Functional assembly of thylakoid ΔpH-dependent/Tat protein transport pathway components in vitro


K. Cline, Horticultural Sciences Department, Box110690, University of Florida, Gainesville, Florida 32611, USA.
Fax: + 1 352 392 5653, Tel.: + 1 352 392 4711 extn 219,


Assembly of the components of the thylakoid ΔpH-dependent/Tat protein transport machinery was analyzed in vitro. Upon incubation with intact chloroplasts, precursors to all three components, Hcf106, cpTatC and Tha4, were imported into the organelle and assembled into characteristic endogenous complexes. In particular, all of the imported cpTatC and approximately two-thirds of the imported Hcf106 functionally assembled into 700 kDa complexes capable of binding Tat pathway precursor proteins. The amounts assembled into thylakoids by this procedure were moderate. However, physiological quantities of mature forms of Tha4 and Hcf106 were integrated into isolated thylakoids and a significant percentage of the Hcf106 so integrated was assembled into the 700 kDa complex. Interestingly, a mutant form of Hcf106 in which an invariant transmembrane glutamate was changed to glutamine integrated into the membrane but did not assemble into the receptor complex. Analysis of energy and known pathway component requirements indicated that Hcf106 and Tha4 integrate by an unassisted or ‘spontaneous’ mechanism. The functionality of in vitro integrated Tha4 was verified by its ability to restore transport to thylakoid membranes from the maize tha4 mutant, which lacks the Tha4 protein. Development of this functional in vitro assembly assay will facilitate structure–function studies of the thylakoid Tat pathway translocation machinery.

p and m

precursor and mature forms of proteins


blue native polyacrylamide gel electrophoresis


light-harvesting chlorophyll a–b complex

Most thylakoid proteins are encoded in the nucleus and synthesized in the cytosol as precursor proteins (reviewed in [1]). Studies of a variety of different thylakoid proteins support a two-step assembly pathway in which precursors are first imported across the chloroplast envelope into the aqueous stroma. In the stroma, their chloroplast-targeting peptides are removed by a stromal processing peptidase, releasing intermediate precursors that are recognized and incorporated into thylakoids by translocation machinery present in stroma and thylakoids. Three thylakoid translocation machines (or translocases) have been identified; a chloroplast Sec-dependent system, a chloroplast SRP-dependent system, and a ΔpH-dependent system also called the chloroplast Tat pathway (reviewed in [1–3]). In addition, a subset of thylakoid membrane proteins is inserted into the membrane by an unassisted or ‘spontaneous’ mechanism (reviewed in [4]). All of the identified components of thylakoid translocases are encoded in the nucleus. Although the import and assembly pathways of substrates of these systems have been worked out in some detail, virtually nothing is known regarding the pathways and mechanisms involved in localizing the membrane components of the translocases. One important reason for understanding their assembly pathways regards the origins and identity of the thylakoid membrane. Thylakoid translocases serve as receptors for newly synthesized thylakoid proteins and therefore determine the unique protein makeup of the thylakoid membrane and lumen. Because thylakoids are not present in progenitor plastids, but seem to derive from the inner envelope membrane during chloroplast development [5,6], understanding the manner by which translocase proteins are targeted to and inserted into the membrane may provide insight into the manner by which thylakoid identity is established.

A second reason to examine translocase component assembly is to generate tools for dissecting their mechanism of action. The ability to reconstitute and analyze the proper integration of components into the membrane is a prerequisite for biochemical studies of structure--function relationships of the individual components. Thylakoids are particularly amenable to in vitro incorporation of proteins. Not only are proteins integrated into the membrane or transported into the lumen in vitro, but many also appear to be correctly assembled into endogenous complexes (reviewed in [7]). This offers the opportunity to biochemically replace missing or inactivated components.

We are especially interested in the thylakoid Tat pathway translocase. This system transports folded proteins across the lipid bilayer using only the thylakoidal pH gradient as energy source. Precursors transported by this pathway contain essential twin arginine residues in their signal peptides; hence the designation Tat for ‘twin arginine translocation’. Three components of the machinery have been identified in thylakoids: Hcf106, Tha4 and cpTatC [2]. Hcf106 and Tha4 are homologous proteins with similar structures; they appear to be anchored to the membrane by an amino proximal transmembrane domain and expose a predicted amphipathic helix and an acidic C-terminal domain to the stroma. Hcf106 and Tha4 share sequence similarity in the transmembrane domain and amphipathic helices. Particularly striking is the presence of certain highly conserved motifs in both proteins. For example, they both possess a conserved glutamate residue in their predicted transmembrane domain, which theoretically should destabilize transmembrane helix insertion unless it is neutralized in some manner. Despite their structural similarities, Hcf106 and Tha4 seem to participate in different steps of the translocation process [8,9]. cpTatC is an integral membrane protein with six predicted membrane spanning helices and its amino and carboxyl termini exposed to the stroma [10,11]. Bacteria and certain archaea possess protein transport systems that are homologous to the thylakoid Tat system and appear to operate by similar principles [12,13].

Here we show that in vitro synthesized thylakoid Tat components assemble into isolated chloroplasts and thylakoids in functional form. Hcf106 and Tha4 are imported across the chloroplast envelope and then insert into thylakoids. They are also very efficiently assembled by presenting the mature forms of Hcf106 and Tha4 to isolated thylakoids. Integration occurs by an apparently spontaneous mechanism. cpTatC was assembled into thylakoids when the precursor protein was presented to intact chloroplasts, although the pathway taken to the thylakoids is unclear. cpTatC was not capable of integrating directly into isolated thylakoids under a variety of different conditions. Our data show that in vitro integrated Hcf106 and cpTatC assemble into a functional 700-kDa receptor complex. In vitro integrated Tha4 was also functionally assembled as evidenced by its ability to biochemically complement the Tat transport activity of thylakoids from maize tha4 plants, which are devoid of Tha4. This offers a powerful tool for unraveling the mechanism Tat-pathway transport.

