X-ray Structures of the Signal Recognition Particle Receptor Reveal Targeting Cycle Intermediates

The signal recognition particle (SRP) and its conjugate receptor (SR) mediate cotranslational targeting of a subclass of proteins destined for secretion to the endoplasmic reticulum membrane in eukaryotes or to the plasma membrane in prokaryotes. Conserved active site residues in the GTPase domains of both SRP and SR mediate discrete conformational changes during formation and dissociation of the SRP·SR complex. Here, we describe structures of the prokaryotic SR, FtsY, as an apo protein and in two different complexes with a non-hydrolysable GTP analog (GMPPNP). These structures reveal intermediate conformations of FtsY containing GMPPNP and explain how the conserved active site residues position the nucleotide into a non-catalytic conformation. The basis for the lower specificity of binding of nucleotide in FtsY prior to heterodimerization with the SRP conjugate Ffh is also shown. We propose that these structural changes represent discrete conformational states assumed by FtsY during targeting complex formation and dissociation.


INTRODUCTION
The functional core proteins of the signal recognition particle (SRP) and the SRP receptor (SR) (called Ffh and FtsY in bacteria) contain GTPase domains and form a distinct subfamily of GTPases. These GTPases mediate cotranslational targeting of secretory and membrane proteins to the endoplasmic reticulum (ER) membrane in eukaryotes or the plasma membrane in prokaryotes (for a review, see [1]). The classical GTPase motifs I-IV (also referred to as G1-G4) [2] are present in both SRP GTPases and show marked conservation with p21Ras. Present in and unique to SRP and SR are four additional elements, the insertion box domain (IBD), the closing loop, the 'DARGG' motif and the 'ALLEADV' motif. These elements contain essential structural functionality for SRP GTPases (for a review, see [3]). In contrast to the 'classical' model of GTPase regulation by external factors, SRP GTPases interact directly to reciprocally stimulate GTP hydrolysis and neither requires an exchange factor for nucleotide release [4][5][6]. The SRP family of GTPases thus provides a unique variation to the 'classical' GTPase model and the elucidation of the underlying mechanisms involved in regulating the targeting reaction is at the core of current structural and biochemical studies.
A family of crystal structures of prokaryotic SRP GTPases illustrates this unique mechanism of activation. In particular, apo structures of Ffh-NG, a truncated version of the prokaryotic SRP core protein containing the amino-and GTPase domains, and FtsY, the SR protein, show the stabilization of an ''open'' state through interactions of the GTPase and SRP conserved sequence motif residues [7,8]. ''Open'' state conformations in the presence of either bound nonhydrolyzable GTP substrate analog guanine 59-imidotriphosphate (GMPPNP) or product GDP have been shown for Ffh-NG [9,10]. Similarly a structure of FtsY with the product, GDP has been obtained [11]. In this structure, the GDP is coordinated with canonical binding interactions and reveals the importance of the C-terminal helix in the NG packing interface. In addition, three additional apo structures of FtsY have been solved and show distinct properties from FtsY in complex with Ffh [12][13][14]. Low measured intrinsic GTPase activities of Ffh and of FtsY (0.09 min 21 , 0.01 min 21 ) [15] imply that proteins bind GTP and remain in ''open'' conformational states. Low specificity for nucleotide in monomeric FtsY has also been shown and further suggests a novel structural regulation of GTPase activity for FtsY [16].
The structures of the FtsY?Ffh-NG complex in presence of the non-hydrolyzable substrate analog GMPPCP [17,18], GMPPNP [19], or GDP:AlF 4 [20] show the formation of a composite, active site sequestered from solvent, through the catalytic interactions between the classical GTPase and SRP specific conserved elements from both GTPases and the bound nucleotides. Nucleotide hydrolysis in each active site drives dissociation of the SRP?SRP receptor complex, allowing the SRP and SRP receptor components to be recycled [21,22]. Interestingly, the FtsY?Ffh-NG complex structures also suggest that (i) the structures observed represent a ground state of the GTP hydrolysis reaction, and (ii) that additional conformational changes in the active site are necessary to progress to the transition state. Recently, mutational studies of FtsY have revealed that site-specific mutations can modulate discrete conformational changes during Ffh?FtsY complex formation [23]. These specific conformational states involve the sequential activation of the Ffh and FtsY active sites after binding of GTP and during the formation of the targeting complex.
Here we report the structures of two conformations of FtsY in complex with the substrate analogue GMPPNP. These structures reveal two novel active site architectures for SR GTPasenucleotide complexes. These structures, along with the structure of apo FtsY and FtsY in the Ffh-NG?FtsY complex, can be interpreted as a series of discrete conformational states along the pathway of step-wise activation of the SRP GTPases during the formation of the SRP?SR complex and provide structural explanations for the biochemical differences observed for FtsY along the targeting cycle.

