6. Cells were pelleted by centrifugation, washed with 20 mm Tris, pH 7.5, and resuspended in M9 minimal medium lacking l-Met·selenomethionine was then added to a final concentration of 0.05 μg/ml along with 50 μg/ml kanamycin. Cultures were then induced with 1 mm isopropyl-β-d-thiogalactopyranoside followed by incubation for 8 h at 25 °C. The protein was purified using the same method described for the native protein.

Crystallization and Data Collection

Crystallization screening was carried out with CRP·cAMP as well as CRP·N⁶-cAMP preincubated with 2.5 mm of each ligand for 2 h at 25 °C. Initial CRP·cAMP hits were optimized, and diffraction quality crystals were obtained at 25 °C when 4-μl (2 μl of protein:2 μl of reservoir solution) drops were equilibrated against 500 μl of well solution containing 1 m sodium citrate, 100 mm CHES, pH 9.5, using a hanging drop, vapor-diffusion setup. The binary CRP·N⁶-cAMP derivative complexes crystallized in 0.4 m NaH₂PO₄/1.6 M K₂HPO₄, 0.1 m imidazole, pH 8.0, 0.2 m NaCl₂. Diffraction quality crystals were obtained after 3–4 days. The crystal data are presented in Table 1.

TABLE 1.

X-ray crystallographic data for Mtb-CRP·cAMP and Mtb-CRP·N⁶-CAMP structures

The values in parentheses are for high resolution shells.

	cAMP-bound Mtb-CRP	N6-cAMP-bound Mtb-CRP
Protein Data Bank ID	3I54	3I59
Data collection
Wavelength (Å)	0.9793	0.9653
Space group	P21	C2
Number of molecules in asymmetric units (Z)	4	2
Resolution (Å)	50-2.2	20-2.3
Unit cell a, b, c (Å)	68.25, 96.33, 79.26	113.73, 75.72, 63.64
Unique angle (°)	113.46	110.91
Redundancy	3.8 (3.3)	3.6 (3.1)
Observations	47,764	22,400
Observations test set	2366	1149
Completeness (%)	96.7 (79.1)	97.9 (89.0)
Rsyma (%)	4.6 (36.9)	5.8 (41.9)
I/Iσb	37.76 (3.01)	20.4 (2.60)
Refinement
Rworkc	21.1	23.41
Rfree	26.46	28.37
Number of atoms
Protein	6565	3096
Ligand	88	62
Ramachandran analysis
Most favorable (%)	95.82	95.29
Most favorable plus allowed (%)	99.1	99.0
r.m.s.d.
Bond lengths (Å)	0.016	0.006
Bond angles (°)	1.822	0.802

^aR_sym = Σ_hΣ_i|I_hi − 〈I_h〉|/Σ_hΣ_iI_hi, where I_hi is the ith observation of the reflection h, whereas 〈I_h〉 is the means intensity of reflection h.

^b I/Iσ = the mean I/sigma for the unique reflections in the output file.

^c R_work = Σ|F_o| − |F_c|/|F_o|. R_free was calculated with a fraction (5%) of randomly selected reflections excluded from refinement.

Structure Determination, Model Building, and Refinement

The, experimental phases from the selenomethionine incorporated CRP·cAMP crystal were obtained using single-wavelength anomalous dispersion phasing (Table 1). The selenomethionine CRP·cAMP crystal diffracted to a resolution of 2.2 Å at APS-23ID Advanced Photon Source, Argonne National Laboratory. Twelve selenium sites were found using SHELXD (31). SOLVE and RESOLVE (32) were used to refine the sites, calculate the initial protein phases, and build the model. Further phase improvement with solvent flattening in AUTOSHARP (33) resulted in high quality density-modified maps that showed clear electron density. A final model was obtained after several cycles of manual model building using XTALVIEW (34). Water molecules were manually added during iterative cycles of model building and refinement using an F_o − F_c map.

Diffraction data from CRP complexed with cAMP were nearly isomorphous to the selenomethionine CRP crystal model. Bias-minimized electron density maps were obtained using the Shake & Warp (SNW) protocol (35). Clear electron density for cAMP was visible in the SNW map prior to any model building. Several cycles of manual model building and NCS restrained maximum likelihood refinement in REFMAC5 (36) were performed until R-factors converged. See Table 1 for the CRP·cAMP complex. The data set for CRP·N⁶-cAMP was collected at 2.3-Å resolution at APS-23ID Advanced Photon Source, Argonne National Laboratory. The structures of the CRP·N⁶-cAMP complex were determined by refining the CRP·cAMP model against the data for the complex. cAMP and cAMP-analog were clearly defined in an F_o − F_c electron density map contoured at 3 σ and manually fitted. The final refinement statistics are given in Table 1. The CRP·cAMP- and N⁶-cAMP-bound structures were geometrically validated using Molprobity (37).

Electrophoretic Mobility Shift Assay

A DNA gel shift assay was used to probe CRP ligand binding. We selected the binding site upstream of Mtb SerC-phosphoserine aminotransferase (Rv0884), which is putatively regulated by CRP as identified previously (14). Complementary synthetic double-stranded 28-mer DNA oligonucleotide (CTTGCATGTGAGCTTGTTCACACTACGC) present upstream of Rv0884 (SerC) was end-labeled with [γ-³²P]ATP using T4 polynucleotide kinase. Two nanomolar of labeled oligonucleotides was incubated with 50 nm Mtb-CRP recombinant protein in 20 μl of CRP binding buffer (10 mm Tris-HCl, pH 8.0, 50 mm KCl, 25 mm MgCl₂, 1 mm EDTA, 55 μg/ml bovine serum albumin, 1 mm dithiothreitol, 0.05% Nonidet P-40, 1 μg of nonspecific competitor DNA poly(dI-dC), Amersham Biosciences) containing (75 and 150) μm freshly made cAMP or N⁶-cAMP for 30 min at 23 °C. After incubation, 3 μl of loading buffer (CRP binding buffer containing 50% glycerol and 0.1 mg/ml bromphenol blue) was added, and samples were immediately loaded onto 5% polyacrylamide gels and run at 6–8 V/cm at 23 °C for 90 min. Following electrophoresis, the CRP·DNA complexes were detected by autoradiography or exposure to phosphorimaging storage screens.

