The Thylakoid ΔpH-dependent Pathway Machinery Facilitates RR-independent N-Tail Protein Integration*

The thylakoidal ΔpH-dependent and bacterial twin arginine transport systems are structurally and functionally related protein export machineries. These recently discovered systems have been shown to transport folded proteins but are not known to assemble integral membrane proteins. We determined the translocation pathway of a thylakoidal FtsH homologue, plastid fusion/protein translocation factor, which is synthesized with a chloroplast-targeting peptide, a hydrophobic signal peptide, and a hydrophobic membrane anchor. The twin arginine motif in its signal peptide and its sole integration requirement of a ΔpH suggested that plastid fusion/protein translocation factor employs the ΔpH pathway. Surprisingly, changing the twin arginine to twin lysine or deleting the signal peptide did not abrogate integration capability or characteristics. Nevertheless, three criteria argue that all three forms require the ΔpH pathway for integration. First, integration was competed by an authentic ΔpH pathway precursor. Second, antibodies to ΔpH pathway component Hcf106 specifically inhibited integration. Finally, chloroplasts from the hcf106 null mutant were unable to integrate Pftf into their thylakoids. Thus, ΔpH pathway machinery facilitates both signal peptide-directed and N-tail-mediated membrane integration and does not strictly require the twin arginine motif.

The thylakoidal ⌬pH-dependent and bacterial twin arginine transport systems are structurally and functionally related protein export machineries. These recently discovered systems have been shown to transport folded proteins but are not known to assemble integral membrane proteins. We determined the translocation pathway of a thylakoidal FtsH homologue, plastid fusion/ protein translocation factor, which is synthesized with a chloroplast-targeting peptide, a hydrophobic signal peptide, and a hydrophobic membrane anchor. The twin arginine motif in its signal peptide and its sole integration requirement of a ⌬pH suggested that plastid fusion/ protein translocation factor employs the ⌬pH pathway. Surprisingly, changing the twin arginine to twin lysine or deleting the signal peptide did not abrogate integration capability or characteristics. Nevertheless, three criteria argue that all three forms require the ⌬pH pathway for integration. First, integration was competed by an authentic ⌬pH pathway precursor. Second, antibodies to ⌬pH pathway component Hcf106 specifically inhibited integration. Finally, chloroplasts from the hcf106 null mutant were unable to integrate Pftf into their thylakoids. Thus, ⌬pH pathway machinery facilitates both signal peptide-directed and N-tail-mediated membrane integration and does not strictly require the twin arginine motif.
Export type pathways target proteins to the bacterial plasma membrane, the endoplasmic reticulum, the mitochondrial inner membrane, and the chloroplast thylakoid membrane (1). Where known, component and mechanistic similarity support the hypothesis that the organelle pathways are descendent from those of the prokaryotic endosymbiont. This is most evident when comparing export pathways of chloroplasts and bacteria. Chloroplasts share with bacteria at least four distinct pathways: a Sec pathway, a ⌬pH/Tat 1 pathway, a signal rec-ognition particle (SRP) pathway, and a spontaneous insertion pathway (2)(3)(4). Chloroplasts and bacteria also appear to employ homologues of the mitochondrial Oxa1p export protein (5).
The ⌬pH/Tat pathway in thylakoids and bacteria is the most recently recognized protein translocation pathway. The ⌬pH pathway was first described as a system that transported a subset of thylakoid lumenal proteins using only the thylakoidal ⌬pH as an energy source (6). Substrates of this pathway possess cleavable, amino-terminal signal peptides with an invariant twin arginine motif amino-terminal to the hydrophobic core (7). Where examined, substitution of one or both arginines abolishes ⌬pH pathway transport (7,8). Identification of a component of the ⌬pH pathway, Hcf106 (9), and the recognition that a subset of bacterial periplasmic proteins possesses twin arginine-containing signal peptides (10,11) led to the description of a homologous bacterial pathway. Disruption or mutations of Hcf106 homologues impaired transport of a range of twin arginine-bearing precursors (12,13). In addition, disruption of a Tat operon gene for the multispanning membrane protein TatC also impaired protein transport (14). A plant chloroplast TatC homologue has been identified and shown to be necessary for transport of ⌬pH pathway substrates in vitro. 2 Recent studies from several laboratories indicate that the ⌬pH and Tat systems do not employ components of the Sec system and vice versa (15)(16)(17)(18).
An intriguing question is why two pathways, Sec and ⌬pH/ Tat, co-evolved to target proteins to the same location. One possible explanation is that unique capabilities of each system match the translocation problems of the respective substrates. For example, substrates of the ⌬pH and Tat pathways appear unable to utilize the Sec pathways, even when provided with a Sec signal peptide (8,19,20). This may be because substrates of the ⌬pH/Tat pathways appear to be transported in a folded conformation (20,21,22), whereas the Sec pathway requires its substrates to be unfolded during translocation (23). Conversely, the Sec pathway is able to recognize transmembrane domains and integrate them into the lipid bilayer, whereas previously identified ⌬pH pathway precursors are extrinsic proteins, suggesting that the ⌬pH pathway might be unable to insert membrane proteins.
