Escherichia coli Signal Recognition Particle Receptor FtsY Contains an Essential and Autonomous Membrane-binding Amphipathic Helix*

Escherichia coli membrane protein biogenesis is mediated by a signal recognition particle and its membrane-associated receptor (FtsY). Although crucial for its function, it is still not clear how FtsY interacts with the membrane. Analysis of the structure/function differences between severely truncated active (NG+1) and inactive (NG) mutants of FtsY enabled us to identify an essential membrane-interacting determinant. Comparison of the three-dimensional structures of the mutants, combined with site-directed mutagenesis, modeling, and liposome-binding assays, revealed that FtsY contains a conserved autonomous lipid-binding amphipathic α-helix at the N-terminal end of the N domain. Deletion experiments showed that this helix is essential for FtsY function in vivo, thus offering, for the first time, clear evidence for the functionally important, physiologically relevant interaction of FtsY with lipids.

Escherichia coli FtsY is a membrane-associated homologue of the eukaryotic SRP 4 receptor ␣-subunit. Together with the universally conserved SRP particle, FtsY is responsible for the cotranslational targeting of many integral membrane proteins that are consequently inserted into the membrane via the SecYEG translocon (1,2). For proper function, FtsY interacts with the SRP protein Ffh in a nucleotide-dependent manner (reviewed in Ref. 3). The structure of the complex between the two NG domains of Ffh and FtsY has recently been resolved by x-ray crystallography (4,5). 5 Targeting of ribosomes to the cytoplasmic membrane in E. coli is dependent on the expres-sion of FtsY (6). In addition, under FtsY-depletion conditions, the expression of polytopic membrane proteins such as LacY (7), SecY (6), and MdfA 6 is repressed. Other studies demonstrated that the SRP receptor forms a complex with membranebound ribosomes at the endoplasmic reticulum (8) and in E. coli, as well as in the absence of SRP or the translocon (9). In addition, recent studies identified interactions between the SRP receptor and the translocon in the endoplasmic reticulum (10) and in E. coli (11,12). These observations thus underscore the central role of FtsY in ribosome targeting and biogenesis of membrane proteins (13). However, despite extensive genetic, biochemical, and structural studies, important aspects of the function of FtsY are not yet fully understood. FtsY contains three distinct domains: the C-terminal N-and G-domains (together 302 residues long), which constitute a universally conserved SRP-GTPase (14), and an N-terminal A-domain (195 residues long) that was shown to be dispensable under various growth conditions (15).
At steady state, FtsY is distributed between the cytoplasm and the membrane (16), and it has no known membrane anchor partner homologous to the mammalian ␤-subunit of the SRP receptor (SR-␤), which coordinates the transfer of ribosomes translating SRP substrates to the translocon (17). Nevertheless, various studies have suggested that FtsY functions as a membrane-bound receptor (18,19). Interestingly, A-domain-truncated FtsY versions exhibit strong affinity for membrane lipids (20,21), possibly through the N-domain (22), an interaction that seems to be dominated by electrostatic forces (21). However, the precise lipid-interacting domain in FtsY has not been defined, and the mechanism and functional role of lipid binding to the receptor remain elusive.
Interestingly, an A-domain-truncated version of FtsY (termed NGϩ1) that is functional in vivo was identified by deletion analysis (15). Removal of the N-terminal amino acid Phe 2 (hereafter termed Phe 196 according to the sequence of wildtype FtsY) from NGϩ1 inactivated the receptor (termed NG). In the present study, we describe comparative structure/function analyses of the two mutants, which shed light on the mechanism underlying FtsY interaction with membrane lipids and revealed an essential and autonomous lipid-binding domain of the receptor.

EXPERIMENTAL PROCEDURES
Materials-Antibodies to FtsY were described previously (6). Bacterial Strains-E. coli HB101 was used for propagation and preparation of various plasmid constructs. E. coli FJP10 (23), harboring ftsY under control of the araB promoter, was used for FtsY-depletion experiments. E. coli BL21 was used for the overexpression of SRP and FtsY mutants.