RESULTS AND DISCUSSION

The x-ray crystal structure of the Mtb-CRP·cAMP binary complex was solved using the single-wavelength anomalous dispersion method (32) with co-crystals of selenomethionylated protein in the space group P2₁ at 2.2-Å resolution. Data collection and refinement statistics are summarized in Table 1. The asymmetric units of the Mtb-CRP·cAMP complex structure contains two dimers, denoted as AB and CD, in which three N-terminal regions of the subunits take on different conformations (Fig. 1, left panel, and supplemental Fig. 1a). Each of these subunits contained one cAMP molecule in anti conformation (Fig. 1).

FIGURE 1.

The x-ray crystal structure of the Mtb-CRP·cAMP binary complex. Top left, structure of the Mtb-CRP·cAMP dimer; the crystallographic asymmetric unit contained two independent dimers, the CD dimer (blue and cyan ribbons) is superposed on the AB dimer (yellow and golden yellow ribbons). The cAMP molecules of the CD dimer are shown in a Corey, Pauling, and Koltun model. The N-terminal domain of each subunit is formed by helices N1–N3, helices A and B, and strands β1–β8; the C-terminal domain is formed by helices D–G and strands β9–β12; helix C forms the dimer interface connecting N- and C-terminal domains. Ribbons of the subunit B (D) of the CRP dimer are indicated by primed letters. The HTH helices E and F and E′ and F′ form the DNA binding region of the CRP dimer. Top right, ribbon representation of the subunit D of the CRP dimer; the cAMP molecule is shown in a Corey, Pauling, and Koltun model. The helices are colored in green, β strands in blue, and loops in pale green. The relative positioning of N-domain hairpin (β4–β5), cAMP, and C-domain is also depicted. Bottom, electron density of cAMP complexed with Mtb-CRP is shown with the nearby secondary structural elements and residue Arg-130 of helix C. The single-wavelength anomalous dispersion experimental maps are contoured at the 1 σ level.

The overall fold and dimer organization of the binary complex resembles that of the previously published apoMtb-CRP and E. coli CRP structures. Each subunit of the dimer is composed of an N-terminal cAMP-binding domain (residues 1–114), a C-terminal DNA-binding domain (146–223), and a hinge region defined by a long helix (residues 117–144 form most of the intersubunit interactions), which connects the N- and C-terminal domains; a hairpin structure of the N-terminal domain (β4–β5 hairpin, residues 54–73) is also defined in Fig. 1 which directly interacts with the C-domain helices (see below). All subunits of the binary complex structure contain three additional short helices in the N-terminal domain that are not present in E. coli CRP, similar to the apoMtb-CRP structure. To adopt the standard nomenclature the new helices were labeled as helices N1, N2, and N3. The two conserved N-terminal helices are called helices A and B, the long central helix is called helix C, the adjacent helix is D, the two helices in the helix turn helix (HTH) DNA binding region are E and F, and the most C-terminal helix is G (Fig. 1, left and right panels, and supplemental Fig. 1a).

The presence of two independent dimers in the crystallographic asymmetric unit of the Mtb-CRP·cAMP binary complex structure was not anticipated, because the apoMtb-CRP contained only one dimer per asymmetric unit (18). The majority of the CRP structures reported to date either contained one CRP subunit per asymmetric unit (the dimer is formed through crystallographic symmetry) (38) or one CRP dimer per asymmetric unit (39). However, in the present structure, the AB and CD dimers showed high structural similarity with only a few differences as shown in Fig. 1, left panel. The r.m.s.d. in the Cα coordinates between subunit A of the AB dimer and subunit C of the CD dimer is only 0.41 Å for 193 Cα atoms (0.71 Å for subunit A Versus D); the deviation between subunit B of the AB dimer and subunit D of the CD dimer is 0.61 Å for 210 Cα atoms (0.71 Å for B versus C). As shown by the superimposition of the individual subunits in supplemental Fig. 1a, the most significant differences appear at the flexible N-terminal helices N1, A, and B. In fact, within this region, residues 21–26 of subunit A, 22–25 of subunit B, 1–17 of subunit C, and 13–14 of subunit D are completely disordered. Apart from the N-terminal region, the surface exposed N-domain hairpin loop shows some noticeable flexibility in subunits A, C, and D as seen by the slightly different conformations with respect to each other (Fig. 1, top left panel, and supplemental Fig. 1a); residues 60–66 of this region in subunit B are disordered. These differences, however, do not affect the structural integrity of the rest of the subunit. Therefore, the AB and CD dimers are not discussed independently here, and the CD dimer is mostly used for the discussion and comparison with the apoMtb-CRP structure, because the N-domain hairpin loop region of B subunit is disordered.

