Interdomain Communication in the Mycobacterium tuberculosis Environmental Phosphatase Rv1364c

Abstract

An “environmental phosphatase” controls bacterial transcriptional responses through alternative sigma factor subunits of RNA polymerase and a partner switching mechanism has been proposed to mediate phosphatase regulation. In many bacteria, the environmental phosphatase and multiple regulators are encoded in separate genes whose products form transient complexes. In contrast, in the Mycobacterium tuberculosis homolog, Rv1364c, the phosphatase is fused to two characteristic regulatory modules with sequence similarities to anti-sigma factor kinases and anti-anti-sigma factor proteins. Here we exploit this fusion to explore interactions between the phosphatase and the regulatory domains. We show quantitatively that the anti-sigma factor kinase domain activates the phosphatase domain, the kinase-phosphatase fusion protein autophosphorylates in Escherichia coli, and phosphorylation is antagonized by the phosphatase activity. Small angle x-ray scattering defines solution structures consistent with the interdomain communication observed biochemically. Taken together, these data indicate that Rv1364c provides a single chain framework to understand the structure, function, and regulation of environmental phosphatases throughout the bacterial kingdom.

One third of the world population is seropositive for Mycobacterium tuberculosis (Mtb).³ Whereas most anti-microbial therapeutics target actively growing bacteria, Mtb can evade current drugs by converting to a persistent, metabolically suppressed state (1). Moreover, Mtb survives in distinct environments in vivo by adjusting transcriptional programs. Little is known about the transcriptional cues that mediate such developmental transitions, and defining their underlying mechanisms remains the focus of much mycobacterial research (2–6). The Mtb genome encodes 13 sigma factor homologs, several of which play essential roles in disease (7).

Diverse bacteria employ sigma factors to mediate the transcriptional programs that drive life cycle transitions (8, 9). Alternative sigma factors, unique RNA-polymerase subunits that mediate promoter choice, are known to respond to environmental cues such as heat and limited energy (10) to drive the transition into spore formation (11) and stationary phase (12). As reviewed in Ref. 9, anti-sigma factors exemplified by the Bacillus subtilis regulator of sigma B W (RsbW) are capable of binding sigma factors and sequestering them away from RNA polymerase (Fig. 1). Through a partner switching mechanism, anti-anti-sigma factor proteins such as RsbV bind the anti-sigma factors, freeing the cognate sigma factors to activate transcription. Anti-sigma factors, which display homology to histidine kinases, phosphorylate their anti-anti-sigma factor antagonists on a serine or threonine residue (9). A master “environmental phosphatase,” such as RsbU in B. subtilis, reverses this phosphorylation and restores the complexes of anti- and anti-anti-sigma factors (13).

FIGURE 1.

Model of alternative sigma factor regulation in B. subtilis. RsbW, an anti-sigma factor (blue), binds and inactivates sigma factor B (gray). RsbV, an anti-anti-sigma factor (purple), binds and inactivates RsbW, functionally relieving sigma factor B inhibition. RsbW phosphorylates RsbV, which dissociates and allows RsbW to bind sigma factor B. The RsbU phosphatase, with an N-terminal PAS domain and C-terminal PP2C phosphatase domain, is the B. subtilis homolog of Mtb Rv1634c. RsbU dephosphorylates RsbV, allowing it to bind RsbW. RsbU is activated by binding to RsbT, a protein with sequence homology to RsbW (blue). RsbS (purple) binds and sequesters RsbT in the high molecular weight RsbR-RsbS “stressosome” complex (purple). Despite the distinct roles of RsbT and RsbW, they are homologous at the sequence level. The RsbR proteins, RsbV and RsbS are also paralogs. The figure is adapted from Ref. 36.

These RsbU-like Ser/Thr phosphatases are regulated by proteins with remarkable homologies to the anti- and anti-anti-sigma factors in a manner that shows parallels to and differences from sigma factor regulation (Fig. 1 and Ref. 13). An upstream anti-sigma factor homolog, such as RsbT, binds the phosphatase and, in contrast to the downstream sigma factor antagonism, stimulates the phosphatase activity. Proteins with homology to anti-anti-sigma factors, like RsbS, bind the upstream anti-sigma factor and prevent phosphatase stimulation. RsbS does not bind RsbT in isolation, but rather forms a large complex with multiple upstream anti-anti-sigma factors, such as RsbRA, RsbRB, RsbRC, and RsbRD in the so-called “stressosome” (14), to sequester RsbT. Phosphorylation of RsbR and RsbS by the cognate anti-sigma factor kinase releases RsbT. RsbS is dephosphorylated by the Ser/Thr phosphatase RsbX, which displays homology to RsbU and the PP2C phosphatases (15).