Structures of FtsY?GMPPNP complex
Two crystal forms of FtsY from Thermus aquaticus (T. aquaticus) were crystallized and structures determined. Crystal form one initially grew out of a purified complex of FtsY, full length Ffh from T. aquaticus, GMPPNP and Mg 2+ . Subsequent crystals were grown in the absence of Ffh. The structure of this form was determined to 2.2 Å resolution and contained a single non-crystallographic (NCS) two-fold related dimer per asymmetric unit with one monomer in apo-FtsY and the other FtsY monomer bound with GMPPNP ( Figure 1A, 1B, 1C). A second form containing the apo/GMPPNP bound dimer was also initially crystallized from the purified FtsY & Ffh & GMPPNP complex in the absence of Mg 2+ in different crystallization conditions from crystal form one ( Figure 1D, E) ( Table 1). In addition, apo crystals of FtsY prederivatized with chloromercurial nitrophenol (CMNP) were grown and solved by multiple wavelength anomalous diffraction (MAD) and used as a molecular replacement model for the apo/ GMPPNP dimer forms. For the purposes of this paper, we will refer to the GMPPNP bound monomer from crystal form one as F1 and from crystal form two as F2.
The FtsY dimer is related to the FtsY?Ffh-NG complex dimer by a ,20u rotation of the NCS two-fold. The apo monomer from crystal form one and crystal form two are nearly identical and show a relative shift in the NG interface compared to F1 and F2. The overall fold of FtsY observed for the various conformations, are nearly identical. Although the F1 crystal form of FtsY was cocrystallized with magnesium chloride, identification of an ordered magnesium ion in the structure is uncertain. An ordered water is observed coordinated with the c-phosphate and four additional waters. This would be a likely candidate for a magnesium ion, except that the bonding distance is more than 0.5 Å greater than expected for a coordinated magnesium ion. Therefore, the molecule was modeled as water and no magnesium is present in the final model. It should be noted that a magnesium ion is also not observed in the structure of Ffh-NG bound with GMPPNP [10]. Present in F1 and F2 are the first 27 N-terminal residues not seen in the FtsY monomer of the FtsY?Ffh-NG complex; they are structured in a loose helical manner. Overall, the FtsY?GMPPNP structures and the apo structure demonstrate varying degrees of an oversized ''open'' conformation in comparison to the FtsY structure from the FtsY?Ffh-NG complex ( Figure 2). The most significant differences between the active form (from Ftsy?Ffh-NG), the two GMPPNP bound forms and the apo form, are localized in the conserved sequence motifs.

Nucleotide specificity is poor
The guanine base is coordinated by residues in motif IV which contains the conserved SRP GTPase 'TKXD' sequence motif.
Aspartate 258 encodes specificity for the guanine base while lysine (Lys256) is involved in maintaining the active site cavity spacing through interactions with the P-loop and an aspartate in the a3 helix ( Figure 3A). A concerted sequestering of the nucleotide is shown in the conformation of FtsY in complex with Ffh-NG: Asp258 coordinates the N1 and N2 amino groups of the guanine base at an average bond distance of 2.7 Å , the closing loop Glu284 packs against the ribose moiety of GMPPCP (2.6 Å ), and Lys256 forms a bridging interaction with both the carbonyl oxygen of Gly112 of the P-loop at 2.8 Å and a 2.7 Å hydrogen bond with Asp229 of helix a3. In the F1 form with GMPPNP, Lys256 also forms a bridging hydrogen bond with the sidechain of Asp229. The interaction distance between the a3 helix Asp229 and motif IV Lys256 is on the same order as seen in the active conformation. However, the lysine is now at a distance of 3.6 Å from the P-loop. The effect of this loosening is a movement of motif IV away the Ploop and in turn away from the nucleotide such that Asp258 is now on average 3.4 Å from the amino groups of the guanine base. The closing loop also has retracted from the nucleotide, although Glu284 and the O3 of the ribose ring in GMPPNP are at the same interacting distance as in active FtsY as the glutamate extends away from the backbone. F2 also lacks the interaction between Lys258 and the P-loop (6.8 Å ) and the distance from Asp258 to the amino groups extends even greater than in F1 to an average distance of 4.6 Å . Movement out of the active site by motif IV now decouples the interaction between the a3 helix Asp229 and motif IV Lys256 (now at a distance of 4.5 Å ). In addition, the closing loop no longer packs against the nucleotide and Glu284 is 4.7 Å away from the ribose oxygen. Finally, in the ground state apo FtsY, the position of the guanine specifying Asp258 is further extended than in F1 or F2. Interestingly, the interaction distance between Lys256 and Asp229 is maintained but the interaction with the the P-loop is decreased by more than 1 Å . Overall, starting with the apo conformation and ending with the complex conformation, these interactions serve to organize a closure of the active site around the nucleotide to allow for increasing coordination of the guanine base by residues from motif IV; a closing which in turn is translated to the a3 helix.

