Crystal structure of a complex between the phosphorelay protein YPD1 and the response regulator domain of SLN1 bound to a phosphoryl analog

Summary

The crystal structure of the yeast SLN1 response regulator domain bound to both a phosphoryl analog (BeF₃⁻) and Mg²⁺ ion in complex with its downstream phosphorelay signaling partner YPD1 has been determined at a resolution of 1.70 Å. Comparisons between the beryllium fluoride-activated complex and the unliganded (or apo) complex determined previously reveal modest but important differences. The SLN1-R1•Mg²⁺•BeF₃⁻ structure from the complex provides evidence for the first time that the mechanism of phosphorylation-induced activation is highly conserved between bacterial response regulator domains and this example from a eukaryotic organism. Residues in and around the active site undergo slight rearrangements in order to form bonds to the essential divalent cation and fluorine atoms of BeF₃⁻. Two conserved switch-like residues (Thr 1173 and Phe 1192) occupy distinctly different positions in the apo- versus BeF₃⁻-bound structures consistent with the “Y-T” coupling mechanism proposed for activation of CheY and other bacterial response regulators. Several loop regions and the α4-β5-α5 surface of the SLN1-R1 domain undergo subtle conformational changes (∼1-3 Å displacements relative to the apo-structure) that lead to significant changes in terms of contacts that are formed with YPD1. Detailed structural comparisons of protein-protein interactions in the apo- and BeF₃⁻-bound complexes suggest at least a two-state equilibrium model for formation of a transient encounter complex, in which phosphorylation of the response regulator promotes the formation of a phosphotransfer-competent complex. In the BeF₃⁻-activated complex, the position of His 64 from YPD1 is within ideal distance and near linear geometry with Asp 1144 from the SLN1-R1 domain for phosphotransfer to occur. The ground state structure presented here suggests that phosphoryl transfer will likely proceed through an associative mechanism involving formation of a pentacoordinate phosphorus intermediate.

Introduction

In two-component regulatory systems¹, phosphoryl transfer between a sensor histidine kinase (HK) and its cognate response regulator (RR) is the chemical basis for signal transduction. Multi-step His-Asp phosphorelay signaling systems have evolved from the simpler two-component systems and are typically composed of a hybrid HK containing a C-terminal RR domain that transmits signals to a histidine-containing phosphotransfer (HPt) protein, which then relays phosphoryl groups to downstream RRs^1-3. Phosphorylation of the cytoplasmic RRs results in activation of either associated effector domains or downstream signaling partners.

While structures of RR proteins (or domains) in their unphosphorylated inactive states have been well documented, structural studies of phosphorylated RRs have been somewhat limited due to the intrinsic lability of the phospho-aspartyl linkage. However, there have been numerous reports in recent years describing the use of beryllium fluoride, a non-covalent mimic of the phosphoryl group, in order to obtain NMR or X-ray structures of stably activated RR domains (reviewed in^3-5). These have shown that phosphorylation of a conserved aspartate residue and binding of an essential divalent cation induces localized structural rearrangements within the active site of RR domains that lead to subtle long range conformational changes affecting primarily one surface of the protein. RR domains have a central five-stranded β-sheet surrounded by five α-helices (overall α₅/β₅ fold). Three carboxylate-containing residues, including the aspartic acid that is the site of phosphorylation, and an invariant lysine residue comprise the active site located at the C-terminal edge of the central β-sheet. Two additional highly conserved residues in the vicinity of the active site (Thr/Ser from β4 and Tyr/Phe from β5) have been implicated as switch-type residues that occupy distinctly different positions in the phosphorylated RR in comparison to the unphosphorylated state. Hence, the orientation of these two residues is highly indicative of whether the RR is in an active or inactive conformation. The combined effect of the active site rearrangement upon phosphorylation and the so-called “Y-T” conformational coupling of the switch residues results in a modest alteration of the α4-β5-α5 surface of the protein^6-8. Despite the growing numbers of RR domain structures that support a simple two-state model, there are two notable exceptions^9,10, suggesting the possibility that multiple conformational states occur along the pathway to RR activation.