Crystallization and Structure Determination of NGϩ1-Crystals of NGϩ1 were obtained by the vapor-diffusion method as described previously (24). Structure was solved by molecular replacement with the NG structure (Protein Data Band code 1FTS) as a search model (14) and using CCP4 (25). The structure was manually built using the program O (26) and refined with CNS (27). Details of the refinement statistics are given in supplemental Table S2.
For mutating the amphipathic helix in NGϩ1 (NGϩ1/ quadro), R198L, K200L, K205L, and K207L were replaced. This construct was created by one-step PCR using PfuTurbo DNA polymerase (Stratagene) and mutagenic primers that together encode the four mutations: sense, 5Ј-ctgttactgaccctggaaaatctcggttccgga and antisense, 5Ј-gctgcgcagcagcagcgcgaacatggaatactg. The PCR product was then treated with DpnI endonuclease for the digestion of methylated, non-mutated parental DNA templates, purified from 0.8% agarose gel, and ligated. Next, the ligation mixture was transformed into E. coli HB101, and purified plasmids were verified by DNA sequencing. A similar procedure was utilized for deleting the helix from full-length FtsY (Phe 196 -Lys 207 ).
Growth Experiments and FtsY Complementation Studies-Cultures were grown at 37°C in LB medium, supplemented with ampicillin (100 g/ml), kanamycin (10 g/ml), chloramphenicol (10 g/ml), and spectinomycin (33 g/ml) when necessary. For growth experiments in liquid media, cells were grown overnight at 37°C, and then diluted to 0.01 A 600 . In experiments with E. coli FJP10, cells were diluted to A 600 of 0.025. FtsY complementation experiments were executed by plating transformed E. coli FJP10 (23), harboring various constructs, on LB agar plates with or without arabinose, the inducer of the chromosomal ftsY.
Cell Fractionation Gel Electrophoresis and Immunoblotting-Harvested cultures were washed in buffer A (50 mM Tris-HCl, pH 8, 100 mM NaCl, 1 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride) and resuspended in the same buffer. Cell suspensions were sonicated, and cell debris was removed by centrifugation (5 min at 13,000 rpm). Membranes were collected by ultracentrifugation (45 min at 150,000 ϫ g). SDS-PAGE and immunoblotting were performed as described previously (6).
Construction and Purification of Thioredoxin Mutants-Plasmid petM20 containing the sequence for E. coli thioredoxin (Trx-a, kindly provided by Gunter Stier, EMBL, Germany Heidelberg) was digested with MscI and XbaI. The small fragment released containing the Trx sequence was used as a template for a PCR utilizing a forward primer, 5Ј-ggaaggcccatgggtagcgataaaattattcacct, and a reverse primer, 5Ј-cgggatccattagtggtggtggtggtggtggccctgaaaataaagattctcggccaggttagcgtcgagga. Via the PCR, a tobacco etch virus (TEV) protease site, a 6-histidine tag, and a BamHI cleavage site were introduced in the 3Ј-end, and an NcoI site was introduced at the 5Ј-end. The resulting PCR product was digested with NcoI and BamHI and ligated into vector pETM13 (kindly provided by Gunter Stier) digested with the same enzymes. NGϩ1-Trx was constructed by using the PCR template described above and the primers 5Ј-ggaaggcacatgttcgcgcgcctgaaacgcagcctgttaaaaaccaaagaaaatctcggtagcgataaaattattcacctg and 5Ј-cgggatccattaatggtgatggtgatggtgggccaggttagcgtcgagg, which introduce a 5Ј-end PciI restriction site and a 3Ј-end BamHI site and a 6-histidine tag. The PCR product was digested with PciI and BamHI and then ligated into the pET24d Vector (Novagen) cut by NcoI and BamHI. NG-Trx was constructed by using the PCR template described above and the primers 5Ј-ggaaggccatggcgcgcctgaaacgcagcctgttaaaaaccaaagaaaatctcggtagcgataaaattattcacctg and 5Ј-cgggatccattaatggtgatggtgatggtgggccaggttagcgtcgagg, which introduce a 5Ј NcoI cleavage site, a 3Ј BamHI cleavage site and a 6-histidine tag. The PCR product was digested with NcoI and BamHI and ligated into the pET24d vector (Novagen) cut by NcoI and BamHI. The final constructs were confirmed by DNA sequencing. Proteins were overexpressed in E. coli BL21/DE3, nucleic acids were precipitated after cell lysis with protamine sulfate, and the proteins were purified by nickel affinity chromatography and gel filtration as described for NG FtsY (24).

