Activating mutations in quorum-sensing regulator Rgg2 and its conformational flexibility in the absence of an intermolecular disulfide bond

Rap/Rgg/NprR/PlcR/PrgX (RRNPP) quorum-sensing systems use extracellular peptide pheromones that are detected by cytoplasmic receptors to regulate gene expression in firmicute bacteria. Rgg-type receptors are allosterically regulated through direct pheromone binding to control transcriptional activity; however, the receptor activation mechanism remains poorly understood. Previous work has identified a disulfide bond between Cys-45 residues within the homodimer interface of Rgg2 from Streptococcus dysgalactiae (Rgg2Sd). Here, we compared two Rgg2Sd(C45S) X-ray crystal structures with that of wild-type Rgg2Sd and found that in the absence of the intermolecular disulfide, the Rgg2Sd dimer interface is destabilized and Rgg2Sd can adopt multiple conformations. One conformation closely resembled the “disulfide-locked” Rgg2Sd secondary and tertiary structures, but another displayed more extensive rigid-body shifts as well as dramatic secondary structure changes. In parallel experiments, a genetic screen was used to identify mutations in rgg2 of Streptococcus pyogenes (rgg2Sp) that conferred pheromone-independent transcriptional activation of an Rgg2-stimulated promoter. Eight mutations yielding constitutive Rgg2 activity, designated Rgg2Sp*, were identified, and five of them clustered in or near an Rgg2 region that underwent conformational changes in one of the Rgg2Sd(C45S) crystal structures. The Rgg2Sp* mutations increased Rgg2Sp sensitivity to pheromone and pheromone variants while displaying decreased sensitivity to the Rgg2 antagonist cyclosporine A. We propose that Rgg2Sp* mutations invoke shifts in free-energy bias to favor the active state of the protein. Finally, we present evidence for an electrostatic interaction between an N-terminal Asp of the pheromone and Arg-153 within the proposed pheromone-binding pocket of Rgg2Sp.

pheromone and pheromone variants while displaying decreased sensitivity to the Rgg2 antagonist cyclosporine A. We propose that Rgg2 Sp * mutations invoke shifts in free-energy bias to favor the active state of the protein. Finally, we present evidence for an electrostatic interaction between an N-terminal Asp of the pheromone and Arg-153 within the proposed pheromone-binding pocket of Rgg2 Sp .
Extracellular chemical communication (quorum sensing, QS 5 ) in the low-G ϩ C Gram-positive bacteria (Firmicutes) is best understood to utilize peptide messengers (pheromones) as signals to coordinate gene expression among populations of cells. Pheromone receptors fall into two functional groups, those that detect peptides outside the cell by integral membrane sensors (e.g. ComD, AgrB, BAP, and others), and those located in the cytoplasm that rely on transport of pheromones into the cell (1, 2). The latter are members of the RRNPP family of receptors, which employ the concave pocket of a C-terminal helical repeat domain to bind the peptide ligands (3)(4)(5)(6)(7). Most Rap/Rgg/NprR/PlcR/PrgX (RRNPP) proteins, like NprR, PlcR, PrgX, and Rgg, contain an N-terminal DNA-binding domain (DBD), and they regulate transcription conditionally as a function of pheromone binding. Rap proteins are also modulated by peptide ligands, but control gene regulation through interactions with protein mediators of transcription (8 -10). As RRNPP proteins serve as communication hubs within bacterial cells, an understanding of their switch-like properties in the process of gene regulation has been a focus of several studies (3,6,7,(11)(12)(13). Despite a similar overall structure among RRNPP proteins, allosteric modulation of activity by ligands is reported to have disparate mechanisms of function, where conformational and oligomeric differences have been described.
The obligate human pathogen Streptococcus pyogenes (Group A Streptococcus (GAS)) is genetically programmed to utilize several different QS systems (14), including four differ-This work was supported in part by National Institutes of Health Grants AI125452 and AI091779 and a grant from Burroughs Wellcome Fund Investigators of the Pathogenesis of Infectious Diseases. The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This article contains supplemental Figs. S1-S4 and Tables S1-S3. The atomic coordinates and structure factors (codes 5W4M and 5W4N) have been deposited in the Protein Data Bank (http://wwpdb.org/). ent and conserved Rgg paralogs. Two of these paralogs, Rgg2 and Rgg3, share significant protein sequence similarity to one another but control gene expression by opposite mechanisms (Fig. 1). Rgg3 occupies a DNA site at target promoters and inhibits transcription until pheromone peptides directly bind to the receptor, causing Rgg3 to release DNA and de-repress transcription. Concurrently, Rgg2 requires binding to pheromone peptides to activate transcription. Together, Rgg2 and Rgg3 tightly control transcription of two identified promoters in response to small hydrophobic peptide pheromones, called SHP2 (DILIIVGG) and SHP3 (DIIIIVGG). Though differing at one position, in vitro and in vivo experiments using synthetically generated peptides (sSHP) indicate sSHP2 and sSHP3 have equivalent binding affinities for both Rgg2 and Rgg3, and thus are equivalent in their ability to induce transcription of Rgg2/3-dependent promoters. Stimulation of GAS with SHP2 or SHP3, whether obtained from GAS culture supernatants or synthetically derived, leads to the induction of several phenotypes observed in GAS wild-type strain NZ131: enhanced lysozyme resistance, cell aggregation, biofilm formation, and induced sensitivity to aminoglycoside antibiotics (15)(16)(17)(18)(19). The crystal structures of apo Rgg2 from Streptococcus dysgalactiae (Rgg2 Sd , 85% identical to Rgg2 of S. pyogenes, Rgg2 Sp ), and Rgg2 Sd bound to the competitive inhibitor cyclosporine A (CsA), were recently determined and provide the first available structures of an Rgg-type protein (20). Structural and biochem-ical analysis revealed that Rgg2 Sd forms a homodimer, with approximately 2-fold non-crystallographic symmetry. The structures indicated several residues likely to be important for SHP-Rgg2 Sd interaction, including Arg-153 in a putative peptide-binding pocket, here proposed to form electrostatic interactions with SHP necessary for full activation (20). By these structural findings, combined with previous in vivo and in vitro characterization of Rgg2, we propose that the protein possesses four primary structural features that define its function: a peptide-binding groove within the repeat domain establishing an allosteric regulatory site; an XRE family helix-turn-helix DBD; a dimerization interface that includes the Cys-45-Cys-45 cysteine pair allowing for intermolecular disulfide bond formation; and an undefined region that provides the ability to promote transcription, likely by recruiting RNA polymerase when Rgg2 is bound to SHP. These features can be interrogated by enlisting combinations of in vitro and in vivo methodologies, including direct assessments of peptide and DNA binding using fluorescent substrates, determination of the dimerization status of purified protein, observation of transcriptional activation using luciferase reporter assays, and the receptor conformations in X-ray crystal structures.
Here, we present X-ray crystallographic, biochemical, and genetic studies that reveal mechanistic insights into Rgg2 function. Prior inquiries of the impact that the disulfide bond provides to Rgg2 Sd were unclear (20), and therefore, to better study its contribution to the Rgg2 Sd structure, we determined X-ray crystal structures of Rgg2 Sd (C45S). To begin to understand how Rgg2 Sp conformational changes regulate its activity, we utilized a genetic selection approach to identify mutations in rgg2 conferring constitutive activity in vivo. These mutants, termed Rgg2 Sp *, were assessed for transcriptional activity in the absence of SHP, and a subset was further characterized and found to be more sensitive than wild-type Rgg2 Sp to both the SHP peptide as well as several SHP variants. In line with these findings, we found that Rgg2 Sp * mutants were less susceptible to inhibition by the competitive inhibitor cyclosporine A. In vitro studies revealed that the enhanced activity of Rgg2 Sp * mutants is not due to increased peptide pheromone-binding affinity; instead, the mutations appeared to alter the free-energy bias of Rgg2 Sp to favor transcriptional activation. Furthermore, results of site-directed mutational analysis of Arg-153 provides evidence that electrostatic interactions between this residue and the SHP pheromone are central in activation of Rgg2 Sp .