The cAMP Binding Pocket

Each subunit of the Mtb-CRP·cAMP binary complex structure contained one cAMP molecule, in an anti conformation (Figs. 1 (bottom panel) and 2). The anti conformation of the cAMP in Mtb-CRP is consistent with the previously published E. coli CRP·cAMP crystal structures (39). The cAMP binding pocket of Mtb-CRP is located in the N-terminal domain interacting with the short helix N3 region, the N-domain hairpin loop (β4–β5), and the long helix C (Fig. 2). The conformations of the binding pocket residues and cAMP in all subunits are similar (supplemental Fig. 1b); therefore only subunit D is used for distance calculations unless specified. In each subunit, the hydrophobic interactions from the residues of the N-domain region play a dominant role in binding cAMP to Mtb-CRP; residues Phe-38, Phe-78, Leu-81, Gly-79, and Ile-57 pack against the ribose ring and Met-72, Leu-69, and Thr-70 interact with the adenine ring of the cAMP (all residues lie within at least 4.0 Å from cAMP) as seen in Fig. 2. In addition, Arg-130 of helix C runs parallel to the plane of the adenine ring of cAMP (3.5 Å from the C2 atom of the ligand), forming additional hydrophobic interactions. This conformation of Arg-130 appears to be critical in placing the side chains of Met-77, Phe-78, Met-72, and Glu-80 residues of the N3 helix region of the cAMP binding pocket. Both the NH1 and NH2 atoms of Arg-130 form hydrogen bonds with the carbonyl oxygen atom of Met-77 (at 3.0 and 2.8 Å distances), and the NH1 atom of Arg-130 forms hydrogen bonds with both Oϵ1 and Oϵ2 atoms of Glu-80 (at 2.8 and 3.2 Å distances). In addition, the Cζ atom of Arg-130 is at 3.7 Å from the Met-72 CE atom. As a result, the Oϵ1 and Oϵ2 atoms of Glu-80 face toward the sugar of cAMP at hydrogen bonding distances (2.6 and 3.3 Å). The O2′ atom of cAMP also makes a hydrogen bond with the backbone amide nitrogen atom of the nearby residue Gly-79 at a distance of 2.9 Å, whereas the O3′ hydrogen bonds with the backbone nitrogen of Leu-81. The carbonyl oxygen atom of Gly-79 hydrogen bonds with a second arginine residue (Arg-89) of the cAMP binding pocket at a distance of 2.8 Å, and Arg-89 makes a hydrogen bond with the axial phosphate oxygen atom of cAMP hydrogen (at 3.0 Å distance). The axial phosphate oxygen atom also hydrogen bonds with Ser-82 Oγ (2.6 Å) and the backbone nitrogen atom of Ser-82 of helix N3 (3.0 Å), whereas the equatorial phosphate oxygen atom of cAMP hydrogen bonds with the backbone nitrogen atom of Thr-90 (2.7 Å). Taken together, Arg-130, the residues of the N3 helix region (Met-77, Phe-78, Gly-79, Glu-80, Leu-81, and Ser-82), as well as the Arg-89 and Thr-90 residues enclose the sugar moiety of cAMP through hydrophobic and hydrogen bond interactions. Due to this packing, the N3 helix and the adjacent loop region stabilize the sugar moiety while the residues from the adjacent hairpin (the N-domain hairpin connected to the short helix N3 loop), Leu-69, Thr-70, and Met-72 of β5, and Ile-57 of β4, interact with the base of cAMP.

FIGURE 2.

The cAMP binding pocket of Mtb-CRP. Left, stereo view of the key residues of the cAMP binding pocket of both the subunits of the Mtb-CRP dimer is shown (top view of the CD dimer); residues of subunit D are primed. The relative location of the cAMP binding pocket at the interface of the N-domain hairpin loop, the long helix C, and the short N-domain helix N3 is also depicted. The key hydrogen-bonding interactions are shown as green solid lines. One subunit is colored in cyan ribbons, and the other is blue. Right, close-up view of the cAMP binding pocket of subunit C.

Two threonine residues interact with cAMP: one near the N6 atom of the base of cAMP (Thr-134) and the other near the equatorial phosphate oxygen atom (Thr-90). The hydroxyl groups of Thr-134 of helix C hydrogen bond with the N6-amino group of cAMP (at 2.9 Å). Interestingly, two side-chain conformations are equally possible for Thr-90; it can either place its Oγ atom close to the equatorial phosphate atom of cAMP to form a hydrogen bond or its Cγ atom can form hydrophobic interactions with C3′ (3.5 Å), C5′ (3.5 Å), and C8 (3.2 Å) of cAMP (supplemental Fig. 1b, superposition of the cAMP binding pockets of all subunits). In the hydrogen bonding networks, the Oγ of Thr-90 hydrogen bonds with the phosphate oxygen atom of cAMP as well as Ser-91 Oγ (3.1 Å). As a result, Ser-91 Oγ is forced away from O4′ of cAMP (4 Å), and the Cβ of Ser-91 compensates through hydrophobic interactions with C5′ (3.7 Å). In the case of the hydrophobic interaction of Thr-90, the Cγ atom is found close to the phosphate oxygen atom (3.2 Å), forming hydrophobic interactions with C3′ (3.5 Å), C5′ (3.5 Å), and C8 (3.2 Å) of cAMP; as a result, the side chain of Ser-91 is oriented toward O4′ to form a new hydrogen bond at 3.1 Å. It is possible that this interplay between Thr-90 and Ser-91 plays a role in adopting minor conformational changes of cAMP. All other protomer residues, with the exception of Thr-90 and Ser-91, adopt similar conformations across all the subunits of the asymmetric unit and form similar interactions with the cAMP binding pocket.