In this paper, we present evidence that the ⌬pH pathway can integrate membrane proteins. Pftf is an integral thylakoid membrane protein and a AAA family member related to mitochondrial and bacterial zinc-dependent proteases, e.g. FtsH, that chaperone protein biogenesis (24). The Pftf precursor has two hydrophobic segments, similar to other FtsH homologues (Fig. 1). However, the first hydrophobic segment (H1) of the Pftf precursor is relatively short, like a signal peptide and is preceded by a twin arginine. In fact, all known plant chloroplast Pftf homologues possess an RRXFLK motif (25), a conserved motif among bacterial Tat pathway precursors (10). Here, we show that Pftf is integrated into thylakoid membranes with characteristics indistinguishable from other ⌬pH pathway substrates. Pftf integration required only the thylakoidal ⌬pH as energy source. It was competed by overexpressed precursor substrates of the ⌬pH pathway and inhibited by antibodies to components of the ⌬pH pathway. Furthermore, Pftf integration was blocked in hcf106 null mutant chloroplasts.
Nevertheless, Pftf integration does not strictly require the typical targeting elements of the ⌬pH pathway, as they are currently understood. Evidence for this conclusion is that substitution of KK for the signal peptide RR did not change integration characteristics. Surprisingly, deleting the entire signal peptide did not alter integration characteristics of the resulting mPftf, including its competition by ⌬pH precursors and inhibition by antibodies to Hcf106 and Tha4. Thus, Pftf lacking the ⌬pH targeting elements still requires components of the ⌬pH machinery for insertion. These results significantly expand the substrate range of the ⌬pH pathway and suggest a re-evaluation of the nature of the ⌬pH pathway and the cis elements thought to govern its targeting.

EXPERIMENTAL PROCEDURES
Preparation of Radiolabeled Precursors-Plasmids for pPftf (pepper), pLHCP, pOE17 (maize), iOE23 (pea), pOE33 and iOE33 (wheat), and pPSII-T (cotton) used for in vitro transcription and translation have been described (8, 26 -28). The sequence of Pftf, originally reported by Hugueney et al. (29), was amended and can be found at GenBank TM accession number AJ012165. Cloning and analysis of DNA products were by standard molecular biology procedures (30). Deletion clones of pPftf were constructed by PCR with Pfu polymerase (Stratagene) using the cloned pPftf as template. The forward primer for synthesis of mPftf was 5Ј-CTCACAATGGCTGATGAGCAAGGTGTTTCTAACTCAAGGT-TGTCT-3Ј, which gave an amino-terminal sequence of MADEQG-VSNSRLSYSI. Besides adding an initiator methionine to the predicted amino terminus, the methionine at position 12 of the mature protein was changed to leucine to prevent internal initiation in vitro. The reverse primer for amplifying mPftf was 5Ј-GTTCAGGGTCTGTTCTCT-TTCATC-3Ј. pKKPftf was constructed by splicing by overlap extension-PCR (31) in which the internal primers contained sequences to modify arginine residues 60 and 61 of pPftf to lysines. The internal forward primer was 5Ј-CGGATGAAGGAAAAAAGGCTTTCTTAAATTATTG-3Ј and the reverse primer was the exact complement of the forward primer. The flanking forward and reverse primers, used for the second round of PCR, were 5Ј-AAGAGCTCATATGGCTACTTCATCAG-3Ј and the above reverse primer used for mPftf amplification, respectively. mKK Pftf was constructed by SOE-PCR with mPftf (above) as template. The forward internal primer 5Ј-CTTTTCCTGCTATCAAAGAAGTCTA-ACGGAGG-3Ј and reverse internal primer 5Ј-CCTCCGTTAGACTTCT-TTGATAGCAGGAAAAG-3Ј contained sequences to modify the two arginines carboxyl proximal to the membrane anchor (H2) to two lysines. All PCR products were digested with EcoICRI and XhoI and ligated into the EcoICRI/XhoI-digested pPftf plasmid and transformed into XLI-Blue (Stratagene) cells, and the resulting clones were sequenced entirely on both strands by the University of Florida Interdisciplinary Center for Biotechnology Research DNA sequencing core. In vitro transcription with SP6 RNA polymerase (Promega) and translation with rabbit reticulocyte lysate (Promega) or coupled transcription/ translation with wheat germ TnT (Promega) in the presence of [ 3 H]leucine (NEN Life Science Products) was performed following the manufacture's guidelines. Unless otherwise stated, translation products were diluted with one volume of 60 mM leucine in 2ϫ import buffer (1ϫ ϭ 50 mM HEPES, KOH, pH 8.0, 0.33 M sorbitol) prior to use.
