INTRODUCTION

Virulence gene activation in Bordetella pertussis is mediated by the BvgAS two-component system (1, 2, 3, 4, 5). Two-component systems use phosphorylation cascades to effect adaptive responses and typically contain sensor and response regulator components (6). The sensor protein, which autophosphorylates at a conserved transmitter domain, is the initial site of signal processing. Phosphotransfer from the sensor protein is catalyzed by a conserved receiver domain in the response regulator. Responses such as transcriptional activation and protein-protein interaction are then mediated by the phosphorylated response regulator.

BvgS autophosphorylates with the -phosphoryl group of ATP and can phosphorylate BvgA (7). Phosphorylated BvgA has an increased affinity for bvg-activated promoters and is competent in an in vitro transcription assay (8, 9, 10). Although these properties of BvgS and BvgA fit well with the paradigm of two-component systems, BvgS employs a mechanism of BvgA phosphorylation that differs from the typical transmitter to receiver phosphotransfer reaction. As shown in Fig. 1, BvgS contains three domains that participate in the phosphorylation cascade: the transmitter, receiver, and C terminus (7, 11). The C terminus is specifically responsible for BvgA phosphorylation (11). Mutational studies of BvgS indicated that His-729 of the transmitter is required for autophosphorylation and that Asp-1023 of the receiver is essential for phosphorylation of the C terminus. His-1172 of the C terminus is necessary for BvgA phosphorylation. These amino acids are all required in vivo for expression of bvg-activated genes (7, 11, 12).

Although Asp-1023 of the BvgS receiver is essential for the phosphorylation cascade and virulence gene activation, the exact role of the receiver had not been determined (11). The BvgS receiver could act as an autoinhibitory domain, with phosphorylation relieving inhibition of phosphotransfer from the BvgS transmitter to the C terminus. Mutation of Asp-1023, the proposed site of phosphorylation in the receiver, would therefore lock the receiver in an inhibitory state. Alternatively, the BvgS receiver could have a positive role in the phosphorelay, serving as a phosphorylated intermediate between the transmitter and C terminus. In this report, we demonstrate that the BvgS receiver exerts both positive and negative influences on the BvgAS phosphorelay. The BvgS receiver can directly transfer a phosphoryl group to the C terminus, but the receiver also mediates dephosphorylation of BvgS by removing phosphoryl groups from both the transmitter and C terminus.

EXPERIMENTAL PROCEDURES

Escherichia coli strain DH5 (13) was utilized for protein overexpression. For quantitation of bvg activity, E. coli strain JFMC3 (MC4100 recA fhaB-lacZYA) was employed (14). Cultures were grown in Luria-Bertani medium (13); when necessary, medium was supplemented with antibiotics at concentrations of 100 µg/ml for ampicillin. -galactosidase activities were determined with cells permeabilized with SDS/CHCl₃ as described (13).

Except for where noted below, construction of all plasmids has been detailed elsewhere (11). pMU714, which contains bvgS with a precise deletion of the regions encoding the receiver and C terminus (BvgSRC), was constructed by deleting a 700-base pair BsrGI fragment from pMU677. pMU677 was created by inserting the oligomers DeRC-1 (5-CTGTACATCTGAC-3) and DeRC-2 (5-GTCAGATGTACAG-3) into the unique SnaBI site of pMU228. pMU228 is identical to pJM26 (14) except that it contains a BsrGI site at the junction of the BvgS transmitter and receiver. Introduction of the mutations that created this site did not affect BvgS function in vivo (data not shown). The vector for overexpression of BvgSRC, pMU540, was created by inserting an NcoI fragment from pMU714 encoding a portion of the BvgS transmitter into pMU100 (11).

The maltose-binding protein-BvgS receiver (MBP-R)¹ plasmid, pMU542, was created by taking an NcoI (blunted by treatment with the Klenow fragment of DNA polymerase) and HindIII fragment from pHB50delC (11) and inserting it into pMalc (New England Biolabs) that had been cut with HindIII and EcoRI (blunted with the Klenow fragment of DNA polymerase).