Experimental procedures


Reagents were obtained from commercial sources. Antibodies to pea Hcf106, Tha4, cpTatC, cpSecY and cpOxa1p have been described [8,10,14]. Antibodies to maize Hcf106 were as described [14] and antibodies to maize Tha4 were the generous gift of A. Barkan [15]. Antibodies to Toc75 and Toc110 were the generous gift of A. Barkan (University of Oregon, Eugene, OR, USA) and D. Schnell (University of Massachusetts, Amherst, MA, USA).

Preparation of precursor proteins

Cloning and analysis of DNA products were by standard molecular biology procedures. Amplifications were performed with Pfu polymerase (Stratagene). Cloned constructs were verified by DNA sequencing of all clones on both strands at the University of Florida Interdisciplinary Center for Biotechnology Research DNA Sequencing Core Facility. The mature form of pea Hcf106 (mHcf106) was cloned by PCR amplification from pHcf106 [10] based on the transit peptide cleavage site predicted by ChloroP [16] and alignment with other orthologous proteins. The 5′ primer (including an engineered EcoRI site) was used to mutate the nucleotides encoding tyrosine 86 to encode methionine such that the amino terminus of the resulting protein began MASLFGVGAPEALVI…; the 3′ primer bound in the pGEM 4Z vector. The resulting product was ligated into pGEM 4Z at the EcoRI and SstI sites in the SP6 direction. mHcf106 residues are numbered beginning with the initiator methionine. An altered form of mHcf106 (mHcf106 E11Q) was derived by PCR amplification using a 5′ primer that mutated nucleotides encoding glutamate 11 of the engineered mHcf106 to glutamine. The mature form of pea Tha4 (mTha4) was cloned by PCR amplification from pTha4 [14] based on the predicted transit peptide cleavage site from a combination of ChloroP [16] and alignment with orthologous proteins. The 5′ primer (including an engineered KpnI site) was used to mutate the nucleotides encoding asparagine 56 to encode methionine such that the resulting protein started MAFFGLGVPELVV…; the 3′ primer bound in the pGEM 4Z vector. The resulting product was ligated into pGEM 4Z at the KpnI site in the SP6 direction. mTha4 residues are numbered beginning with the initiator methionine. An altered form of mTha4 (mTha4 E10Q) was derived by PCR amplification using a 5′ primer that mutated nucleotides encoding glutamate 10 of the engineered mTha4 to glutamine. The mature form of pea TatC was cloned by PCR amplification from pTatC as described in Mori et al. [10]. The 5′ primer (including an engineered EcoRI site) was used to mutate the nucleotides encoding residues 49 and 50, leucine/valine, to encode methionine/alanine such that the resulting protein began MACFAVDDEIRE…; the 3′ primer bound in the pGEM 4Z vector. The resulting product was ligated into pGEM 4Z at the EcoRI and BamHI sites in the SP6 direction.

Preparation of radiolabeled precursors

In vitro 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. For some experiments, transcripts were produced separately by transcription with SP6 polymerase and translation with a homemade wheat germ translation system [17]. Translation products were diluted with 1 vol. 60 mm leucine in 2× import buffer (1× = 50 mm Hepes/KOH pH 8.0, 0.33 m sorbitol) prior to use unless otherwise indicated in the figure legend.

Preparation of chloroplasts, thylakoids and lysate

Intact chloroplasts were isolated from 9- to 10-day-old pea seedlings [18] and were resuspended in import buffer at 1 mg·mL−1 of chlorophyll. Maize plants were grown at 20 °C in a 12 h light/12 h dark cycle for 7–10 days. Mutant tha4/tha4 maize seedlings were selected by their pale green phenotype and by high chlorophyll fluorescence with a hand-held UV lamp. Maize chloroplasts were isolated as described [14]. Chloroplast lysate, washed thylakoids and stromal extract were prepared from isolated chloroplasts [18]. Chlorophyll concentrations were determined according to Arnon [19]. Protein was determined by the BCA method according to the manufacturer's instructions (Pierce).

Chloroplast import and thylakoid protein integration assays

Import of radiolabeled precursors into isolated chloroplasts or integration into washed thylakoids or chloroplast lysate was conducted in microcentrifuge tubes in a 25 °C water bath illuminated with 70 µE·m−2·s−1 white light in the presence of 5 mm MgATP [18] for the times indicated in the figure legends. Assays were terminated by transfer to 0 °C. Where indicated, recovered chloroplasts or thylakoids were protease post-treated with thermolysin [18]. Chloroplasts were repurified on Percoll cushions and washed in import buffer. Chloroplasts recovered from import assays were subfractionated by lysis in 100 µL 10 mm Hepes/KOH pH 8 for 5 min followed by addition of 20 µL of 2× import buffer. Thylakoids were pelleted in a swing-out microcentrifuge at 5000 g for 30 s followed by washing in import buffer. Envelope membranes were recovered from the 5000 g supernatant by centrifugation at 50 000 g for 30 min. Where designated, thylakoid membranes were washed with 0.5 mL 0.2 m Na2CO3 or 0.1 m NaOH for 60 min on ice and the thylakoids were then recovered by centrifugation at 30 000 g for 15 min.

Quantitative immunoblots

Immunoblots were developed by the ECL procedure (Amersham). For quantification of in vitro integrated proteins, translation products were run on SDS/PAGE in parallel with dilution series of Hcf106 stromal domain or Tha4 stromal domain standards [10]. Proteins were electroblotted to nitrocellulose membranes and then immunodecorated with the appropriate antibodies. The density of scanned bands on X-ray film was determined using alphaease software and protein quantities were estimated by comparison to standards in the linear exposure range of the film. Samples of the same translation products and thylakoids recovered from the corresponding integration assays were separated by SDS/PAGE and the bands visualized by fluorography. Bands in the linear range of the film were quantified as above. The amounts of Hcf106 and Tha4 associated with thylakoids were then calculated from their relative band density and from the ratio of micrograms protein per unit band density of the translation products.

Blue native gel electrophoresis

Washed thylakoids were dissolved in 1% digitonin and subjected to blue native (BN) PAGE as described by Cline and Mori [8]. Gels were analyzed by fluorography or subjected to immunoblotting as described [8]. Molecular markers used for blue native gels were ferritin (880 kDa and 440 kDa) and BSA (132 kDa and 66 kDa).