In addition to these intrasubunit interactions, the residue Asn-135′ of helix C′ of the neighboring subunit forms two direct hydrogen bonds with the N6 and N7 atoms of the base of cAMP (Nδ2 of Asn-135 is 2.9 Å from the N7 atom of cAMP, and Oδ1 of Asn-135 is 3.0 Å from the N6 atom of cAMP). The residue Leu-131′ of the helix C′ of the neighboring subunit also makes hydrophobic interactions with the base of cAMP at a distance of 4.3 Å.

Asymmetry/Symmetry in the Mtb-CRP Dimer: Comparison of Apo- and cAMP-bound Structures

The apoMtb-CRP crystal structure demonstrates a pronounced asymmetry between the subunits within the dimer (18). This suggests that one subunit of the apo protein is in an off state while the other adopts the typical on state. More importantly, based on the solution structure E. coli CRP, it has also been proposed that the allosteric switch from being apo- to cAMP-bound follows a mechanism in which the hinge region of the long helix C in the apo structure unwinds in both the subunits (near the hinge region close to N6 atom of the adenine moiety of cAMP), giving rise to an off state conformation for the HTH helices of both the subunits. To evaluate the proposed on- and off state mechanism of the Mtb-CRP structure further, we carried out a detailed comparison between the subunits of apoMtb-CRP and Mtb-CRP·cAMP dimers.

The Mtb-CRP·cAMP binary complex structure reveals that there is no obvious asymmetry within the dimer, in contrast to the apoMtb-CRP structure. This is illustrated in supplemental Fig. 1a, right panel, showing the superimposition of the individual subunits. Subunit A of the present structure superimposes on subunit B with an r.m.s.d. of 0.66 Å (193 Cα atoms compared), on subunit C of the CD dimer with an r.m.s.d. of 0.44 Å, and on subunit D with 0.76 Å. Similarly, subunit A of Mtb-CRP·cAMP superposes on the subunit A of apoMtb-CRP structure with an r.m.s.d. of only 0.70 Å (same 193 Cα atoms compared). However, for the same number of Cα atoms, subunit A of cAMP-bound Mtb-CRP showed an r.m.s.d. of 3.60 Å with subunit B of the apoCRP structure, indicating that all the subunits, including subunit A of the apo structure, belong to one conformation, with the exception of subunit B of the apo structure. We investigated three types of superimpositions for further analysis: apo dimer on cAMP-bound dimers (Fig. 3, top left panel), apo subunits on cAMP-bound subunits (Fig. 3, top middle panel), and individual domains of subunit B of apo on subunit A of the cAMP-bound structure (Fig. 3, top right panel). The superimpositions further confirmed that all subunits belong to one structural conformation with the exception of apo subunit B. The apoCRP subunit B, which stands out among all the subunits, displays several conformational differences in the C-terminal domain of the protomer, particularly in the DNA binding region (Fig. 3), assuming the off state structure of Mtb-CRP (Fig. 3, bottom panel). The minor conformational changes observed in the HTH region for the rest of the subunits are insignificant compared with subunit B of the apo structure. This is consistent with the previously reported cAMP-bound (39) or cAMP·DNA-bound structures (40), which typically show flexibility in this region (supplemental Fig. 2).

FIGURE 3.

Comparison of apo- and cAMP-bound structures of Mtb-CRP. Top: ribbon representation of the superposition of the apo dimer on cAMP-bound Mtb-CRP (left), apo subunits on cAMP-bound subunits (middle), and individual domains of subunit B of apo on subunit A on the cAMP-bound structure (right). Color code: blue and cyan, Mtb-CRP·cAMP (CD dimer); yellow and golden yellow, Mtb-CRP·cAMP (AB dimer); red, subunit B of apoCRP; and magenta, subunit A of the apoCRP structure. Bottom: the side-by-side comparison of on state and off state subunits of Mtb. Left, the closed, on state structure of Mtb-CRP; the blue ribbon is a representative subunit of cAMP-bound CRP, and magenta is subunit A of apoCRP. Right, the open, off state structure of Mtb-CRP; subunit B of the apoMtb-CRP crystal structure is shown.

The native state conformation of E. coli apoCRP in solution may be defined by both the subunits in a symmetric, open conformation in contrast to the apoMtb crystal structure. Is it possible that the asymmetry in the apoMtb crystal structure is caused by a crystallization condition? It is still not clear how apoMtb-CRP can precisely coordinate asymmetry between the subunits, or whether the asymmetry hypothesis based on the apoCRP is valid in solution. It has been argued based on the apoMtb structure (18) that the possibility of steric hindrance in the hinge region (between the long helices C, D, and F), particularly around Arg-149, leads to the asymmetry between the subunits. Although the second subunit should undergo some local conformational changes with respect to the conformation of its closely interacting partner subunit, it is unlikely for the off state subunit to influence the overall fold of the neighboring subunit. Perhaps, the off state conformation of Arg-149 can be accommodated either through the flexible nature of this amino acid or through some conformational changes around the hinge region. In fact, the solution structure of apo E. coli CRP (22) suggest that C helix unwinds, forming a coil conformation near this hinge region, symmetrically accommodating two off state subunits within the dimer. However, in the apoMtb-CRP structure, the long helix C does not undergo a helix coil transition in any of the subunits (Figs. 1 and 3). It is possible that, in the absence of cAMP, CRP can adopt two low energy states (one on state and one off state), and due to the crystal packing one subunit adopted the on state conformation while the other, due to lack of cAMP, adopted the off state. Perhaps, due to this asymmetry and additional interactions, helix C does not undergo a conformational change to become coiled in the crystal packing environment. If this is the case, one cAMP-bound on state subunit might trigger conformational changes in the neighboring subunit, even in solution. In addition, the apoCRP structure from Thermus thermophilus (PDB ID: 2ZCW) shows higher similarity to subunit B of the apoMtb structure (r.m.s.d. 2.40 Å versus 3.70 Å to subunit A) implying that it has crystallized in the closed form. As with the Mtb structure, in solution the T. thermophilus CRP may be sampling various low energy conformations and is finally locked into a closed conformation by crystal contacts. Following this model, the CRP apo structure would be very flexible and, perhaps, disordered in many regions because it samples multiple, low energy conformations in solution, and is locked into a rigid and less flexible closed conformation upon CRP binding. Thus, like the E. coli CRP, Mtb-CRP might also form a symmetric off state dimer.