RESULTS
Site-directed Mutagenesis of the N Terminus of NGϩ1-As described previously (15), the functional NGϩ1 mutant of FtsY was created by insertion of a single phenylalanine residue at position 2 of the inactive mutant NG (Phe 196 ). The dramatic functional difference between NG and NGϩ1 was now investigated by mutating this phenylalanine in an attempt to explore the importance of the chemical properties at the N terminus of NGϩ1. The expression of selected mutants harboring hydrophobic and large (leucine, isoleucine, and tryptophan), small (alanine), or charged (lysine) residues at position 196 was examined by immunoblotting, and their activity was assayed by complementation of FtsY depletion (Fig. 1). The results (summarized in Table 1) indicated that a large hydrophobic residue is required at position 196. However, because additional truncation studies and characterization of other mutants indicated that the length of the N terminus might also be important (data , we investigated f-Met processing of selected mutants. Briefly, NG, NGϩ1, and several of the active and nonactive mutants were purified using a C-terminal His 6 tag, and the purified proteins were analyzed by N-terminal sequencing. As shown in Table 1, NGϩ1 and the functional constructs retained their N-terminal methionine. In contrast, the inactive mutants lost this methionine (NG, NGϩ1(F196A)). Together, the results suggest that the length of the N terminus of NG might be important, as well as a large hydrophobic residue at position 196. To evaluate these conclusions and investigate possible structural requirements at the N terminus of NG, we crystallized NGϩ1. Structural Differences between NG and NGϩ1-In the crystal structure of NG from E. coli FtsY the N terminus is not ordered (14). For a direct assessment of the observed differences between NG and NGϩ1, we crystallized and solved the structure of NGϩ1 at 1.9-Å resolution (supplemental Table S2). A comparison with the NG structure (1FTS (14)) revealed that, unlike NG ( Fig. 2A), the N terminus of NGϩ1 is well ordered and forms a regular ␣-helix (Fig. 2B). Apart from this N-terminal peptide, the overall structures of NG and NGϩ1 are identical (the root mean square deviation between the 295 C␣ atoms of residues 201-495 is 0.54 Å). Interestingly, the structure of NGϩ1 also reveals a polar interaction between Arg 198 and Asp 479 (Fig. 2C). Arg 198 is located at the ordered N terminus of NGϩ1, and Asp 479 is part of the closing loop (G5) of the GTPase. The N-and C-terminal helices are in proximity (Fig. 2D), and this interaction might be involved in regulation of the GTPase, as suggested earlier (29 -31). Taken together, comparison of the three-dimensional structures and the results presented above enabled us to draw the following conclusions: (i) the Arg 198 -Asp 479 interaction might be important for the functional assembly of the receptor, (ii) this interaction might also facilitate the formation of a regular ␣-helix, and (iii) the ␣-helical N terminus might be important for NGϩ1 function. These possibilities were explored further (see below). Importantly, the recently solved structure of the full-length FtsY 7 revealed that the N-terminal ␣-helix of NGϩ1 is part of an authentic structural element of the wild-type receptor, where the helix is even longer. Moreover, a comparison with recent structures of FtsY from Mycoplasma mycoides (29) and Thermotoga maritima (Joint Center for Structural Genomics, PDB entry code: 1vma) indicates that this helix is conserved in FtsY homologues.