Chloroplast Protein Import and Thylakoid Protein Transport and Integration Assays-Chloroplasts were isolated from 9 to 10-day-old pea seedlings (Pisum sativum cv. Laxton's Progress 9) as described (32). Maize plants were grown at 26°C in a 16 h light/8 h dark cycle for 7-10 days. Mutant hcf106 mum3 maize seedlings were selected by their pale green phenotype and confirmed by high chlorophyll fluorescence with a hand-held UV lamp. Maize chloroplasts were isolated by essentially the same procedure as pea chloroplasts except that intact chloroplasts were purified on a step density gradient consisting of 8 ml of 75% Percoll and 15 ml of 35% Percoll in GR buffer. The gradients were centrifuged at 3200 ϫ g for 15 min, and intact chloroplasts were collected at the 35%/75% interface. Chloroplast lysate, washed thylakoids, and stromal extract were prepared from intact chloroplasts as described (26). Import of precursors into chloroplasts was as described (26). Briefly, in vitro translated precursors were incubated with intact chloroplasts (0.33 mg chlorophyll/ml) and 5 mM MgATP in import buffer at 25°C in a light bath (70 microeinstein/m 2 /s) for the times indicated in the figure legends. Reactions were initiated by the addition of translation products equivalent to one-sixth the assay volume and terminated by transfer to an ice bath. Intact chloroplasts were recovered, with or without protease treatment, by centrifugation through a 35% Percoll cushion. For fractionation, recovered chloroplasts were lysed by resuspension in 20 mM Hepes, KOH, pH 8, and incubation on ice for 5 min. The thylakoid membranes were separated from the stromal fraction by centrifugation for 8 min at 3200 ϫ g. Thylakoids were either washed with import buffer, extracted with 0.1 M NaOH, or treated with thermolysin as described (32) except that for protease treatment, samples were resuspended in import buffer containing 0.2 mg/ml thermolysin for 60 min at 4°C followed by washing in import buffer containing 50 mM EDTA.
Assays for protein translocation into isolated thylakoids were essentially as described (26). Assays contained 0.33 mg/ml chlorophyll equivalent of thylakoids, lysate, or reconstituted lysate, 1/6 volume radiolabeled translation product, and where indicated, 5 mM MgATP. Reaction mixtures were incubated for the indicated times at 25°C in the presence or absence (foil-wrapped tubes) of light. Reactions were terminated by transfer to 0°C, and the thylakoids were either washed in import buffer plus 10 mM MgCl 2 , extracted with 0.1 M NaOH, or treated with thermolysin as described above.
Competition for integration into isolated thylakoids was performed essentially as described (26). Inclusion bodies of Escherichia coli-expressed tOE23 were dissolved in fresh 10 M urea, 10 mM dithiothreitol to 90 M protein for 3 h at room temperature, diluted to appropriate 3ϫ stocks in 0.5 M urea, 0.5 mM dithiothreitol, import buffer plus 10 mM MgCl 2 , and added to otherwise complete reaction mixtures. All reactions contained 0.33 mg of chlorophyll/ml of chloroplast lysate, 5 mM MgATP, radiolabeled translation product, 0.167 M urea, and 0.16 mM dithiothreitol in import buffer plus 10 mM MgCl 2 . Reactions were started immediately after the addition of competitor by incubation at 25°C for 20 min in the light. The assays were terminated by the addition of nigericin and valinomycin (0.5 and 1.0 M final concentration, respectively) and transferred to ice. Recovered thylakoids were then protease-treated.
In organello competition was conducted essentially as described (26). Inclusion bodies of pOE23 or pOE33 were dissolved in fresh 10 M urea, 10 mM dithiothreitol for 3 h at room temperature. Chloroplasts in import buffer plus 5 mM MgATP were preincubated with unlabeled competitors for 7 min in the light at 25°C. Competitors were aliquotted from stocks, such that the final competitor concentration was 0.75 M pOE23, 0.6 M pOE33, or no competitor, and the urea concentration was 0.25 M in all assays. Radiolabeled precursors (1/6 volume) were then added, and the incubation was continued for an additional 10 min. Competition assays were stopped by adding an equal volume of 1.5 M nigericin and 3 M valinomycin in import buffer and transferred to ice. Chloroplasts were then repurified by centrifugation through Percoll cushions.
Use of Inhibitors and Antibody Inhibition of Thylakoid Protein Transport-Integration assays were conducted in the presence of apyrase by pretreatment of thylakoids, lysate, and translation products with 0.04 units apyrase/ml as described (6). Import and integration assays were conducted in the presence of ionophores nigericin and valinomycin at 0.5 M and 1.0 M final concentration, respectively, as described (6). Thylakoids were treated with anti-maize Hcf106 and anti-pea Tha4 IgGs in the presence or absence of antigens as described (17). Thylakoids were treated with anti-pea Hcf106 and anti-pea plant chloroplast TatC homologue as described (17). 2 Miscellaneous-Chlorophylls were determined according to Arnon (33). Proteins were quantified using the BCA method (Pierce) with bovine serum albumin as standard. Immunoblotting was by the ECL method as described (17). Unless indicated, chemicals were purchased from Sigma.