Measurement of the pH gradient across maize thylakoid membranes

The ΔpH generated across maize thylakoid membranes was measured by the 9-aminoacridine method essentially as described by Mills [20]. Intact chloroplasts were lysed by dilution into 10 mm Hepes/KOH pH 8, 10 mm MgCl2 and after 5 min they were adjusted with an equal volume of 2× import buffer containing 20 mm dithiothreitol, 30 µm 9-aminoacridine, and 20 µm methyl viologen. Fluorescence was measured in a Shimadzu RF-5000 fitted with a light emitting diode to generate actinic light at 643 nm. The fluorescence excitation wavelength was set to 360 nm and the emission wavelength to 490 nm. Fluorescence quenching was measured in the presence of actinic light; the sample then received 6 mm Mg-ATP, and the additional fluorescence quenching was remeasured with a correction for direct quenching by ATP. The ΔpH was calculated from fluorescence quenching as described by Mills [20], assuming a lumenal volume of 20 µL per mg chlorophyll [21].


In vitro translated ΔpH-dependent/Tat components are integrated into thylakoid membranes

As reported previously [10,14], in vitro translated pHcf106 and pTha4 are imported into intact chloroplasts, processed to mature size, and integrated into thylakoids (Fig. 1A, lanes 1–7). Several additional features of in vitro integration are demonstrated below. First is that small amounts of imported and processed Hcf106 and Tha4 are recovered with the envelope fraction (Fig. 1A, lane 3). Experiments with Hcf106 that included markers for envelope and thylakoid membranes showed that thylakoid contamination could not account for the envelope-associated Hcf106 (data not shown).

Figure 1.

In vitro-translated Hcf106 and Tha4 become integrally associated with thylakoids. (A)  In vitro translated 3H-labeled pTha4 and pHcf106 were incubated with pea chloroplasts (Import) and 3H-labeled mTha4 and mHcf106 were incubated with chloroplast lysate (Integration) and 5 mm ATP for 25 min in the light at 25 °C. Recovered chloroplasts were lysed and subfractionated into envelope (E), stroma (S), and thylakoids as described in Experimental procedures. Recovered thylakoids were washed with import buffer (T), with 0.2 m Na2CO3 (TC) or 0.1 m NaOH (TOH), or treated with thermolysin (T+) as designated above the panels. Samples were analyzed by SDS/PAGE and fluorography. The positions of pTha4, mTha4 pHcf106, and mHcf106 are designated to the left of the panels. Lanes: tp, translation product equivalent to 0.15% of that added to the assay; lanes 2–12, soluble or membrane fractions equivalent to 5% of the assay. (B,C) Proteolysis of in vitro integrated mHcf106 and mTha4 to detect membrane-embedded segments. Thylakoid membranes recovered from integration assays with mTha4 (B, lanes 1–6), mTha4 E10Q (B, lanes 7–12), mHcf106 (C, lanes 1–7), or mHcf106 E11Q (C, lanes 8–14) conducted as described in (A) were resuspended in import buffer at 0.167 mg chlorophyll·ml−1. Protease reactions were initiated by adding thermolysin or trypsin to a final concentration of 80 µg·mL−1. Reactions were conducted on ice for times designated above each panel (in min). Mock-treated samples (B, lanes 1, 7; C, lanes 1, 8) were incubated without protease for 40 min. Reactions in B, lanes 6 and 12 and C, lanes 6 and 13 were sequential treatments in which thylakoids were treated with thermolysin for 20 min, the thylakoids pelleted and resuspended in import buffer containing 80 µg·mL−1 trypsin, and the reaction continued for an additional 20 min. Samples in C, lanes 7 and 14 (*) represent an aliquot of the sequential treatment removed before addition of trypsin. Thermolysin treatments were terminated with 3 vols 14 mm EDTA in import buffer; trypsin treatments were terminated with 2 mm phenylmethanesulfonyl fluoride 150 µg·mL−1 soybean trypsin inhibitor, and 150 µg·mL−1 aprotinin. Recovered membranes were analyzed on 16% Tricine/SDS gels followed by fluorography. Radiolabeled proteins were extracted from gels slices and quantified by liquid scintillation counting [17]. Numbers below the bands represent the percentage of radiolabel contained in each band and are average values obtained from two identical experiments. Radiolabel in the mock-treated band was arbitrarily set to 100%.

Immunoblot analysis of chloroplast subfractions was used to assess the distribution of endogenous Hcf106 and Tha4 [Fig. 2]. Lanes were loaded with enriched fractions on an equal protein basis (lanes 1–4) and also in the approximate stoichiometric ratio that these membranes are present in chloroplasts (lanes 5–8). Both Tha4 and Hcf106 are primarily localized in thylakoids (lanes 5–8) but are also present in envelope fractions. This is especially apparent when equal amounts of protein are compared (lanes 1–4). Surprisingly, both components are present in the outer envelope fraction (lanes 4, 9). Cross-contamination of envelope subfractions, especially outer envelope in the inner envelope fraction, is common [22] and is seen in Fig. 2. However, cross-contamination does not account for the presence of Tha4 in the outer envelope fraction (compare lanes 7 and 8 for Tha4, the outer envelope marker Toc75, and the inner envelope marker Tic110).

Figure 2.

Distribution of endogenous components in chloroplast subfractions. Isolated intact chloroplasts were subfractionated into thylakoids (T), stroma (S), inner envelope membrane (IE), and outer envelope membrane (OE) by a combination of differential and sucrose gradient centrifugation as described by Keegstra and Yousif [36] with the exception that after freezing and thawing, the chloroplast suspension was subjected to five strokes of a glass homogenizer. A second preparation (not shown) omitted the freeze–thaw step and the chloroplasts were ruptured by 20 strokes of a glass homogenizer. Essentially the same immunoblot results were obtained for both preparations. Samples were loaded such that each lane contained the same quantity of total protein (left half of panels) or in the approximate stoichiometric ratio that each fraction represents in chloroplasts (right half of panels). Antibodies used for immunoblotting and their target proteins are shown to the left of panels. Toc75 and Tic110 are integral proteins of the outer and inner envelope membranes, respectively. The inset shows immunoblots of cpSecY, cpOxa1p, and cpTatC, respectively, with higher levels of envelope proteins (8 µg of T, S, IE and 5 µg OE protein) loaded per lane.

