Insertion of PsaK into the thylakoid membrane in a "Horseshoe" conformation occurs in the absence of signal recognition particle, nucleoside triphosphates, or functional albino3.

The photosystem I subunit PsaK spans the thylakoid membrane twice, with the N and C termini both located in the lumen. The insertion mechanism of a thylakoid membrane protein adopting this type of topology has not been studied before, and we have used in vitro assays to determine the requirements for PsaK insertion into thylakoids. PsaK inserts with high efficiency and we show that one transmembrane span (the C-terminal region) can insert independently of the other, indicating that a "hairpin"-type mechanism is not essential. Insertion of PsaK does not require stromal extract, indicating that signal recognition particle (SRP) is not involved. Removal of nucleoside triphosphates inhibits insertion only slightly, both in the presence and absence of stroma, suggesting a mild stimulatory effect of a factor in the translation system and again ruling out an involvement of SRP or its partner protein, FtsY. We, furthermore, find no evidence for the involvement of known membrane-bound translocation apparatus; proteolysis of thylakoids destroys the Sec and Tat translocons but does not block PsaK insertion, and antibodies against the Oxa1/YidC homolog, Alb3, block the SRP-dependent insertion of Lhcb1 but again have no effect on PsaK insertion. Because YidC is required for the efficient insertion of every membrane protein tested in Escherichia coli (whether SRP-dependent or -independent), PsaK is the first protein identified as being independent of YidC/Alb3-type factors in either thylakoids or bacteria. The data raise the possibility of a wholly spontaneous insertion pathway.

Studies in bacteria and plant thylakoids have demonstrated the operation of a complex signal recognition particle (SRP) 1 -dependent pathway for the insertion of membrane proteins. This pathway was first demonstrated in bacteria (reviewed in Ref. 1) where the insertion of several plasma membrane proteins was shown to require the activity of two soluble/extrinsic proteins, SRP and FtsY. Escherichia coli SRP is a complex comprising a 4.5 S RNA molecule together with a homolog of the 54-kDa subunit of eukaryotic SRPs, which binds to particularly hydrophobic regions such as nascent or newly synthesized membrane proteins. The SRP-substrate complex interacts, at least in some cases, with the SecYEG translocon in the plasma membrane and, in a poorly understood sequence of events, the substrate is transferred into the translocon with the assistance of an additional factor, FtsY (2)(3)(4)(5)(6)(7)(8). Both SRP and FtsY are GTPases and the insertion process depends totally on GTP hydrolysis.
A broadly similar pathway has been characterized for the insertion of the major light-harvesting chlorophyll-binding protein, Lhcb1, into the thylakoid membrane of plant chloroplasts. This conservation of insertion pathway is perhaps not surprising, given that chloroplasts are widely accepted to have evolved from endosymbiotic cyanobacterial-type organisms. After import from the cytosol, Lhcb1 binds stromal SRP to form a soluble targeting complex (9). Integration of the polytopic Lhcb1 protein into the thylakoid membrane further requires FtsY (10,11), GTP, and a membrane-bound translocase. The translocase has yet to be fully characterized; antibodies to SecY strongly inhibit the insertion of SecA-dependent lumenal substrates but do not block the insertion of Lhcb1 (12,13). This raises the possibility that the thylakoid SecYEG complex is not required, although this work could not rule out the possibility that the SRP interacts with a different SecY determinant that is unaffected by antibody binding. Interestingly, the chloroplast SRP particle differs from that of E. coli in that (i) no RNA is present and (ii) this SRP possesses a novel 43-kDa subunit (14), which binds to a novel SRP recognition element in Lhcb1 that is found in other members of the light-harvesting chlorophyll a/b -binding protein family (LHC proteins) (15,16). Nevertheless, there are clear parallels with the E. coli insertion pathway and the similar natures of the pathways are reinforced by recent findings concerning the essential involvement of Oxa1-type proteins. Oxa1 is a mitochondrial inner membrane protein that is involved in the biogenesis of a range of mitochondrial inner membrane proteins, specifically those that insert from the matrix side of the membrane (17,18). An Oxa1 homologue, termed Alb3, is also essential for the insertion of Lhcb1 in thylakoids (13) and recent work has shown an E. coli homologue, YidC, to be essential for the insertion of at least some SRP-dependent membrane proteins in this organism (19).