The high degree of sequence similarity between downstream and upstream anti- and anti-anti-sigma factors despite their distinct functions has presented challenges for deciphering the mechanisms of anti-anti- and anti-sigma factor functions. Moreover, biochemical analysis has been impeded because most of these regulators, when expressed in isolation, are insoluble. This property reflects the functions of these regulators in multi-protein complexes. Co-expression studies are difficult to design because binding partners cannot be predicted reliably based on homology.

In the RsbU-like environmental phosphatase of M. tuberculosis (Rv1364c), however, the catalytic domain is fused in a single polypeptide chain to regulatory modules. A PAS domain occurs at the N terminus, and an anti-sigma factor and an anti-anti-sigma factor domain follow the phosphatase module. Such a combination of domains within a single polypeptide has not been described outside of Rv1364c and close mycobacterial orthologs. We hypothesize that the anti- and anti-anti-sigma factor domains of Rv1364c function like the B. subtilis RsbT/RsbS pair to regulate the phosphatase domain (16). These intramolecular interactions in Rv1364c may be representative of the transient and biochemically intractable multiprotein complexes found in other bacteria (17). Thus, Rv1364c may serve as a model system of the environmental phosphatase regulatory circuit ultimately responsible for sigma factor regulation.

To explore the mechanisms of environmental phosphatase regulation, we probed the function of Rv1364c variants biochemically. We found that Rv1364c encodes an active phosphatase that is stimulated by the anti-sigma factor domain. Moreover, contrary to a recent report (18), the anti-sigma factor domain functions as an active kinase. Mutational inactivation of the phosphatase catalytic site stabilized phosphorylated Rv1364c, while mutations of the kinase active site or the predicted phospho-acceptor residue blocked phosphorylation. To experimentally characterize the structural changes driven by phosphorylation, we employed small angle x-ray scattering (SAXS) to define the shapes and conformations of the unphosphorylated and phosphorylated forms of Rv1364c. Rv1364c typifies functionally flexible systems that are suitable for SAXS analyses (19, 20). The solution structural envelopes suggest that Rv1364c forms an elongated dimer. Stable phosphorylation resulting from a mutation in the phosphatase active site caused a structural change consistent with the rearrangement of domains within the dimer. This conformational change provides a molecular context for the interdomain communication and phosphoregulation implied by the biochemical studies.

EXPERIMENTAL PROCEDURES

DNA Constructs

Start and end points for each domain of the Rv1364c gene were determined by alignment with homologs identified using PHYRE (21) and BLAST. Using Mtb H37Rv genomic DNA as a template for PCR amplification, gene segments encoding the Rv1364c gene were inserted into the Gateway vector pHMGWA (22), which included N-terminal His₆ and maltose-binding protein (MBP) tags, followed by a tobacco etch virus (TEV) protease site. Mutations were introduced using the QuikChange method (Stratagene). All constructs were confirmed by DNA sequencing.

Protein Expression and Purification

BL21(DE3) Codon Plus cells (Stratagene) cells harboring each expression plasmid were grown at 37 °C to an absorbance (A₆₀₀) of 1.8 in Terrific Broth (Research Products International), moved to 18 °C for 10 min, induced using 300 μm isopropyl 1-thio-β-d-galactopyranoside, and grown for an additional 4 h. The cells were harvested by centrifugation and resuspended in 100 mm Tris-HCl (pH 7.8), 0.3 m NaCl, 10% (v/v) glycerol, and 0.5 mm TCEP. Cells were lysed by sonication on ice, and the lysate was clarified by centrifugation. The supernatant was loaded onto a 5-ml chelating Sepharose HP column (Amersham Biosciences) equilibrated with 0.1 m NiSO₄ and eluted in lysis buffer with 300 mm imidazole. Protein was dialyzed into 50 mm HEPES (pH 7.65), 150 mm NaCl, 0.5 mm TCEP, and 5 mm MnCl₂ and simultaneously proteolyzed with TEV protease overnight at 4 °C. Each protein was further purified by gel filtration chromatography using a HiLoad 26/60 Superdex 75 column (Amersham Biosciences) or Sephacryl S-200 column (Amersham Biosciences). Elution profiles were compared with gel filtration protein standards (Bio-Rad) for apparent molecular mass calculations. Protein was dialyzed into 100 mm NaCl, 15 mm HEPES 7.6, 0.5 mm TCEP, 2 mm MnCl₂, and 10% glycerol prior to DLS and SAXS analysis.