Incubation of in vitro translated mature Tha4 (mTha4) and mature Hcf106 (mHcf106) with isolated thylakoids resulted in their tight association with the membrane (Fig. 1A, lanes 8–12). Previous analysis of endogenous components established that mHcf106 and mTha4 are resistant to a 0.2 m sodium carbonate wash and are largely degraded by protease treatment, suggesting that these components are inserted into the thylakoid bilayer via a single predicted transmembrane domain [10,14]. As shown in Fig. 1A and as reported previously [23], integrated Hcf106 is also largely resistant to the more stringent 0.1 m NaOH extraction procedure (lanes 6, 11). In contrast, Tha4, either imported into chloroplasts or integrated into isolated thylakoids was largely extracted from the membrane by 0.1 m NaOH (lanes 6, 11). Endogenous Tha4 exhibits this same differential resistance to Na2CO3 and NaOH (E. H. Summer and K. Cline, unpublished results). This raised the question of whether Tha4 is truly anchored in the bilayer or only firmly bound to the surface of the membrane.

In order to answer this question, thylakoids were treated with protease and then analyzed on 16% Tricine/SDS gels for the presence of the predicted protease resistant transmembrane domains of Tha4 and Hcf106. It was not possible to analyze the endogenous proteins because our antibodies were raised only to the Tha4 and Hcf106 stromal domains. Therefore this analysis was conducted with thylakoids recovered from integration assays with radiolabeled mTha4 and mHcf106. Two different proteases were used. Thermolysin has numerous predicted cleavage sites within the transmembrane and amphipathic helical domains of Hcf106 and Tha4. However, because thermolysin sites in the amphipathic helices might be inaccessible, trypsin was also used to cleave at multiple sites on the charged side of the amphipathic helices.

Both thermolysin and trypsin produced a 2.5–3 kDa degradation product from integrated Tha4 (Fig. 1B, lanes 2–5). The estimated size of the Tha4 transmembrane domain and N terminus is 2.2 kDa. Sequential treatment with thermolysin followed by trypsin produced the same size band, suggesting that both enzymes digest the entire stromal domain of Tha4, leaving its imbedded transmembrane domain. When thermolysin treatment was conducted in the presence of 1% Triton X-100, Tha4 was completely degraded (data not shown). Based on the numbers of leucine residues in the transmembrane domain relative to the total number of leucines in the mature protein, 60% of the radiolabel should be present in the Tha4 degradation product. The degradation product produced by 10 min of proteolysis contained ≈ 50% of the radioactivity of mock-treated mTha4 (Fig. 1B, lanes 2, 4), but the percentage of radiolabel diminished with extended treatment time to less than one-third of the theoretical (lanes 3, 5, 6).

Thermolysin treatment produced an ≈ 4 kDa degradation product from integrated Hcf106 (Fig. 1C, lanes 2, 3). Trypsin produced a predominant product at 2.5–3 kDa, similar to the Tha4 degradation product, and a minor band at ≈ 8 kDa (lanes 4, 5). Sequential treatment with thermolysin followed by trypsin similarly yielded major and minor products at 2.5–3 kDa and 8 kDa, respectively (lane 6). The larger product may result from degradation of an Hcf106 aggregate that doesn't enter the gel because a sample removed after the thermolysin reaction prior to the trypsin reaction showed only the ≈ 4-kDa band (Fig. 1C, lane 7). The major product is most likely the protected Hcf106 transmembrane domain, which is predicted to be 2.3 kDa. The Hcf106 transmembrane domain contains 31% of the leucine resides of mHcf106. The major degradation product of trypsin or thermolysin plus trypsin contained about 30% of the radiolabel and appeared to be stable to extended protease treatment. As with Tha4, Hcf106 was completely degraded when thermolysin plus trypsin treatment was conducted in the presence of 1% Triton X-100 (data not shown). These results indicate that in vitro integrated Tha4 and Hcf106 are anchored in the membrane by their predicted transmembrane domains. Given the similar behavior of the in vitro integrated and endogenous proteins with respect to alkaline extractions and other characteristics (below), it is likely that the endogenous proteins are similarly anchored in the membrane.

In vitro translated cpTatC assembles into thylakoids when imported into chloroplasts, but not when presented directly to isolated thylakoids

Incubation of pcpTatC with intact chloroplasts resulted in its import, processing to mature size, and localization to the thylakoids (Fig. 3, lanes 1–5). Similar to Hcf106 and Tha4, some imported cpTatC was usually recovered in the envelope and stromal subfractions (lanes 2, 3). In contrast to endogenous Hcf106 and Tha4, endogenous cpTatC appears to be largely confined to the thylakoid membrane (Fig. 2). Only upon extended exposure of immunoblots containing greater amounts of envelope protein (5–8 µg) could trace amounts of cpTatC be detected in the inner envelope preparation (Fig. 2, inset). Whether the envelope and/or stromal cpTatC observed in vitro are assembly intermediates or off-pathway dead ends is currently under investigation.

Figure 3.

Import and integration assays with the precursor and mature form of cpTatC.In vitro assays for import of pcpTatC into chloroplasts were conducted as described in Fig. 1. Integration assays were conducted with chloroplasts lysate and ATP (as in Fig. 1) either with mcpTatC translation product alone (lanes 6–8) or with a mixture of mcpTatC and mHcf106 translation products either mixed after translation (lanes 12, 13) or translated in the same reaction mixture (lanes 14–16). For comparison, an integration assay with mHcf106 alone is included (lanes 9–11). The positions of the cpTatC precursor (pcpTatC), mature form (mcpTatC), two previously described degradation products (DP1 and DP2), and mHcf106 are designated on the sides of the panels. Samples designations shown above the lanes are as in Fig. 1.