All protein purification methods were as previously stated for BvgS (the cytoplasmic portion of BvgS; see Fig. 1) and glutathione S-transferase (GST) fusions to BvgS (11). GST-BvgS, GST-BvgSRC, GST-BvgSC, and MBP-R were purified from insoluble inclusion bodies, dissolved in 6 M guanidine HCl, 50 mM Na₂HPO₄ (pH 7.2), and renatured in 20 mM HEPES (pH 7.4), 50 mM KCl, 10 mM MgCl₂, 5 mM dithiothreitol, 50% glycerol. GST-R (the BvgS receiver) and GST-C (the BvgS C terminus) were purified by glutathione-Sepharose affinity chromatography.

For purification of phosphorylated proteins from ATP, phosphorylation reactions were performed as below except that GST-BvgS derivatives were present at 1-3 µM and ATP was at 60 µM. The reaction was stopped after 10 min by the addition of 50 mM EDTA, and 50-100 µl of glutathione-agarose was added. After binding for 10 min on ice, the pelleted agarose was washed six times with 5 volumes of phosphate-buffered saline (10 mM buffered phosphate (pH 7.3), 120 mM NaCl, 2.7 mM KCl). Proteins were eluted with 50 mM Tris-HCl (pH 8.0), 10 mM glutathione for 10 min on ice. The yield following purification was typically 25%.

The MBP-R fusion was purified from BvgS and ATP by amylose-agarose affinity chromatography. The purification scheme was identical to glutathione-agarose purification methods except that the wash buffer was 20 mM Tris, pH 7.4, 200 mM KCl, 1 mM EDTA, and the elution buffer was 20 mM Tris, pH 7.4, 100 mM KCl, 1 mM EDTA, and 10 mM maltose. Purification yields of the MBP-R fusion were 10%.

Phosphorylation assays were essentially as described before (11). Except where otherwise indicated in the figure legends, BvgA and BvgS wild-type or mutant derivatives (GST-BvgS, GST-BvgSRC, and GST-BvgSC) were present at 0.3 µM. GST-R and GST-C were present at 1 µM. The reaction buffer contained 30 mM HEPES (pH 7.2), 50 mM KCl, 10 mM MgCl₂, and 30 µM ATP (specific activity, 0.30 Ci/mmol). Reactions were terminated by the addition of 4 × sample buffer (0.32 M Tris (pH 6.8), 40% glycerol, 0.1 M EDTA (pH 8.0), 8% SDS, 0.4 M dithiothreitol) and heated to 55 °C for 5 min. Samples were separated by SDS-polyacrylamide gel electrophoresis and transferred to an Immobilon membrane (Millipore Corp.). In the case of samples that were analyzed by thin layer chromatography, reactions were terminated by the addition of 50 mM EDTA.

As described by Hess et al. (15), thin layer chromatography was utilized for resolving phosphorylated proteins from ATP and inorganic phosphate. Reactions were normalized to standards by cpm and then spotted onto a polyethyleneimine-cellulose plate (Brinkmann) and resolved with 0.8 M LiCl and 0.8 M acetic acid. The polyethyleneimine-cellulose plate was dried and wrapped in polyvinyl chloride before autoradiography.

RESULTS

Multiple BvgS domains are involved in the BvgAS phosphorylation cascade (Fig. 1; Refs. 7 and 11). While the phenotypes of mutations indicated a requirement for the BvgS receiver in vivo, the contribution of this domain to the phosphorelay was unknown (7, 11, 12). We speculated that the receiver might mediate dephosphorylation of BvgS, since the receiver had been shown to dephosphorylate the isolated C terminus (11). We tested the role of the receiver in dephosphorylation by using BvgS and BvgS with an asparagine substituted for the conserved aspartic acid in the receiver, Asp-1023 (BvgS D1023N). The proteins were phosphorylated and purified from ATP, and the stability of the phosphoryl group was measured by incubation in buffer containing Mg²⁺. Phosphorylated BvgS decayed rapidly, with a t1/2 of approximately 30 s (Fig. 2A, open circles). In contrast, BvgS D1023N was stably phosphorylated under the same conditions, and the t1/2 of the phosphorylated form was nearly 20 min (Fig. 2A, open squares). Dephosphorylation of BvgS was accompanied by generation of inorganic phosphate as measured by thin layer chromatography (Fig. 2B). However, BvgS D1023N did not dephosphorylate to a substantial degree, and only very low amounts of inorganic phosphate were generated (Fig. 2C). The conserved aspartic acid in the receiver, Asp-1023, is therefore required for efficient dephosphorylation of BvgS.