Pftf Is Imported into Isolated Intact Chloroplasts and Properly Integrated into Thylakoids by a Two-step Mechanism-
Upon incubation with chloroplasts, the 74-kDa Pftf precursor (pPftf) (Fig. 2, lane 1) was imported into the organelles and processed to an ϳ65-kDa protein (mPftf) that was protected from protease treatment of the intact chloroplasts ( Fig. 2A,  lanes 2 and 3). Upon subfractionation of the chloroplasts following import, mPftf was found exclusively in the thylakoid fraction (lane 5). It was resistant to extraction with 0.1 M NaOH, demonstrating that it was integrally associated with the bilayer (Fig. 2A, compare lanes 5 and 7) and partially degraded by protease treatment of thylakoids (lane 6), resulting in an ϳ13-kDa degradation product (Fig. 2B, lane 2). These characteristics are identical to those previously determined for the endogenous protein of chromoplast internal membranes and pea thylakoid membranes (28 and data not shown). Furthermore, they indicate that the ϳ13-kDa protease resistant fragment is a suitable indicator of properly integrated Pftf.
If pPftf possesses a bipartite transit peptide consisting of a stromal targeting domain followed by a lumen-targeting signal peptide (Fig. 1), then pPftf would be processed in two steps, first by a stromal processing protease to form a species intermediate in size, and second by a thylakoidal protease to generate mPftf. Two lines of evidence support this model. The first is the result of conducting a chloroplast import assay in the presence of ionophores that dissipate the proton motive force. Chloroplast protein import requires ATP, but not ion or pH gradients, whereas most thylakoid translocation is stimulated by the thylakoidal pH gradient (see Ref. 2). When pPftf was incubated with chloroplasts, ATP, and ionophores, it accumulated in chloroplasts as a protein slightly larger than mPftf ( Fig. 2A, compare lanes 8 and 7). This intermediate-sized Pftf (iPftf) was predominantly associated with thylakoids, but it was not integrated into the membrane because it was largely extracted by 0.1 M NaOH ( Fig. 2A, lane 13, compare with lane 7) and degraded by protease treatment (Fig. 2B, lane 3). These characteristics are similar to those of lumenal proteins previously shown to localize by a two-step pathway, except that lumenal intermediates generally accumulate in the stromal fraction (e.g. see Ref. 26). A second line of evidence is that pPftf was processed by clarified stromal extract to a species inter-mediate in size (data not shown) and processed by washed thylakoids to mPftf (see Fig. 3). These data support the idea that the H1 domain of pPftf is a signal peptide that is removed upon integration into thylakoids and that the mature sized mPftf has a single transmembrane anchor and a lumenal tail. Additional evidence for this point will be presented below.
Pftf Requires Neither the Twin Arginine Nor the Entire Signal Peptide for Integration into Thylakoids-Preliminary results demonstrated that pPftf integrates into isolated thylakoids in a ⌬pH-dependent manner. This suggested that Pftf employs the ⌬pH pathway for integration. One means of addressing whether Pftf is using the ⌬pH pathway is to modify the cis topogenic elements for ⌬pH pathway targeting. The two arginines just prior to H1 in pPftf were replaced with two lysines ( Fig. 1; pKKPftf). As a more severe test, the entire signal peptide of pPftf was removed producing mPftf. These modified Pftf proteins were assayed with intact chloroplasts as well as with isolated thylakoids. As expected, pKKPftf was imported into isolated chloroplasts (Fig. 3A, lanes 7 and 8), but mPftf remained on the outside of the chloroplasts where it was degraded by protease (Fig. 3A, lanes 12 and 13). Imported pKKPftf became integrally associated with the thylakoid membranes, similar to pPftf, as determined by resistance to extraction with NaOH and by the production of the characteristic FIG. 1. Topology of pPftf and mPftf and constructs. pPftf is synthesized with two hydrophobic domains, H1 from 66 to 82 and H2 from 169 to 193 (28). The predicted stromal processing site, 49 -50, and thylakoid processing site, 86 -87, are indicated by gaps. pKKPftf was constructed by mutating the arginines at positions 60 and 61 (in the RRXFLK motif) to lysines. mPftf was constructed by deleting the entire bipartite transit peptide, i.e. residues 1-86 as described under "Experimental Procedures." mKKPftf was constructed by replacing the two arginines following the H2 of mPftf to two lysines. The topology of mPftf in the thylakoids was deduced from experimental results presented in this paper. Charged residues in the lumenal tail are indicated with a "ϩ" (K or R) or "Ϫ" (D or E). 13-kDa degradation product following lysis of the chloroplasts and protease treatment of the membranes (Fig. 3A, lanes 9 and  10 compare with lanes 4 and 5). This indicates that the twin arginine motif is not essential for Pftf integration.