The Binding of cAMP Switches Mtb-CRP from Off State to On State: Comparison between the Apo Mtb-CRP and Mtb-CRP·cAMP Structures

To investigate how cAMP binding to the N-terminal domain influences the DNA-binding HTH motifs of the C-terminal domain, both the subunits of the apo structure were superimposed on subunit D of cAMP-bound structure using the long C helix as common point of reference (Fig. 4). Helix C was selected as a common point of reference to represent the relative positioning of the subunits within the dimer and to minimize the superimposition bias toward the structural similarity of the N-terminal domains. As shown in Figs. 3 and 4, subunit A of the apo structure showed only minor differences compared with the cAMP-bound structure (colored magenta and blue in Figs. 3 and 4), in contrast to subunit B of the apo structure (shown in pale red). Taken together, it is very clear that the cAMP-bound structure curtails the asymmetry observed within the dimer of the apoCRP structure and pushes the HTH regions of both the subunits into an almost identical active state, preparing the dimer for DNA binding.

FIGURE 4.

The binding of cAMP switches Mtb-CRP from off state to on state. The superposition of subunit A (magenta ribbon) and subunit B (pale red ribbon) of the apoMtb-CRP structure on a representative subunit of cAMP-bound structure using the long C helix as common point of reference. In the off state (pale red ribbon and the residues labeled with red color), the secondary structural elements of the N-terminal along with the N-domain hairpin domain are shifted away from the central C helix and indicated by a black arrow. Upon binding of cAMP, the entire N-terminal domain moves toward the helix C to pack around the cAMP to augment the closed structure of CRP. The key residues involved in this transition are shown by the stick model along with cAMP (see panel, for close-up view). The closed positioning of N-domain region in the cAMP-bound structure allosterically augments the active conformation of the he C-domain helices, particularly the HTH motifs, indicated by inward arrows. The close proximity of the N-domain hairpin and HTH motif helices is indicated by a dotted arrow.

A long standing question has been whether or not cAMP binding alone can signal the switch from off to on state. Although this cannot be answered fully without the CRP·cAMP·DNA ternary complex structure from the same organism, it is clear from the previous structures of E. coli CRP·cAMP and E. coli CRP·cAMP·DNA that the subunits undergo only small adjustments upon DNA binding after cAMP is bound (see supplemental Figs. 2 and 3 for the overlay). The relative position and orientation of helix F of the HTH motif within the dimer seems to be critical for DNA binding to CRP. An overlay of Mtb-CRP·cAMP on top of E. coli CRP·cAMP·DNA suggests that helix F of Mtb-CRP of both subunits is symmetrically placed in an optimum position and orientation, at a distance of ∼38 Å from each other and roughly aligned with the E. coli CRP HTH motif, suggesting that it can fit into the major grooves of the DNA substrate (supplemental Fig. 3). Consequently, the side chains of residues Asp-174, Thr-176, Gln-177, Glu-178, Arg-188, Glu-189 (disordered), Asn-192, Lys-193 (disordered), His-200, Glu-207, and Lys-209 of the HTH region are surface-exposed, suggesting they may be involved in the CRP·DNA-binding interactions and therefore may need to undergo only slight conformational adjustments. The subunits of Mtb-CRP·cAMP dimer both adopt the active conformation placing their HTH motifs symmetrically accessible to the DNA substrate. Thus, we propose that cAMP binding alone is sufficient to create an on state-like conformation.

These structural comparisons also enabled us to provide a structural basis for the transition of off state to on state upon binding of cAMP. The off state subunit, subunit B of the apoMtb structure, must undergo several conformational changes upon binding of cAMP (Fig. 4, left and right panels) as described in detailed below. Upon the binding of cAMP, the side chain of Arg-130 residue flips away from the cAMP binding pocket (the Cζ atom moves by 4.8 Å while Cα is in the same position), making room for the adenine moiety of the cAMP. Additionally, the short helix and the Glu-80 residue also shift away from the cAMP binding pocket (the Cα moves by 3.2 Å), making room for the sugar moiety of the cAMP (Fig. 4, right panel). In the cAMP-bound structure, both Arg-130 and Glu-80 are involved in ligand-binding interactions. These rearrangements also allow Ser-82 to move closer to one of the phosphate oxygen atoms of cAMP (the Cα atom moves by 1.8 Å and Oγ moves by 2.1 Å). In the cAMP-bound structure, the entire N-domain along with the hydrophobic pocket residues (Phe-38, Tyr-48, Ile-57, Phe-78, and Ala-93) move toward the central C helix to form tight packing around cAMP; the Cα atom of Phe-38 moves 7.8 Å in the on state compared with its off state, the Cα of Phe-78 moves 6.1 Å, Thr-90 moves 5.1 Å, Ile-57 moves 7.5 Å, and Arg-89 moves 5.5 Å.