Decay of BvgS and BvgS D1023N.

Decay of BvgS in the Presence of ATP

When phosphorylated BvgS and BvgS D1023N were incubated with excess ATP in pulse-chase reactions, the phosphorylated proteins had similar rates of decay (7). However, in the absence of excess ATP, BvgS and BvgS D1023N have vastly different decay rates (Fig. 2). These results suggested that the presence of excess ATP may influence the rate of dephosphorylation of BvgS. We purified phosphorylated BvgS and BvgS D1023N, added excess ATP, and measured phosphorylation as a function of time following ATP addition. The decay rate of the phosphoryl group on BvgS was not measurably affected (Fig. 2A, closed circles). However, BvgS D1023N dephosphorylated at a significantly increased rate (~7-fold) in the presence of excess ATP, with a t1/2 of approximately 2.5 min (Fig. 2A, closed squares). Nearly identical results were seen upon the addition of ADP or ATPS to BvgS D1023N (data not shown).

If binding of ATP or ADP to BvgS D1023N stimulates a phosphatase activity of the transmitter, then generation of inorganic phosphate should accompany dephosphorylation. Alternatively, dephosphorylation of BvgS could be due to donation of the phosphoryl group to ADP, forming ATP. Reversibility of transmitter phosphorylation has been demonstrated for the NR_II (NtrB) sensor protein (16). As shown in Fig. 3, dephosphorylation of BvgS D1023N in the presence of unlabeled ATP, ATPS, or ADP is coincident with formation of labeled ATP. Negligible amounts of inorganic phosphate were detected. Formation of labeled ATP after incubation of phospho-BvgS D1023N with ATP and ATPS is most likely due to the presence of ADP, which is either generated by BvgS-mediated hydrolysis of ATP or is present as a contaminant in the ATP or ATPS preparation. We conclude that hydrolysis of the -phosphoryl group of ATP by BvgS to form phospho-BvgS and ADP can be reversed but is detectable only in the absence of the receiver function.

Mechanism of GST-BvgS D1023N dephosphorylation.

The BvgS transmitter and C terminus are both potential targets for dephosphorylation by the receiver. To examine the interaction of the receiver with these domains in more detail, we determined the rates of receiver-mediated dephosphorylation of the isolated transmitter and C terminus. The transmitter and C terminus, purified as fusions to glutathione S-transferase, were relatively stable in the absence of the receiver (t1/2 of 10 and 15 min, respectively; Fig. 4). Both proteins were rapidly dephosphorylated in the presence of a 10-fold molar excess of the BvgS receiver. However, the C terminus was dephosphorylated at faster rate (t1/2 = 6 s) than the transmitter (t1/2 = 30 s). As measured by affinity chromatography and thin layer chromatography, dephosphorylation of the transmitter and C terminus by the receiver is due to transfer of the phosphoryl group to the receiver. The receiver phosphoryl group then undergoes autohydrolysis, generating inorganic phosphate (data not shown).

We had previously determined that phosphorylation of the C terminus occurred by a phosphotransfer reaction, but we could not distinguish whether the transmitter or receiver served as a phosphodonor. To determine which domains of BvgS are required or sufficient for particular steps of the phosphorelay, we precisely deleted specific BvgS domains and assayed the resulting mutant proteins. As shown in Fig. 5, BvgS was able to autophosphorylate, transphosphorylate the BvgS C terminus (+GST-C), and phosphorylate BvgA (+BvgA). BvgS wild type was also slightly dephosphorylated by the addition of the BvgS receiver in trans (+GST-R), and faint labeling of the receiver was detected. BvgS with deletion of the receiver and C terminus (RC) retained autophosphorylation ability (Fig. 5) and was dephosphorylated by the BvgS receiver (+GST-R). However, BvgSRC was deficient for BvgA phosphorylation and transphosphorylation of the C terminus (Fig. 5, +BvgA and +GST-C). From these results, we conclude that the transmitter is not sufficient for phosphorylation of the BvgS C terminus. The inability of the transmitter to phosphorylate BvgA indicates that the only relevant BvgS domain identified thus far for BvgA phosphorylation is the C terminus (11).