Phosphatase Activity Assays

Kinetic assays were performed with 860 nm enzyme diluted into 100 mm Tris pH 8, 50 mm NaCl, 0.5 mm TCEP, 10% glycerol, 10 mm MgCl₂, and 10 mm MnCl₂ with 0.78–50 mm p-nitrophenyl phosphate (pNPP) in a total volume of 100 μl. pNPP hydrolysis kinetics were monitored continuously for 60 min at 405 nm in triplicate. Curve fitting and kinetic constant calculations were performed using SigmaPlot (Systat Software, Inc). For pH dependence end point assays, 860 nm full-length Rv1364c was incubated in 100 mm NaCl, 10% glycerol, 10 mm MgCl₂, 10 mm MnCl₂, 25 mm pNPP, and 50 mm buffer. The following buffers were used: sodium cacodylate, pH 5.6, 6.0, and 6.4; HEPES pH 6.8, 7.2, and 7.6; and Tris pH 8.0. Reactions were allowed to proceed for 10, 20, 30, or 40 min before quenching with the addition of 25 μl of 500 mm EDTA, pH 8.0 and 25 μl of 1 n sodium hydroxide.

Phosphorylation State Detection

Phosphorylation reactions were performed in 15 mm HEPES pH 7.6, 0.1 m NaCl, 2 mm MgCl₂, 2 mm MnCl₂, 2 mm ATP, 10% glycerol for 90 min. Each Rv1364c construct was desalted prior to electrophoresis on a Tris-glycine gel (Invitrogen) and stained in parallel with either Coomassie Blue or the Diamond Pro-Q stain (Invitrogen). Fluorescence was quantified on a Typhoon 8600 with ImageQuant (GE Healthcare). Serial dilutions of the phosphorylated Mtb PknB kinase domain were used to ensure phosphorylation was detected in the linear range.

SAXS Data Collection and Evaluation

SAXS data were collected at the SIBYLS beamline at the Lawrence Berkeley National Laboratory Advanced Light Source (19). Using a wavelength, λ, of 1.03 Å with the sample-to-detector distance set to 1.5 m resulted in scattering vectors, q, ranging from 0.01 Å⁻¹ to 0.30 Å⁻¹. The data were acquired at 20 °C, and short and long time exposures (0.5 s, 5 s) were merged for the calculations using the entire scattering profile. Data were processed as described (19). The experimental SAXS data were measured at different protein concentrations to explore multimer formation and aggregation using Guinier plots. The radius of gyration R_G was derived by the Guinier approximation I(q) = I(0) exp(−q²R_G²/3) with the limits qR_G < 1.3. The interference-free SAXS profile was estimated by extrapolating the measured scattering curves to infinite dilution.

The SAXS curves measured for different concentrations (1–4 mg/ml) of wild-type and D328A Rv1364c displayed a concentration dependence arising from self-association in the concentrated samples. The estimated infinite dilution SAXS represents a form factor of the interference-free state. The GNOM programs (23) were used to compute the pair-distance distribution functions, P(r), and the maximum dimension of the macromolecule, D_max. The overall shapes were restored from the experimental data using the program GASBOR (24). Sixteen low-resolution models obtained from different runs were averaged using the program DAMAVER (25) to construct the average model representing the general structural features of each reconstruction. SAXS bead models were converted to volumetric Situs format with the pdb2vol kernel convolution utility (26).

RESULTS

Rv1364c Encodes Predicted Regulatory, Phosphatase, Kinase, and Substrate Domains

Four domains showing homology to known folds were identified within the Rv1364c sequence (Fig. 2A (18)). The N-terminal domain, from amino acids 1–397, is an RsbU-like phosphatase domain. The RsbU domain is composed of an N-terminal regulatory PAS domain from residues 1–137 followed by a PP2C Ser/Thr phosphatase domain from residues 138–397. Based on similarities to anti-sigma factors, we hypothesize that amino acids 398–544 encode a domain like RsbT that binds and activates the phosphatase and exhibits Ser or Thr kinase activity. The C-terminal domain (residues 545–653) exhibits homology to anti-anti-sigma factors, including a conserved phospho-acceptor site at Ser-600. This homology suggests that the C-terminal domain acts like RsbS to bind in an unphosphorylated state to the anti-sigma factor domain and that phosphorylation may trigger phosphatase activation.

FIGURE 2.

