The Structure of a pH-Sensing Mycobacterial Adenylyl Cyclase Holoenzyme

See allHide authors and affiliations

Science  13 May 2005:
Vol. 308, Issue 5724, pp. 1020-1023
DOI: 10.1126/science.1107642


Class III adenylyl cyclases contain catalytic and regulatory domains, yet structural insight into their interactions is missing. We show that the mycobacterial adenylyl cyclase Rv1264 is rendered a pH sensor by its N-terminal domain. In the structure of the inhibited state, catalytic and regulatory domains share a large interface involving catalytic residues. In the structure of the active state, the two catalytic domains rotate by 55° to form two catalytic sites at their interface. Two α helices serve as molecular switches. Mutagenesis is consistent with a regulatory role of the structural transition, and we suggest that the transition is regulated by pH.

Adenylyl cyclases (ACs) synthesize the universal second messenger 3′,5′-cyclic adenosine monophosphate (cAMP) (1). Most ACs belong to class III, such as all mammalian and many bacterial enzymes (2), and are multidomain proteins (2, 3). In the genome of the bacterium Mycobacterium tuberculosis (4), 15 putative class III ACs (5) with eight different domain compositions have been identified. For comparison, the similarly sized genome of Escherichia coli contains a single AC gene, and even in the human genome only 10 AC genes have been identified (6, 7). This suggests that mycobacteria can respond to changing extra- and intracellular conditions by cAMP formation.

The mycobacterial AC Rv1264 is auto-inhibited by its N-terminal domain (8). A knockout of the single Streptomyces AC, which has an identical domain composition to Rv1264, abolishes the bacterial response to an acidic milieu that affects differentiation processes (9). Because M. tuberculosis must counteract acidification of phagolysosomes during host invasion for intracellular survival (10, 11), we examined the pH sensitivity of Rv1264 (Fig. 1A) (12). At pH 8, AC activity was 3 nmol of cAMP·mg-1·min-1 at 0.5 mM adenosine triphosphate (ATP) with a maximal velocity (Vmax) of 34 nmol of cAMP·mg-1·min-1 and a substrate affinity (SC50) of 1.5 mM ATP. At pH 6, AC activity increased almost 40-fold to 115 nmol and Vmax increased 12-fold to 420 nmol of cAMP·mg-1·min-1. The substrate affinity increased slightly to 0.8 mM ATP. The Hill coefficient of 1.9 was unaffected. In contrast, the isolated catalytic domain (Rv1264211-397) displayed uniformly high AC activity between pH 5.5 and 8 (Fig. 1A). Thus, in Rv1264, pH sensitivity is mediated by a distinct regulatory domain, and the activation by far exceeds the usual pH dependence of an enzyme. Biochemically, Rv1264 qualifies as a pH-sensing AC and is a likely candidate for mycobacterial pH sensing.

Fig. 1.

The pH dependence of the AC activity of the Rv1264 wild type and mutants. AC activities of purified recombinant enzymes were measured from pH 4.8 to 8.0, with 0.5 mM ATP as a substrate. Standard deviation (SD) is given by error bars, if they exceed the size of the symbols. The symbol size itself corresponds to an SD of 10%. (A) Rv1264 catalytic domain (Rv1264211-397) (◼) and holoenzyme (⚫). To facilitate comparisons, these curves are included as dotted lines in the other panels. (B) Rv1264 M193P/M194P (◼). (C) Rv1264 R309A (⚫) and E195A (◼). (D) Rv1264 H192A (◼) and H192E (▢).

To understand the molecular mechanism of pH sensing and AC regulation, two crystal forms of Rv1264 were analyzed (12). Anisotropic crystals in a hexagonal space group with a diffraction limit of 3.3 Å were grown from Li2SO4, and the resulting model was designated the active form of Rv1264 (Fig. 2A); the 2.3 Å resolution structure obtained from monoclinic crystals grown from polyethylene glycol was designated the inhibited form (Fig. 2B and table S1). The designation was based on the structural and ensuing mutagenic analysis (13). Both structures have three discrete segments, a rigid N-terminal domain (amino acids 14 to 191), a flexible linker region (amino acids 192 to 213) (Fig. 2, red), and the C-terminal catalytic domain (amino acids 214 to 377).

Fig. 2.

Overall structure of Rv1264 in the active and inhibited states. (A) The active dimer, green, and (B) the inhibited dimer, blue. Monomers are distinguished by dark and light colors. The regulatory domains remain essentially unchanged upon enzyme activation, but the interface with the catalytic domains differs substantially. Secondary structure elements are labeled in the ribbon diagrams (for one monomer); C- and N termini are also indicated. Structural switch regions, red, are found in the linker region and in the catalytic dimer. The boxed regions are shown in detail in Fig. 4. Ribbon diagrams in all figures were drawn with PyMOL (22).