All attempts to obtain significant integration of mcpTatC into isolated thylakoids were unsuccessful. Figure 3 shows that mcpTatC was not integrated into isolated thylakoids in the presence of stromal proteins, ATP and light (lanes 6–8). It has been reported [24] that Escherichia coli TatC is unstable in the absence of TatB. Accordingly, we attempted integration assays with a mixture of cpTatC and mHcf106 translation products (lanes 12, 13) and even translated mcpTatC and mHcf106 together prior to incubating with thylakoids (lanes 14–16). Although mHcf106 integrated efficiently (compare with lanes 9–11), there was no evidence that cpTatC became integrated into thylakoids (compare with lanes 6–8). The inability of cpTatC to integrate into isolated membranes makes it more difficult to determine its integration pathway.

Association of in vitro translated components with endogenous complexes

One important characteristic of endogenous components is their organization in complexes. cpTatC and a substantial percentage of Hcf106 are part of an ≈ 700-kDa complex [8]. A portion of Hcf106 and all of Tha4 is present in independent lower molecular mass complexes that vary in size with the concentration of digitonin used to solubilize the membranes. To determine if the in vitro integrated components assemble into comparable complexes, membranes recovered from import and integration assays were dissolved in 1% digitonin and subjected to BN/PAGE and fluorography. As shown in Fig. 4, cpTatC and Hcf106 imported into chloroplasts became associated with an ≈ 700 kDa complex (lanes 1, 2). A smaller but significant amount of the imported Hcf106 also migrated at ≈ 250 kDa (lane 2). Imported Tha4 migrated at ≈ 240 kDa (lane 3). These are the same profiles obtained for endogenous components solubilized under comparable conditions [8].

Figure 4.

Incorporation of in vitro-translated components into native complexes. Substrates were generated by coupled transcription--translation in wheat germ extract. Samples were analyzed by BN/PAGE and fluorography. Chloroplasts (Import) were incubated with ATP and translated precursors pcpTatC, pHcf106, and pTha4 as shown above the panel for 15 min in the light at 25 °C. Lysate (Integration) was incubated with ATP and translated mature proteins mHcf106, mTha4 mHcf106 E11Q, mTha4 E10Q as shown above the panel for 15 min in the light at 25 °C. Chloroplasts were repurified, lysed, and the thylakoids recovered by centrifugation. Recovered thylakoids from assays were washed, solubilized with 1% digitonin, and analyzed by BN/PAGE and fluorography (Experimental procedures). Positions of molecular weight markers are indicated to the left of the panel. Lanes labeled TP were loaded with translation product in BN sample buffer. Lanes labeled TP + Membr were loaded with translation product and solubilized membranes in BN sample buffer.

mHcf106 integrated into isolated thylakoids was also associated with a ≈ 700 kDa complex and with a ≈ 250 kDa band (lane 4). Two minor bands migrating between the 700 kDa and 250 kDa bands can be seen in lane 4, but these bands were not present in other similar experiments. Tha4 integrated into isolated thylakoids was predominantly present in a band at ≈ 240 kDa (lane 5). As controls for this experiment, mHcf106 and mTha4 translation products in 1% digitonin and translation products mixed with solubilized membranes were loaded in separate lanes. Translation products by themselves migrated at the top of the gel, presumably as aggregates (lanes 8, 12). mHcf106 translation product mixed with solubilized membranes migrated predominantly at ≈ 250 kDa but not at ≈ 700 kDa (lane 10). This result indicates that assembly of Hcf106 into the ≈ 700 kDa cpTatC–Hcf106 complex requires prior integration into the membrane. The mTha4 translation product mixed with solubilized membranes migrated at ≈ 240 kDa (lane 14).

Integration reactions and BN/PAGE analysis were also conducted with mHcf106 and mTha4 in which the conserved transmembrane glutamate was replaced by the structurally conserved but uncharged glutamine (mHcf106 E11Q and mTha4 E10Q, respectively). mHcf106 E11Q and mTha4 E10Q integrated into thylakoids and displayed similar characteristics as the wild-type proteins including protection of the transmembrane domain from proteolysis (Fig. 1). Membrane integrated mHcf106 E11Q migrated at ≈ 250 kDa on the blue native gel, but did not associate with the 700 kDa complex (Fig. 4, lane 6). This indicates that Hcf106 assembly into the ≈ 700 kDa receptor complex requires the conserved glutamate in its transmembrane domain. mTha4 E10Q migrated at 240 kDa similar to wild-type Tha4 (lane 7).

Twin arginine precursor binding by in vitro assembled ≈ 700 kDa complex

As a first test of the functionality of in vitro inserted components, we examined the ability of complexes containing in vitro integrated components to bind precursor proteins. Previous work established that the ≈ 700 kDa cpTatC–Hcf106 complex functions as a receptor for twin arginine-containing precursor proteins [8]. This was shown by several approaches, but is also indirectly evident from a shift in the molecular mass of the complex on blue native gels following precursor binding. The shift of endogenous complexes was detected following binding of the unlabeled precursor DT23 by BN/PAGE and immunoblotting. DT23 is a modified form of the OE23 precursor that binds tightly to the cpTatC–Hcf106 complex [8,25]. Binding resulting from increasing concentrations of DT23 resulted in a small shift in the apparent molecular mass of cpTatC (50–100 kDa; Fig. 5A). Likewise, the ≈ 700 kDa Hcf106 band experienced a similar shift in molecular mass upon binding DT23, whereas the lower Hcf106 bands were not affected by precursor (Fig. 5B). The shift in molecular mass first occurred between 5 and 25 nm DT23 (Fig. 5A,B, lanes 4, 5). This is consistent with our finding that 25 nm unlabeled DT23 competed ≈ 50% of the binding of radiolabeled DT23 (data not shown). The specificity of the band shift is demonstrated by the fact that the Sec pathway precursor, pOE33, had no effect on the migration of any component on the BN/PAGE gel (Fig. 5A,B, lane 9).

Figure 5.

