Chloroplast SecA and Escherichia coli SecA Have Distinct Lipid and Signal Peptide Preferences

Abstract

Like prokaryotic Sec-dependent protein transport, chloroplasts utilize SecA. However, we observe distinctive requirements for the stimulation of chloroplast SecA ATPase activity; it is optimally stimulated in the presence of galactolipid and only a small fraction of anionic lipid and by Sec-dependent thylakoid signal peptides but not Escherichia coli signal peptides.

The chloroplast Sec pathway for protein transport is particularly interesting because it sits at the evolutionary crossroads between eukaryotic systems which have Sec-dependent pathways that do not involve SecA (e.g., endoplasmic reticulum [ER] transport) and prokaryotic Sec systems that do require SecA. In bacteria, the components of this pathway include the SecYEG complex, which constitutes the translocation channel (8, 11), and SecA, an ATPase that powers the translocation of the polypeptide across the cytoplasmic membrane (16, 31, 32). In Escherichia coli, the efficiency of protein translocation is directly proportional to the amount of anionic phospholipids (17, 20, 25) and this reflects the effect of these lipids on SecA ATPase activity (19, 29). In plants, genes encoding chloroplast homologues of SecY (15) and SecE (10) have been identified, as has a SecA homologue from several chloroplast sources (22, 24, 26, 28). The lipid composition of the thylakoid membrane is dominated by the neutral galactolipid digalactosyldiacylglycerol (DGDG) (70 to 80%) and anionic dioleylphosphatidylglycerol (DOPG) (30). The extent to which these lipids influence the activity of the thylakoid transport machinery is not known.

Sec-dependent proteins that cross the eukaryotic ER or the bacterial cytoplasmic membrane are synthesized with a single amino-terminal signal peptide whose highly hydrophobic nature plays a critical role in facilitating protein transport (5) through interactions with SecA (4). Proteins destined for the Sec pathway in the chloroplast thylakoid differ in that they are synthesized with presequences that are bipartite; the most amino-terminal portion corresponds to a transit sequence for passage through the chloroplast envelope and the second region is reminiscent of bacterial and ER signals. Does chloroplast SecA (cpSecA) directly interact with this signal, and is it tailored to discriminate thylakoid Sec signals from the variety of targeting signals for other chloroplast transport pathways?

Here, we address these issues by using cpSecA ATPase activity as an earmark of ligand interactions. In marked contrast to E. coli SecA, cpSecA ATPase activity is enhanced by a high concentration of DGDG and only a small amount of DOPG. Furthermore, cpSecA ATPase activity is preferentially stimulated by weakly hydrophobic chloroplast, but not bacterial, Sec signal peptides.

Helical content and ATPase activity of cpSecA is retained in lipid vesicles.

To set the stage for examining the preferences of cpSecA for two key ligands, lipids and signal peptides, we first established that SecA retained secondary structural elements and ATPase activity upon lipid vesicle integration.

Plasmid pET-chlSecA was used to express a modified cpSecA in which the 60 amino-terminal residues corresponding to the chloroplast transit peptide are deleted and the carboxyl terminus is fused with a hexahistidine tag (23). E. coli strain BLR(DE3) (Novagen), grown at 37°C, was cotransformed with pET-chlSecA and RIG (1). The RIG plasmid carries genes encoding tRNAs that recognize codons used infrequently in E. coli, thus increasing the expression levels of heterologous proteins. Following growth and induction, cpSecA was solubilized (7) and purified by Ni-nitrilotriacetic acid metal affinity chromatography.

Circular dichroism spectroscopy of cpSecA (Fig. 1A) in aqueous solution yields a spectrum with minima at 208 and 220 nm and a maximum at 194 nm, characteristic of a protein with substantial α-helical secondary structure. The spectrum was similar for cpSecA in lipid vesicles of DGDG-DOPG (8:2). An analysis of the spectra yielded values of 76 and 79% α-helix for cpSecA in aqueous solution and lipid vesicles, respectively. The compatibility of cpSecA in DGDG-DOPG vesicles is underscored by comparison with cpSecA in sodium dodecyl sulfate (SDS) micelles, which gave a spectrum suggestive of a large loss of helicity (to 12%), with a concomitant large increase in random coil (to 54% from 10%) and some elements of β-sheet.