Non-canonical positions of c-phosphate and active site residues
A comparison of the FtsY in complex with Ffh-NG, GMPPNP and apo structures reveal that the largest conformational differences localize to the SRP conserved sequence elements that coordinate the phosphate groups of the nucleotide. The Insertion Box Domain (IBD), a unique motif II sequence element contains two critical residues: Arg142 and Gln148. In complex with Ffh-NG, FtsY residue R142 interacts with c-phosphate and Gln148 helps coordinates the magnesium ion ( Figure 3B). A striking difference between this conformation of the IBD and the conformation in F1 and F2 is observed. In the GMPPNP bound forms, Arg142 has moved away from the active position in a greater than 5 Å movement accompanied by an unfolding and rearrangement of the 'DTFRAGA' sequence in the a1 helix of the IBD. This unfolding event also leads to a counterclockwise rotation (relative to the N to C axis of the a1a helix) of the Gln148 sidechain away from the active site and excludes any interaction with a magnesium ion. In apo FtsY, the IBD has re-structured and as a consequence Arg142 and Gln148 have returned to the active site although their positions differ from the active form. By superimposing bound nucleotide from the active form, F1 and F2, a clash in the van der Waals spacing between the apo position of this motif II arginine sidechain and the c phosphate position of bound nucleotide is observed (not shown). In addition, Gln148 has rotated clockwise (from the same reference axis as above) greater than 2 Å from the active position to locate itself again out of the active site but in the opposite relative direction than in F1 and F2.
Binding of nucleotide is expected to confer certain conformational changes in the protein to both accommodate the nucleotide and to orient the active site moieties for proper hydrolysis. However, as also observed in Ffh-NG bound with GMPPNP, FtsY does not appear to be in a state that would allow for proper nucleotide hydrolysis. Lys115 is displaced from the active site cavity and does not interact with the phosphate groups of GMPPNP ( Figure 3C). Also absent is the interaction between the backbone amide nitrogen of Arg195 and the c phosphate. Thus, the c phosphate appears to be missing two critical contacts. Finally, the amino nitrogen of Asn111 forms a hydrogen bond with the c phosphate oxygen, therefore restraining the c phosphate out of the P-loop cavity. In apo FtsY Arg195 is positioned away from the phosphate-binding cavity. The preceding glycine, Gly194, forms a salt bridge with the P-loop and orients the backbone nitrogens away from the c phosphate. In F1, the sidechain of Arg195 moves into the active site such that it coordinates a water molecule with the c phosphate oxygen and Asn111 sidechain from the P-loop. The amido group of Asn111 also forms a hydrogen bond with the c phosphate oxygen. These interactions act to force the c-phosphate out of the binding pocket and constrict the P-loop. The measured distance between the Ca of P-loop residues Asn111 and Thr116 is 9.8 Å in F1. This constriction is not as great as seen in the Ffh-NG?GMPPNP complex [10] (,8.8 Å ) but still represents a constriction of 0.7 Å as compared to FtsY in complex with Ffh-NG.