Small molecule phosphodonors, such as acetyl phosphate and phosphoramidate, have also been used to phosphorylate RRs in vitro^11,12. However, the observed rate of phosphorylation is several orders of magnitude slower than if the cognate HK or HPt protein served as the phosphodonor^13-15. A similar rate difference was exploited as a means for differentiating cognate from non-cognate HK-RR and HPt-RR pairs in a phosphotransfer profiling assay^16,17. This raises two important questions: how do cognate HK, HPt and RR proteins specifically interact with each other and what aspect(s) of the association contribute to phosphotransfer efficiency?

To date, there are only two examples of structures of RR domains in complex with their cognate HPt protein, namely Spo0B/Spo0F^10,18 and YPD1/SLN1-R1¹⁹. No high resolution structures have been determined, thus far, of complexes between an HK and RR. The reason for the paucity of structural and biochemical data on RR/HPt and RR/HK interactions most likely stems from the transient and presumably weak nature of the interaction, which is typical of signaling partners, and the additional influence that phosphorylation may have on recognition and/or binding.

In the yeast Sacchromyces cerevisiae, YPD1 functions as an HPt protein that shuttles phosphoryl groups to and from three homologous RR domains associated with SLN1, SSK1 and SKN7 (referred to as SLN1-R1, SSK1-R2 and SKN7-R3, respectively). In the osmoregulatory branch of this phosphorelay system (SLN1-YPD1-SSK1), YPD1 serves two important functions by mediating phosphoryl group transfer between the hybrid HK SLN1 and the SSK1 RR and by stabilizing the phosphorylated state of SSK1^20,21. In its phosphorylated state, SSK1 is incapable of activating the downstream mitogen-activated protein (MAP) kinase cascade and therefore inhibits signaling via the high osmolarity glycerol response (HOG) MAP kinase cascade under normal growth conditions.

We previously reported the first structure of a monomeric HPt domain (YPD1) in complex with its upstream phosphodonor, the SLN1-R1 domain¹⁹. Two crystal forms were obtained (P3₂ and P2₁2₁2₁ space groups) that differed slightly in the relative orientation of YPD1 with respect to the SLN1-R1 domain. The P2₁2₁2₁ crystal form appeared to be favored when beryllium fluoride was included in the crystallization condition²², and from the structure of this complex we observed the two active sites to be better aligned than the P3₂ complex for phosphoryl transfer to occur. However, in the YPD1/SLN1-R1 complex, we did not observe electron density for the phosphoryl analog, BeF₃⁻ or for the divalent metal ion.

Recently, Varughese et al. published the 3.05 Å structure of Mg²⁺- and BeF₃⁻-bound Spo0F in complex with the dimeric Spo0B HPt protein¹⁰. This complex revealed a snapshot of a pre-transition state complex just prior to phosphotransfer. Here we report the crystal structure of beryllium fluoride- and magnesium ion-bound SLN1 RR domain (SLN1-R1) in complex with the HPt protein YPD1 at 1.70 Å resolution. This structure also represents a pre-transition state interaction between “activated” SLN1-R1 and YPD1 and allows for a detailed analysis of the active site configuration, protein-protein interactions, and mechanism of phosphotransfer between HPt and RR proteins. Furthermore, because of the high degree of structural homology among HPt proteins and within the RR superfamily, the yeast YPD1/SLN1-R1•Mg²⁺•BeF₃⁻ quaternary complex can serve as a model for studies of other HPt protein-RR complexes in other organisms.