X-ray crystal structures of Rgg2 Sd (C45S)
X-ray crystal structures of apo Rgg2 Sd (C45S) were determined at 2.39 and 2.20 Å resolution in two crystal forms, named Rgg2(C45S) x1 and Rgg2(C45S) x2 , respectively ( Fig. 2 and supplemental Fig. S1 and Table S1). As described below and under "Experimental procedures", the Rgg2 Sd (C45S) structures were compared with the previously determined apo wild-type Rgg2 Sd structure (PDB code 4YV6). To identify any gross conformational changes in Rgg2 Sd (C45S) relative to wild-type Rgg2 Sd , the entire Rgg2 Sd (C45S) x1 or Rgg2 Sd (C45S) x2 dimer (both protomers A and B, simultaneously) were aligned to apo Figure 1. Rgg2/Rgg3 quorum-sensing system of S. pyogenes. Short hydrophobic peptide genes (shp2 and shp3) encode precursor peptides that are processed and secreted to generate mature SHP pheromones. Pheromones are actively transported to the cytoplasm by the oligopeptide permease transporter, Opp. SHPs are allosteric regulators of the Rgg receptor proteins. In the absence of SHP, transcription of shp promoters is repressed by Rgg3. The presence of SHP accomplishes two events: it causes Rgg3 to release DNA, allowing Rgg2 to bind at the promoter, and SHP stimulates Rgg2 to become a transcriptional activator. The conformational changes in Rgg2 and Rgg3 incurred by SHP pheromones are unknown.

Rgg2* mutant positions correlate with structural shift
wild-type Rgg2 Sd (root-mean-square deviation for pairwise comparisons of structurally aligned C␣ atoms ϭ 0.66 and 0.90 Å, respectively) (supplemental Fig. S1A). These alignments revealed that Rgg2 Sd (C45S) x1 closely matches the structure of wild-type Rgg2 Sd , whereas Rgg2 Sd (C45S) x2 has undergone significant conformational changes. Specifically, residues 170 -173, which were part of a loop (residues 170 -177) connecting wild-type Rgg2 Sd protomer B repeat 3 helix B (R3 B ) and repeat 4 helix A (R4 A ), have adopted ␣-helical structure, extending the Rgg2 Sd (C45S) x2 R3 B approximately one helical turn, and withdrawing what remains of the loop (residues 174 -177) from the dimer interface. Furthermore, protomer A residues 67-70, belonging to the flexible linker region connecting the DBD ␣-helix 5 and repeat 1 helix A (R1 A ), have adopted ␣-helical structure, extending the DBD ␣-helix 5 ϳ1 helical turn. We refer to this linker as flexible because in the wild-type Rgg2 Sd and Rgg2 Sd (C45S) x1 structures, the linker region is a loop or is otherwise disordered. It is also worthwhile to note that in both Rgg2 Sd (C45S) structures, the region containing repeats 1-3 (R1-R3, residues 71-173) in protomer A was shifted slightly relative to the equivalent region in the wild-type protein (as evidenced by the comparatively higher root-mean-square deviation for pairwise comparisons of structurally aligned C␣ atoms in this region) (supplemental Fig. S1A). In all likelihood, this region undergoes a conformational shift upon SHP binding to the concave SHP-binding surface of the repeat domain described previously (20).
To dissect how the C45S mutation enabled the abovedescribed conformational changes in Rgg2 Sd (C45S) x2 , the Rgg2 Sd (C45S) x2 and wild-type Rgg2 Sd structures were aligned using the modeled DBD residues (amino acids 4 -66) of a single protomer (protomer B). The protomer B DBD was selected as the basis for alignment because the secondary and tertiary structures of each are essentially identical (root-mean-square deviation for pairwise comparisons of structurally aligned C␣ atoms ϭ 0.51 Å). This structural comparison revealed that the Rgg2 Sd (C45S) x2 protomer A and repeat domain of protomer B have undergone a rigid-body rotation of 4° (Fig. 2). We propose that the Rgg2 Sd (C45S) x2 rigid body (tertiary structure) and secondary structure changes are enabled by structural flexibility otherwise absent upon formation of the Cys-45-Cys-45 intermolecular disulfide bond that connects the wild-type Rgg2 DBDs. It is important to note that because Rgg2 Sd (C45S) in the Rgg2 Sd (C45S) x1 crystal form adopts tertiary and secondary structure conformations similar to those of the wild-type Rgg2 Sd structure, it is not in fact the presence of the serine residue (the substitution for Cys-45) that enables the conformational change, but rather the lack of the intermolecular disulfide bond. We conclude that the Rgg2 Sd (C45S) x2 crystal form traps one of many conformations that Rgg2 Sd can adopt in the absence of the intermolecular disulfide bond.

Constitutively active mutations of Rgg2 Sp map to the dynamic loop of Rgg2 Sd (C45S) x2
As the Rgg2 Sd (C45S) mutant protein displayed a capacity to adopt a conformation under crystallization conditions not seen for the wild-type protein, we tested the possibility that this residue change would also confer differences in Rgg2 transcriptional activity. A test of this substitution had been conducted for Rgg2 Sd (20) but not for Rgg2 Sp . Thus, we constructed missense mutations at residue 45 of Rgg2 Sp to generate C45S and C45A and transferred these variants to S. pyogenes. The reporter strain used, RVW119 (⌬rgg2 shp2 GGG shp3 GGG ), contains a chromosomal deletion of rgg2, null mutations in coding sequences of shp2 and shp3 (and is therefore unable to produce SHP pheromones), and a chromosomally integrated P shp3 -lux reporter. Luminescence from cultures expressing the C45S, C45A, and wild-type Rgg2 Sp variants displayed similar responses to a range of SHP concentrations (Fig. 3A); thus, in this assay, the apparent conformational stability provided by the intermolecular disulfide bond in vitro did not affect transcriptional activity.
Instead, to begin to map regions of the Rgg2 structure that correlate with the protein's ability to modulate transcription, we adopted a genetic selection strategy to identify mutants of Rgg2 Sp capable of promoting transcription in the absence of an inducing SHP pheromone. We speculated that amino acid changes providing Rgg2 enhanced transcriptional activity may suggest regions of Rgg2 that undergo conformational changes during protein activation. The aphA3 gene, encoding resistance to kanamycin, was placed under the control of the P shp2 promoter to create plasmid pRVW31 (P shp2 -aphA3) as a scheme to select for resistant colonies in the absence of