FIG. 1.

CpSecA properties. (A) Circular dichroism spectra of cpSecA. cpSecA (0.15 mg/ml) in 10 mM Na₂HPO₄ (solid line), in DGDG-DOPG (8:2; 400 μg/ml) lipid vesicles (small dashed line), or in 0.25 M SDS (large dashed line). Spectra were analyzed using the CD Neural Network program (3). (B) Fluorescence spectra of cpSecA in the absence (solid line) and presence (dotted line) of mant-ADP. cpSecA (0.2 μM) was incubated with a 10-fold excess of mant-ADP for 10 min at 25°C in translocation buffer [0.05 M HEPES-KOH, pH 7.0, 0.03 M KCl, 0.5 mM Mg(OAc)₂, 0.03 M NH₄Cl]. The emission spectra were recorded using an excitation wavelength of 290 nm. (C) Change in fluorescence intensity at 450 nm in the absence and presence of ATP. F₀ is the fluorescence of cpSecA alone, and F is its fluorescence in the presence of nucleotide.

Fluorescence resonance energy transfer in aqueous solution between cpSecA (excitation and emission bands at 290 nm and 350 nm, respectively) and 2′(3′)-O-(N-methylanthraniloyl)-ADP (mant-ADP) (excitation and emission bands at 350 and 450 nm, respectively) demonstrates nucleotide binding (Fig. 1B). Using excitation at 290 nm, a fluorescence emission band appears at 450 nm for cpSecA in the presence of mant-ADP. The addition of ATP at levels 20 to 100 times that of mant-ADP reverses the effect, resulting in a loss of the 450-nm fluorescence band (Fig. 1C). This indicates that both nucleotides bind cpSecA and that ATP can displace mant-ADP in a dose-dependent manner indicative of a specific interaction with nucleotide binding sites. Consistent with this observation, cpSecA was found to have substantial ATPase activity in aqueous solution (90 pmol min⁻¹ μg SecA⁻¹), which is further enhanced in the presence of lipid vesicles (120 pmol min⁻¹ μg SecA⁻¹), as has been observed for SecA from other species (2, 17).

The ATPase activity of cpSecA, in contrast to E. coli SecA, indicates a preference for integration in vesicles with high DGDG content.

The effect of lipid composition on the ATPase activity of cpSecA was examined using established assays (17, 19, 29). Keeping the concentrations constant for cpSecA (40 μg/ml) and lipid (400 μg/ml), we investigated the cpSecA-lipid ATPase activity in lipid systems composed of dioleoylphosphatidylcholine (DOPC)-DOPG, dioleoylphosphatidylethanolamine (DOPE)-DOPG, or DGDG-DOPG. The cpSecA-lipid ATPase activity exhibited a clear optimum at 20% DOPG, and the activity fell off rapidly at higher concentrations (Fig. 2A). That cpSecA activity peaks at 20% DOPG is interesting in view of the strong dependence of protein translocation on anionic lipids in vivo in E. coli (17, 27) and the optimum of 40 to 60% observed for signal peptide-stimulated E. coli SecA-lipid ATPase activity in vitro (19). The requirement for 20% DOPG is, however, consistent with the content of anionic lipids found in the thylakoid membranes of higher plants (30). Furthermore, these thylakoid membranes are dominated by neutral galactolipids which can comprise about 80% of the total lipid content. In vitro, we find that the presence of the galactolipid DGDG is clearly favored for cpSecA activity over the zwitterionic phospholipid DOPC or DOPE as a partner with DOPG. This is in marked contrast to E. coli SecA ATPase activity, which is strongly inhibited by even a small amount of DGDG (Fig. 2A).

FIG. 2.