Results

Overall structure of the quaternary complex

The YPD1/SLN1-R1•Mg²⁺•BeF₃⁻ complex was formed by soaking ammonium sulfate grown YPD1/SLN1-R1 apo crystals²² in a buffered PEG solution containing MgCl₂, BeCl₂ and NaF. The exchange of ammonium sulfate with PEG served two purposes in that it removed a sulfate molecule bound at the active site in the apo-crystal complex ¹⁹ and allowed the introduction of an essential Mg²⁺ ion²³. The F_o–F_c difference electron density map (Figure 1) showed clear evidence of an octahedrally-coordinated metal ion and density consistent with beryllium fluoride bound to Asp 1144 with tetrahedral geometry. The asymmetric unit consists of a 1:1 complex between YPD1 (full-length, residues 2-167) and SLN1-R1 (residues 1086-1218) bound to one BeF₃⁻ molecule and one Mg²⁺ ion. The model of this quaternary complex was refined to a resolution of 1.70 Å with a final crystallographic R-factor of 19.6% (R_free = 23.9%) (Table 1). In Ramachandran plots, generated by the PROCHECK program²⁴, 97.8% of residues occupy favored or allowable regions. Only 6 residues, 2.2 % of the model, are in the disallowed region of the Ramachandran plot because of their locations within turns. Electron density was not observed for the first two and final two residues of the SLN1-R1 expressed protein. A number of surface-exposed residues showed electron density for multiple side chain conformations and were therefore modeled as such with partial occupancy. In addition, Tyr 1165 of SLN1-R1 (located on the α3-β4 loop on the opposite face of the molecule from the active site) had additional electron density within 2.0 Å of the side chain hydroxyl that we tentatively interpret to be a sodium ion.

Figure 1.

Stereoview of the F_o-F_c map calculated with active site residues omitted and contoured at 3σ. The final model of the SLN1-R1 active site residues in the presence of Mg²⁺ and BeF₃⁻ is displayed within the electron density (YPD1 has been omitted for clarity). SLN1-R1 carbons are yellow with oxygens and nitrogens colored red and blue, respectively. The magnesium ion is green, beryllium is light green and fluorine atoms light blue. Red spheres represent water molecules.

Table 1.

X-ray Data Collection and Refinement Statistics

A. Crystal parameters
Space group	P212121
Unit cell dimensions (Å)	52.01, 75.54, 98.39
Matthews' coefficient (Å3/Da)	2.81
Solvent content (%)	56.2
B. Data Collection
Beamline at Brookhaven NSLS	X12B
Wavelength (Å)	0.900
Temperature (K)	100
Resolution (Å)	1.70
No. of observations	313590
Unique reflections	51408
Multiplicity	6.1
Completeness (%)	99.9 (99.9)
Mean I/σ(I)	27.0 (3.62)
Rmerge (%)a	4.8 (39.0)
C. Refinement
Resolution range (Å)	20.00 – 1.70
No. of protein atoms	2292
No. of heteroatoms	258 (252 H2O, BeF3−, Mg2+, Na+)
Avg. B-factors (Å2)	37.6
R-factor (%)b	19.6 (31.2)
R-free (%)c	23.9 (39.1)
Rmsd bond lengths (Å)	0.032
Rmsd bond angles (°)	2.492
Ramachandran plot
Most favored region (%)	90.1
Additionally allowed region (%)	6.9
Generously allowed region (%)	0.7
Disallowed region (%)	2.2

^aR_merge = Σ (I-<I>) / Σ(I), where I is the intensity measurement for a given reflection and <I> is the average intensity for multiple measurements of this reflection.

Values in parentheses are for the highest resolution shell.

^bR-factor = Σ | | F_obs | - | F_cal | | / | ΣF_obs |, where F_obs and F_cal are the observed and calculated structure factor amplitudes, respectively.

^cR_free was calculated with 5% of the diffraction data randomly selected and not used throughout refinement.

The overall structure of the YPD1/SLN1-R1•Mg²⁺•BeF₃⁻ complex is very similar to the P2₁2₁2₁ YPD1/SLN1-R1 apo complex solved previously¹⁹. The root mean square deviations (rmsd) for Cα atoms, as determined by LSQMAN²⁵, for the individually superimposed proteins were 0.594 Å and 0.984 Å for YPD1 and SLN1-R1, respectively. YPD1 retains its all-helical structure²⁶ with surface-exposed residues from the αA, αB, αC, and αD helices forming a primarily hydrophobic docking site for SLN1-R1. A ribbon diagram showing the relative orientation of the two proteins in complex with each other is shown in Figure 2. The modest differences that we observe between this complex and the previously reported apo complex¹⁹ appear to be directly attributable to the binding of the essential Mg²⁺ ion and BeF₃⁻ to the SLN1-R1 domain (detailed below).