Site-directed Mutagenesis-The interaction between Arg 198 and Asp 479 (Fig. 2C) was not observed in the NG structure (14) and could be of structural and functional relevance. Sequence alignment shows that Asp 479 is universally conserved, whereas Arg 198 is partially conserved among SRP receptors (data not shown). We performed non-conservative replacements at positions Asp 479 and Arg 198 , independently and simultaneously, to analyze the importance of these residues. The mutants were expressed (Fig. 3A), and their in vivo functions were tested by complementing FtsY depletion.
Surprisingly, replacement of the fully conserved residue Asp 479 for cysteine (D479C), alanine (D479A), or even arginine (D479R) did not have any appreciable effect on NGϩ1 function in vivo (Fig. 3B). In contrast, replacing residue Arg 198 had a dramatic inhibitory effect. Mutation R198C strongly inhibited the function of NGϩ1, and mutant R198D could not even be constructed, possibly due to its toxicity even under non-induced, basal expression levels. These results indicated that the observed interaction between Arg 198 and Asp 479 is not essential for NGϩ1 function, but might contribute to the stability of the NGϩ1 structure. The observation that the functional mutant D479R could not be crystallized, due to aggregation of the purified protein at elevated concentrations, might further support this 7 R. Parlitz and I. Sinning, unpublished data.  (Table 1) have significantly reduced hydrophobicity at their N termini (Fig. 4). Multiple sequence alignments of FtsY show that a motif of 13 amino acid residues, which corresponds to the amphipathic helix observed in the NGϩ1 structure, is conserved in eubacteria, archea, and chloroplasts (Fig. 5A). Yet we noticed that the negative charge (Glu 208 ) is not fully conserved. It was proposed that, although the negative charge is not essential for interaction with lipids, when present, it enhances binding of the amphipathic helix to membranes by repulsion, pushing the helix further down in between the head groups and the carbon chains of the lipid layer (32). Interestingly, a similar functionally important motif (termed membrane-targeting sequence (MTS)) was described previously for the conserved cell division determinant MinD (34,35). MinD is an ATPase closely related to FtsY, because both are members of the SIMIBI class of NTPbinding proteins (36). Notably, certain MTS sequences are similar to the amphipathic helix of FtsY (Fig. 5B). However, unlike in FtsY, MinD proteins harbor the amphipathic helix at their C termini.
Interaction with LUVs-The results described above strongly suggest that the N-terminal helix of NGϩ1 is involved in the documented interaction of the receptor with lipids (21). To test this possibility, we prepared LUVs consisting of 70% phosphatidylethanolamine and 30% phosphatidylglycerol to mimic the composition of E. coli membranes. Various FtsY constructs were purified and incubated with the LUVs, which were then purified by floatation centrifugation. The LUV-associated proteins were analyzed by SDS-PAGE and quantified by densitometry (see "Experimental Procedures") (Fig. 6). FtsY binds to LUVs efficiently (ϳ30%), unlike the A domain alone, which shows only little association with LUVs (ϳ8%) (Fig. 6A), indicating that  the receptor's interaction with membranes requires the NG domain. NGϩ1 bind efficiently, although to a lower extent (ϳ19%), whereas NG binds rather poorly (ϳ5%). This observation seems to contradict our previous study (21), in which however lipid binding was assayed only qualitatively by Western blotting and the composition of the liposomes was different (see "Experimental Procedures"). The observed membrane interaction of fulllength FtsY correlates nicely with the previously observed interaction of 30% FtsY bound to IMVs in an in vitro translation experiment (16). These results suggest that, despite the observation that the A-domain alone does not interact with LUVs, it apparently affects the capacity of the receptor to bind lipids (see "Discussion"). These results confirm the hypothesis that the N-terminal helix of NGϩ1 is required for interaction with lipids. The observed in vivo inactivity of the NG mutants lacking Met 195 and Phe 196 (Table 1 and Fig. 7) supports the importance of the hydrophobic face of the helix for this interaction.
To examine whether this segment is an autonomous lipid-binding element, we attached the conserved N-terminal peptides from NG and NGϩ1 to E. coli Trx, a soluble protein that does not bind to LUVs. The hybrids were then tested for their ability to bind LUVs. Fig. 6B shows that Trx and NG-Trx did not associate with liposomes, whereas NGϩ1-Trx bound efficiently. Taken together, these studies demonstrate that FtsY contains a transplantable, bona fide membrane-interacting sequence in the A/N-domain interface and that the hydrophobic residues at the N terminus of NGϩ1 are important for efficient interaction.