Critically, YidC is also essential for the biogenesis of E. coli membrane proteins that do not require SRP or the Sec apparatus. The insertion mechanism of M13 procoat protein has been characterized in some detail (20) and this protein has been shown to insert efficiently into the plasma membrane in the complete absence of functional SRP, SecA, or the membrane-bound SecYEG translocon. On the basis of these data it had been suggested that this protein may insert spontaneously into the membrane, but depletion of YidC led to a rapid block in its insertion indicating a central role in their insertion. Thus, it appears that one pool of YidC may be associated with the Sec apparatus (21) while another pool may in effect represent a novel form of translocase dedicated for the insertion of some, if not all, SRP-independent membrane proteins. Like M13 procoat, a subset of thylakoid membrane proteins are synthesized with cleavable signal peptides but inserted in the absence of SRP (22,23), but the possible involvement of the YidC homolog, Alb3, remains to be clarified.
All of the previous studies on thylakoid protein insertion have focused either on relatives of the well studied Lhcb1 (LHC proteins), or on proteins that bear cleavable N-terminal signal peptides. In this report we have sought to characterize the insertion of a different type of thylakoid membrane protein, PsaK, which is not synthesized with a cleavable signal peptide and which is unrelated to LHC proteins. We show that this protein inserts with high efficiency into thylakoids by a mechanism that does not involve SRP, NTP hydrolysis, or any of the known translocation machinery in the thylakoid membrane, including Alb3.

EXPERIMENTAL PROCEDURES
DNA Constructs-A full-length cDNA clone encoding the precursor of barley PsaK (pPsaK) in the plasmid pBluescript TM SK(Ϫ) was a gift from B. L. Møller (24). A construct encoding the mature PsaK protein was prepared by polymerase chain reaction amplification of the aforementioned barley cDNA. The forward primer (5Ј-CAGGGGATCCG-CATGGACTACATCGGC-3Ј) introduced a BamHI site between base pairs 287 and 288, and altered cysteine 42, which is the last amino acid of the transit peptide, to an initiating methionine residue. The reverse primer (5Ј-CCCGAAGCTTGCAGAACAGCTATG-3Ј) introduced a Hin-dIII restriction site between base pairs 603 and 604, after the stop codon at base pair 565. The amplified region was cloned 5Ј BamHI-HindIII 3Ј into pGEM TM 4Z, and sequenced completely before being used as a template for transcription and translation. A cDNA clone encoding pea pLhcb1 was provided by N. E. Hoffman (25), while the cDNA clone encoding wheat p23K has been described (26).
Transcription and Translation-The plasmid encoding pPsaK was linearized with XhoI, whereas linearization was found to be unnecessary for the plasmid encoding PsaK. Transcription in vitro was performed according to Promega protocols, using either T3 RNA polymerase (for pPsaK) or SP6 RNA polymerase (for PsaK, pLhcb1 and p23K). Radiolabeled proteins were prepared using a wheat germ lysate system (Promega) in the presence of [ 35 S]methionine (Amersham Pharmacia Biochem), also according to the manufacturers instructions. The translation mixtures were treated with puromycin and centrifuged prior to use in insertion assays, as described by Thompson et al. (27). Import Assays-Assays for the import of precursor proteins by intact pea chloroplasts and isolated thylakoid membranes were essentially as described in Ref. 28 except that the light intensity was 300 mol of photons m Ϫ2 s Ϫ1 . The proton ionophore nigericin (Sigma) was dissolved in ethanol, and used at a final concentration of 2 M in the presence of 10 mM KCl; control samples were identical, except they contained an equivalent volume of ethanol instead of nigericin. Assays to measure the effect of apyrase (Sigma, type VI) were carried out as described in Ref. 27. 10 mM stock solutions of the non-hydrolyzable ATP and GTP analogues, AMP-PNP and GMP-PNP (Sigma) were prepared in 10 mM Hepes, 5 mM MgCl 2 and the pH adjusted to 7. When used, the final concentrations of these analogues were 0.5 mM. Proteolysis of thylakoid membranes, prior to insertion assays, was as described in Ref. 27 and Alb3 antibody-inhibition tests were as described in Ref. 13, except that the thylakoids were incubated with anti-Alb3 antibodies for 2 h instead of 1 h.