The ability of pKKPftf and mPftf to integrate into isolated thylakoid membranes was also examined. When incubated with washed thylakoids (Fig. 3B), pKKPftf was processed to mature size and became integrally associated with thylakoids as judged by resistance to extraction with NaOH, similar to pPftf (Fig. 3B, lanes 6 and 7). mPftf was also integrated into washed thylakoids (lanes 8 and 9) and produced the 13-kDa protease degradation product (Fig. 3C, lane 5), which was further resistant to base extraction, indicating that it was an-chored in the membrane (Fig. 3C, lane 6). These data indicate that Pftf does not require the signal peptide for membrane targeting or integration. A time course of integration determined that mPftf integrates into washed thylakoids at approximately the same rate as pPftf, whereas pKKPftf integrates at only about 50% the rate of pPftf (data not shown).
pPftf, pKKPftf, and mPftf Share Identical Requirements for Integration into Thylakoids-To assess whether pKKPftf and mPftf utilize the same mechanism for integration as pPftf, the stromal and energetic requirements of pKKPftf and mPftf were determined in simultaneous assays with pPftf, pOE17, and iOE33 (Fig. 4). Like pPftf integration (Fig. 4A), pKKPftf (Fig.  4B) and mPftf (Fig. 4C) integration did not require stromal factors (lanes 2 and 7) or ATP (lanes 3 and 7). Integration was dependent on the proton motive force as it was inhibited by ionophores in the light (lane 6) and was ATP-dependent in the dark (lanes 4 and 5), i.e. via reverse action of the ATP synthase (6). In another experiment, the proton motive force requirement of Pftf was further dissected by conducting assays in the presence of increasing levels of nigericin or valinomycin separately. The protonophore nigericin inhibited Pftf integration, whereas the electrogenic ionophore valinomycin had no effect on Pftf integration at levels as high as 1 M (data not shown). This indicates that it is the ⌬pH component of the proton motive force, rather than the ⌬⌿ component, that is required for Pftf integration. Thus all forms of Pftf assayed exhibited the identical translocation requirement, a pH gradient but not ATP, similar to the ⌬pH substrate pOE17 (Fig. 4D) and very different from the Sec pathway substrate iOE33 (Fig. 4E).
Thylakoid Integration of pPftf, pKKPftf, and mPftf Are Competed by the ⌬pH Precursor OE23-Considering the fact that all three Pftf constructs still required only a pH gradient for translocation, it was important to subject them to other criteria used to assess utilization of the ⌬pH pathway. Previous studies (26)  5 and 14), pKKPftf (lanes 6 -10 and 15), or mPftf (lanes [11][12][13] and ATP for 30 min. Following incubation chloroplasts were repurified without (C) or with (CP) protease treatment. Repurified chloroplasts were extracted with 0.1 M NaOH (CN) or used to prepare total membranes, which were treated with protease in the absence (TP) (lanes 5 and 10) or presence (PD) (lanes 14 and 15) of 1% Triton X-100. B and C, thylakoid integration assays with pPftf, pKKPftf, and mPftf. Washed pea thylakoids were incubated with in vitro translated pPftf (lanes 1, 4, and 5), pKKPftf (lanes 2, 6, and 7), or mPftf (lanes 3, 8, and 9) for 20 min in the light. Thylakoids recovered from assays were divided into equal aliquots. B, one aliquot was washed with import buffer (T) and the second was extracted with 0.1 M NaOH (TN). The sample in lane 10 was identical to lane 4 of A and was included as a standard for mPftf obtained from a chloroplast import assay. C, the remaining aliquots were treated with protease. Following proteolysis, samples were washed with buffer (P) or extracted with 0.1 M NaOH (PN). Samples were analyzed by SDS-PAGE and fluorography on either 12.5% gels (A and C) or a 7.5% gel (B). In the figure, translation products (tp) and samples deriving from assays with pPftf, pKKPftf, and mPftf are designated as p, pKK, and m, respectively, above the panels.

FIG. 4. pPftf, pKKPftf, and mPftf integration require the thylakoidal proton motive force but not stromal factors or NTPs.
Integration (pPftf, pKKPftf, and mPftf) and transport (pOE17 and iOE33) reactions were conducted with pea thylakoids under conditions designed to test requirements for stroma, ATP, and proton motive force. Assays were conducted for 15 min at 25°C. Assay conditions, depicted above the panels, included the absence or presence of stromal extract (i.e. assays with washed thylakoids or reconstituted lysate), the absence or presence of white light, 5 mM ATP, nigericin and valinomycin (N/V), and apyrase to scavenge all residual ATP and GTP. Thylakoids recovered following incubation were protease-treated and analyzed by 12.5% SDS-PAGE and fluorography. Gel lanes contained 1 l of translation products (tp) and thylakoids recovered following incubation with 6.25 l of translation product. established conditions by which saturating concentrations of unlabeled overexpressed precursors competed for translocation of precursors employing the same pathway. Fig. 5 shows the results of two competition experiments. In Fig. 5A, translocation into isolated thylakoids was competed with E. coli-produced tOE23, an efficient substrate of the ⌬pH pathway (8). Insertion of pPftf was partially competed by tOE23 (lanes 3-5). Interestingly, mPftf integration was much more strongly competed; elevated concentrations of tOE23 virtually eliminated mPftf integration. The response of the controls, pOE17 for the ⌬pH pathway and iOE33 for the Sec pathway, verified the specificity of this competitor.