Our structural comparisons have also shown that the allosteric off state is defined by a structural relaxation of the secondary structure away from the central C helix. The relative positioning of the N-domain and helix C seems to be the key. This is illustrated by corresponding secondary structural elements, including the β1–β8 strands of the N-terminal domain, which are shifted away from the central C helix (Figs. 3 and 4). More importantly, the N-domain hairpin and the β6-helix-β7 region also shift away from the C-terminal domain, particularly from the smaller helix of the DNA-binding HTH motif (helix E). As a result, all of the C-terminal domain helices, including the DNA binding region, form the off state conformation (Figs. 3 and 4). Upon binding of cAMP, the entire N-terminal domain, including the N-domain hairpin-(N3 helix)-loop-β7 region, pack around the adenine and sugar moieties of cAMP (see above). As a result of these rearrangements, several new inter-domain interactions are established, particularly between the β5 strand of the N-domain hairpin region and helix E of the C-terminal HTH motif (Fig. 4, left panel, and supplemental Fig. 4). The backbone nitrogen atom of Leu-68 of β5 strand forms a hydrogen bond interaction with the carbonyl oxygen atom of the Gln-182 of helix E (at 3.1 Å); the side-chain atoms of Leu-68 are involved in the hydrophobic interactions with the carbon atoms of residues Glu-179, Gln-182, and Leu-183 of helix E (at 3.9 to 4.1 Å distances). The side-chain atoms of Ile-71 of the β5 strand are involved in hydrophobic interactions with Leu-183 of helix E (3.7 and 4.3 Å distances), with Leu-175 at 3.8 Å distance, Leu-157 of helix D at 4.5 Å distance, and with Phe-161 at 4.1 Å distance. In addition, the side-chain nitrogen atom of Lys-56 of β4 strand forms a salt bridge with the Oϵ2 atom of Glu-179 (at 2.6 Å), and Lys-54 hydrogen bonds with Asp-174 (at 3.0 Å); Ile-96 of the β7 strand of the N-terminal domain is involved in hydrophobic interactions with Leu-175 of helix E (at 3.6 Å). These interactions together augment the stabilization of the C-domain in the active state close to the central helix C (Fig. 4 and supplemental Fig. 4).

The transition from the off to on state begins with ligand binding. Subsequently, this ligand binding pocket structural transition augments several inter-domain interactions, particularly between the N-domain hairpin region and the C-terminal DNA-binding HTH motif region. At this stage, we see the critical role of the N-domain hairpin: it transmits the allosteric signal from cAMP binding to positioning the C-terminal DNA-binding HTH motif region into the on state. Interestingly, the solution studies of apo E. coli CRP (22) indicated that the on state may be augmented through the binding interaction near the N6 atom of the cAMP ligand, primarily through the helix-coil transition of the helix C of the hinge region. In contrast, for Mtb-CRP, as detailed above, the dominant network of non-bonding interactions with the cAMP atoms, N-terminal hydrophobic clustering, N-domain hairpin, and HTH helix region strongly suggest that the on state is primarily stabilized through the cAMP-N-domain hairpin-HTH interactions, which, in turn, stabilize the hinge region. This is further corroborated by the fact that the hinge region of helix C of Mtb-CRP is involved in only three hydrogen-bonding interactions with the cAMP (through Thr-134 and Asn-135′) compared with numerous N-domain interactions. Perhaps both events happen simultaneously, affecting each other.

The close proximity of the N-domain hairpin and helix C also suggests a possible connection between the hinge region and the N-domain (Figs. 1 and 3). In the Mtb-CRP·cAMP-bound structure, the loop region of the N-domain hairpin showed marked flexibility, adopting slightly different conformations near the C helix. In one conformation, Pro-62, located at the tip of this loop, is stacked close to Phe-143′ of the C helix of the neighboring subunit (at a 3.8-Å distance). In another subunit, this loop moves away from the C helix of the neighboring subunit placing Pro-62′ at a 6.3-Å distance from Phe-143; however, Arg-65′ compensates through hydrophobic interaction with Phe-143. These conformational changes do not seem to affect helix C or the hinge region. This is further confirmed with the loop region of the N-domain hairpin, which is disordered in subunit B with no significant changes in the nearby helix C. The stabilization of helix C, despite the ability of the nearby loop region to adopt multiple conformations, is made possible by the non-binding interactions between helix C and helix C′ at the dimer interface (Fig. 1, left and right panels). In particular, the Leu-138, Leu-141, Leu-138′, and Leu-141′ side chains interact with each other and stabilize the helix C and C′ at the hinge region. Together, these interactions provide a structural basis for the ability of ligand binding to invoke the allosteric maneuvering of sub-domains through the positioning of N-domain and in turn N-domain hairpin.