In vitro integrated Hcf106 and cpTatC assemble into 700-kDa complexes that bind twin arginine containing precursors. (A,B) Precursor binding to endogenous complexes. Thylakoids were incubated with unlabeled DT23 in a total of 300 µL import buffer. DT23 was prepared by dissolving purified inclusion bodies in 10 m urea, 10 mm dithiothreitol for 3 h at room temperature. pOE33, a Sec pathway precursor, was prepared in urea/dithiothreitol as described for DT23. Assays received 12 µL precursor or 12 µL urea/dithiothreitol and were incubated for 15 min in the dark on ice. Recovered thylakoids were dissolved in 1% digitonin and analyzed by BN/PAGE on 5–13.5% gradient gels, which were processed for immunoblotting with antibodies to cpTatC (A) or Hcf106 (B) as depicted above the panels. (C,D) Precursor binding to in vitro integrated components. In vitro translated 3H-labeled pcpTatC or pHcf106 were incubated with intact chloroplasts in an import assay for 20 min. Intact chloroplasts were repurified, lysed, and the thylakoids isolated and washed with import buffer. Thylakoids were incubated in binding assays with varying concentrations of unlabeled DT23 precursor as above. Thylakoids recovered from assays were analyzed by BN/PAGE and fluorography.

In order to determine whether complexes resulting from assembly of in vitro integrated cpTatC and Hcf106 are capable of binding to precursor, membranes recovered from chloroplast import of radiolabeled pcpTatC or pHcf106 were incubated with unlabeled precursor and then analyzed by BN/PAGE and fluorography. The labeled cpTatC and Hcf106 bands exhibited similar shifts in molecular mass as the endogenous proteins (Fig. 5C,D). This demonstrates that in vitro integrated cpTatC and Hcf106 assemble into complexes capable of binding precursor. Given the large size of the cpTatC–Hcf106 complex and preliminary observations that it contains multiple copies of cpTatC and Hcf106 [8], we cannot conclude that in vitro assembled components bind directly to DT23, only that they become members of functional receptor complexes.

We frequently observe that the precursor-bound complex is darker on BN/PAGE than the unbound complex (Fig. 5A,B,D), although this is not always the case (Fig. 5C). This may result from precursor-induced stabilization of the ≈ 700 kDa complex to detergent because SDS/PAGE immunoblot analysis showed that the detergent extract samples of Fig. 5A,B,D, lanes 1–4 contained as much cpTatC as those in lanes 5–7 (data not shown).

Hcf106 and Tha4 integrate into thylakoids by the spontaneous pathway

The above results demonstrate that in vitro translated components of the thylakoid Tat system faithfully integrate into thylakoids that contain wild-type levels of endogenous components. One objective of this study is to biochemically complement mutant membranes in which a component is missing. As one or more protein translocation systems will be impaired in such mutants, determining the mechanism by which components integrate into the membrane is important. The facility with which mHcf106 and mTha4 integrate into isolated thylakoids allowed a controlled assessment of the mechanism of their association with the membrane. For this analysis translation product was incubated with thylakoids under conditions that varied the supply of energy and stromal proteins. Tight association with thylakoids was assessed by extraction of the membranes with 0.2 m Na2CO3 for Tha4 and 0.1 m NaOH for Hcf106 (Fig. 6). As can be seen, Hcf106 and Tha4 became integrated into thylakoids regardless of the conditions. GTP, ATP, a ΔpH, or stromal proteins were not required for integration (lane 3). Even at 0 °C, a substantial amount of these proteins became integrated into the membrane (lanes 4, 5). This indicated that integration of Hcf106 and Tha4 occurs in the absence of energy or stromal proteins. Thermolysin treatment (for Tha4) and thermolysin/trypsin treatment (for Hcf106) of membranes recovered from assays conducted in the absence of stroma, ΔpH, or ATP/GTP (i.e. as in lane 3) produced the characteristic protease protected fragments that are seen with membranes recovered from assays conducted with stroma and energy (i.e. as in Fig. 1B,C). This confirms that the transmembrane domain becomes imbedded under these conditions. The efficacy of the conditions used in assays of Fig. 6. was verified by light-harvesting chlorophyll a–b complex (LHCP) integration assays. LHCP, which employs the chloroplast SRP pathway, did not integrate into thylakoids unless stroma (the source of cpSRP) and ATP/GTP were present (see LHCP-DP lane 6, compare to lanes 7, 9). LHCP integration was substantially reduced in the absence of a ΔpH (lane 8).

Figure 6.

Tha4 and Hcf106 are integrated into thylakoids by the spontaneous pathway.In vitro translated mTha4, mHcf106, and pLHCP were assayed for integration into isolated thylakoids. Assays in lanes 1–9 contained thylakoids equivalent to 50 µg chlorophyll and, where indicated above the panel, stromal extract, 2.5 mm GTP, 2.5 mm ATP, 6 U apyrase, 0.5 µm nigericin, and 1.0 µm valinomycin in a total volume of 150 µL 50 mm Hepes/KOH pH 8, 0.33 m sorbitol, 6.7 mm MgCl2. Assays were conducted in darkness or white light at 0 °C or 25 °C as shown above the panel. Thylakoids used in assays shown in lanes 10–12 were pretreated with 0, 1, or 10 µg·mL−1 thermolysin at a thylakoid concentration equivalent to 1 mg·mL−1 chlorophyll in import buffer for 30 min at 4 °C in darkness. Proteolysis was terminated with 2.5 vols 14 mm EDTA in import buffer. Thylakoids were pelleted, washed with 14 mm EDTA in import buffer followed by import buffer and were resuspended in import buffer containing 10 mm MgCl2 prior to use. Thylakoids recovered from Tha4 integration assays were washed with 0.2 m Na2CO3; thylakoids from Hcf106 integration assays were washed with 0.1 m NaOH; thylakoids from LHCP integration assays were treated with thermolysin. LHCP-DP is a degradation product that represents correctly integrated LHCP.