Figure 2.

Global view of the YPD1/SLN1-R1•Mg²⁺•BeF₃⁻ quaternary complex. YPD1 is displayed in a green ribbon representation with His 64 shown in stick model. The SLN1-R1 domain is shown in yellow with Asp 1144 and the bound BeF₃⁻ displayed in sticks. The transparent surfaces of the two proteins are shown in their respective colors to illustrate the high degree of surface complimentarity at the interface.

Mg²⁺/BeF₃⁻ induced changes that affect YPD/SLN1-R1 interactions

The major difference observed in the relative orientation of the two proteins upon binding of Mg²⁺ and BeF₃⁻ is a 2.2 Å rigid-body shift of YPD1 with respect to SLN1-R1 (Figure 3A). This realignment of YPD1 positions the active site His 64 side chain within ideal distance (3.13 Å) and near linear geometry of Asp 1144 of SLN1-R1 for phosphotransfer to occur (Figure 3B). In addition, there is a slight reduction in the overall size of the buried surface area at the interface from 970 Ų in the apo complex to 949 Ų in the Mg²⁺/BeF₃⁻ bound complex, a decrease of 2.2% of the surface area based on calculations using AREAIMOL in the CCP4 program suite²⁷. The program LIGPLOT²⁸ as well as the PISA webserver²⁹ were used to determine the specific residues that are involved in forming hydrophobic, hydrogen bond or ionic interactions between the two proteins in both the apo complex and the Mg²⁺/BeF₃⁻ bound complex. A list of the specific interactions is provided as Supplementary Information (Table 2).

Figure 3.

Superposition of the YPD1/SLN1-R1•Mg²⁺•BeF₃⁻ quaternary complex (YPD1 in green and SLN1-R1 in yellow) and the YPD1/SLN1-R1 apo complex (in magenta and cyan, respectively). A) Top view of the overlayed structures in which the αA helix of YPD1 was removed for clarity. BeF₃⁻ is shown in light green and light blue. The alignment was performed using the first 50 residues of SLN1-R1 because there is little difference in these residues between the two complexes. The rigid body shift in YPD1 from the apo complex (magenta) to the BeF₃⁻-bound complex (green) is approximately 2.2 Å and allows for the alignment of the active site residues for phosphotransfer. B) Close-up view of the superimposed complexes showing the active site residues His 64 from YPD1 and Asp 1144 from SLN1-R1. The distance from the His 64 Nε2 atom to the beryllium atom (light green) is 3.19 Å.

In the Mg²⁺/BeF₃⁻ bound complex, there are a total of 26 hydrophobic contacts, a dramatic increase from the 16 observed in the apo complex (Figure 4 and Supplementary Table 2). The number of polar interactions, however, remained about the same (9 H bond interactions and 1 salt-bridge in the Mg²⁺/BeF₃⁻ bound complex, 10 H bonds and 1 salt-bridge in the apo complex). An additional 6 water-mediated H bonds were observed in the Mg²⁺/BeF₃⁻ bound complex, whereas the apo complex had only 2 water-mediated contacts at the interface. Figure 4 highlights the interfacial contacts that are new or altered in the Mg²⁺/BeF₃⁻-bound complex in comparision to the structure of the apo complex.

Figure 4.

Surface maps of YPD1 (left) and SLN1-R1 (right) separated in order to illustrate areas of protein-protein interaction. Yellow represents contact regions that are conserved in the apoand BeF₃⁻ -bound complexes, whereas red represents areas of contact that are new or altered in the BeF₃⁻ -bound complex (specific contacts are listed in Supplementary Table 2). The black color highlights the positions of the phosphorylatable residues His 64 in YPD1 (left) and Asp 1114 in SLN1-R1 (right).