Treatment of Thylakoid Membranes, Post-assay-In order to test if PsaK was correctly inserted in the thylakoid membrane (either after import into intact chloroplasts, or after insertion into isolated thylakoid membranes), the membranes (between 10 and 20 g of chlorophyll, depending on the experiment) were washed with 0.5 ml of ice-cold 10 mM Hepes-KOH, pH 8, 5 mM MgCl 2 (HM) and then reisolated by centrifugation at 18,000 ϫ g and 4°C for 5 min in a microcentrifuge. Next, they were washed with 100 l of 20 mM Tricine-NaOH, pH 8 (TB), and reisolated as above, then subjected to one round of extraction with 6.8 M urea, using a protocol adapted from Breyton et al. (28) and described in detail in Ref. 27. Next, the membranes were resuspended in TB and digested with 0.2 mg ml Ϫ1 trypsin (Sigma, type XIII) in a final volume of 100 l for 30 min on ice. Trypsin digestions were stopped by the addition of 0.5 mg ml Ϫ1 trypsin inhibitor (Sigma, type I-S), followed by centrifugation at 18,000 ϫ g and 4°C for 10 min in a microcentrifuge. Finally, the thylakoid membranes were resuspended in 15 l of TB containing 5 g of trypsin inhibitor, and an equal volume of 2 ϫ protein sample buffer, then immediately boiled for 5 min. Insertion efficiencies were measured by exposing dried SDS-PAGE gels in PhosphorImager cassettes, followed by quantitation in a Molecular Dynamics PhosphorImager.

Structures of the Precursor and Mature Forms of PsaK-
PsaK is a nuclear-encoded subunit of photosystem I that is important for the stable interaction of LHCI with the photosystem I core complex (29). Barley PsaK is synthesized in the cytosol as a 13.7-kDa precursor, imported into the chloroplast and subsequently inserted into the thylakoid membrane (24). The presequence is a typical stroma-targeting peptide that is predicted to be removed by the stromal processing peptidase (SPP). From sequence analysis, the 9-kDa mature protein is predicted to span the membrane twice with the N and C termini both located in the lumen and the positively charged loop region remaining on the stromal (cis) side of the thylakoid membrane according to the "positive-inside rule" (30). Recent high-resolution crystallographic analyses of Synechococcus elongatus photosystem I have confirmed these predictions for the cyanobacterial PsaK, which is highly similar to eukaryotic PsaK. 2 This "horseshoe" configuration is illustrated in Fig. 1, which also indicates the locations of the methionine residues in the mature barley protein (two in the first transmembrane region, one in the second). The stroma-exposed loop region contains numerous positively charged residues and trypsin is therefore predicted to cleave in this region and generate two degradation fragments containing the indicated numbers of methionine residues. The amino acid sequence of pPsaK is also given in Fig. 1 together with the start site of a mature-size construct described below.