Dependence of cpSecA ATPase activity on the presence of various lipids. Each data point represents an average of triplicate assays, performed at 37°C, ± standard error (error bars). (A) Influence of DOPG. The final concentration of lipids was 400 μg/ml. The cpSecA or E. coli SecA activity in the presence of lipids is, in each case, reported as a percent increase over the cpSecA or E. coli SecA activity, respectively, in aqueous solution. (B) Influence of DGDG. Lipid vesicles were composed of 20% DOPG with various ratios of DOPC and DGDG comprising the remaining 80%. The final concentration of lipids was 400 μg/ml. (C) Dependence on lipid concentration. The molar fraction of DGDG and DOPG was kept at 8:2. The cpSecA activity in each case is reported as a percent increase over the activity in aqueous solution.

The importance of a high molar fraction of DGDG for cpSecA activity is further underscored when the DOPG content is held constant at 20% with various ratios of the neutral lipids DOPC and DGDG comprising the remaining 80% lipid (Fig. 2B). A dose-response relationship is established with respect to DGDG content, with optimal activity observed in a mixture of DGDG and DOPG at a molar ratio of 8:2 (no DOPC). That DGDG can form vesicles has been confirmed (30), and the presence of a small amount of bilayer-promoting DOPG should integrate well with DGDG. The observation that cpSecA ATPase activity specifically requires a large fraction of neutral galactolipid corresponding to that present in the thylakoid membrane (30) emphasizes the importance of the membrane environment for ATPase activity and is consistent with the notion that the ATPase activity is intimately involved in the translocation of preprotein across thylakoid membranes.

Furthermore, using vesicles of 80% DGDG and 20% DOPG, we find that the cpSecA ATPase activity is dependent on lipid concentration (Fig. 2C) with an optimum corresponding to an approximately 1,000:1 molar ratio of lipid to SecA. This is consistent with a requirement for SecA to be sufficiently ensconced in lipid to generate the fully active species (19). Flotation analysis (17) confirms that, under these conditions, cpSecA is associated with the membranes (Fig. 3). Interestingly, the association is related to the presence of DOPG and this may explain the requirement for small amounts of this lipid for maximal ATPase activity. Although DOPC with DOPG also promotes membrane association, only DGDG with DOPG stimulates cpSecA-lipid ATPase activity (Fig. 2A).

FIG. 3.

Association of cpSecA with lipid vesicles. Flotation gradient centrifugation of 15 μg cpSecA in the absence or presence (400 μg/ml) of lipid vesicles composed of 8:2 ratios of DGDG-DOPG, DOPC-DOPG, or DGDG-DOPC or of DOPC alone in a reaction mixture volume of 500 μl. Membrane-associated cpSecA will float to the top of the gradient (fraction 1), whereas cpSecA that does not integrate into lipid vesicles will be at the bottom (fraction 5). SDS-polyacrylamide gel electrophoresis shows the presence of cpSecA in each fraction quantified by densitometry.

cpSecA is preferentially stimulated by chloroplast signal peptides that utilize the Sec-dependent transport pathway.

To explore the substrate specificity of cpSecA and the extent to which complexes of cpSecA and particular lipids may be involved in preprotein recognition, we examined the influence of signal peptides on cpSecA ATPase activity (Fig. 4). Signal peptides, made as glutathione S-transferase (GST) fusions, were expressed in E. coli strain BLR(DE3) and purified on a glutathione-Sepharose column. Thirty residues of the mature region of alkaline phosphatase are carboxyl terminal to the signal peptides and have no effect on the in vitro ATPase activity (14). E. coli SecA was expressed and purified as previously described (19, 29).

FIG. 4.