An exact deletion of the BvgS C-terminal domain (BvgSC) retained autophosphorylation ability (Fig. 5). While BvgSC was able to transphosphorylate the C terminus (+GST-C), it was unable to detectably phosphorylate BvgA (+BvgA). The in vitro phenotypes of BvgSRC and BvgSC indicate that the transmitter is sufficient for autophosphorylation, while the receiver is required for transphosphorylation of the C terminus. It is interesting to note that BvgSC and BvgS were not efficiently dephosphorylated by the addition of the receiver in trans, whereas BvgSRC was (Fig. 5, +GST-R lanes). The presence of a receiver in cis appears to decrease the efficiency of receiver-mediated dephosphorylation in trans, and we hypothesize that this is due to either stearic effects or a more efficient interaction of the receiver with the transmitter and C terminus in cis. Coincident with their inability to phosphorylate BvgA, BvgSRC and BvgSC were unable to activate an fhaB::lacZYA fusion in vivo, while wild type BvgS could direct high levels of expression of this fusion (Table I).

Table I. In vivo activities of bvgS and mutant derivatives Plasmids were transformed into the E. coli strain JFMC3, which contains a chromosomal transcriptional fusion between theh bvg-activated gene, fhaB, and lacZYA. A wild-type copy of bvgA was present on all plasmids except pBR322. RC refers to a precise deletion of the BvgS receiver and C terminus, and C is an exact deletion of the BvgS C terminus. The phenotype of bvgSRC has been noticed by others (12). Plasmid name bvgS allele  -Galactosidase units pDM20 Wild type 12,000  ± 1,000 pMU714  RC 9.3  ± 0.3 pHB50delC  C 10.1  ± 0.3 pBR322 Vector 10.9  ± 0.7

The results of Fig. 5 suggested that the BvgS receiver directly mediated the phosphotransfer reaction to the C terminus. This was seemingly contradicted by the fact that we had not been able to detect significant phosphorylation of the BvgS receiver in vitro (7). During experiments in which we were attempting to reconstruct the phosphorylation cascade, we noticed that we could readily purify phosphorylated receiver by affinity chromatography. When the phosphorylated receiver was subjected to SDS-polyacrylamide gel electrophoresis immediately following purification (Fig. 6, lane 1), phosphorylation of the receiver could no longer be detected. This same phenomenon was also observed if the phosphorylated receiver was incubated in buffer containing Mg²⁺ prior to electrophoresis (lane 2). Despite the apparent lability of the phosphorylated receiver to SDS-polyacrylamide gel electrophoresis, we reasoned that transfer of the phosphoryl group from the receiver to the C terminus should be readily detectable, since the phosphorylated C terminus is stable to SDS-polyacrylamide gel electrophoresis (11). When the C terminus was incubated with the purified phosphorylated BvgS receiver, the C terminus was now phosphorylated (lane 3), demonstrating that the receiver can serve as a phosphodonor for the C terminus. In contrast, when the BvgS transmitter was incubated with the phosphorylated receiver, no transfer to the transmitter was detected (BvgSRC, lane 4), suggesting that the phosphotransfer reaction between the transmitter and the receiver is not readily reversible.