Insertion of PsaK Does Not Require SRP or Nucleoside Triphosphates, but Is Stimulated by the ⌬ H ϩ- Fig. 2 shows a chloroplast import experiment conducted with [ 35 S]methionine-labeled barley pPsaK. The precursor protein (pPsaK) is imported into chloroplasts (lane C) where it is processed to the mature form and resistant to added protease (lane Cϩ); the mature protein has an apparent mass of 7 kDa (in agreement with Ref. 24) and is found exclusively in the thylakoid fraction (lane T). The protein is highly resistant to urea extraction (lane Turea), which has been shown to be an effective means of identifying many membrane proteins (28). Trypsin treatment of the thylakoids from lane T (lane 1) leads to the generation of two fragments, one of which contains precisely twice as much 35 S radioactivity as the other (data not shown, from Phosphor-Imager analysis). On the basis of the data described above, the stronger, lower band most likely represents the first trans-membrane region (TM1) and the other (TM2) is the C-terminal transmembrane span. If the thylakoids are washed with urea after trypsin treatment, a proportion of the fragments are extracted (lane 2) whereas the same fragments observed in lane 1 are obtained if the membranes are urea-treated prior to proteolysis (lane 3). Clearly, mature PsaK is far more resistant to urea extraction than either of the single-span degradation fragments and we believe that this reflects the ability of urea to partially extract small, single-span proteins, whereas proteins containing two or more spans are almost totally resistant. We have observed a similar phenomenon with single-span proteins bearing signal peptides: the precursor proteins adopt loop conformations in the membrane that are far more resistant to urea extraction than the single-span mature proteins (27).
The primary aim in this work was to analyze the mechanism by which PsaK inserts into the thylakoid membrane, and for these experiments we used in vitro assays for the insertion of radiolabeled protein into purified pea thylakoids. Our criteria for correct insertion were (i) that the protein should be resistant to urea extraction and (ii) that trypsin should generate the two degradation products observed after import of pPsaK as shown in Fig. 2. Fig. 3A shows the results of incubating the pPsaK translation product (lane Tr) with thylakoids in the presence of stromal extract: the mature protein is efficiently generated by SPP in the extract, a large proportion of mature PsaK becomes associated with the thylakoids (lane T) and urea extraction followed by trypsin digestion produces the two diagnostic degradation products (TM1 and TM2). These results effectively confirm that insertion is taking place. However, the full precursor protein is unsuitable for detailed tests on the insertion requirements because stromal extract (which con- . After a 20-min incubation in the light, the membranes were washed with HM and TB, before being subjected to extraction with urea (lower panel, ϩUrea), the urea-resistant protein being denoted as PsaK. Next the urea-washed membranes were digested with trypsin (upper panel, ϩ Trypsin), where the characteristic degradation products are denoted by TM1 and TM2. An aliquot of the translation mixture alone was also digested with trypsin (lane Trϩ). The samples were analyzed by SDS-PAGE and fluorography. Insertion efficiencies (below the lower panel, quoted as %), relative to the HM buffer control (100%) were calculated by using a PhosphorImager to measure the densities of the urearesistant mature PsaK bands. C, as a control, pea pLhcb1 (lane Tr) was translated in vitro and incubated with isolated pea thylakoids, in the absence (HM) or presence (SE) of stromal extract, and in the absence (Ϫ) or presence (ϩ) of 2 units of apyrase, for 20 min in the light. After the incubation, the thylakoids were washed with HM and TB, before being subjected to both urea extraction and digestion by trypsin, which yielded the normal degradation product (marked DP) when correct insertion occurred. An aliquot of the translation mixture alone was also digested with trypsin (lane Trϩ). Sec/SRP/Alb3-independent Insertion of PsaK tains essentially all of the SRP) cannot be omitted since it also contains almost all of the stromal peptidase required to generate the mature form of PsaK. The N terminus of the mature PsaK is located in the lumen, which means that processing to the mature size must precede insertion. We have found that the full precursor form does not insert correctly in the absence of stromal extract (see below), which is not surprising because the entire presequence would have to be translocated across