M. tuberculosis Rv1364c encodes an active phosphatase domain. A, Rv1364c contains four distinct domains. The N-terminal PAS domain is thought to sense energy levels in the cell and regulate the phosphatase (37). Residues 138–397 encode a PP2C phosphatase domain, and the N-terminal 397 residues comprise a RsbU-like domain. Residues 398–544 comprise an anti-sigma factor domain, which we hypothesize functions like RsbT. The C-terminal domain, which we hypothesize functions like RsbS to bind the RsbT domain, exhibits homology to anti-anti-sigma factors. B, dependence of pNPP hydrolysis by full-length Rv1364c protein as a function of pH. The pH dependence was shallow (18), and activity was greatest at the highest pH tested (8.0). C, kinetic analysis of full-length and domain constructs of Rv1364c (top) using pNPP as the substrate. The PP2C-RsbT domain combination (maroon octagons) comprises the minimal functional unit for full phosphatase activity toward pNPP. Consistent with the hypothesis that the anti-sigma factor domain functions as an RsbT-like phosphatase activator, the PP2C domain alone (green triangles) is significantly less active than the PP2C-RsbT domain combination (red circles).

Rv1364c RsbT Domain Activates the Phosphatase Domain

To test the prediction that Rv1364c encodes an active phosphatase, the protein was expressed and purified from E. coli and the kinetics of phosphate ester hydrolysis were measured. Using the noncognate, colorimetric substrate, pNPP, phosphatase activity was observed for full-length Rv1364c (Fig. 2). Either manganese or magnesium was required for hydrolysis, and activity was maximal with both divalent cations present (data not shown). The pH dependence of this activity was assessed using a base-quenched end point assay, and the upper bound of accessible pH was limited by the solubility of MnCl₂ (Fig. 2B). While activity was maximal at pH 8.0, there was only a modest (∼6-fold) reduction in activity at pH 5.5, as observed for other PP2C Ser/Thr phosphatases (27).

To assess the potential for interdomain communication, we removed multiple domains from full-length Rv1364c and kinetic constants were measured. At the optimum divalent metal concentrations and pH, the phosphatase activity of full-length Rv1364c was compared with that of a PAS domain truncation (ΔPAS), RsbS domain truncation (ΔRsbS), double PAS, and RsbS domain truncation (ΔPASΔRsbS) or PP2C phosphatase domain alone (ΔPASΔRsbSΔRsbT) (Fig. 2C). The truncations that retained the PP2C and RsbT domains were fully or nearly fully active. By comparison, the isolated phosphatase domain alone exhibited reduced k_cat and increased K_m values, which corresponded to an apparent 9–18-fold activation by the RsbT domain (Table 1). The shortest fully active combination, which was slightly more active than full-length phosphatase, was the PP2C-RsbT fusion (ΔPASΔRsbS). These results show that the anti-sigma factor (RsbT) domain activates the phosphatase domain through a direct interaction, while the PAS and RsbS domains are dispensable for pNPP hydrolysis in vitro. Activation of the phosphatase domain by the anti-sigma factor domain is consistent with the RsbU-RsbT interaction paradigm (Fig. 1 and Ref. 28).

TABLE 1.

Apparent kinetic constants for pNPP hydrolysis of Rv1364c variants

K_m, V_max, and k_cat were calculated by nonlinear regression of the data (e.g. Fig. 2C) using SigmaPlot. The unphosphorylated fusions containing the PP2C and RsbT (anti-sigma factor) domains showed the highest activation relative to the isolated PP2C domain (last column).

Domains present	Vmax	Km	kcat	kcat/Km	-Fold activation
	μm/min	mm	min	m/min
All (full-length)	12.2 ± 0.8	19.1 ± 2.9	14.2 ± 0.9	0.74	9.3
PP2C-RsbT	15.7 ± 0.7	13.3 ± 1.5	18.3 ± 0.8	1.4	17.5
PP2C	2.1 ± 0.5	30.6 ± 2.5	2.4 ± 0.6	0.08	1
PP2C-RsbT-RsbS	15.6 ± 0.5	14.5 ± 1.1	18.1 ± 0.6	1.2	15
PAS-PP2C-RsbT	16.6 ± 0.7	20.2 ± 1.7	19.3 ± 0.8	0.95	11.9

Rv1364c RsbT Domain Encodes an Active Kinase

RsbT proteins, in addition to binding and activating RsbU phosphatases, have been shown to catalyze transfer of the γ-phosphate from ATP to a serine residue of RsbS proteins (28). To assess the potential kinase activity of the Rv1364c RsbT domain, the phosphorylation states of the constructs were characterized. After expression in E. coli, Rv1364c mutants were purified by immobilized metal affinity chromatography (IMAC) using Ni²⁺-Sepharose and analyzed by SDS-PAGE. The variants were expressed at similar levels except for the deletion mutant lacking the RsbT and RsbS domains (ΔRsbTΔRsbS) (Fig. 3A, top). ΔRsbTΔRsbS was difficult to express and purify and generally insoluble. Staining these proteins for the presence of phosphoryl groups using Diamond Pro-Q showed that the Rv1364c construct lacking the PAS and phosphatase domains (ΔPAS ΔPhos) was efficiently phosphorylated (Fig. 3B, bottom). In contrast, the constructs containing the phosphatase domain or constructs lacking the RsbS domain were phosphorylated at background levels.