Mg²⁺/BeF₃⁻ induced changes that affect the SLN1 response regulator domain

The essential Mg²⁺ ion is bound in the active site of SLN1-R1 with near perfect octahedral geometry (Figure 5). The metal ion is bound to two water molecules (2.06 Å, 2.12 Å), the backbone O of Gln 1146 (2.14 Å), Oδ2 of Asp 1144 (2.05 Å), Oδ2 of Asp1095 (1.99 Å), and F1 of beryllium fluoride (2.04 Å). There are several rearrangements that occur in and around the active site upon binding of Mg²⁺. In order for the backbone O of Gln 1146 to form a direct bond to the metal ion (Figure 5), the entire β3-α3 loop shifts by ∼1 Å in comparison to the apo complex (Figure 6). In addition, the movement of Asp 1095 in coordinating to the metal ion and Glu 1094 to form a water-mediated contact to the metal ion, alters the conformation of the β1-α1 loop and the N-terminus of α1 helix (Figure 6).

Figure 5.

Close up view of the active site with important sidechains and distances labeled. SLN1-R1 carbons are yellow with oxygens and nitrogens colored red and blue, respectively; BeF₃⁻ is light green and light blue, Mg²⁺ is magenta, and waters are red. Mg²⁺ coordination bonds are shown in yellow and the distances range from 1.99 Å to 2.14 Å. Bonds to the BeF₃⁻ atoms are shown in green with the bond lengths indicated in angstroms.

Figure 6.

A) Superposition of the SLN1-R1 proteins from the BeF₃⁻ -bound complex (yellow) and the apo complex (cyan) showing the site of phosphorylation (Asp 1144) and the residues that undergo a shift in position upon binding of BeF₃⁻. Thr 1173 and Phe 1192 are the two switch residues in the SLN1-R1 protein that occupy distinctly different positions in the active vs. inactive conformation. In the unphosphorylated state, Phe 1192 is exposed to solvent (cyan). Upon binding of BeF₃⁻, Thr 1173 shifts 2.3 Å toward the bound BeF₃⁻ and vacates space that allows Phe 1192 to rotate 90° toward the active site and adopt the active conformation (yellow). B) Plot of the rmsd for the Cα atoms of SLN1-R1 upon binding of BeF₃⁻. The x-axis is residue number and y-axis is atomic displacement in angstroms. Below the residue numbers is the secondary structure of the protein with α-helices shown as cylinders and β-strands shown as arrows. The dashed line represents the average rmsd for the whole protein (0.9 Å). Colored boxes separate the peaks into two groups. Red boxes are changes that are directly applicable to the binding of BeF₃⁻ or Mg²⁺. Gray boxes are the result of the higher resolution of the BeF₃⁻ bound structure that allowed for better modeling of the loop regions that were not as well defined in the apo complex.