These results suggested that Hcf106 and Tha4 are assembled into thylakoids by an unassisted or ‘spontaneous’ mechanism (reviewed in [4]). Another characteristic of spontaneous integration is the ability of proteins to insert into protease pretreated thylakoids [26]. Tha4 and Hcf106 integrated into thermolysin-treated membranes (Fig. 6, lanes 11, 12) as well as into control membranes (lane 10). In the experiment in Fig. 6, a reduced amount of Hcf106 integrated into the membranes treated with the highest level of protease (lane 12). However, such reduction was not observed in other experiments. LHCP integration into protease-treated membranes was undetectable (lanes 11, 12). Immunoblot analysis verified that the protease treatment degraded cpOxa1p, cpSecY, and cpTatC, the core components of the cpSRP, Sec-dependent, and ΔpH-dependent/Tat pathways, respectively (Fig. 6, inset). These results indicate that Tha4 and Hcf106 can integrate into thylakoids even when all of the known protein translocation machineries are disabled.

Hcf106 and Tha4 integrate into thylakoids in amounts comparable to those of the endogenous components

A second requirement for biochemical complementation is that components be incorporated into thylakoids in amounts comparable to endogenous components. An estimate of the amount of mHcf106 and mTha4 integrated into isolated thylakoids was made by quantitative immunoblotting of radiolabeled translation products in parallel with quantification of the amount of radiolabeled component inserted in vitro (Experimental procedures). Approximately 120 000 molecules of mHcf106 translation product were integrated per chloroplast equivalent and about 510 000 molecules of mTha4 translation product were integrated per chloroplast equivalent. Previous analysis estimated endogenous Hcf106 to be present at 95 000 molecules per chloroplast equivalent and endogenous Tha4 to be present at 140 000 molecules per chloroplast equivalent [10]. Thus, in vitro reactions are capable of supplying physiological amounts of Hcf106 and Tha4 to isolated membranes, making biochemical complementation theoretically possible.

Biochemical complementation of maize tha4 mutant thylakoids

To directly test if in vitro produced Tha4 could complement a Tha4 deficiency, thylakoids were isolated from tha4 maize mutant plants [15] and used in protein transport experiments (Fig. 7). Seeds from self-pollinated tha4/+ plants were grown in soil on a light/dark cycle for 10 days. Homozygous mutant plants were distinguished from their normal siblings based on their pale green color. Correct identification was confirmed by immunoblot analysis of leaf tips and of isolated thylakoid membranes (Fig. 7B). Chloroplasts were isolated as described in Experimental procedures and used to produce lysates, which were used in transport assays with the Tat pathway substrate DT23. Wild-type thylakoids transported DT23 to the lumen (Fig. 7A lanes 1, 2) whereas mutant thylakoids did not (lanes 7, 8). However, when preincubated with in vitro translated pea mTha4, mutant thylakoids became competent for DT23 transport (lanes 3, 4). The Tha4 E10Q variant did not complement the Tha4 deficiency (lanes 5, 6). Transport of DT23 achieved by tha4 membranes supplemented with in vitro translated Tha4 was significantly less than transport by the wild-type membranes. This may be due to a reduced capability of tha4 thylakoids to generate a pH gradient. In a separate experiment, we found that tha4 thylakoids generated a ΔpH of only ≈ 2.2 in the presence of 70 µE m−2·s−1 light and 6 mm ATP, i.e. the transport assay conditions, whereas wild-type thylakoids generated a ΔpH of ≈ 2.8 under the same conditions.

Figure 7.

In vitro complementation of Tha4 deficient thylakoid membranes from maize. (A) Chloroplasts were purified from sibling wild-type and tha4 maize seedlings as described in Experimental procedures. Chloroplast lysates equivalent to 50 µg of chlorophyll in 50 µL were incubated with 25 µL 35 mm Mg-ATP and 35 mm dithiothreitol and 50 µL of in vitro translated pea mTha4, mTha4 E10Q or mock translation mix for 15 min at 25 °C in the dark. Precursor DT23 (50 µL) was then added to each reaction mixture and protein transport reactions initiated by transfer of assay mixtures to the light. After 20 min, thylakoids were recovered by centrifugation, resuspended in 300 µL import buffer and divided into two equal aliquots. The aliquots were treated with (+) or without thermolysin for 40 min at 4 °C. Proteolysis was terminated with an equal volume of import buffer, 14 mm EDTA; the thylakoids were recovered by centrifugation, and washed with import buffer, 5 mm EDTA. Samples were subjected to SDS/PAGE and analyzed by fluorography. (B) Thylakoid membranes obtained from the Percoll gradient during chloroplast purification were analyzed by immunoblotting with antibodies to maize Tha4 and maize Hcf106 as shown.

A similar experiment was conducted with thylakoids from hcf106 mutant plants [23]. Thylakoids from wild-type siblings were capable of Tat pathway transport, whereas mutant thylakoids were deficient in Tat transport. Incubation of in vitro translated mHcf106 from either pea or maize failed to complement the mutation even though significant amounts of Hcf106 integrated into the membrane (data not shown).


In this study we reconstituted the assembly of Tat system components into thylakoids in vitro. For cpTatC, this required import into intact chloroplasts (Fig. 3). For Tha4 and Hcf106, efficient integration was achieved with isolated thylakoids (Figs 1 and 6). In vitro integrated components displayed all of the characteristics of the endogenous components. These include localization to thylakoids and resistance to alkaline extraction of the membrane (Figs 1 and 3 and [10]). For Tha4 and Hcf106, it could be shown that they are anchored into the membrane by a single transmembrane domain as predicted (Fig. 1B and C). Furthermore, in vitro integrated Hcf106 and cpTatC were assembled into a characteristic ≈ 700-kDa complex that previous work has identified as a receptor complex for twin arginine-containing precursors (Figs 4 and 5). Band shift experiments verified that these in vitro produced complexes were capable of binding precursors (Fig. 5). Binding of saturating amounts of the precursor DT23 resulted in an upward shift in the apparent molecular weight of endogenous as well as in vitro integrated cpTatC and Hcf106. The fact that this shift was only 50–100 kDa was surprising, considering that the cpTatC–Hcf106 complex and the orthologous E. coli TatC–TatB complex seems to contain multiple copies of the two components [8,27].