Competition experiments were also carried out in organello, wherein competition for thylakoid translocation occurs within intact chloroplasts. In these assays, competitive precursors are imported into the chloroplasts at a rate much faster than thylakoid transport, allowing saturating levels of intermediates to accumulate. In organello competition more closely represents physiological conditions as all competing intermediates have first been imported across the plastid envelope and are completely competent for thylakoid transport (26). The results of in organello competition of pPftf and pKKPftf are shown in Fig. 5B. Competition with the ⌬pH pathway precursor pOE23 inhibited Pftf integration as evidenced by the accumulation of about 50% of the imported Pftf as iPftf (lanes 5-7). A significant percentage of iPftf accumulated in the stromal fraction (lane 6). pOE23 competition of pKKPftf virtually eliminated thylakoid integration and nearly all of the resulting iKKPftf accumulated in the stroma (lanes 5-7). Interestingly, iKKPftf also accumulated in the stroma when thylakoid integration was blocked with ionophores (lanes 11 -13). As can be seen, thylakoid localization of Pftf and KKPftf was unaffected by the Sec pathway competitor pOE33 (lanes 8 -10) compared with the "no competitor" control (lanes 2-4). Competition experiments with radio-labeled pOE17 and pOE33 (Fig. 5B, lower panels) verified the efficacy and specificity of the competitors.
Antibodies to Components of the ⌬pH Pathway Machinery Inhibit pPftf and mPftf Integration-We recently showed that antibodies to Hcf106 and Tha4, two components of the ⌬pH pathway, specifically inhibit translocation of ⌬pH pathway substrates but not those of the Sec or SRP pathway (17). As shown in Fig. 6, antibodies to either Hcf106 (lane 2) or Tha4 (lane 5) inhibited integration of pPftf or mPftf. Inhibition was prevented by including 20 M of the respective antigens, hcf106sd (lane 3) or tha4sd (lane 7), during the antibody-binding step. Inhibition was greater for mPftf than pPftf. The controls for this experiment, iOE23 and pPSII-T for the ⌬pH pathway and pLHCP for the SRP pathway, demonstrated the specificity and efficacy of the treatments. In other experiments (data not shown), integration of pPftf, pKKPftf, and mPftf was unaffected by antibodies to cpSecY under conditions that virtually eliminated transport of Sec pathway substrates in parallel assays.
Pftf Was Imported into hcf106 Chloroplasts, but Failed to Integrate into the Membranes-Hcf106 was originally identified through genetic experiments in which mutant maize plants were specifically defective in thylakoid transport of ⌬pH pathway precursor proteins both in vivo and in vitro (9,34). As an additional test of whether Pftf employs ⌬pH machinery, chloroplasts were isolated from hcf106 maize seedlings and used in chloroplast import assays (Fig. 7). The hcf106 chloroplasts were devoid of the Hcf106 protein, but contained normal amounts of Tha4 (Fig. 7B). pPftf was imported into wild type maize chloroplasts and accumulated as the integrated form, as determined by NaOH extraction (Fig. 7A). In contrast, although pPftf was imported normally into hcf106 chloroplasts, it was not integrated into thylakoids as it was extracted from the thylakoid fraction with NaOH (Fig. 7A). In the experiment FIG. 5. Integration of pPftf, mPftf, and pKKPftf is competed by saturating levels of ⌬pH pathway precursors. A, competition for integration into isolated thylakoids. Integration of radiolabeled pPftf and mPftf and transport of pOE17 and iOE33 into isolated pea thylakoids was conducted in the presence of increasing concentrations of unlabeled E. coli-synthesized tOE23 (depicted above the panels) as described under "Experimental Procedures." Recovered thylakoids were protease-treated prior to analysis by 12.5% SDS-PAGE/fluorography. The radiolabeled substrate in each reaction is designated to the left of each fluorogram. Gel lanes contained 0.5 l of translation product, and thylakoids recovered following incubation with 3.12 l of translation product. B, in organello competition for thylakoid transport and integration. Intact pea chloroplasts were incubated with 5 mM ATP and 0.75 M unlabeled pOE23, 0.6 M pOE33, or without competitor (Control) for 7 min at 25°C in the light. Radiolabeled precursors (shown to the left of the panels) were then added to the reaction mixtures, and the incubation was continued for another 10 min. Parallel assays were conducted with radiolabeled precursors and nigericin/valinomycin (N/V)-treated chloroplasts for 10 min for comparison with thylakoid integration blocked by dissipating the ⌬pH. Chloroplasts were recovered from the assays by centrifugation through Percoll cushions and were analyzed directly (C) or subfractionated into stroma (S) and thylakoids (T). For assays with pOE17 and pOE33, only the recovered chloroplasts are shown. Assays were analyzed by SDS-PAGE and fluorography on 7.5% gels (Pftf and pKKPftf) or 11.5% gels (pOE17 and pOE33). Gel lanes contain the equivalent of 0.25-l translation products and samples following incubation with 2.2-l translation products.