We determined the structures using molecular replacement techniques with the dimer of N-terminal domains as a search model (determined previously to 1.6 Å resolution) (14). In all three structures, the N-terminal dimer comprises 20 α helices and has a disc-like shape of 80 by 50 by 20 Å3, with an internal contact surface of 5500 Å2. Dimerization involves a crossover of protein chains by threading of the αN10 helix of one monomer through the central coiled coil of the other monomer (helices αN4, αN7, αN8, αN9, and αN10) (Fig. 2, A and B). This results in a domain arrangement for Rv1264 in which the regulatory domain of monomer A contacts the catalytic domain of monomer B and vice versa (Figs. 2 and 3A). Residues 60 to 191 of the regulatory domain are essentially identical in both states of the holoenzyme and can be superimposed with a root mean square deviation (rmsd) of 1.42 Å over 264 Cα positions.

Fig. 3.

Active site formation of Rv1264. (A) Schematic diagram of domain rearrangements upon activation. Regulatory and catalytic domains are labeled N and C, respectively. Numbers in Å2 refer to buried surface areas between domains. (B) Superposition of the active sites of Rv1264 (active conformation, green with black labels) and a mammalian AC (19) (1CJK, yellow with gray labels). ATP was modeled into Rv1264 by superposition with ATPαS-RP of 1CJK. The side chains of seven key residues, ATP, and sulfate are shown in stick representation with oxygen, red; nitrogen, blue; phosphate, mauve; sulfur, orange; and Mg2+ and Mn2+, light blue. (A) and (B) after residue names refer to the two monomers. (C) Surface representation of the catalytic domains of Rv1264 in the inhibited (blue) and active (green) states. Activesite residues for one active site are shown in stick representation. For clarity, residues from the second of the two identical active sites are shown as lines. (D) Detailed view of the α1-switch region (red) from a superposition of the catalytic domains of the active (green) and inhibited (blue) states of Rv1264 (rmsd of 1.66 Å over 157 out of 163 Cα atoms).

The catalytic domains in the active state (Fig. 2A) are arranged head-to-tail in the wreath-like quaternary structure known from the C2 homodimer of rat AC type II (15) (fig. S1), as well as from the G protein–activated heterodimer of canine AC type V C1 and rat AC type II C2 catalytic domains (16). Although sequence identities between the Rv1264 and mammalian catalytic domains were low, with 22% to canine C1 and 25% to rat C2, the catalytic dimer of Rv1264 superimposes well with the mammalian heterodimer (rmsd of 1.76 Å over 256 out of 324 Cα positions). It is more compact than the mammalian catalytic domain because of several slightly shortened loops (fig. S2). A 7–amino acid gap obliterating the so-called dimerization arm (2, 8, 16) (fig. S2) is the major reason for a reduction of the interface from 3800 Å2 in the mammalian heterodimer to 1900 Å2 in the Rv1264 active-state homodimer. The known catalytic residues of class III ACs (2, 17, 18) in Rv1264 (Asp222, Lys261, Asp265, Arg298, Asp312, Asn319, and Arg323) (Fig. 3, B and C, and fig. S2) are in almost identical positions to those in the mammalian heterodimer (19). A superposition of these residues of one active site gave an rmsd of 0.69 Å over the Cα positions and 1.17 Å over all atoms (Fig. 3B). The phosphate-coordinating residues Arg298 and Arg323 in Rv1264 bind to a sulfate ion located in the position of the β phosphate of the substrate ATP, as confirmed by modeling ATP into the active site of Rv1264 by superposition with noncylizable adenosine 5′-(1-thiotriphosphate) (ATPαS-Rp) from the mammalian heterodimeric structure (19). Taken together, these data demonstrate that the Rv1264 structure shown in Fig. 2A represents the active state of the enzyme.

The inhibited state of Rv1264 is characterized by a disassembled active site (Fig. 3C). The transition from the active to the inhibited state is accompanied by a marked change in tertiary structure (Fig. 2). Each of the two catalytic domains is rotated by 55° and translated by 6 Å with respect to the fixed regulatory domains (Fig. 3A and movie S1). The catalytic residues move by up to 25 Å during this transition (Fig. 3C). In the inhibited state, residues from a single active site were ∼40 Å apart. Additionally, the interface between the two catalytic domains was reduced from 1900 Å2 in the active state to 930 Å2 in the inhibited state, whereas that between regulatory and catalytic domains was increased from 1500 Å2 in the active state to 2800 Å2 in the inhibited state.