The phosphoryl mimic, BeF₃⁻, binds to the Oδ1 atom of the active site aspartate (Asp 1144). In comparison to the apo complex the Oδ1 atom of Asp 1144 is shifted about 1.6 Å and rotated downward (∼60°) about the Cα-Cβ bond (Figure 3B). This change in position, along with the displacement of the sulfate ion from the active site, sets up near ideal geometry for phosphotransfer, with an angle of 163° from Oδ1 of Asp 1144 to the Be atom to Nε2 of His 64 on YPD1. The binding of BeF₃⁻ is stabilized by a network of hydrogen bonds and a direct bond to the magnesium ion (Figure 5). The F1 atom forms a water-mediated contact to Glu 1094 and is also coordinated directly to the Mg²⁺ ion. The F2 atom forms H bonds to the Oγ1 atom of Thr 1173 and the backbone N of Gln 1146. The F3 atom forms H bonds to the Nζ atom of Lys 1195 and the backbone N of Ala 1174. In comparison to the apo structure, several significant conformational changes occur upon binding BeF₃⁻ that are consistent with what has been observed in a significant number of bacterial phosphorylated- or BeF₃⁻-activated RR domains^8,30-40. Specifically, Thr 1173 undergoes a shift in its Cα position of 2.23 Å in order for its side chain Oγ1 atom to bond to one of the fluorine atoms of BeF₃⁻ (Figures 5 and 6). This movement allows the other so-called “switch residue”, Phe 1192, to rotate ∼90° inward toward the active site into a solvent inaccessible conformation. Because of the extended nature of the loop region immediately following Thr 1173, residues in this loop (β4-α4) as well as in the α4 helix exhibit the most significant shifts in Cα position between 1-3 Å (Figure 6). The conformational change associated with the interaction of the Nζ atom of Lys 1195 with the F3 atom of BeF₃⁻ causes a shift in the β5-α5 loop and is partially responsible for the ∼1.5 Å shift seen in the α5 helix (Figure 6B). The movement of Thr 1173 towards the active site brings Leu 1172 into hydrophobic contact with Ile 1197, and this appears to play a role in the displacement of the α5 helix. Other differences in backbone displacements (gray-shaded areas in Figure 6B) are attributed to the higher resolution of the BeF₃⁻-bound structure, which allowed for better modeling of two loop areas (residues 1134-1138 and 1163-1165) that were not as well defined in the apo complex.

Discussion

Effect of BeF₃⁻ binding on YPD1/SLN1-R1 interactions

The YPD1/SLN1-R1•Mg²⁺•BeF₃⁻ complex described herein is the first report of a high resolution structure of an HPt protein-RR complex in an “activated” state due to the presence of the phosphoryl analog beryllium fluoride. In comparison to the apo-complex, the largest global change observed is a rigid body shift of 2.2 Å in the orientation of YPD1 with respect to SLN1-R1. Conformational changes in SLN1-R1 induced by the binding of Mg²⁺ and BeF₃⁻ also contribute to a significant number of new or altered protein-protein interactions observed in comparison to the apo-complex (Figure 4 and Supplementary Table 2). In contrast, very few changes were observed for the Spo0B/Spo0F•Mg²⁺•BeF₃⁻ complex (PDB ID 2FTK) relative to the inactive state (PDB ID 1F51)¹⁰. Superposition of the two structures revealed only ∼0.5 Å rmsd for backbone atoms and a net increase of one hydrophobic interaction.

It is noteworthy that multiple binding modes (two different space groups) were observed for the apo-complex¹⁹, and in the BeF₃⁻-activated complex most of the new or altered contacts formed between the two proteins are centered around the active site residues (His 64, Asp1144) (Figure 4). This supports the notion that an equilibrium may exist in solution that involves formation of transient non-specific “encounter complexes” that can then reorient in a more optimal binding mode resulting in a functionally competent complex^41,42. Furthermore, the subtle differences that we observe in complexes of YPD1 and SLN1-R1 in the presence and absence of a phosphoryl analog suggest that fine tuning of protein-protein interactions may contribute to signaling specificity in two-component systems.

Mg²⁺/BeF₃⁻-induced conformational changes in SLN1-R1

From structural studies of phosphorylated or BeF₃⁻-activated bacterial response regulator domains^8,30-40, a common mechanism of activation has emerged involving a concerted shift in position of a conserved Thr (or Ser) residue and a conserved aromatic residue (Tyr or Phe) towards the active site which results in alterations of the α4-β5-α5 surface area. The BeF₃⁻-bound SLN1-R1 structure reported here provides evidence for the first time that the same mechanism of activation is conserved in eukaryotic RR domains as well. Also required for phosphotransfer is a Mg²⁺ ion in the active site to neutralize the charge of the phosphoryl group or in this case BeF₃⁻, in the active site. As shown in Figure 5, three protein atoms, two water molecules and a F1 atom from BeF₃⁻ form bonds to the Mg²⁺ ion in nearly perfect octahedral geometry. In the presence of BeF₃⁻, other residues also shift position to form direct contacts to F2 and F3, notably Lys 1195, Thr 1173, Ala 1174 and Gln 1146. The largest change in position observed between the apo- and bound-complexes is the movement of the Thr 1173 side chain ∼2.3 Å towards the active site in order to directly interact with BeF₃⁻ (Figures 5 and 6). In addition, there is ∼90° rotation of the Phe 1192 side chain from an outward facing, solvent-exposed position to an inward-facing solvent inaccessible position (Figure 6). These changes are consistent with the proposed two-state equilibrium model involving a concerted movement of two switch-like residues (Ser/Thr from β4 and Tyr/Phe from β5) indicative of RR domain activation⁸.