Rgg2* mutant positions correlate with structural shift
SHP. Optimization of kanamycin concentrations needed to select for Rgg2 activation was conducted and led to the finding that 100 g/ml was sufficient (see under "Experimental procedures").
Plasmid pRVW31 was then integrated into the chromosome of a GAS NZ131 derivative that contains the deletions ⌬rgg2 ⌬rgg3 to generate strain RVW89. As expected, without rgg2 this strain was unable to grow on THY plates containing 100 g/ml kanamycin. pJC217, a plasmid capable of replication in GAS and containing rgg2 with its endogenous promoter, was treated with variable concentrations of ethyl methanesulfonate (EMS), a chemical agent that alkylates guanine bases, as a means to generate point mutations in the plasmid. EMS-treated pJC217 was transferred to RVW89 by electroporation, and kanamycinresistant mutants were selected on THY agar. THY medium inhibits SHP signaling because it supplies a high concentration of nonspecific nutrient peptides that compete with SHP pheromones for transport into the cytoplasm through the oligopeptide transporter Opp, rendering GAS non-responsive to SHP (17). Mutant selection was conducted on THY to enhance the likelihood that kanamycin-resistant clones would not arise because of unintended induced SHP signaling, but rather only from mutations in rgg2 that cause SHP-independent activation. Plasmids were recovered from 47 resistant clones, and the rgg2 gene was sequenced. Of these, 31 were found to have mutations in the coding sequence of rgg2 and 19 were of known unique Figure 3. A, Rgg2Sp(C45S) and Rgg2Sp(C45A) mutants display only minor changes in transcriptional activation in response to sSHP peptides. The maximum luminescence, normalized to cell density (RLUs), attained by NZ131 ⌬rgg2 shp2 GGG shp3 GGG P shp3 -luxABCDE cells (strain RVW119), complemented with indicated Rgg2 variants on a multicopy plasmid, are reported here as a function of sSHP2 concentration, ranging from 0 to 1,000 nM. B, location of Rgg2* mutations depicted as spheres on Rgg2 Sd (C45S) x2 protomers A (green) and B (blue). Rgg2 Sd (C45S) x2 regions of conformational flexibility in protomer A (residues 67-70) and protomer B (residues 170 -177) are colored magenta and yellow, respectively. C, mutated Rgg2* proteins support varying levels of constitutive Pshp3-lux activity. RLUs attained by RVW119 cells complemented with indicated Rgg2 variants on a multicopy plasmid are shown. Cells were treated with either 100 nM sSHP2 or equivalent concentration of reverse sequence peptide (revSHP). D, Rgg2* mutants demonstrate significant alterations in peptide sensitivity. RLUs attained by RVW119 cells complemented with indicated Rgg2 variants on a multicopy plasmid are reported here as a function of sSHP2 concentration, ranging from 0 to 1000 nM. Wild-type Rgg2 treated with equivalent concentrations of revSHP peptide is included as a control (black line). All data are representative of three biological replicates. Error bars show standard deviation. E, frequencies of rgg2* mutations. Total hits indicate the total number of times a given mutation was found. Unique hits indicate the total number of times the mutation was identified in a separately mutagenized pool.

Rgg2* mutant positions correlate with structural shift
provenance as they had arisen in separately mutagenized pools of plasmid (Fig. 3E). As these mutations tentatively conferred gain of function properties, we designate them rgg2*. Interestingly, a majority of the mutations (L163F, M167I, S174L, Q176K, and V184M) were located near the loop (residues 170 -177) identified as dynamic region in Rgg2 Sd (C45S) x2 (described above). The correlation of these mutations to a region where dynamic movements were observed suggested to us that a conformational movement of the loop is an important step for Rgg2 transcriptional activation.

Rgg2 Sp * mutants are sensitized to peptides in cultures
To ensure that constitutive activation of the Rgg2/3 quorumsensing pathway observed in our selections was due to observed changes to the amino acid sequence of Rgg2 Sp and not secondary mutations elsewhere on the plasmid, point mutants were regenerated in pJC217 by site-directed mutagenesis (except for Rgg2 Sp * V184M, see under "Experimental procedures"). The resulting plasmids, or a plasmid containing unaltered rgg2 Sp , were transferred into RVW119 (⌬rgg2 shp2 GGG shp3 GGG ), which contains a chromosomally integrated P shp3 -lux reporter. This rgg3 ϩ strain was chosen as an ideal strain to test the ability of Rgg2 Sp * mutants to function in the presence of Rgg3, the known negative regulatory counterpart to Rgg2. It was also imperative to conduct experimentation in a shp-null genetic background to uncouple exogenous pheromone responses from endogenous pheromone production (which is strongly influenced by positive-feedback regulation), and it was previously demonstrated that multicopy expression of wild-type rgg2 in a shp-replete background was sufficient to induce QS activity (17). Initially, the luciferase reporter activity of each strain was assessed under non-inducing conditions (i.e. without sSHP), but to control for general effects that addition of a similar peptide may have on cells, we added to the cultures a peptide comprising the reversed-sequence of SHP (revSHP), known not to stimulate the Rgg2/3 pathway (17). All Rgg2 Sp * variants displayed a basal level of bioluminescence that was greater than that generated by wild-type Rgg2 Sp (Fig. 3C). Despite there being only modest changes (ϳ2-3-fold increases) for many of the mutants (D60N, W140L, S174L, Q176K, V184M, and G232R), activity levels were statistically significant for each except D60N (p Ͻ 0.05, Student's t test). Two mutants, L163F and M167I, displayed substantially higher levels of luminescence than wild type in the absence of sSHP3. When treated with 100 M sSHP3, nearly all Rgg2 Sp * mutants responded by inducing transcription of luciferase to levels comparable with or higher than wild type, except W140L and G232R, which were induced less than 1.5-fold (p Ͻ 0.05, Student's t test). Overall, the panel of Rgg2 Sp * mutants provide multiple examples of Rgg2 Sp activation that will be used to elucidate mechanisms of regulation. For further analysis, we chose four mutants (D60N, L163F, S174L, and G232R) displaying distinct activities and locations within Rgg2 Sp .
Considering that the variability in response to a single concentration of sSHP3 was disparate among Rgg2 Sp * mutants, we generated dose-response curves with a titration of sSHP3 for each selected mutant (Fig. 3D). These results indicated that L163F and S174L were 10 -30 times more sensitive to sSHP3 than wild type, or in other words, less sSHP3 was required to reach equivalent luminescence levels observed in wild type. This was in contrast to D60N, which displayed a similar pattern of response to pheromone as the wild type. Previously, a collection of peptides designed to test the primary features of SHP were assessed for their ability to activate Rgg2 Sp (17). To summarize previous findings, the negative charge of aspartic acid at the N terminus was an important feature of SHP3, because changing this position to asparagine abrogated function, whereas a substituted glutamic acid remained active. Two other pheromone variants were synthesized to shorten the length by one residue from either end of the peptide, and each displayed poor or no activity at 50 nM (17). We titrated each peptide variant in cultures of Rgg2 Sp * mutants to test the possibility that selectivity for peptides might be altered (Fig. 4, A-E). Remarkably, D60N, L163F, and S174L each displayed increased responses to the peptide variants. L163F showed the greatest degree of sensitivity to all tested peptides, although the narrow dynamic range of luminescence activation made it difficult to precisely determine the concentration at which half-maximal induction (EC 50 ) occurred. S174L also displayed a substantial increase in sensitivity to the peptide variants as compared with wild-type Rgg2 Sp , and it even responded to revSHP, although only at the highest concentration tested, 1 M. G232R displayed limited responsiveness across the range of peptides and concentrations tested, and there was no appreciable increase over basal levels in RLUs, even at the highest concentrations of peptides tested. These results indicated that Rgg2 Sp * D60N, L163F, and S174L mutants were more amenable to activation, even by ligands that are not optimal inducers.
CsA is known to inhibit Rgg2 Sp by competing with SHPs for binding to the receptor and blocking transcriptional activation (15). We tested Rgg2 Sp * mutants D60N, L163F, S174L, and G232R for the ability of CsA to inhibit activation of luciferase expression. Dose-response curves of the four rgg2 Sp * alleles in strain RVW119 (⌬rgg2 shp2 GGG shp3 GGG ) stimulated with 50 nM sSHP2 were generated with titrations of CsA that ranged between 0.16 and 10 M (Fig. 3F). Rgg2 Sp * mutants D60N, L163F, and S174L displayed partial sensitivity to CsA but required higher concentrations than seen for wild type to be inhibited, and complete inhibition of transcription was not possible under the conditions employed. In concordance with the observation that G232R does not respond to sSHP2, CsA had no appreciable inhibitory effect on this mutant. To determine whether CsA inhibition worked only by competing with SHP pheromone for the binding pocket of Rgg2 or by also altering the conformation of our Rgg2 Sp * to resemble wild type more closely, we assessed transcriptional activity in the absence of sSHP2 of Rgg2 Sp * L163F, our most constitutively active mutant, following treatment with CsA. Using the same approach as described above, but in the absence of sSHP2 pheromone, CsA reduced constitutive Rgg2 Sp * L163F transcriptional activity in luciferase assays at higher concentrations (Ͼ2.5 M) (Fig. 4G). At 10 M, CsA reduced L163F activity to approximately that of wild-type Rgg2 Sp in the absence of peptide pheromone. These results suggest that not only does CsA compete with the SHP pheromones at the binding pocket, but as our previous struc-