The Lipid-binding, Amphipathic ␣-Helix Is Essential for Function of FtsY-The results described above clearly show that, unlike in NG, the identified N-terminal amphipathic helix of NGϩ1 enables the receptor to interact with liposomes in vitro. To address the physiological relevance of these results further (see also Ref. 15) and to examine whether this helix is required for the in vivo activity of NGϩ1 and FtsY, we performed the following mutagenesis experiments. In FtsY the helix was deleted (Phe 196 -Lys 207 ), and in NGϩ1 the positively charged face of the helix was neutralized by replacing the corresponding four basic residues (Arg 198 , Lys 200 , Lys 205 , and Lys 207 ) by leucines. The mutated FtsY (FtsY⌬196 -207) exhibits a wild-type-like expression level (Fig. 7A), whereas the expression of the mutated NGϩ1 construct (NGϩ1/quadro) was very low (Fig. 7B), although at a level comparable to that of the chromosomally expressed FtsY in wild-type cells (Fig. 7B, compare NGϩ1quadro in the right lane with that of FtsY in the left lane). Importantly, however, Fig. 7 shows that both mutants are unable to complement FtsY depletion, indicating that the helix and its amphipathic character are indeed required for the receptor function in vivo.

DISCUSSION
We have previously shown that the FtsY-truncated mutant NGϩ1 is fully active in vivo (15) and that shortening its N terminus by a single residue abolishes its function. We have now  demonstrated that truncation, together with in vivo f-Met processing, destroys an amphipathic helix in NGϩ1. This helix is an essential, autonomous membrane-interacting sequence, thus defining for the first time, a lipid-binding domain in FtsY that is required for function in vivo.
The issue of how FtsY associates with the cytoplasmic membrane has received great attention since its first functional characterization (16). Already then, it was proposed that the receptor has a limited number of binding sites on the membrane (16), implying that lipids may not provide the major docking site of the receptor (see also Ref. 20). This proposal has been confirmed several times by demonstrating that FtsY assembles on membranes via interactions with a trypsin-sensitive component (22,37). Finally, the most compelling evidence for FtsYmembrane protein association has recently been offered by studies of the interaction between FtsY and the SecYEG complex in E. coli (11,12). Whether the translocon is the only proteinaceous binding site for FtsY remains unknown (38,39). Nevertheless, in addition to convincing evidence for FtsYmembrane protein interaction, other studies demonstrated unequivocally that the receptor also interacts with lipids, although such interaction probably does not dictate docking of the receptor. 8 Detailed characterization of this phenomenon was initially put forth by de Leeuw et al. (21), who demonstrated that FtsY interacts directly with E. coli phospholipids, with a preference for anionic ones, and that this interaction stimulates the GTPase activity of FtsY. Moreover, at least one determinant of lipid interaction was identified as the NG domain. In contrast to acidic lipids, other studies proposed that FtsY assembles on membranes via interactions with phosphatidylethanolamine, through the AN domain (37). Regardless of the crucial differences in identifying the specific lipids (see later), the combined studies implied a role for the N-domain in lipid binding. Taken together, the established interactions of the receptor, both with membrane proteins and lipids, raised two major questions as follows: (i) Which structural elements in FtsY are responsible for the association with proteins and lipids? Whereas the mechanism of association with the translocon and/or other membrane proteins remains to be investigated, our work discovered a functionally important lipid-interacting determinant of FtsY at the A/N-domain interface. Notably, the LUV-binding assays indicated that, although the A-domain is not essential, it contributes to this interaction. Possible interpretations are as follows: First, the positively charged, N terminus of the full-length receptor (16) might be able to enhance lipid binding only in the context of the full-length receptor. Second, the structure of full-length FtsY 7 revealed that the helix is in fact longer than seen in NGϩ1, providing a possible explanation for the enhanced LUV interaction of FtsY, compared with NGϩ1. (ii) What is the role of the individual interactions? Whereas at this stage one can only speculate on the role of FtsY interaction with membrane proteins, which is probably required for docking, our recent work 8 provides direct evidence for a role of lipid binding in regulating downstream events along the SRP pathway, by stimulating the GTPase activity of the membranebound FtsY-SRP complex. Such communication between the N-terminal helix of NGϩ1 and the GTPase G-domain has been suggested on the basis of the M. mycoides FtsY-NG structure (29) (see later). Importantly, the idea that the amphipathic helix regulates downstream, membrane-associated steps in the pathway is supported by our results showing that it is not required for membrane targeting and docking of the receptor. 