DISCUSSION
The structures of FtsY described here represent snapshots of FtsY and nucleotide in various active site conformations. We interpret these structures as representing discrete states along the pathway of activation and dissociation of FtsY in the targeting cycle. These structures can also be compared with the product complex of FtsY with GDP bound [11]. The latter structure crystallizes in a different space group (I4) from the structures reported here, reflecting the conformational change upon going from the substrate bound, to product bound conformations. This step-wise activation potentially provides specific regulatory points along the protein-targeting pathway.

Structural Basis for Nucleotide Promiscuity Revealed
The most obvious result in our study relates the variation in the position of conserved motif IV with a structural explanation for the poor base specificity and promiscuous nucleotide hydrolysis reported in nucleotide binding studies of FtsY by Shan and Walter [24]. In solution, FtsY displays little specific affinity for GTP versus ATP, XTP and other nucleotides. However, complex formation with Ffh increases nucleotide specificity 10 23 -fold [15]. An examination of the coordinating distance of motif IV aspartate to the guanine ring provides a structural account for these observations. The coordination distance observed between Asp258 and the guanine nucleotide is most optimal in the complex with Ffh yet dissipates in  the absence of Ffh as observed in F1. In addition, F2 shows bound nucleotide despite no coordination from Asp258 ( Figure 4). This is a simple structural explanation for the low specific affinity.
Nucleotide Binding Alone does not Initiate the Catalytic Cycle of the Targeting Complex.
The conserved structural elements in F1 and further in F2 suggest a sequential ordering of the active site during complex formation. The increasing coordination between Asp258 and GMPPNP from apo FtsY to F2 and further in F1 relates to a closing of the active site and an increase in the coordination of the guanine base of the nucleotide. Complex formation with Ffh-NG directs FtsY IBD residues Arg142 and Gln148 to interact with the band cphosphate groups while catalytic Asp139 activates the proposed nucleophilic water. In addition, Asn111 forms the only sidechain contact with the trans substrate of Ffh-NG, a hydrogen bond with the 39O of the GMPPCP. Analogous conserved residues act in the same manner in the Ffh-NG component of the complex. Further, the nucleotides themselves act in trans to stabilize the negative change on the c-phosphate formed during GTP hydrolysis in contrast to most other GTPases that require an external activation residue from their respective GAPs. Both F1 and F2 show a total disruption in these interactions and in fact exhibit a coordination that positions the c-phosphate out of the active site. Without these intra-and interactions, hydrolysis of GTP is not favored.

N-terminal Helix Represses GTPase Activity
The A-domain is an N-terminal domain which has been proposed to anchor FtsY to the membrane [25,26] although it is not essential for SRP-mediated targeting [27]. T.aquaticus FtsY lacks the A-domain, but retains an N-terminal helix, termed N1. The N1 helix becomes susceptible to proteolysis upon formation of targeting complex with Ffh [28] and deletion of N1 causes a ,4-5 fold increase in GTPase activity. In contrast to the structures presented here, the N1-deleted form of FtsY from T. aquaticus binds GDP in a canonical manner including the specificitydetermining hydrogen bonding interactions with Asp258 [11]. However, deletion of N1 has a pronounced affect on the position of the N-domain relative to the G-domain upon nucleotide binding when compared to either apo FtsY, FtsY:GMPPNP or FtsY:GMPPCP from the targeting complex with Ffh ( Figure 5 and Figure 6). It is also important to note that the C-terminal helix (aC) assumes a similar position in the apo, GMPPNP and the GDP structures of FtsY, but rearranges upon complex formation with Ffh. Given (i) the proteolytic susceptibility of N1 upon complex formation, (ii) increase in GTPase activity upon deletion of N1 and (iii) the observation of a canonical active site for monomeric FtsY:GDP, the N1 helix might act to repress GTPase activity in the monomeric form of FtsY. This could be accomplished by favoring an N/G domain organization that favors a more open active site. Upon complex formation, N1 rearranges for proper complex formation and active site optimization. In this manner, the N1 helix would act as a negative regulatory element; maintaining low activity in the non-targeting, monomeric form of FtsY as observed in the structures presented in this paper. This regulation might also be affected by interactions with the membrane and membrane-bound translocon.
Step-Wise Activation and Dissociation of the SRP-Mediated Targeting Complex GTP-hydrolysis by GTPases is intrinsically slow but can be accelerated by binding of specific external factors. SRP GTPases are unique in that activation of GTP-hydrolysis occurs upon formation of a complex of two GTPases. Our data support and expand an emerging model that explains some of the complexity in the catalytic cycle of FtsY and Ffh. SRP GTPases are regulated not at the point of nucleotide binding or exchange, but rather through a 'step-wise' mechanism. FtsY and Ffh bind GTP in an ''open'' or ''primed'' state. An ''active'' conformational state is reached upon the concerted rearrangements of the SRP GTPase conserved sequence elements after complex formation with SRP. A concerted step-wise formation would also imply that nucleotide hydrolysis and release might also be step-wise. Nucleotide release in GTPases is typically regulated by guanine nucleotide exchange factors by disruption of Mg 2+ coordination, perhaps by displacement of Mg 2+ from the active site [29][30][31] . It may be that the IBD gates coordination of the active site Mg 2+ and in turn contributes to the low affinity for nucleotide through disruption of proper coordination architecture in the monomer form. The N1 helix might also contribute to regulation of the GTPase cycle and promote a step-wise activation by repressing GTPase activity in the monomeric form. The FtsY GMPPNP structures described here depict conformational states along the activation pathway for SRP GTPases in the targeting reaction and reveal the structural basis for the futile nature of GTP hydrolysis in monomeric FtsY. In addition, these structures provide insight into the step-wise ability of FtsY to self-regulate its GTPase activity as well as reset to the apo state through release of the nucleotide for subsequent rounds of targeting.