Rgg2* mutant positions correlate with structural shift
tural studies suggest, it locks Rgg2 in an inactive (pheromonefree) conformation.

Rgg2 Sp * mutants do not have enhanced SHP affinity
To better understand the altered QS activity of our selected Rgg2 Sp * mutants, Rgg2 Sp wild type, D60N, L163F, S174L, and G232R were purified via affinity chromatography. As it had previously been shown not to interfere with in vitro assays, an N-terminal MBP tag used for purification purposes was retained during all ensuing experiments to prevent protein aggregation (18). We first sought to confirm that MBP-Rgg2 was functional in vivo and did not possess any unexpected Rgg2 Sp *-like activity, secondary to the expression of the MBP tag. We expressed the MBP-Rgg2 Sp fusion under the native P rgg2 promoter in multicopy on plasmid pMBP-Rgg2, and observed peptide responsiveness in RVW119 (⌬rgg2 shp2 GGG shp3 GGG ), which contains a chromosomally integrated P shp3lux reporter. Compared with wild-type Rgg2 Sp , the MBP-Rgg2 Sp construct is less responsive to peptide at all tested concentrations (supplemental Fig. S2). However, the construct is functional in vivo, and the MBP tag did not cause constitutive QS activation.
We conducted fluorescence polarization measurements using FITC-labeled sSHP2 to assess the peptide's affinity for MPB-Rgg2 Sp * proteins. Given the increased transcriptional activity we had observed in luminescence assays using Rgg2 Sp * variants, we anticipated that the proteins would bind peptide with greater affinity than wild-type MBP-Rgg2 Sp ; however, the A-E, wild-type Rgg2 and select Rgg2* variants show different levels of induction when treated with peptide variants. The maximum light levels, normalized to cell density (RLUs) attained by GAS NZ131 ⌬rgg2 shp2 GGG shp3 GGG P shp3 -luxABCDE (strain RVW119) cells complemented with indicated Rgg2 variants on a multicopy plasmid, are reported here for increasing concentrations of indicated peptides, ranging from 0 to 1000 nM. F, Rgg2* mutants demonstrate reduced sensitivity to the competitive inhibitor of Rgg-SHP interaction, CsA. The maximum light level, normalized to cell density (RLUs), attained by RVW119 cells complemented with indicated Rgg2 variants on a multicopy plasmid were treated with 50 nM sSHP2 and concentrations of CsA ranging from 10 to 0.16 M. Data were normalized based on maximum readings obtained in the absence of CsA. G, in the absence of sSHP2 peptide, Rgg2 Sp * constitutive activity can be reduced by CsA. Data were based on maximum RLU readings obtained at A 600 reading closest to 0.15. All data shown here are representative of at least three biological replicates. Error bars show standard deviation.

Rgg2* mutant positions correlate with structural shift
results did not support this hypothesis. MPB-Rgg2 Sp * D60N and S174L demonstrated a similar affinity for FITC-sSHP2, as seen in the overlapping polarization curves (Fig. 5A). Intriguingly, MBP-Rgg2 Sp * L163F and G232R failed to display any polarization. Because purification yields of MBP-Rgg2 Sp * L163F and G232R had been notably lower than other variants, we were concerned that these mutations may cause protein instability. Although thermal stability determined by microscale thermophoresis did not indicate any significant differences in unfolding-transition midpoint temperatures (T m ) or in aggregation onset temperatures (T agg ) (supplemental Fig. S3), circular dichroism (CD) spectra of all five purified MBP-Rgg/ Rgg2 Sp * proteins indicated that L163F and G232R were somewhat altered, suggesting conformational differences existed (supplemental Fig. S4). Though perhaps overly cautious, we continued studies only with D60N and S174L variants.
To explore the possibility that Rgg2* mutants may have decreased affinity for CsA when compared with wild-type Rgg2, we conducted competitive fluorescence anisotropy experiments to determine the apparent K d value for CsA in wild-type Rgg2 versus the S174L mutant (Fig. 5B). Although both wildtype Rgg2 Sp and Rgg2 Sp * S174L demonstrated equivalent affinities for sSHP2 in direct binding experiments, in light of the decreased sensitivity for CsA inhibition in vivo by the Rgg2 Sp * mutants, we wondered whether changes to the conformation of Rgg2 Sp * mutants might alter affinity for CsA. When compared with previously published results (21), the calculated K d value for wild-type Rgg2 Sp was equivalent (0.53 M, 95% confidence interval, 0.25-1.08 M); however, CsA's affinity for Rgg2 Sp * S174L was increased to 2.15 M (95% confidence interval, 1.46 -3.16 M). These results suggest that the conformational changes incurred by the S174L mutation decreased the affinity of CsA for Rgg2 Sp *.
As neither MBP-Rgg2 Sp * D60N nor S174L showed altered SHP-binding capacity in vitro when compared with wild-type MBP-Rgg2 Sp , we considered the possibility that the point mutations might enhance the DNA-binding affinity of the protein, although prior studies have not indicated that increased access to the promoter would result in constitutive transcriptional activity (18). Nevertheless, to test this possibility, we conducted electrophoretic mobility shift assays (EMSA) using fluorescently labeled double-stranded DNA containing the minimum Rgg2/3-binding sequence (18). As shown previously, titration of wild-type MBP-Rgg2 Sp between 0 and 400 nM led to a complete shift of the probe, indicating in vitro MBP-Rgg2 Sp -DNA binding, even in the absence of SHP peptide (Fig. 5C). Figure 5. A, MBP-Rgg2 Sp , D60N, and S174L bind peptide with equivalent affinity. Direct fluorescence polarization was conducted utilizing FITC-labeled sSHP2. Wild-type Rgg2 Sp , D60N, and S174L all displayed equivalent affinity for labeled peptide resulting in overlaying binding curves. MBP-Rgg2* L163F and G232R failed to display any significant degree of polarization. Data are representative of three technical replicates; error bars show standard deviation. B, CsA has lower affinity for Rgg2 Sp * S174L. Competitive fluorescence polarization was conducted utilizing FITC-labeled sSHP2. Indicated MBP-Rgg2 species were incubated with FITC-labeled sSHP2 for 10 min prior to the addition of increasing concentrations of CsA. The reaction proceeded for 20 min before assessment. Data are representative of two technical replicates. C-E, MBP-Rgg2* D60N and S174L DNA binding are limited in the absence of sSHP peptide. Electromobility shift assays were conducted using 20 nM fluorescently labeled DNA and with 0, 50, 100, 200, or 400 nM MBP-Rgg2 proteins of the indicated species. 400 nM wild-type MBP-Rgg2 is included in all gels, indicated by (ϩ). No protein control is indicated by Ϫ.