shown in Fig. 7, iPftf was not clearly resolved from mPftf. However, in other experiments, it was apparent that only iPftf accumulated in hcf106 chloroplasts, whereas it was largely mPftf that accumulated in wild type chloroplasts (data not shown). The ⌬pH pathway control for this experiment, pOE17, was imported into wild type chloroplasts and accumulated normally as the mature, lumenal form, whereas it accumulated in hcf106 chloroplasts primarily as the stromal intermediate iOE17 (Fig. 7A, lower panel). The Sec pathway control, pOE33, was imported and accumulated as the mature, lumenal form in both hcf106 and wild type chloroplasts (Fig. 7A, lower panel).
A Conserved Twin Arginine Carboxyl Proximal to the H2 Anchor Is Not the ⌬pH Pathway cis-Targeting Element-Examination of Pftf sequences revealed the presence of an RR flanking the hydrophobic anchor on the carboxyl side (Fig. 1). We assessed the possibility that the ⌬pH machinery can recognize a twin arginine motif on either end of a hydrophobic segment by constructing a modified mPftf (mKKPftf) in which the carboxyl flanking RR was changed to KK. mKKPftf integration was analyzed in the absence or presence of ionophores and with thylakoids preincubated with antibodies to ⌬pH pathway components (Fig. 8). Similar to mPftf, mKKPftf integration was prevented by dissipation of the thylakoidal pH gradient (Fig. 8,  lanes 2 and 3). We recently prepared antibodies to pea Hcf106 and pea chloroplast TatC. Preincubation of thylakoids with either of these antibodies specifically inhibits the transport of ⌬pH pathway substrates but does not affect the proton gradient. 2 As shown in Fig. 8, either antibody inhibited mPftf or mKKPft integration (lanes 5 and 7). Specificity of the antibody effect is evident by the lack of effect of pre-immune IgGs (lanes 4) and prevention of antibody inhibition by co-incubation with the corresponding antigens (lanes 6 and 8). In addition, parallel assays verified the lack of effect of antibodies on transport of the Sec pathway substrate OE33 and the SRP pathway substrate LHCP (data not shown).

DISCUSSION
In this work we analyzed Pftf integration into thylakoids and found that it employs the ⌬pH pathway machinery. This is the first membrane protein found to be a substrate for the ⌬pH pathway; all other known substrates are soluble proteins of the thylakoid lumen. Several lines of evidence argue for this conclusion. First Pftf integration did not require stromal factors and relied entirely on the pH gradient as an energy source for integration (Fig. 4). Second, integration was competed by saturating quantities of ⌬pH pathway precursors but not by Sec pathway precursors (Fig. 5). Third, antibodies to the ⌬pH machinery specifically inhibited Pftf integration (Figs. 6 and 8). Finally, hcf106 mutant chloroplasts were unable to integrate Pftf into thylakoids, even though they efficiently imported the precursor into the organelle (Fig. 7). These results indicate FIG. 6. Antibodies to zmHcf106 and psTha4 inhibit integration of pPftf and mPftf. Pea thylakoids were incubated in import buffer plus 10 mM MgCl 2 and 1% bovine serum albumin with or without anti-pea Tha4 IgG (0.5 mg/ml) or anti-maize Hcf106 IgG (1 mg/ml) in the presence or absence of 20 M antigen, tha4sd, or hcf106sd, for 1 h on ice as described (17) and depicted above the panels. Thylakoids were then washed with import buffer plus 10 mM MgCl 2 and assayed for insertion of pPftf, mPftf, iOE23, pPSII-T, or pLHCP in a reaction containing ATP and stroma in the light at 25°C for 30 min. Reactions were stopped on ice, and the samples were subjected to protease posttreatment. Samples were analyzed by 12.5% (pPftf, mPftf, LHCP, and iOE23) SDS-PAGE and fluorography. PSII-T assay samples were acetone-extracted prior to analysis by 16.5% tricine gel electrophoresis. Gel lanes contain equivalent quantities of thylakoids recovered from the assays.

FIG. 7. Insertion of Pftf is inhibited in hcf106 mutant plants.
Chloroplasts were isolated from wild type and hcf106-mum3 maize leaves and described under "Experimental Procedures." A, chloroplasts were assayed for import and localization of pPftf, pOE17, and pOE33. Intact chloroplasts were repurified following the assays by centrifugation through Percoll cushions and analyzed directly (C) and also subfractionated into stroma (S) and thylakoids (T). Thylakoids were also extracted with 0.1 M NaOH (TN). For assays of pOE17 and pOE33 import, only the recovered chloroplasts are shown. B, chloroplasts from mutant hcf106 and wild type seedlings were subjected to immunoblotting with antibodies to Hcf106 and Tha4. Each lane contained chloroplasts equivalent to 1.75 g of chlorophyll.
that, like the Sec system, the ⌬pH pathway can distinguish between segments of hydrophilic peptide to be completely transported to the lumen and segments of hydrophobic peptide to be released into the bilayer.