The instability of the phosphorylated BvgS receiver to SDS-polyacrylamide gel electrophoresis distinguishes it from the BvgA receiver and many other receivers. Because the BvgS receiver dephosphorylates in the presence of SDS and urea and in the absence of added Mg²⁺ (data not shown), it is unlikely that extensive secondary structure is required for this phenotype. We suspected that an amino acid adjacent to the proposed site of phosphorylation in the BvgS receiver may be partially responsible for dephosphorylation during SDS-polyacrylamide gel electrophoresis. Interestingly, in a comparison of amino acid similarity among 79 receiver domains (17), a negatively charged amino acid is never found adjacent to the aspartic acid corresponding to Asp-1023, and BvgS is the only receiver that contains a cysteine adjacent to the conserved aspartic acid. When the receiver is incubated for extended periods in the presence of 0.1% SDS, buffers with pH values of 8.5 or above result in dephosphorylation to background levels, which indicates that electrophoresis per se is not required for dephosphorylation (data not shown). This sensitivity of the phosphorylated BvgS receiver to pH values above 8.5 corresponds with the pK value of the -SH side group of cysteine of 8.3 (18). We propose that the presence of a negatively charged amino acid adjacent to the site of phosphorylation in the BvgS receiver is destabilizing and that the pH value of 9.5 of the Tris/glycine front during SDS-polyacrylamide gel electrophoresis (19) is the reason for the apparent instability of the receiver in vitro. Despite the extreme base lability of the BvgS receiver, we have been able to clearly define two roles for this domain: dephosphorylation of BvgS and phosphotransfer to the C terminus.

DISCUSSION

The BvgAS phosphorylation cascade can be modeled as a stepwise process involving an initial autophosphorylation event followed by three phosphotransfer reactions (Fig. 7). In our model, signal inputs are relayed through the membrane to the transmitter. The transmitter then autophosphorylates at His-729 with the -phosphate of ATP to form phospho-BvgS and ADP. This reaction is reversible in vitro, but whether the reversibility is significant in vivo is not known. His-729 donates the phosphoryl group to Asp-1023 of the receiver. Asp-1023 can then donate the phosphoryl group to water to form inorganic phosphate, or it can transfer to His-1172 of the BvgS C terminus. The C terminus can then transfer back to the BvgS receiver, or it can phosphorylate BvgA (11). Phosphorylated BvgA then binds to specific DNA repeats upstream of bvg-activated promoters and effects gene regulation (8, 9, 10, 20). In all cases, mutations that interrupt the phosphorelay in vitro have a corresponding defect in virulence gene activation in vivo (7, 11). The BvgS receiver is a pivotal component of the phosphorelay. It appears to act as a biochemical checkpoint, mediating phosphorylation and dephosphorylation of the C terminus as well as dephosphorylation of the transmitter. We currently do not know what conditions influence these activities of the receiver or whether auxiliary proteins are involved in regulation of receiver activity such as in chemotaxis in E. coli or sporulation in Bacillus subtilis (21, 22, 23).

A new type of two-component regulatory system that utilizes a His  Asp  His  Asp phosphorelay is emerging, as first described in the regulation of sporulation initiation in B. subtilis and recently described for osmoregulation in yeast (24, 25). This subclass contains a receiver that acts as a phosphorylated intermediate (the BvgS receiver) as well as a histidine phosphotransfer domain (the BvgS C terminus), which acts as an intermediate between two receiver domains. These features distinguish this class from traditional two-component systems. Some of these phosphorelay components can be linked in a single protein, as in BvgS, or unlinked as in the B. subtilis sporulation cascade (24). Another putative member of this subclass includes the ArcB/ArcA proteins, which repress genes involved in aerobic metabolism (26, 27, 28, 29). There are several other defined sensor proteins besides ArcB and BvgS that have a transmitter/receiver/C terminus architecture; these are expected to follow the same multidomain phosphorylation strategy (see Refs. 28 and 30 for listings). No specific biological theme common to this subclass of two-component systems has yet been described.

We have been characterizing the biochemistry of BvgAS signal transduction to gain a better understanding of the biological interactions of Bordetella with its animal hosts. bvg is the key regulatory locus controlling a transition between two distinct phases of Bordetella. Several operons are either activated or repressed by BvgAS in B. pertussis and Bordetella bronchiseptica (2, 31). In B. bronchiseptica, bvg-activated genes are necessary for colonization of the respiratory tract, while at least one class of bvg-repressed genes interferes with infection (32, 33). bvg-repressed genes in B. bronchiseptica include the motility operon and genes required for production of a siderophore (34, 35). It has been shown that the Bvg phase is advantageous for survival under nutrient-limiting conditions (33). The involvement of the BvgAS signal transduction system in several aspects of the Bordetella life cycle suggests that BvgAS function is highly pleotropic and coupled to other regulatory pathways. The multiple domains required for BvgAS signal transduction may allow for coordination of several intracellular and extracellular inputs into a central phosphorelay.