Rgg2* mutant positions correlate with structural shift
However, the DNA migration shift was substantially decreased or not apparent at all for D60N or S174L at concentrations up to 400 nM Rgg2 Sp *, and even at 1,600 nM Rgg2 Sp *, the shift remained incomplete (Fig. 5, D and E, and data not shown). These paradoxical findings, which indicated a decreased ability to bind DNA, led us to suspect that the mutations were affecting either proteins' conformational stability (i.e. shifting their conformational equilibrium) or their propensity to bind DNA in vitro. We hypothesized that if this were true, then binding peptide ligands could stabilize the receptors; therefore, the EMSA was repeated in the presence of either the sSHP2 active peptide or the reverse-peptide (revSHP) at 2 M concentrations. The addition of the sSHP2 peptide increased the DNA-binding affinity of both MBP-Rgg2 Sp * D60N and S174L. Notably, MBP-Rgg2 Sp * S174L, when sSHP2 was provided, demonstrated a shift that was similar to wild-type MBP-Rgg2 Sp across all concentrations of protein. As hypothesized, the addition of revSHP or the vehicle control (DMSO) failed to improve the Rgg2 Sp *-DNA interaction, consistent with the idea that peptide pheromone can bind to the MBP-Rgg2 Sp * mutants and stabilize a conformation that binds tightly to DNA with a similar affinity as wild-type Rgg2. As discussed below, whereas peptides can enhance Rgg2 Sp * binding to DNA in vitro, they are not necessarily the key factor driving the binding of Rgg2 Sp * mutants with DNA in vivo.

Rgg2 residue Arg-153 is essential for SHP responsiveness
In previous structure-function studies to identify the putative Rgg2 peptide-binding pocket, it was demonstrated that Arg-153 was essential for S. dysgalactiae Rgg2 (Rgg2 Sd ) to respond to pheromone in a GAS test-bed. Furthermore, the Rgg2 Sd -CsA crystal structure showed that the competitive inhibitor of SHP binding, CsA, can H-bond with Rgg2 Sd -Arg-153 (20). These findings are consistent with other works relating to pheromone-binding pockets of RRNPP-type regulators, many of which rely upon polar and electrostatic interactions to coordinate their peptide ligands (22). We therefore tested the formal hypothesis that Rgg2 Sp Arg-153 forms a salt bridge with the aspartic acid of the SHP peptides by substituting residues in Rgg2 Sp and in SHP peptides. First, we generated R153A and R153E mutants in Rgg2 Sp , changing the large, positively charged arginine residue to a small, neutral alanine, and to the oppositely charged glutamate, respectively. Transcriptional activity was then assessed in a ⌬rgg2 shp2 GGG shp3 GGG P shp3 -lux background. When Rgg2 Sp (R153A) was treated with wild-type synthetic pheromone (sSHP3(WT)), an ϳ15-fold reduction in maximum transcriptional activity compared with wild-type Rgg2 Sp was observed (Fig. 6A); this result matched well with the previously reported equivalent mutation in Rgg2 Sd (20). Stimulation of Rgg2 Sp (R153E) with sSHP3(WT) saw a further decrease in transcriptional response, reduced by 100-fold compared with the response seen with wild-type Rgg2 Sp (Fig. 6A). Next, we predicted that if sSHP3 variants with altered C-terminal residues (sSHP3(D16N) and sSHP3(D16R)) were used to stimulate the Rgg2 Sp alleles, patterns of transcriptional activation would correlate with the expected electrostatic interactions between Rgg2 Sp residue 153 and SHP ligands. In other words, we anticipated that if residue 153 provides an interaction with SHP3, then attractive electrostatic interactions would induce an enhanced transcriptional response, and repulsive interactions would have decreased responses. Luciferase activity of each Rgg2 Sp allele was measured following stimulation with the sSHP3 variants and compared with the responses stimulated by sSHP3(WT) (Fig. 6B). For wild-type Rgg2 Sp , activity induced by sSHP3(D16N) and sSHP3(D16R) was comparatively 4-and 60-fold less, respectively, than seen with sSHP3(WT). Rgg2 Sp (R153A) displayed equivalent responses to both sSHP3(WT) and sSHP3(D16N), whereas sSHP3(D16R) treatment was 2.9-fold reduced. Finally, Rgg2 Sp (R153E) showed a 6-and 3-fold increase in activity over sSHP3(WT) when treated with sSHP3(D16N) and sSHP3(D16R), respectively. Thus, electrostatic interaction between Rgg2 Sp position 153 and the N terminus of the mature Figure 6. Charge interaction between Rgg2 residue Arg-153 and the N-terminal amino acid of the SHP peptide helps mediate Rgg2 transcriptional activation. A, wild-type Rgg2, Rgg2 R153A, and Rgg2 R153E were treated with 100 nM sSHP2 in GAS NZ131 ⌬rgg2 shp2 GGG shp3 GGG P shp3 -luxABCDE complemented with indicated rgg2 variants on a multicopy plasmid. The maximum light units relative to growth (RLUs) is reported here. B, maximum relative light activity attained by the indicated Rgg2 mutants following treatment with indicated sSHP peptide variant. D16N and D16R were normalized to the same Rgg2 mutant treated with 100 nM sSHP2. All data shown here are representative of at least three biological replicates.

Rgg2* mutant positions correlate with structural shift
SHP provides an important contribution to peptide recognition and allosteric regulation.