FIGURE 3.

The phosphatase domain antagonizes Rv1364c autophosphorylation. A, Rv1364c contains active phosphatase and kinase domains. Purified domain-truncation mutants of Rv1364c were stained with Coomassie Blue to measure protein levels (top) or Diamond Pro-Q, which preferentially stains phosphoproteins (bottom). The ΔRsbTΔRsbS mutant was unstable during expression and purification, and this construct is represented by less total protein. The ΔPASΔPP2C variant (lane 3), containing the predicted anti-sigma factor kinase domain and the phosphoacceptor in the anti-anti-sigma factor domain, is significantly phosphorylated after expression in E. coli. While the PP2C domain alone exhibited a typical level of background Diamond staining associated with non-phosphorylated proteins, the full-length protein appeared slightly phosphorylated when purified using a single column, one-day protocol. B, phosphorylation levels observed by Diamond staining were quantified for ΔPASΔPP2C (positive control), PP2C domain alone (negative control), full-length Rv1364c purified rapidly as in A, and Rv1364c after more extensive purification in the presence of magnesium for 2 days. Rv1364c, after the 1-day purification, stained at 11% of the positive control, while the same protein treated with magnesium stained at −1% of the positive control. C, effects of amino acid substitutions in the kinase, phosphatase, or phosphoacceptor sites on the phosphorylation state of Rv1364c. Asp-328 is predicted to coordinate the catalytic hydroxyl nucleophile in the RsbU-like phosphatase domain; Glu-444, Asn-448, and His-452 are predicted to contact ATP in the RsbT-like kinase domain; and Ser-600 matches the phospho-acceptor in RsbS-like anti-anti-sigma factors. The SDS gel was stained to detect phosphorylation (Pro-Q Diamond, top) and protein levels (Coomassie Blue, bottom). Lanes 1–3 contain the Peppermint Stick molecular mass markers and Mtb PknB positive controls for phosphoprotein staining. Lanes 4–7 contain full-length Rv1364c, and lanes 8–13 contain the RsbT-RsbS (ΔPASΔPP2C) construct. Reduction of phosphatase activity in the D328K mutant unmasked the kinase activity of the wild-type full-length phosphatase (compare lanes 4–7). Phosphorylation of the RsbT-RsbS (ΔPASΔPP2C) construct required Ser-600 (compare lanes 8 and 9), as well as the wild-type RsbT kinase domain (lanes 10–12). The S600A mutant containing an active kinase domain failed to phosphorylate the H452A mutant containing the wild-type phosphoacceptor in the presence of 2 mm Mg²⁺ATP (lane 13), suggesting that the kinase reaction is predominantly intramolecular.

To define the basis for the kinase activity, we characterized the effects of mutations in the predicted phosphoacceptor and kinase active sites (Fig. 3C). Phosphorylation of the RsbT-RsbS construct (ΔPAS ΔPhos) was abolished by the S600A mutation of the predicted phosphorylation site in the RsbS domain. Further, the mutations E444K, N448K, or H452A in the predicted kinase active site in the RsbT domain also eliminated phosphorylation. The S600A RsbT-RsbS (ΔPAS ΔPhos) mutant, containing a wild-type RsbT kinase domain, did not efficiently phosphorylate the H452A mutant containing a wild-type phosphoacceptor site but lacking kinase activity, consistent with the conclusion that the kinase functions in an intramolecular reaction. These results suggest that the RsbT domain functions as a classic anti-sigma factor-like kinase to phosphorylate the RsbS domain at the canonical regulatory site, and the phosphatase domain antagonizes this activity.