SecA-lipid ATPase activity in the presence of GST-conjugated signal peptides. The final concentration of lipid in all assays was 400 μg/ml, and the final concentration of GST and signal peptides was 20 μM. Each data point represents an average of triplicate assays ± standard error (error bars). (A) cpSecA-lipid ATPase activity. The AP- and PPC-GST fusions carry the E. coli alkaline phosphatase and pea plastocyanin thylakoid signal peptides, respectively. Lipid ratios were DGDG-DOPG-SL (6:2:2) and DGDG-DOPG (8:2). (B) cpSecA-lipid ATPase activity. AP and β-lac are GST fusions with signal peptides of the E. coli Sec pathway preproteins alkaline phosphatase and β-lactamase, respectively. PPC and OE33 are GST fusions with the chloroplast Sec signal peptides pea plastocyanin and the 33-kDa subunit of the wheat oxygen-evolving complex, respectively. Suf and OE23 are GST fusions with the signal peptides for TAT preproteins, Suf I and the 23-kDa subunit of the oxygen-evolving complex, in E. coli and chloroplasts, respectively. The SecA activity in DGDG-DOPG (8:2) in each case is calculated relative to that in aqueous solution without signal peptide. (C) E. coli SecA-lipid ATPase activity. The SecA activity in E. coli phospholipids in each case is calculated relative to that in aqueous solution without signal peptide. Signal peptides are as defined for panel B.

The requirement for cpSecA in DGDG-DOPG at 8:2 is emphasized by comparing the ATPase activity of cpSecA in this lipid composition (Fig. 4A) with that in E. coli lipids and in aqueous solution, regardless of the ligand present. Similar results were obtained when plant sphingolipid (SL) was included (DGDG-DOPG-SL [6:2:2]). Remarkably, we also find that only the chloroplast signal peptide stimulates ATPase activity beyond the level observed in its absence. The E. coli alkaline phosphatase signal peptide has no effect on cpSecA; the ATPase activity was comparable to that in the absence of any signal peptide or GST alone. We considered the possibility that this may reflect an unfavorable interaction due to the presence of chloroplast lipid as opposed to cpSecA itself. However, no stimulation with GST-alkaline phosphatase (AP) is observed for cpSecA in E. coli lipids (Fig. 4A). Moreover, this difference in cpSecA and E. coli SecA substrate specificity is underscored with E. coli and chloroplast signal peptides. The ATPase activity of cpSecA is stimulated by only the Sec-dependent chloroplast signal peptides of pea plastocyanin and the 33-kDa subunit of the oxygen-evolving complex of wheat. E. coli signal peptides of both the Sec-dependent (alkaline phosphatase and β-lactamase) and TAT (Suf I) pathways do not stimulate cpSecA, nor does a chloroplast ΔpH-dependent signal (OE23) (Fig. 4B). While E. coli Sec-dependent signal peptides stimulate the E. coli enzyme (14) (Fig. 4C), neither chloroplast signal peptide, plastocyanin, nor the 23-kDa subunit of the oxygen-evolving complex does (Fig. 4C).

The signal peptides for pea plastocyanin and E. coli alkaline phosphatase are comparable in length (25 and 21 residues, respectively) and contain charged amino-terminal segments, central hydrophobic cores, and polar carboxyl segments. The plastocyanin signal peptide, like other thylakoid Sec signal peptides, however, includes an amino-terminal negative charge and is significantly less hydrophobic than the alkaline phosphatase signal peptide and other E. coli Sec signals. Perhaps the efficient import of a thylakoid protein first across the chloroplast envelope precludes the use of a Sec signal that is too hydrophobic and which otherwise might bind cytosolic factors, delaying or aborting its chloroplast targeting.

Once in the stroma, proteins may encounter multiple thylakoid-targeting routes. These routes include the spontaneous insertion of proteins (18), and signal recognition particle-dependent (9), ΔpH-dependent (6, 21), and SecA-dependent pathways (22, 33). A twin-arginine motif is specifically recognized by components of the ΔpH-dependent pathway, and hydrophobic sequences are favored by the signal recognition particle-dependent pathway (12, 13). The specificity of cpSecA for the Sec-dependent chloroplast signals observed here provides a mechanism for avoiding precursor proteins with those features. The distinctive requirements for the stimulation of cpSecA suggest that it has evolved to be specifically well suited for the environment of the chloroplast thylakoid and to recognize thylakoid Sec-dependent proteins.