Discussion
A model for Rgg2 activation draws upon prior work investigating the Rgg2 Sp and Rgg3 Sp regulatory system in Group A Streptococcus (Fig. 1). A peculiar feature of this system is the dual use of both a transcriptional activator (Rgg2) and a repressor (Rgg3) to bind the same DNA sequences, control the same promoters, and respond similarly to two similar pheromones (SHP2 and SHP3) (17)(18)(19). Previous studies showed that recombinant Rgg2 Sd forms homodimers in solution, as well as when in complex with the inhibitory molecule CsA (20). Because Rgg2 Sp affinity for DNA is not influenced by pheromone binding (18), we propose that SHP binding triggers an Rgg2 Sp conformational change enabling a productive interaction with RNA polymerase. The proteins show a high degree of similarity and synteny with one another (52% identical, 72% similar). Of the eight residue changes accounting for Rgg2* mutants, seven are conserved in Rgg3 (with the exception of Gln-176). How these proteins, which in all likelihood have a similar three-dimensional structure, control gene expression by opposite mechanisms remains an outstanding question, driving our ongoing studies to understand the fundamental allosteric properties of RRNPP proteins. Given what is known about other members of the RRNPP family of regulators, it is likely that peptide binding causes an allosteric conformational change in Rgg2 Sp at sites relatively distant from the pheromone-binding pocket. However, we stress that in the absence of evidence showing SHP-mediated Rgg2 Sp conformational changes or Rgg2 Sp interactions with RNA polymerase, these conclusions are inferred but not proven. The disproportionally high frequency of Rgg2 Sp * mutations occurring between residues 163 and 184, and the proximity of these mutations to both the region of conformational flexibility (residues 170 -177) and the pheromone-binding pocket (Fig. 2) (20), led us to hypothesize that the 170 -177 region mediates conformational changes that transmit pheromone-binding signal input to polymeraseactivating signal output.
How might Rgg2* mutations function to drive constitutive receptor activation? In the RRNPP family member PlcR, binding of the ligand PapR results in a conformation change in a helix that triggers a large reorientation of the PlcR DNA-binding domains, allowing for DNA binding and transcriptional activation (12). But unlike PlcR, Rgg2 binds promoter DNA even in the absence of bound pheromone. In fact, increased occupancy of the promoter by Rgg2 is not sufficient to induce transcription since prior work showed that Rgg2 requires pheromones for activation, even in the absence of the Rgg3 repressor (19). Therefore, even if Rgg2 Sp * mutants are more able than wild type to outcompete Rgg3 for promoter sites, it would not explain their increased transcriptional activity (17,19). Thus, Rgg2 Sp * mutations must incorporate, at the very least, an enhanced ability to recruit RNA polymerase and/or stimulate initiation of transcription. Surprisingly, Rgg2 Sp * D60N and S174L mutants displayed a decreased affinity for DNA, at least in vitro. Despite a reduced ability to bind DNA, these mutants were selected by having enhanced basal transcriptional activity, conferring enhanced kanamycin resistance in the absence of pheromone in our selection system and, independently, constitutive luminescence activity in transcriptional reporters. Interestingly, addition of sSHP to recombinant Rgg2 Sp * D60N and S174L in vitro restored DNA-binding ability to levels seen for wild-type Rgg2 Sp . It is possible that the D60N and S174L mutations affect Rgg2 Sp by driving their conformational equilibrium away from a DNA-binding proficient conformation but toward one more amenable to pheromone-and/or polymerasetriggered activation and DNA binding. This concept is expanded on below. Regardless of the mechanism, the negative effects imposed by mutations on DNA binding must be compensated by other positive attributes that result in a net increased ability to induce transcription, likely through positive interactions with RNA polymerase.
Rgg2 Sp response to SHP pheromone was disparate among the Rgg2 Sp * mutants. Two of the Rgg2 Sp * mutants, L163F and G232R, were refractory to pheromone binding. Although L163F was more responsive to peptides and reached higher activation levels, G232R remained primarily in a low, constitutive state. In contrast, S174L displayed an enhanced sensitivity to sSHP2 peptide (i.e. the EC 50 values for sSHP2 were lower than that seen for wild-type Rgg2). Rgg2(L163F), Rgg2(S174L), and Rgg2(D60N) even responded to SHP derivatives that were previously shown to be poor inducers of wild-type Rgg2 Sp (Fig.  4) (17). However, surprisingly, affinity of the Rgg2 Sp * mutants for peptides was not enhanced in vitro, as determined in fluorescence anisotropy studies (Fig. 5). How can the paradox of mutants displaying enhanced sensitivity to pheromones be reconciled with the observation that they present equivalent affinities for the ligands? As mentioned above, we propose that the Rgg2 Sp * mutations have an underlying effect of shifting the free energy of Rgg2 Sp toward a state amenable to SHP-or polymerase-dependent activation. Generation of Rgg2 Sp * mutations is reminiscent of LuxN mutants displaying altered sensitivities to autoinducer molecules (pheromones) of the AI-1 quorumsensing system of Vibrio harveyi (23). In these studies, mutations in LuxN conferred measurable shifts in the biases of the receptor's basal kinase and phosphatase enzymatic activities (23). In the case of Rgg2, the protein does not have kinase/ phosphatase activity like LuxN, but instead it functions as a transcriptional activator. LuxN* mutants (having a free-energy shift toward the phosphatase-ON state) displayed enhanced sensitivity toward the pheromone, accounting for lower EC 50 values, just as we describe for Rgg2 Sp * mutations. Interestingly, three of the four LuxN* mutations were located outside of the putative AI-1-binding site of LuxN, and when accounting for the free-energy shift of the enzyme, their affinities for the pheromone were unchanged.
We feel it is important to point out that although an antagonist compound of LuxN also retained equivalent IC 50 values for the LuxN* alleles in the cited report (23), in the case of Rgg2 Sp * mutants, we observed diminished IC 50 values for CsA as compared with its effect on wild-type Rgg2 Sp (Fig. 5). The binding affinity for CsA was also modestly reduced, at least for the tested mutant S174L (Fig. 5B). It remains unclear why affinities for pheromone were unaltered in the Rgg2* mutants but were reduced for CsA. It is possible that CsA's size (nearly twice the Rgg2* mutant positions correlate with structural shift size of SHP) or relative rigidity (CsA is cyclical although SHPs are linear) may hinder interactions with the binding pocket if subtle changes in pocket shape exist in Rgg2* mutants that do not interfere with SHP binding. Most importantly, it was interesting to find that CsA could inhibit the basal activity of Rgg2 Sp * L163F in the absence of pheromone, suggesting that CsA stabilizes Rgg2 in an inactive conformation. This could have important implications for developing Rgg2 inhibitors that function in the presence of constitutively active Rgg2 mutations. If CsA or its derivatives are pursued as antivirulence lead compounds against Rgg2/3-containing streptococci, selection of Rgg2* alleles should be taken into consideration. Debate has emerged whether targeting quorum-sensing pathways by small-molecule antagonists offers a viable alternative to antibiotic therapeutics (25)(26)(27)(28)(29). Receptor mutants (i.e. Rgg2 Sp *) identified in this report obviate the potential of natural mutations that give rise to altered protein activities that resist inhibitor action. However, whether such mutants would confer extended fitness to the organism will depend on the role the regulatory system delivers in an organism's life cycle and the advantages that social regulation of behavior provides.
To summarize our findings, we have generated a model that integrates structural and genetic information pertaining to Rgg2 (Fig. 7). The results herein describe regions of Rgg2 conformational flexibility, which may indicate movements that initiate or result in activation of the protein. Because culturing conditions did not indicate transcriptional activation by the Rgg2 Sp (C45S) mutation, we employed a genetic screen that identified active Rgg2 Sp * mutants. Several of these mutants' locations in Rgg2 Sp correspond to the same regions seen to move in the Rgg2 Sd (C45S) x2 crystal structure. We suggest these spatial correlations provide supportive evidence to the concept that the observed intramolecular movements are related to protein activation. Rgg2 Sp * mutants display altered ranges of activity in response to pheromone, but not altered affinity toward SHP, indicating a shift in their free-energy bias.
Ideally, Rgg proteins and the RRNPP family could stand as new therapeutic targets whose allosteric misregulation would effectively modulate bacterial behavior. The ability to predict how an RRNPP regulator responds to a pheromone or to a modulatory compound would be a powerful tool for antivirulence or biotechnology strategies that would benefit greatly by manipulating bacterial gene expression through use of chemical signals.