Corroborating this conclusion, the full-length Rv1364c exhibited a lower level of phosphorylation than the RsbT-RsbS fragment. Using the phosphatase domain alone (ΔPASΔRsbSΔRsbT) as a negative control, full-length (FL) Rv1364c exhibited 11% phosphorylation compared with the RsbT-RsbS fusion (ΔPAS ΔPP2C) (Fig. 3B). This apparently incomplete phosphorylation of full-length Rv1364c was observed (Fig. 3A) when the proteins were expressed in the ATP-rich environment of E. coli for 24 h and purified in divalent cation-free buffer in 5 h. On the other hand, including magnesium in the buffers during a 2-day, multicolumn purification in the absence of ATP eliminated phosphorylation of full-length Rv1364c. This protein exhibited −1% phosphorylation (Fig. 3B), which, within experimental error, indicated the absence of phosphate modifications. In contrast, the D328A mutation in full-length Rv1364c, which inactivates the phosphatase domain by eliminating a conserved carboxylate predicted in PP2C enzymes to coordinate the hydroxyl nucleophile, led to the accumulation of the phosphorylated protein (Fig. 3C). These results reflect a dynamic antagonism between the PP2C and kinase domains in Rv1364c.

Small Angle X-ray Scattering of Full-length Rv1364c

The ability to isolate the homogeneously dephosphorylated wild-type protein and the phosphorylated D328A mutant afforded the opportunity to use SAXS to identify key structural features of the Rv1364c protein responsive to phosphorylation. The protein samples used for SAXS were purified in the presence of magnesium over three columns (IMAC, gel filtration, and anion exchange chromatography) and analyzed using dynamic light scattering to ensure the lack of aggregation.

The SAXS profiles of both phosphorylated and unphosphorylated Rv1364c in a concentration range of 1–4 mg/ml exhibited features of a well-behaved protein (data not shown), but scattering in the low-angle region indicated particle interference at high concentrations. Consequently, we determined the interference-free SAXS profile by extrapolating the measured scattering curves to infinite dilution (Fig. 4A). Linear Guinier plots (29) of the interference-free profiles (Fig. 4A, inset) indicated that the protein adopts a distinct aggregation-free state. The molecular mass of the unphosphorylated protein calculated from these SAXS data using the Porod volume (30) was 150 kDa. The calculated molecular mass of the Rv1364c construct used in this study was 69.7 kDa, so the mass of a dimer (139.4 kDa) was within the 20% standard error associated with such calculations (20). The existence of a dimer in solution also was observed by gel filtration chromatography (Table 2). The radius of gyration (R_G) of the unphosphorylated form calculated from the SAXS data was ∼51 Å, which is unusually large for a protein of this size. These results suggest that Rv1364c contains extended elements or adopts a hollow conformation.

FIGURE 4.

SAXS reveals key Rv1364c structural features and a phosphorylation-dependent conformational change. A, interference-free experimental scattering curve for full-length Rv1364c (black) and the phosphorylated D328A mutant (red). A Guinier plot (inset) with the linear fit (blue line) within the limits qR_G >1.3 was used to calculate R_G values of 51 Å for the unphosphorylated protein and 54 Å for the phosphorylated mutant. B, SAXS profiles shown in the Kratky plot (I(q)*q² versus q). Differences in the Kratky plots highlighted by black and red arrows indicate phosphorylation-induced conformational changes. C, pair distribution function, P(r), indicates an elongated particle, with a maximum dimension, D_max, of ∼190 Å, and broadening of the structure (highlighted with arrows) consistent with reorientation of the domains upon phosphorylation. D, five representative models of unphosphorylated (gray, top) and phosphorylated (red, bottom) Rv1364c reconstructed in P2 symmetry using GASBOR (24) are shown in a surface representation calculated using the SITUS package (26). Consistency between the independently generated models was high, with χ² values of 1.2–1.5. The averaged envelopes show flat extended dimers with a globular core. Global differences between the models of the unphosphorylated and phosphorylated proteins provide evidence for partner switching of the domains. A representative view of the GASBOR average model reconstructed with a P1 symmetry operator (right) shows a similar overall shape to models constrained by the P2 operator.

TABLE 2.

Apparent molecular mass, as observed by size exclusion chromatography, of Rv1364c variants

Whereas the phosphatase domain alone was monomeric, all other constructs formed apparent dimers or tetramers in solution.

Domains present	Theoretical mass	Observed mass	Predicted state
	kDa	kDa
All (full-length)	69.5	148.3	Dimer
PP2C-RsbT	42.9	91.4	Dimer
PP2C	27.0	30.6	Monomer
PP2C-RsbT-RsbS	54.0	213.8	Tetramer
PAS-PP2C-RsbT	58.4	245.7	Tetramer

Comparison of the x-ray scattering profiles from phosphorylated (D328A) and unphosphorylated Rv1364c revealed differences over the entire scattering range in the Kratky plot (Fig. 4B). The phosphoprotein showed broadening of the distribution function, P(r) (Fig. 4C), and an increase in R_G from ∼51 to ∼54 Å. These results provide evidence for a change in the overall shape of the dimer in response to phosphorylation of Ser-600.