One surprising and novel feature of Pftf integration is that it required neither the RR motif of the signal peptide nor even the hydrophobic core domain. pKKPftf integrated faithfully into thylakoids (Figs. 3 and 4) and exhibited all of the hallmarks of ⌬pH pathway translocation. mPftf, which lacks the entire signal peptide, was similarly integrated into thylakoids by the ⌬pH pathway (Figs. 3 and 4). Nevertheless, several observations indicate that the RR motif and the hydrophobic core of the signal peptide contribute to efficient integration. pKKPftf integrated into isolated thylakoids with only ϳ50% of the efficiency of pPftf. Although mPftf exhibited the same rate of integration into isolated thylakoids, as did pPftf, other characteristics demonstrate a weaker interaction with the ⌬pH pathway machinery. For example, mPftf and pKKPftf were competed by lower concentrations of overexpressed ⌬pH pathway precursors than pPftf (Fig. 5). Because effective concentrations for competition are related to K m values of the competing substrates, this indicates that mPftf and pKKPftf interact with ⌬pH pathway machinery more weakly than pPftf. In support of this conclusion, when pKKPftf was imported into chloroplasts under conditions that prevented integration, iKKPftf accumu-lated in the stroma, rather than bound to the membrane surface as did iPftf (Fig. 5B). We interpret these data to mean that the RR-containing signal peptide of pPftf promotes a strong interaction with the ⌬pH pathway machinery but is not essential for integration. This differs from previous analyses of lumenal protein substrates, where conservative substitution of the RR eliminates transport (7,8). Studies of the bacterial Tat pathway have shown that conservative substitution of a single arginine of some substrates reduces but does not eliminate transport (11,35). However, this is the first example where deleting the entire signal peptide did not eliminate translocation.
The fact that mPftf integrates with a reasonable efficiency suggests the presence of a second targeting signal. This could be a redundant signal that serves to simply increase the interaction of the precursor with the machinery or it might be a cryptic signal that is unmasked by deletion of the signal peptide. The most likely second signal is the H2 membrane anchor of Pftf, which presumably would promote translocation of the amino terminus into the lumen (N-tail translocation). Interestingly, chloroplast Pftf proteins possess conserved twin arginines flanking H2 on the carboxyl proximal side. Although RR has never been shown to function on the carboxyl side of a hydrophobic core, we examined integration of a modified mPftf in which the RR was replaced with two lysines. mKKPftf was still integrated into thylakoids by the ⌬pH pathway (Fig. 8). Thus, the role of H2 in targeting remains to be dissected in future studies.
We also considered the possibility that the mPftf-targeting element allows Pftf to access another translocation pathway that conceivably shares ⌬pH pathway components. Antibody inhibition experiments argued against the involvement of the Sec translocon. The fact that mPftf must translocate a large N-tail to the lumen suggested that the chloroplast Oxa1p homologue might be involved. In mitochondria, Oxa1p has been shown to facilitate N-tail translocation of moderately sized amino-flanking peptides across the inner membrane (5,36,37), and antibody inhibition studies show that the chloroplast homologue is involved in integration of the membrane protein LHCP into thylakoids (38). However, recent experiments in our lab did not find an effect of antibodies to cpOxa1p on mPftf integration. 3 We therefore conclude that the ⌬pH protein transport system is an exceptionally versatile translocation system, i.e. it can transport signal peptide-containing precursors, large folded polypeptide domains, and large, charged N-tails. In addition to significantly expanding the range of known capabilities of the ⌬pH system, our studies refocus attention on the puzzling multiplicity of export systems across prokaryote-derived membranes.
FIG. 8. A conserved RR carboxyl proximal to the H2 anchor is not required for ⌬pH pathway integration of mPftf. Integration mPftf and mKKPftf was assayed with washed thylakoids or thylakoids preincubated with IgGs. Thylakoids were pretreated with 1 mg/ml preimmune IgGs (PI), 0.2 mg/ml anti-psHcf106 IgGs (␣106), or 1 mg/ml anti-pea plant chloroplast TatC homologue IgGs (␣TatC), respectively, as described under "Experimental Procedures" and depicted above the panels. Where designated, thylakoid preincubation was conducted in the presence of 20 M antigen, either pshcf106sd or plant chloroplast TatC homologue peptide. Assays were conducted for 25 min in the absence or presence of nigericin and valinomycin (N/V) as designated above the panels. Assays were terminated with nigericin and valinomycin, and the recovered thylakoids were treated with thermolysin. Analysis was by SDS-PAGE and fluorography on a 12.5% gel. Gel lanes contain equivalent quantities of thylakoids recovered from the assays.