Two regions, called switch elements, undergo major structural changes (Fig. 2, red). The α1-switch helix (residues 226 to 231), located within the catalytic domain (Figs. 2A and 3D), is also present in the mammalian structures and contributes to binding of the β/γ-phosphates (18). There, the α1-switch helix and the adjacent loop to α2 form the binding site for the heterotrimeric G protein subunits Gsα and Giα that regulate the activity of mammalian ACs (15). In the active state of Rv1264, the α1-switch is in a helical conformation as in the mammalian enzymes. However, in the inhibited state, it changes into a random coil with the Cα atoms moving by up to 11 Å (Fig. 3D).

The second switch element, the αN10 switch, comprises residues 192 to 206 in the linker region and undergoes the most noticeable changes between the two states (Fig. 2). In the active state, it assumes a random coil conformation with a short helical segment (residues 202 to 206) (Fig. 2A) loosely connecting regulatory and catalytic domains. High B factors in this region are consistent with a high mobility that allows the catalytic domains to assemble both active sites. In the inhibited state, this segment forms an α helix with low B values, extending the αN10-helix of the regulatory domain by 24 Å or four turns (Fig. 2B) to keep the catalytic domains apart. The newly formed helical turns participate in the interface between regulatory and catalytic domains. To test whether the integrity of the helical segment is a prerequisite of pH regulation, two consecutive prolines were introduced at the beginning of the αN10-switch. Because of their restricted Φ and Ψ-values, this was not compatible with an α-helical conformation (20). In the M193P/M194P double mutant, pH regulation was lost, and the activity of the mutant holoenzyme over the pH range of 5.5 to 8 is similar to that of activated wild-type protein at pH 5.5 (Fig. 1B). Thus, the extension of the αN10 switch helix is a determinant of the inhibited state.

Two residues in the αN10-switch helix, identified by mutagenesis, are important for the interaction between the regulatory and catalytic domains (Fig. 4C). One site of interaction comprises Asp62, Tyr66, Glu195, and Arg309 (Fig. 4A). The arginine points into a crevice in the regulatory domain and organizes the other residues through formation of hydrogen bonds. An R309A mutation renders the holoenzyme active and unregulated (Fig. 1C). The same mutation in the isolated catalytic domain (Rv1264211-397R309A) has no effect (21). Additionally, an E195A mutation partially relieves the inhibition, results in a fourfold increase in activity at pH 8.0, and shifts the pH optimum from pH 5.8 to pH 6.5 (Fig. 1C). A further site of interaction at the beginning of the αN10-switch helix forms around His192 (Fig. 4B). The catalytic residues Lys261 and Asp312 are held 14 Å and 21 Å away from their respective positions in the active state by an interaction with His192 (Figs. 3C and 4B). A H192A mutant shows the wild-type phenotype at pH 8, but the slope of activation is shifted by 0.5 pH units toward the acidic pH (Fig. 1D). At acidic pH, His192 is expected to be protonated, and the positive charge may result in electrostatic repulsion of Lys261. Similarly, a H192E mutant has a 10-fold higher activity at pH 8.0 than the wild-type protein, indicating that a negative charge at this position interferes with Asp312 (Fig. 1D). The data indicate that there is no designated, single amino acid residue that acts as a pH receptor, but that a network of hydrogen and ionic bonds in the interface between catalytic and regulatory domains mediate pH sensitivity and responsiveness.

Fig. 4.

Detailed view of two interaction sites of the αN10-switch helix with the catalytic domain in the inhibited state. Water atoms are indicated as red spheres; distances between polar atoms are given in Å. (A) and (B) after residue names refer to the two monomers, distinguished by dark and light blue colors; the αN10-switch helix is shown in red. The positions of these interaction sites within the holoenzyme are indicated in Fig. 2B. (A) Site around Arg309 involving Asp62, Tyr66, and Glu195. (B) Site around His192 involving Asp312 and Lys261. (C) View along the αN10-switch helix showing the interactions of this helix with the catalytic domain.

Because of the structural similarities between the catalytic domains of both AC Rv1264 and mammalian ACs, the α1 and the αN10 switch regions may be prototypical regulatory elements for class III ACs. In the active state of Rv1264, the catalytic domains align even in the absence of bound nucleotide, whereas the regulatory domains effectively impede this alignment in the inhibited state. The comparison with mammalian enzymes shows that, in all class III ACs, activity is tuned by the alignment of the catalytic dimer. The mode of interaction may differ, but the repositioning of catalytic domains organized by a common structural switch may prove to be a general principle as more AC holoenzyme structures become available.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Table S1

References and Notes

Movie S1

References and Notes

View Abstract

Navigate This Article