Crystallization
T. aquaticus FtsY was expressed in E. coli strain BL21 (DE3) (Novagen) using a pET21 plasmid and purified to homogeneity. FtsY was concentrated to 9 mg ml 21 in 50 mM Hepes pH 7.5, 150 mM potassium acetate pH 7.5, 2.5mM magnesium acetate and 2 mM b-OG, octyl-b-d-glucoside. Ffh was prepared as previously described [7]. Crystallization attempts of FtsY and Ffh with 3x molar excess of GMPPNP, and 10mM MgCl 2 were carried out as hanging-drops with the vapor diffusion method at room temperature and crystals initially grew out of two conditions: 2.0 M ammonium sulfate and 2.0 M ammonium sulfate, 100 mM sodium acetate. Identification of FtsY as the sole component of the crystals was evaluated by gel electrophoresis and mass spectrometry. Additional crystals were grown from 18% PEG 2000, Tris pH 7.5, 100mM NaCl . For MAD, a mercury derivative of FtsY was used. Purified FtsY was incubated with 10 mM TCEP-HCl (Tris(2-Carboxyethyl) Phosphine, Hydrochloride) for 4 hours to reduce oxidation on the cysteine (Cys136). Following buffer exchange with a Hepes buffer to eliminate reducing agent, 10mM of the Hg compound CMNP (6-chloromercuri 2,4-dinitrophenol-Kodak) was added to the protein and incubated for 4 hours. The FtsY-Hg complex was then buffer exchanged with Hepes buffer using Centricon YM-10K (Amicon) to eliminate unbound Hg.
The covalent binding of Hg was confirmed with mass spectrometry.

Data collection
Data collection parameters and statistics for all data are listed in Table 1. FtsY-Hg crystals diffracted to 1.8 Å and a three wavelength MAD data set was collected at the Advance Light Source in Berkeley (ALS) beamline 5.0.2. Data were measured at three wavelengths, l f0 = 1.006 Å , l f9 = 1.009 Å and l high remote = 0.993 Å . Data collected at SSRL were integrated and reduced with MOSFLM and SCALA [32] and data collected at ALS were processed with HKL2000 [33]. All reduced data were truncated and placed on an absolute scale with Wilson B-factor estimation with TRUNCATE from the CCP4 program suite [32].

Structure determination
Two mercury sites were resolved by hand and confirmed using an automated heavy atom search procedure in CNS [34]. Heavy atom refinement and phasing was done with SHARP and resulted in an overall FOM of 0.51 for MAD and 0.75 for combined SIR and MAD. Iterative rounds of solvent flattening, NCS averaging, and phase combination of SIR and MAD phases were calculated with the CCP4 program DMmulti. The calculated phases were used in phased molecular replacement to position a poly-alanine version (loops omitted) of the previously determined E. coli FtsY structure into the electron density. Initial density in the active site from the MAD phased maps as well as simulated annealing omit maps calculated from the molecular replacement model showed clear density for the tri-phosphate nucleotide. Figures were generated using SPDBV and PyMol.