8 As shown in Fig. 5A, the amphipathic helix of FtsY is relatively conserved in organisms or compartments lacking a membrane protein homologous to the eukaryotic ␤-subunit of the receptor (bacteria, archaea, and chloroplasts). Comparison of the NGϩ1 structure to other documented structures of SRP receptors (PDB code 1vma (29)) suggests that not only the N-terminal sequence of NGϩ1 is conserved but also its structure. Using the program Amphipaseek (40), we predict that the N-terminal peptide of NGϩ1 forms an in-plain membrane anchor. For FtsY we propose that lipid interaction initially involves electrostatic attraction between basic residues located at the polar face of the amphipathic helix and negatively charged lipid head groups. In the second step, hydrophobic interactions lead to the insertion of the helix into the membrane. This mode of interaction results in orienting the helix parallel to the membrane surface.
An interesting similarity exists between FtsY and other nucleotide-binding, peripheral membrane proteins in their functional interaction with membranes. For example, the peripheral translocon subunit SecA also requires acidic phospholipids for its high affinity binding to the translocon (41,42). Another example, probably the most striking one, is the previously characterized MTS in proteins of the MinD family. These proteins are structurally related to FtsY, because they are also members of the SIMIBI class of NTP-binding proteins (36). This family contains GTPases and ATPases characterized by the formation of nucleotide-dependent dimers. MinD proteins play a crucial regulatory role in selecting division sites in eubacteria, chloroplasts, and archaea. Apparently, in addition to being related in structure and function, the MTS of MinD proteins show some sequence similarity to the amphipathic helix of FtsY (Fig. 5B). Because the membrane composition in bacteria can vary significantly, the observed variation in the sequences of the membrane-binding helix in FtsY and in MinD (Fig. 5) might correlate with the membrane composition. For the MTS in MinD such correlation was indeed shown (35). The different locations of the membrane-binding domains in these proteins, the N terminus of the NG domain in FtsY and the C terminus of MinD, further support the proposal that the amphipathic helix is autonomous and able to link N-or C-terminally fused proteins to membranes, as shown for MinD (35) and FtsY (the present study). Moreover, as shown previously for FtsY (21) and recently summarized for other peripheral proteins, including MinD (43), activation of these proteins is mediated by membrane domains rich in anionic lipids, in accordance with the positively charged face of their amphipathic helices, thus lending support to the notion that FtsY interaction with acidic lipids (21) rather than neutral ones (37) is essential for the receptor function.
Previously, we reported that the GTPase of FtsY is activated by its interaction with anionic phospholipids (21), but the functional relevance of this stimulation was not understood. Now we localize an autonomous membrane-binding site to a conserved amphipathic helix at the N terminus of NGϩ1 and in a parallel study 8 we provide evidence that the stimulation of the FtsY GTPase by lipids is required for a later step in the SRP pathway. Productive FtsY-lipid interaction requires a certain length and hydrophobicity of the helix. The motif present in NGϩ1 represents the minimal regulatory element, because NG is not able to function in vivo. Comparison of the structures of SRP receptors shows that the conserved amphipathic helix in NGϩ1 packs between the C terminus and motif IV of the G domain in the conserved hydrophobic N/G interface (Fig. 2D) (see also Ref. 44). Such a configuration would indeed facilitate a lipid-mediated information transfer from the N-terminal helix to the catalytic domain, as recently proposed (29).