We determined solution structural envelopes of full-length Rv1364c to generate models for higher order complex formation between RsbU-, RsbT-, and RsbS-like domains. The P(r) functions calculated by GNOM (23) (Fig. 4C), marked by a broad P(r) maximum, indicated that Rv1364c adopts an elongated conformation with maximal dimension (D_max) of ∼190 Å. Using the program GASBOR (24) to fit the interference-free data in the q range of 0.01–0.30 Å⁻¹, we calculated envelopes constrained by P2 symmetry for both the unphosphorylated (wild-type) and phosphorylated (D328A) proteins. The commonalities among individual reconstructions of the SAXS envelopes reflect the similarities between independent runs (Fig. 4D). To ensure the symmetry operator was not inappropriately biasing our calculations, the overall shape was re-calculated without a symmetry operator (Fig. 4D), and similar structural features were observed.

DISCUSSION

This report describes and quantifies the overall structure and regulation of Rv1364c, the RsbU-like, environmental Ser/Thr phosphatase of Mtb. The functional equivalence of Rv1364c and RsbU and its regulators in B. subtilis is supported by previous reports implicating Rv1364c in the Mtb sigma factor-F pathway (18). As bacterial genomes often encode functionally related proteins in close proximity, it is also relevant to note that Rv1364c neighbors rsfA, the gene for an anti-anti-sigma factor that regulates sigma factor F in response to cellular redox levels (31). Homologies alone, however, are not sufficient to distinguish the disparate functions of such domains. Some anti-sigma factor proteins, for example, bind and inactivate sigma factors while others bind and activate a Ser/Thr phosphatase. The physical connection between the RsbU-like phosphatase and anti-sigma factor/anti-anti-sigma factor pair in Rv1364c suggested a likely interaction, providing clues to domain function and making biochemical studies tractable.

The full-length Rv1364c protein showed pNPP hydrolysis activity in vitro, consistent with previous results (18). In addition, we found that fusion to the RsbT domain activated the phosphatase domain ∼18-fold (Fig. 2C), consistent with a direct regulatory interaction. Neither the PAS domain nor the RsbS domain was essential for this stimulation. The apparently dispensable roles for the terminal domains in vitro should not be extrapolated to assume similarly dispensable roles in vivo. PAS domains, for example, are thought to bind small-molecule cofactors and interact with other proteins. While dephosphorylated B. subtilis RsbS has been shown to bind RsbT, effectively blocking the RsbU-stimulating activity of the RsbT, this RsbS-RsbT binding is significantly diminished in the absence of RsbR and stressosome formation (32). Consistent with this essential role for RsbR, the RsbS domain of Rv1364c did not antagonize RsbT domain-dependent phosphatase stimulation in the in vitro assays (Fig. 2C). In the absence of a PAS domain ligand or the intact stressosome, the roles of these terminal domains cannot be fully assessed.

In addition to activating the phosphatase domain, the Rv1364c RsbT domain phosphorylated the RsbS domain. This robust activity (Fig. 3A) contrasts with the recent report that the RsbT domain of Rv1364c lacks kinase activity (18). This apparent discrepancy likely results from the use in the previous study of the full-length (active) phosphatase, isolated single domains or noncognate substrates to assay for phosphoryl transfer. In addition, the previous in vitro kinase assays employed low concentrations (8 μm) of Mg²⁺-ATP (18), rather than physiological concentrations in the mm range. We found that the kinase activity is antagonized in the full-length protein by the phosphatase domain (Fig. 3), which limits the accumulation of phosphorylated product. The D328A mutation in the phosphatase active site led to the accumulation of phosphorylated full-length protein (Fig. 3C), consistent with the idea that the interplay of kinase and phosphatase domain activities controls the phosphorylation state. Deletion or inactivation of the phosphatase domain unmasked the kinase activity.

The previously observed lack of kinase activity also was based in part on phosphorylation assays using the noncognate substrates, myelin basic protein and histones, which lack competent recognition sites (18). In addition, phosphorylation of the Rv1364c RsbS-like anti-anti-sigma factor domain was tested only in trans using the isolated domains (18). Consistent with these intermolecular kinase assays, we found that the S600A RsbT-RsbS construct did not phosphorylate the H352A mutant lacking a functional kinase (Fig. 3C). In contrast, the Rv1364c RsbT-RsbS domain fusion showed stable phosphorylation in E. coli, consistent with the idea that this segment contains both the kinase domain and the site of intramolecular phosphorylation.