In terms of the combined effects of binding of Mg²⁺/BeF₃⁻ to SLN1-R1, we observe no obvious changes to the electrostatic surface and no net increase in the hydrophobic surface area. However, a slightly lower gap index⁴³ can be calculated for the Mg²⁺/BeF₃⁻-bound YPD1/SLN-R1 complex, which is indicative of a slightly higher degree of surface complementarity. Overall, there does not appear to be dramatic changes in the molecular surfaces of the two proteins, rather slight changes in contact residues at the interface and the orientation of the two proteins relative to each other (Figures 3 and 4) results in a more optimal fit for phosphoryl transfer.

Mechanism of phosphoryl group transfer

The combined effects of Mg²⁺ and BeF₃⁻ binding to the SLN1-R1 domain alter the way in which it binds to YPD1 and result in an almost linear alignment of the active site residues for productive phosphotransfer. The observed angle between the nitrogen atom (Nε2) of YPD1, the beryllium atom and the SLN1 Asp 1144 oxygen atom (Oδ2) is 163°, very close to the ideal linear alignment required for phosphoryl transfer⁴⁴.

It was first proposed by Mildvan in 1979 that for the transfer of a phosphoryl group using an associative mechanism, the attacking nucleophile and the electrophile were required to be within 3.3 Å of each other^44,45. In support of a dissociative mechanism, the same atoms must be ≥ 4.9 Å apart in order to accommodate a metaphosphate intermediate. In the beryllium fluoride-bound YPD1/SLN1-R1 complex, the Nε2 atom of His 64 and the Be atom are 3.19 Å apart (Figure 3B), within the range required for an associative mechanism. Furthermore, for an associative mechanism there would be a larger charge distribution on the equatorial atoms of the transition state or reaction intermediate. Therefore the protein would have to provide electrophilic regions to neutralize the charge on those atoms. In the ground state or pre-transition state structure presented here, the bound Mg²⁺ ion as well as multiple hydrogen bond donors as seen in Figure 3B (Lys 1195, Thr 1173, two waters, and backbone nitrogen of Ala 1174) are proposed to supply the necessary charge neutralization.

Thus, based on the geometry of the active site residues and surrounding highly conserved residues, we believe the structure of the BeF₃⁻-bound YPD1/SLN1-R1 complex supports an associative mechanism for phosphoryl group transfer, in agreement with the conclusion drawn from the activated Spo0F/Spo0B complex ¹⁰. We propose a model for a pentacoordinate trigonal bipyramidal transition state (Figure 7) in which charge neutralization by the Mg²⁺ ion, Lys 1195, Thr 1173 helps to facilitate formation and stabilization of the transition state.

Figure 7.

Proposed transition state structure representing an associative mechanism for phosphotransfer. Two sidechains from residues Lys 1195 and Thr 1173 and the Mg²⁺ ion are shown interacting with the equatorial oxygens to neutralize the negative charge on the pentacoordinate phosphorus intermediate.