Structural Basis of N⁶-cAMP Binding to Mtb-CRP

The concept of helix-to-coil transition (near the N6 atom of cAMP) proposed as the mechanism as seen through the solution structure of apo E. coli CRP does not seem to be the key for Mtb structure activation. Moreover, some of the key residues involved in the helix-to-coil to off state transition of E. coli are very different in Mtb: Trp-85 is Ser-92 in Mtb, Gln-80 is Gly-87 in Mtb, and Gln-125 is Arg-132 in Mtb. Despite all these differences, even in E. coli, the involvement of numerous interactions of the N-domain upon binding of cAMP may be the key factor for the active positioning of the HTH motif. It is not clear, however, why mutations in the hinge region of E. coli lead to inactivation (41) and why the syn conformation of cAMP and its steric hindrance of helix C leads to inactivation (6). To investigate this further, we tested the binding ability of Mtb-CRP to DNA in the presence of N⁶-cAMP with a bulky methylphenylethyl extension from its N6 position using a previously reported biochemical analysis (30). Binding of Mtb-CRP with the 28-bp putative SerC upstream sequence was demonstrated in the absence and presence of cAMP or N⁶-cAMP. Migration of the DNA fragment was retarded without cAMP (lane 3 of Fig. 5), whereas the affinity (relative abundance) of the protein·DNA complex increased with 100 μm and 150 μm of cAMP. However, in the presence of 75 μm or 150 μm (lanes 4 and 5) N⁶-cAMP, a decreased migration of complex as compared with cAMP was observed. The specificity of binding of CRP is further evident from the disappearance of the complex in lane 2 using a 100-fold excess of unlabeled probe (lane 2). Based on the DNA gel shift assay, the binding ability of Mtb-CRP to DNA is significantly inhibited upon the binding of N⁶-cAMP (Fig. 5). It is possible that the loss of activity observed in lanes 4 and 5 might be due to the reduced binding of N⁶-cAMP to Mtb-CRP under experimental conditions, mimicking the apo structure. However, it has been demonstrated previously that N⁶-cAMP competes with cAMP and binds to E. coli CRP to induce a conformational change, but does not promote transcription (30). More importantly, the Mtb-CRP·N⁶-cAMP crystal structure presented here provides direct evidence for the inhibitor binding (see below).

FIGURE 5.

EMSA showing binding of Mtb-CRP to SerC DNA template. Gel mobility-shift experiment performed with a 28-bp (³²P end labeled) double-stranded DNA oligonucleotide containing CRP-binding site of the upstream region of Mtb-SerC. The radiolabeled fragments (2 nm) were mixed with Mtb-CRP (50 nm), and then resolved on a 5% polyacrylamide gel. Lane 1, DNA probe only; lane 2, 100-fold excess of homologous unlabeled DNA probe; lane 3, CRP without cAMP; lane 4, 75 μm N⁶-cAMP; lane 5, 150 μm N⁶-cAMP; lane 6, 75 μm cAMP; and lane 7, 150 μm cAMP. The upper band shows the CRP·DNA complex and the lower band refers to free DNA.

To explore the mechanism of action of this inhibition, we solved the x-ray crystal structure of Mtb-CRP complexed with N⁶-cAMP. The crystal parameters of Mtb-CRP·N⁶-cAMP binary complex structure are not isomorphous compared with the apo- and cAMP-bound structure (Table 1 and supplemental Fig. 5). The structure was solved by molecular replacement using the cAMP-bound subunit as a search model. The asymmetric unit contained two subunits; subunit A showed a very close resemblance to the CRP bound to cAMP structure while subunit B showed significant differences in the relative positioning of the C-terminal domain with respect to the helix C within the dimer. In other words, the superposition of individual domains or subunits of both molecule A and B of the N⁶-cAMP-bound CRP structures mostly resembles that of cAMP-bound CRP (Fig. 6). There was an exception, however, as demonstrated in the superposition of subunit B, which showed either the helix C or the rest of the structure tilted ∼7° from the CRP·cAMP structure (Fig. 6 and supplemental Fig. 6). The r.m.s.d. of subunit A of N⁶-cAMP-bound CRP on CRP·cAMP is 0.85 Å and demonstrates the close resemblance between the two structures. In the case of subunit B, the r.m.s.d. is 0.95 Å for all the residues, excluding helix C, indicating that this subunit is similar with the exception of the relative positioning of helix C and the rest of the structure.

FIGURE 6.

Left, the superposition of the N⁶-cAMP-bound Mtb-CRP dimer (red ribbon) on top of the cAMP-bound Mtb-CRP dimer (cyan and blue ribbons); the N⁶-cAMP molecules are shown as a CPK model. The DNA template (green ribbon) is modeled by superposing the E. coli CRP·cAMP·DNA ternary complex structure (PDB ID: 1ZRF) on top of the cAMP-bound Mtb-CRP structure; the CRP part of the E. coli structure is not shown here for clarity. The DNA binding region of the subunit B of the Mtb-CRP·N⁶-cAMP structure showed significant differences compare with the rest of the structure and is highlighted. Right, close-up view of the HTH motif, near the major groove of the modeled DNA template; the N⁶-cAMP is shown as a stick model, and its N6 extension is colored in yellow.

Each subunit of the complex contained one N⁶-cAMP molecule, in anti (for the cAMP)-R (for the N6 extension) conformation for subunit A and anit-S conformation for subunit B (Fig. 7 and supplemental Fig. 5). The overall fold of the complex structure, the anti conformation of the N⁶-cAMP, and majority of the binding interactions around the sugar and base region of cAMP are consistent with the cAMP-bound CRP structure (Fig. 7), however, with the exception of the N6 extension. In both subunits, the methylphenylethyl extension protrudes into helix C of subunit A and C′ of subunit B, adopting a different conformation in one subunit compared with the other; S in subunit B and R in subunit A with respect to the chiral center near the N6 atom (Fig. 7). In subunit A, the cAMP analog places the phenyl extension somewhat parallel to the plane of the N6 atom in an R conformation, between Leu-69 of the β5 strand of the hairpin and Asn-137, Leu-138, and Leu-141 of helix C. Thus, the N6 extension in subunit A is mainly stabilized by hydrophobic interactions; the Cβ atom of Asn-137 is 3.3 Å from the phenyl ring, CD1 of Leu-141 is 3.3 Å, CD1 of Leu-138 is 4.1 Å, and CD2 of Leu-69 is 3.9 Å. To accommodate the hydrophobic extension of the adenine moiety, particularly the methyl group, the nearby residue Asn-135′ of helix C′ of subunit B and Thr-134 of subunit A are forced away from the adenine moiety of N⁶-cAMP. As a result, the N6 atom loses two hydrogen bonds as compared with the cAMP-bound structure (N6 is now at 3.3 Å from Oδ1 of Asn-135′ and at 3.8 Å from Oγ1 of Thr-134), whereas the N7 atom of the adenine moiety retains its hydrogen bond with Asn-135′ at 2.7 Å distance. Interestingly, the position of the phenyl ring of the N6 extension lies very close to the side chain of Leu-138′ of helix C of the neighboring subunit (∼1 Å based on CRP·cAMP structure), as shown in Fig. 7, right panel. Because of this steric hindrance, the Cα atom of Leu-138′ moves away from the phenyl moiety by ∼3.8 Å, and its side-chain atoms are disordered. As a result, helix C′ of subunit B partially unwinds. The loop region (residues 61–65) of the N-domain hairpin of subunit A is also disordered.