Bacterial strains and culture conditions
Strains used in this study are described in supplemental  Table S1. S. pyogenes was grown in Todd Hewitt medium (BD Biosciences), supplemented with 0.2% (w/v) yeast extract (Amresco). Solid media contained 1.4% agar. All luciferase experiments were conducted in a chemically defined medium containing 1% glucose (w/v) (17,30). Antibiotics were used at the following concentrations: chloramphenicol, 3 g/ml; erythromycin, 0.5 g/ml; kanamycin, 100 g/ml; spectinomycin, 100 g/ml. Cloning was conducted in Escherichia coli strain BH10C. E. coli XL10-Gold cells (Agilent Technologies) were used for the generation of Rgg2 point mutations. BL21(DE3) cells were used for protein expression. E. coli cells were cultivated in Luria-Bertani medium (BD Biosciences), supplemented with the following antibiotics as appropriate: ampicillin, 100 g/ml; chloramphenicol, 10 g/ml; erythromycin, 500 g/ml; kanamycin, 100 g/ml; spectinomycin, 100 g/ml. Peptide pheromones were purchased from Neo Scientific (Woburn, MA). Lyophilized peptides were dissolved in DMSO to a stock concentration of 2 mM based on the preparation's purity and lyophilized weight.

Genetic selection design
Plasmid pRVW38 (P shp2 -aphA3) was generated from pSAR56 (P shp2 -luxAB) (17) by digesting pSAR56 with NdeI and EcoRI. aphA3 was amplified from pOsKaR using primers RW110 and RW111 (31). The plasmid was assembled using the NEBuilder HiFi assembly kit (New England Bioscience) prior to Figure 7. Summary of transcriptional regulation by Rgg2 and Rgg3 pheromone receptors. A, Rgg2 and Rgg3 compete for DNA sites (orange) at P shp promoters. In the absence of SHP, Rgg3 repression dominates the promoter. At or above threshold SHP concentrations, SHPs bind to Rgg2 and Rgg3. This results in a conformational change in Rgg3 that causes it to release DNA, which allows Rgg2 to access the promoter. SHP binding to Rgg2 causes a conformational change that results in positive transcriptional activity (possibly by generating positive interactions with RNA polymerase). B, Rgg2 Sd (C45S) mutants can adopt a conformation in which secondary structures of the repeat domain undergo a position shift compared with the wildtype protein that contains an intermolecular disulfide bond. We propose this conformational change is enabled by structural flexibility otherwise absent, whereas the Cys-45-Cys-45 disulfide bond is intact. C, schematic representation of Rgg2 Sp allele activation states. Horizontal lines represent the spectrum of transcriptional activation states in the absence (yellow points) and presence (green points) of SHP pheromones. CsA inhibits Rgg2 Sp activation by interfering with SHP binding. Rgg2 Sp (C45S) had only modest effects on transcription under tested growth conditions. Rgg2 Sp * mutants (e.g. L163F, G232R, S174L) have elevated transcription activity in the absence of SHP.

Rgg2* mutant positions correlate with structural shift
transformation into electrocompetent E. coli strain BH10C. The plasmid was verified by sequencing. To validate the selection system, pRVW31 was integrated into the genome of wild-type GAS NZ131 and cultured in the presence or absence of a synthetic SHP peptide (sSHP3), and bacteria were challenged with a range of concentrations of kanamycin. As predicted, cells treated with sSHP3 were able to divide and grow at normal rates to high density at concentrations of kanamycin up to 1,000 g/ml, whereas cells treated only with the solvent used to apply pheromones (DMSO vehicle) failed to grow in concentrations at or above 100 g/ml kanamycin. pRVW31 was subsequently transferred to RVW89 (NZ131 ⌬rgg2 ⌬rgg3 P shp2 -aphA3) for use in the selection. As the selection was conducted on THY medium, a medium where Rgg2/3 QS is non-functional, this strain was effectively rendered SHP-non-responsive as well. The plasmid pJC217, encoding rgg2 under its own promoter, was treated with EMS at various concentrations (ranging from 1 l of neat EMS/g of plasmid to 1 l of a 1:10 dilution of EMS/g of plasmid). Following sodium thiosulfate treatment to neutralize the EMS, 100 ng of plasmid was introduced into RVW89 via electroporation and plated on THY agar with kanamycin 100 g/ml and spectinomycin 100 g/ml. Isolated mutants were outgrown in THY broth under selection, and genomic DNA was obtained. Primers specific to pJC217, JC180, and JC424 were used to amplify rgg2, and samples were sequenced via Sanger sequencing using the reverse primer JC237.

Mutant rgg2 strain generation
Mutations of interest, with the exception of Rgg2 V184M, were regenerated de novo in pJC217 using a QuickChange Lightning site-directed mutagenesis kit (Agilent) and verified by sequencing. Rgg2 V184M was unable to be obtained via sitedirected mutagenesis. Rather, the rgg2 coding sequence from an EMS-treated sample was amplified by primers RW214 and RW215 (supplemental Table S3) and cloned into pLZ12 digested with EcoRI and BglII using Gibson Assembly with NEBuilder HiFi master mix (New England Bioscience). Plasmids bearing mutant rgg2 were transformed into a P shp3 luciferase reporter strain lacking rgg2, shp2, or shp3, RVW119 (⌬rgg2 shp2 GGG shp3 GGG P shp3 -luxABCDE).

Additional cloning
pMBP-Rgg2 was generated by amplifying the MBP-Rgg2 fusion from pCA104 using primers RW277 and RW288 (18). The pLZ12-sp backbone containing the Prgg2 fragment was generated by inverse PCR from pJC217 using primers RW287 and RW290. The plasmid was assembled with NEBuilder HiFi master mix (New England Bioscience) and verified by colony PCR. All primers are described in supplemental Table S2.

Luminescence assays
Luminescence assays were conducted as described previously (32). In brief, overnight cultures of GAS were diluted 1:100 into chemically defined medium containing appropriate antibiotics. Cells were grown at 37°C until reaching an A 600 of ϳ0.1. 100 l of bacteria were transferred to each well of 96-well plates, and indicated sSHP peptides or inhibitors were added at appropriate concentrations. Plates were incubated at 37°C in a BioTek Synergy 2 plate reader (Winooski, VT); growth was assessed by absorbance at 600 nm, and luminescence was recorded as counts/s by the luminometer. Relative light units (RLUs) were calculated by dividing luminescence readings by A 600 values.

Rgg2* purification
Nucleotide changes to the rgg2-coding sequence were made to the maltose-binding protein-rgg2 expression plasmid pCA104 (MBP-Rgg2), described previously (18) using Quick-Change Lightning site-directed mutagenesis (Agilent) as described previously. Following outgrowth, mutations were verified by Sanger sequencing. Plasmids encoding wild-type Rgg2 or Rgg2* mutations (D60N, L163F, S174L, or G232R) were introduced into BL21(DE3) cells by electroporation and selected on ampicillin-containing LB agar. Protein purification proceeded as described previously (18,21). In brief, 500-ml cultures of BL21(DE3) cells were grown to early log (A 600 ϭ 0.5) and induced with 0.5 mM IPTG for 6 h at 25°C. Cells were pelleted and lysed via sonication in Buffer A (20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM EDTA, 10 mM ␤-mercaptoethanol) with 1ϫ EDTA-free protease inhibitor (Thermo Fisher Scientific) and 1 mg/ml lysozyme (Sigma). Supernatants were clarified by centrifugation and applied to an amylose column (GE Healthcare). The column was washed 12 times with Buffer A, and Rgg2 was eluted in 3 column volumes of Buffer B (Buffer A ϩ 10 mM maltose). Fractions containing purified protein (determined by SDS-PAGE to be Ͼ95% by eye) were pooled; Buffer B was exchanged for Buffer A, and proteins were concentrated via a 10-kDa MWCO Amicon spin column (Millipore). Protein concentration was measured on a NanoDrop 1000 (Thermo Fisher Scientific) and calculated using a predicted extinction coefficient for absorbance at 280 nm. Proteins were stored in Buffer A ϩ 20% glycerol at Ϫ80°C until use.