In summary, the structure of the YPD1/SLN1•Mg²⁺•BeF₃⁻ complex allowed us to identify the changes that occur upon binding of a phosphoryl analog to the response regulator domain SLN1-R1 and how these changes affect the binding of SLN1-R1 to its phosphorelay signaling partner YPD1. We propose that phosphorylation of SLN1-R1 induces conformational changes, consistent with those observed in activated bacterial RR domains, which alters the manner in which P∼SLN1-R1 interacts with its downstream signaling partner YPD1. In the presence of the phosphoryl analog, SLN1-R1 appears to form a phosphotransfer-competent complex with YPD1 in which an associative mechanism for phosphoryl transfer is favored.

Materials and Methods

Protein expression, purification and crystallization

The Saccharomyces cerevisiae YPD1 (full-length) and SLN1-R1 (residues 1084-1220) proteins were over-expressed in Escherichia coli strain DH5α, purified and co-crystallized as descdribed previously^22,46,47. Crystals of the apo complex were grown in 2.2 – 2.6 M ammonium sulfate, 50 mM sodium acetate (pH 5.6), under hanging drop vapor diffusion conditions at room temperature. The size of the crystals reached an average of 250 μm × 250 μm × 300 μm after about one week. They were transferred to 40% PEG 4000, 60 mM sodium acetate (pH 5.6), 20 mM MgCl₂, 10 mM BeCl₂ and 70 mM NaF and allowed to soak at room temperature for 10-12 hours. The soaked crystals were then flash frozen in liquid nitrogen or directly in a nitrogen cryostream at 100 K prior to data collection.

Data collection and processing

Crystals were initially screened at our in-house X-ray facility (RigakuMSC RUH3R generator, Raxis IV⁺⁺ image plate detectors, and Oxford Series 700 cryosystem). Approximately 1 out of 10 crystals were of suitable quality for data collection and showed evidence of difference electron density in F_o-F_c maps in the vicinity of the active site residues. However, the resolution using our home X-ray source ranged only from 2.3-2.5 Å on the best crystals. Therefore, additional crystals were screened and the best were sent to the National Synchrotron Light Source (Beamline X12b) mail-in program at Brookhaven National Laboratory for higher resolution data collection. The data were collected in two passes from a single crystal using 0.900 Å wavelength radiation, first a high-resolution pass of 138 frames with a 1° oscillation and an exposure time of 40 seconds, then a low-resolution pass of 139 frames with a 1° oscillation and an exposure time of 40 seconds. The images were merged and processed with HKL2000 ⁴⁸. The X-ray data collection and processing statistics are listed in Table 1.

Structure determination and refinement

The structure of the YPD1/SLN1-R1complex bound to Mg²⁺ and BeF₃⁻ was solved by molecular replacement using the program PHASER⁴⁹. The search models were YPD1 and SLN1-R1 from the 2.3 Å P2₁2₁2₁ crystal structure of the YPD1/SLN1-R1 complex (1OXB) solved previously in our lab¹⁹. After molecular replacement, the initial R-factor was 36.03% with an R_free of 37.06%. Following the first restrained refinement step and initial map calculation using REFMAC5⁵⁰, as a part of the CCP4 suite of programs²⁷, beryllium fluoride and magnesium were modeled into clearly defined density of an F_o-F_c electron density map (Figure 1a). After further interactive cycles of maximum likelihood and isotropic temperature factor refinement using TLS restraints⁵¹ and model building using Coot⁵², the R-factor dropped to 23.23% with an R_free of 25.24%. Water molecules were initially modeled using Arp_waters from the ARP/wARP routine⁵³ implemented in REFMAC5⁵⁰ and subsequently only water molecules with Fourier difference peaks greater than 2σ were included in the final model. The final model was refined to 1.70 Å with an R-factor of 19.6% and an R_free of 23.9%. Refinement statistics of the final model are listed in the Table 1.

All structure figures were generated using Pymol (http//pymol.sourceforge.net). Graphical representations were produced using Microsoft Excel 2004 version 11.2.3 (Microsoft corporation).

The atomic coordinates and structure factors for the YPD1/SLN1-R1•Mg²⁺•BeF₃⁻ complex have been deposited in the RCSB Protein Data Bank with a PDB ID code of 2R25.