FIGURE 7.

Left, stereo view of the key residues of the N⁶-cAMP binding pocket of both the subunits of Mtb-CRP dimer is shown (view of the DNA substrate). The cAMP part of the ligand is black, and the extension from its N6 atom is yellow. The relative location of the N⁶-cAMP binding pocket at the interface of the N-domain hairpin loop, the long helix C, and the short N-domain helix N3 is also depicted. The key hydrogen-bonding interactions are shown as green solid lines, and the distances are shown in broken green lines. One subunit is colored in red ribbon and the other orange red. Right, close-up view of the relative position and conformation of the N⁶-cAMP binding pockets, within the Mtb-CRP dimer. The long C helices of the cAMP-bound Mtb-CRP structure are also shown here as reference, in cyan and blue ribbons.

In subunit B, both the N6 and N7 atoms of the adenine moiety retain their hydrogen bonds with Asn-135 of subunit A as compared with the cAMP structure, which accommodates the S conformation of N6 extension. This allows the N6 extension of subunit B to be closer to helix C′ in the S conformation, as compared with the N⁶-cAMP of subunit A in the R conformation and helix C. In this position, the phenyl moiety overlaps with the original position of Asn-137′ of helix C′, thus imposing further unwinding of helix C′. Therefore, to accommodate both of the N⁶-cAMP molecules in different isomeric conformation, one of the helices at the dimer interface (helix C′ of subunit B) has to undergo unwinding. As a consequence, the side chains of residues 137–141 of helix C′ and residues 142–144 become completely disordered. Surprisingly, the disorder and unwinding of helix C′ does not create any local conformational differences on individual secondary structural elements of the C-terminal domains. Instead, the entire structural arrangement of subunit B, including the N-terminal domain and C-terminal domain, is tilted away from its own helix C, toward the minor groove of a modeled DNA substrate (Fig. 6). Therefore, most of the cAMP region of the N⁶-cAMP makes similar interactions as compared with the typical cAMP binding to stabilize the structural integrity and relative positioning of the N- and C-terminal domains within subunit B. However, to accommodate the extension, the entire N- and C-domains tilt away from the central helix; using the base of the helix C as a common point of reference, the tilt is estimated to be ∼7°. Thus, the base of the tilt is affected less and the DNA binding region (the HTH motif region of the C-terminal domain) is affected more. Taken together, this rearrangement affects the positioning of most of the helices and strands of the C-domain of subunit B while also shifting the HTH motif in an unfavorable position to accept the major groove of the DNA (Fig. 6, based on the overlay of DNA-bound E. coli CRP); the maximum shift is observed at the tip of the helix F where the Cα of Arg-201 is shifted 9.3 Å as compared with the cAMP-bound CRP structure (supplemental Fig. 6). Although the N⁶-cAMP disrupts the dimer interface and forces helix unwinding in subunit B, subunit A of the ordered helix C of the N⁶-cAMP structure resembles that of Mtb-CRP·cAMP.

Once again, the crystal structure is pointing toward significant asymmetry within the dimer, induced by the binding of N⁶-cAMP. The role of asymmetry in the inhibition of Mtb-CRP, based on the snapshot of N⁶-cAMP in the crystal packing environment, however, needs to be explored further using a solution structure. What is clear, though, is that a single molecule of N⁶-cAMP alone is not sufficient for the inactivation of CRP, single inhibitor binding can most likely be accommodated by minor changes in the hinge region. It is likely that the binding of N⁶-cAMP in one subunit has a distinct impact on the inhibitor binding in the second subunit and subsequent inactivation through the HTH rearrangements.

Additionally, the N⁶-cAMP-bound CRP structure is almost identical to the cAMP-bound on state, excluding helix C. However, the off state, due to the N6 extension, is achieved through the rearrangement and relative positioning between helix C and the rest of the structure, in subunit B. This off state mechanism is strikingly different from the off state of apoCRP in which the entire N-terminal, N-domain hairpin underwent dramatic displacement leading to conformational changes in the individual helices of the C-domain. It is possible that the previously suggested steric hindrance of syn cAMP might also follow a similar mechanism as that of N⁶-cAMP, integrating the N- and C-domain, while still positioning the HTH in an inactive state. Perhaps, in solution this may lead to more local conformational changes. The solution structure of apoMtb-CRP and crystal structure of Mtb-CRP·DNA complex can further expand the understanding of the precise regulation of CRP.

In the absence of the N6 extension, i.e. in a physiological state, cAMP binds to the N-domain, which stabilizes the C-domain HTH motifs of Mtb-CRP, most likely through non-bonding interactions between cAMP, the N-domain hairpin, and C-domain helices, which are subsequently followed by the stabilization of the helix C region.