Rgg2 Sd (C45S) production for crystallization
S. dysgalactiae rgg2(C45S) was cloned in pTB146 as described previously (20). His-Sumo-Rgg2(C45S) was overexpressed in E. coli strain BL21(DE3) by first growing the cells at 37°C in LB medium containing 100 g/ml ampicillin to A 600 ϭ 0.5 and then inducing expression with 0.5 mM IPTG for 16 h at 16°C. The cells were collected by centrifugation and lysed in Buffer C (20 mM Tris-HCl (pH 8.0), 400 mM NaCl, 10% glycerol) supplemented with 20 g/ml DNase. Lysate supernatant was applied to His-60 nickel resin (Clontech) equilibrated in Buffer C. His-Sumo-Rgg2(C45S) was eluted by washing the column with increasing amounts of imidazole and analyzed for purity using SDS-PAGE. Eluted protein was combined with 1.25 mg of the SUMO protease Ulp1 and dialyzed against 2 liters of Buffer D (20 mM sodium phosphate buffer (pH 8.0), 150 mM NaCl, 10 mM ␤-mercaptoethanol, and 0.1% Triton X-100) overnight at 25°C. The next day, the protein was passed over His-60 nickel resin to separate the cleaved tag from the Rgg2(C45S) protein.

Crystallization, X-ray diffraction data collection, and structural alignments
Rgg2 Sd (C45S) x1 crystals were obtained by the vapor diffusion method at 20°C with 2-l hanging drops of 4.0 mg/ml Rgg2 Sd (C45S) mixed 1:1 with mother liquor containing 100 mM Bistris propane (pH 7.75), 350 mM KSCN, and 18% PEG 3350. Rgg2 Sd (C45S) x2 crystals were obtained by the vapor diffusion method at 20°C with 0.4-l sitting drops of 90 M Rgg2 Sd (C45S) and 480 M sSHP2 mixed 1:1 with mother liquor containing 170 mM NH 4 OAc, 85 mM sodium citrate (pH 5.6), 25.5% PEG 4,000, and 15% glycerol. Prior to X-ray diffraction data collection, the Rgg2 Sd (C45S) x1 crystal was moved to a solution of the mother liquor supplemented with 10% glycerol and immediately flash-cooled in liquid nitrogen. X-ray diffraction data for Rgg2 Sd (C45S) x1 were collected using single crystals mounted in nylon loops that were then flash-cooled in liquid nitrogen before data collection in a stream of dry N 2 at 100 K. X-ray diffraction data were collected at the Stanford Synchrotron Radiation Lightsource (SSRL) beamline 14-1 at 1.1808 Å with a MARmosaic 325 CCD detector. X-ray diffraction data for Rgg2 Sd (C45S) x2 were collected in an identical manner using our in-house Raxis IV ϩϩ image plate detector at 1.5408 Å. X-ray data for both crystals were processed using HKL2000 (33). Initial crystallographic phases were determined by molecular replacement using Phaser (34), and the previously determined structure of apo Rgg2 Sd (PDB code 4YV6) as a search model (20). The final models were generated using iterative cycles of model building in Coot (35) and refinement in phenix.refine (36). Initial refinement included simulated annealing as well as rigid body, individual atomic coordinate, and individual B-factor refinement. Later rounds of refinement employed individual atomic coordinate, individual B-factor, and TLS refinement. TLS groups were selected using the TLSMD server (37). During the final rounds of refinement, the stereochemistry and ADP weights were optimized. Insufficient electron density was observed for the following residues in flexible regions of the protein structures, and they were omitted from the model: Rgg2 Sd (C45S) x1 residues 1-2 and 280 -284; Rgg2 Sd (C45S) x2 1 and 275-284. Ramachandran statistics were calculated in Molprobity (38). Molecular graphics were produced with PyMOL (PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC).
Pairwise alignment of the wild-type Rgg2 Sd and Rgg2 Sd (C45S) structures revealed a subtle asymmetry that exists within Rgg2 Sd dimers, i.e. the conformational difference between Rgg2 protomers A and B. The asymmetry reflects the conformation that the DBDs would adopt to bind the non-palindromic rgg2-box DNA, the flexible linker that connects the DBDs to the repeat domains, and the different crystal forms (supplemental Fig. S1B). Through pairwise structural alignment, the protomers in Rgg2 Sd (C45S) x1 and Rgg2 Sd (C45S) x2 that most closely fit the wild-type Rgg2 Sd protomer A or B were identified, and these were consistently paired in all of the structural comparisons. Sequence-independent structural superpositions and were carried out using PyMOL. Analysis of protein domain motion was carried out using DynDom (24) and PyMOL.

Fluorescence polarization assays
Direct fluorescence polarization (FP) assays were conducted as described previously (15). In brief, purified MBP-Rgg2 (or indicated mutant variant Rgg2*) was serially diluted in FP Buffer (Buffer A supplemented with 0.01% Triton and 0.1 mg/ml BSA). A master mix of FITC-sSHP2 was prepared in FP Buffer and aliquoted into a black, 1 ⁄ 2 area 96-well plate (Corning). Diluted MBP-Rgg2 was added to each well, and allowed to equilibrate for 30 min at 25°C in the dark. The protocol was modified to assess CsA competition as follows: 230 nM MBP-Rgg2 or MBP-Rgg2 S174L was incubated in FP Buffer with 10 nM FITC-sSHP2 for 10 min. CsA was added to the indicated concentration and allowed to equilibrate for an additional 20 min at 25°C in the dark. Plates were assessed on a BioTek Synergy 2 plate reader, and in-system software was used to calculate polarization values.

Electrophoretic mobility shift assays
EMSAs were conducted as described in Chang et al. (17). A master mix of EMSA Buffer (20 mM HEPES (pH 7.9), 100 mM KCl, 12.5 mM MgCl 2 , 0.2 mM EDTA (pH 8.0); 0.5 mM dithiothreitol, 50 mg/ml sheared salmon sperm DNA, 0.001 units/l poly(dI-dC), 100 mg/ml bovine serum albumin, 0.5 mM CaCl 2 , 12% (v/v) glycerol) with 20 nM double-stranded FITC-P shp2 probe (CCATTTTTCCCACTTTCACAACAA) was prepared and aliquoted into a 96-well plate. Purified Rgg2/Rgg2* proteins were diluted in EMSA Buffer and added to the probe master mix to the indicated final concentration. sSHP2, revSHP, or DMSO was also added to reactions at that time. The reaction was allowed to proceed at 25°C for 30 min in the dark. The resulting product was loaded and run on a 5% native-PAGE gel, buffered with 20 mM phosphate. The resulting gels were imaged on a Typhoon Trio imaging system (GE Healthcare).

Circular dichroism
Circular dichroism (CD) spectra were collected on a JASCO J-810 (Easton, MD) system using a 2-cm cuvette. Samples were diluted in CD Buffer (20 mM potassium phosphate (pH 7.6), 100 mM NaF) to a final concentration of 5 M. Scans were conducted between 260 and 190 nM, using 0.5 nM increments.

Protein thermostability assays
To assess the thermal stability of MBP-Rgg2, purified proteins were subjected to heat denaturation using the Prometheus NT.48 nanoDSF (NanoTemper Technologies). Samples were subjected to a temperature ramp of 1.0°C/min from 20 to 95°C, and fluorescence was constantly monitored. Data were analyzed with the PR.ThermControl software.