Our analyses indicate that Tha4 and Hcf106 integrate into the membrane by a ‘spontaneous’ mechanism (Fig. 6). One feasible way this could occur is that their amphipathic domains fold into helices at the membrane surface, embed themselves with their axes parallel to the plane of the membrane, and facilitate insertion of the transmembrane domain. Examples of amphipathic helical folding at the membrane interface are found among the antimicrobial peptides [28]. Furthermore, our unpublished studies show that the Hcf106 and Tha4 stromal domains lack secondary structure in aqueous solution but attain a high percentage of alpha helical structure as the polarity of the solution is decreased with trifluoroethanol. Additionally, studies of E. coli TatA (Tha4 ortholog) show that it transitions from random coil to helix upon incubation with liposomes [29]. The fact that Hcf106 possesses a substantially longer amphipathic helix than Tha4 may account for its stronger association with the membrane (Fig. 1). Such an integration mechanism could also account in part for the distribution of Hcf106 and Tha4 in chloroplasts, as the lipid composition would be the likely determinant for membrane specificity. Thylakoids and the inner envelope membrane possess nearly identical polar lipid compositions [30]. Thus, upon import into the chloroplast and cleavage of their transit peptides, these components might insert into inner envelope and thylakoid bilayers in a ratio that reflects the relative abundance of these membranes. On the other hand, it is difficult to understand the presence of Tha4 in the outer envelope membrane. The fact that outer envelope Tha4 was mature in size indicates that it had gained access to the stromal transit peptidase. It's conceivable that Tha4 redistributed during the fractionation procedure, as its association with the membrane is more tenuous than Hcf106. However, this seems unlikely as essentially the same localization results were obtained with two different methods of chloroplast lysis (Fig. 2).

The mechanism of cpTatC integration into thylakoids is presently unclear. The facts that virtually all endogenous cpTatC is in the thylakoid membrane and that neither mcpTatC (Fig. 3) nor the cpTatC precursor [31] integrated into isolated thylakoids make it unlikely that cpTatC integrates by the spontaneous mechanism. Rather it is more likely that some sort of machinery is involved in the assembly of cpTatC, possibly one that involves the envelope as an intermediate location. Inhibitor studies with chloroplast import assays suggest that cpTatC does not use the cpSec, cpSRP or Tat pathways (K. Cline, unpublished data). It is important to determine the nature of the routing machinery and the pathway to the thylakoids. The fact that endogenous cpTatC is the only Tat component confined to thylakoids suggests that cpTatC plays the major role as Tat pathway receptor. In fact, cpTatC has been shown to make direct contact with bound precursors [8]. Thus the mechanism of cpTatC assembly into thylakoids will directly relate to the manner by which thylakoids establish their identity. Recently it was shown that chloroplast SecE, a component of the Sec translocase, integrates into thylakoids by a spontaneous or unassisted mechanism [32]. This suggests that cpSecY plays the dominant role as receptor for the thylakoidal Sec pathway. Similar to cpTatC, the mechanism by which cpSecY inserts into thylakoids has been difficult to determine.

A major objective of this study was to examine the possibility of biochemical complementation of components for structure–function studies. We were unable to complement thylakoids from hcf106 maize plants despite the fact that in vitro translated mHcf106 efficiently integrated into the mutant membranes. One possible explanation is that cpTatC failed to accumulate in mutant thylakoids in the absence of Hcf106. It has been reported that TatC is unstable in tatB deletion mutants of E. coli[24]. It was not possible to directly test for the presence of cpTatC in these membranes because antibody to pea cpTatC reacts poorly with maize cpTatC. However, the observation that maize mHcf106 integrated into isolated maize thylakoids did not migrate at ≈ 700 kDa on BN/PAGE is consistent with this idea [31]. Unfortunately, efforts to supply substantial amounts of cpTatC by import into hcf106 chloroplasts were not successful (V. Fincher & K. Cline, unpublished data).

On the other hand, the assembly of Hcf106 and cpTatC into a 700 kDa complex that can bind precursor proteins in pea thylakoids provides a method for examining the requirements for assembly of the receptor complex. In that regard, the failure of mHcf106 E11Q to assemble into the ≈ 700 kDa complex suggests that transmembrane glutamate is important for interaction of Hcf106 and another member of the ≈ 700 kDa complex, e.g. cpTatC. It also predicts that mHcf106 E11Q should not be functional. This possibility needs further examination because two recent studies of E. coli Tat B (Hcf106 ortholog) found that a conserved transmembrane glutamate in a comparable position is not essential for bacterial Tat activity [33,34].

A more direct demonstration of functional assembly comes from the ability to complement the Tat transport activity of tha4 thylakoids with in vitro translated pea mTha4. The somewhat reduced activity of in vitro complemented thylakoids may be due to differences in the Tha4 content of the membranes or to the fact that pea Tha4 was used rather than maize Tha4. A more likely explanation is that tha4 membranes are impaired in their ability to generate and maintain a substantial pH gradient (see above). Nonetheless, the ability to provide active Tha4 to Tha4-lacking membranes in vitro is a first step towards unraveling the roles that Tha4 plays in the translocation of folded proteins. Such a study [35] using a modified biochemical complementation approach has performed a more detailed analysis of our preliminary results − that the Tha4 transmembrane glutamate is essential for activity – and has shown that the transmembrane glutamate is important for assembly of the translocase. This is an interesting finding considering the above results that the Hcf106 transmembrane glutamate is also essential for assembly.


We thank R.-A. Monde and A. Barkan for the generous gift of tha4 seed and antibodies to the maize Tha4 protein, which were obtained through funding by the National Institutes of Health (R01 G48179) and the National Science Foundation (DBI 0077756) and for critical review of the manuscript, and M. Settles for the generous gift of hcf106 seed. We thank M. McCaffery for excellent technical assistance and M. McCaffery and F. Gerard for critical review of the manuscript. This work was supported in part by National Institutes of Health grant R01 G46951 to K.C. This manuscript is Florida Agricultural Experiment Station Journal series No R-09824.