Homologies to active anti-sigma factor kinases suggested that conserved residues Glu-444, Asn-448, and His-452 in Rv1364c are located in the ATP binding site. Similarly, Ser-600 in the Rv1364c RsbS domain matches the phosphoacceptor residues in sequence alignments with functionally phosphorylated anti-anti-sigma factor proteins. Mutations of either the predicted kinase active site or the phosphoacceptor site blocked phosphorylation of the RsbT-RsbS construct (ΔPASΔPP2C; Fig. 3C). Rather than requiring an entirely new mechanism of regulation in Mtb (18), these results suggest that the RsbT-like kinase domain of Mtb Rv1364c fits the paradigm established in B. subtilis (13) for regulation of the environmental phosphatase.

Diminished phosphorylation was observed in the constructs containing the Rv1364c phosphatase domain, and the intact protein purified from E. coli was partially phosphorylated (Fig. 3). Incubation of partially phosphorylated full-length protein with Mg²⁺ and Mn²⁺ removed the remaining phosphates. We therefore conclude that the phosphatase domain of Rv1364c acts on the same substrate as the kinase domain. It remains to be determined if this reaction is stimulated by other components of the stressosome, as observed in B. subtilis.

To characterize the structural consequences of phosphorylation, we used SAXS to determine the overall shape of Rv1364c and the changes in domain organization in response to phosphorylation. The SAXS data in this report characterize, for the first time, the interactions between RsbU-, RsbT-, and RsbS-like domains. Unlike other bacteria in which unphosphorylated RsbS is expected to sequester RsbT and block binding to the RsbU phosphatase, these domains are held together in Mtb by fusion into the single Rv1364c polypeptide chain. Despite differences in the x-ray scattering curves, the SAXS patterns demonstrate that both the phosphorylated and unphosphorylated Mtb Rv1364c proteins form extended dimers. The radii of gyration, 51 Å for the unphosphorylated protein and 54 Å for the D328A phosphoprotein, are large for a dimer of this molecular mass, reflecting the presence of arms extended from the central globular core (Fig. 4D). Both dimers are relatively flat, with a possible reduction in density around the 2-fold rotation axis.

Attempts to fit high resolution crystallographic models of homologs of the individual domains into the SAXS envelope were unsuccessful. This difficulty likely has multiple causes, including the sequence divergence of the homologs with structures in the Protein Data Bank (PDB). The PAS domain, for example, exhibited only 17% homology to the closest homolog in the PDB. Additional challenges to fitting atomic models into the SAXS envelopes also may be presented by the possible flexibility and conformational disorder of the extended regions, the low resolution of the envelopes, uncertainty about the location of the dimer interface and potential heterogeneity due to differences in aggregation, phosphorylation, and conformational switching.

Valuable information about the structure, however, can be gleaned from the combination of gel filtration elution profiles, SAXS envelopes, and biochemical assays. All of the most active phosphatase constructs were multimers in solution (Table 2 and Fig. 2C). Only the phosphatase domain in isolation was a monomer, and this monomeric RsbU-like domain showed 18-fold lower catalytic activity compared with the dimeric RsbU-RsbT domain fusion. While the presence of a short coiled-coil sequence has been predicted in the phosphatase segment (18), this region contains many noncanonical predicted interface residues (33) and clearly is insufficient to mediate association (Table 2). Fusion of the RsbT anti-sigma factor domain to the RsbU domain was sufficient to mediate dimerization. This association parallels the finding that a number of anti-sigma factors from different species form dimers in isolation (16, 34, 35). The SAXS envelope of full-length Rv1364c not only confirmed the dimeric arrangement, but also placed limits on the location of the three domains. An arrangement in which the anti-sigma factor domain dimerizes and interacts with the phosphatase domain of the opposing monomer in the central globular region of this envelope would maintain the physical interactions necessary to account for the biochemical activity and the structural data. Notably, the phosphorylation of Rv1364c results in opening of the central region of the dimer (Fig. 4). This conformational change may alter the arrangement of the RsbU-, RsbT-, and RsbS-like domains and provides evidence for phosphorylation-dependent partner switching of domains that ultimately controls the bacterial transcriptional response.

Higher order complex formation and partner switching play essential roles in sigma factor regulatory cascades. The ability of the Rv1364c anti-sigma factor kinase domain to stimulate phosphatase activity and to phosphorylate the RsbS domain in an intramolecular reaction suggests that this multi-domain protein provides a tractable system to characterize the general features of environmental phosphatase activity and regulation across the bacterial kingdom.