the thylakoid membrane. Further experiments were therefore conducted with a mature size PsaK translation product synthesized from a truncated cDNA as described under "Experimental Procedures." Fig. 3B shows assays for the insertion of this construct (denoted PsaK) into isolated thylakoids. After each assay, samples of the thylakoids were either treated with trypsin (upper panel) or urea-extracted (lower panel). PsaK again inserts into isolated thylakoids and the inserted protein is again converted to the diagnostic degradation products TM1 and TM2 ("ϩ trypsin" panel). As a control, we incubated the translation product with trypsin in the absence of membranes (lane Trϩ) and this procedure leads to an almost complete degradation of PsaK, with no evidence for either the TM1 or TM2 bands (providing further evidence that TM1 and TM2 represent membrane-integrated regions). In this assay, TM1 should contain three times as much [ 35 S]Met as TM2 because of the initiation methionine introduced during the cloning procedure, which is located in the lumen after insertion and hence protected from proteolysis. In repeat assays such as those shown in Fig. 3B, the TM1:TM2 [ 35 S]Met ratio was always near 3 (within 15% in each case) confirming correct insertion, and this is supported by the observation that the mature protein is equally resistant to urea extraction ("ϩ urea" panel). We have routinely found that PsaK inserts with very high efficiency in this assay system; 34% of translation product was found to be inserted in this particular experiment (based on the recovery of [ 35 S]Met in TM1 and TM2) and insertion efficiencies of over 40% have been observed in other experiments (data not shown).
The left hand panel of Fig. 3B shows assays carried out in the presence of stromal extract (SE) or Hepes-magnesium buffer (HM), and in each of these cases insertion was assayed either with or without pretreatment of the assay mixture with apyrase (indicated as Ϯ above the lanes). This enzyme hydrolyzes nucleoside triphosphates in the mixture and totally blocks insertion by the SRP route (9,11). These lanes show that insertion is most efficient in the HM incubations, i.e. in the absence of stromal extract (and hence SRP). The presence of apyrase does, however, lead to a reduction in insertion efficiency (to 71% of the control value) but this is not due to SRP involvement because the presence of stromal extract also leads to a slight reduction in insertion efficiency (down to 81% of the HM value). Stromal extract contains essentially all of the SRP (see below) and this result rules out an SRP involvement.
In the same experiment we also assessed the effects of including GMP-PNP and AMP-PNP (non-hydrolyzable analogs of GTP and ATP, respectively) in an attempt to identify the cause of the reduction observed with apyrase. GMP-PNP, in particular, is an effective inhibitor of SRP (11). However, neither analog inhibits insertion to any marked extent, either in the absence or presence of stromal extract (right-hand panel in Fig. 3B).
As a control for these experiments we used insertion assays with Lhcb1, a known substrate for the SRP pathway. After insertion into thylakoids, Lhcb1 becomes highly resistant to trypsin digestion and a characteristic stable degradation product is generated (9, 11-13). Fig. 3C shows that insertion is completely dependent on stromal extract, as shown by the appearance of the degradation product in the SE panel. Insertion is also completely dependent on the presence of NTPs in the mixture as pretreatment with apyrase (lanes indicated by ϩ) totally blocks insertion, as found previously (9,11). On the basis of these data we conclude that PsaK does not require SRP for its insertion and, because insertion is not even stimulated by the presence of stroma, we propose that insertion is indeed completely independent of SRP. Insertion is, however, slightly inhibited by apyrase treatment even in the absence of stromal extract, which indicates a mild stimulatory influence of an ATP or GTP-hydrolyzing factor in the wheat germ translation system. We believe the most likely explanation to be an antiaggregation effect of chaperones in the wheat germ extract which may prevent some PsaK molecules from adopting an insertion-incompetent conformation.
The above studies rule out a central role for NTP hydrolysis in the insertion of PsaK but to obtain a more comprehensive picture of the energy requirements we tested whether the thylakoidal proton motive force stimulates insertion. Fig. 4A shows the effect of nigericin (a proton ionophore) on the import and sorting of pPsaK in intact chloroplasts. In the control assay, the protein is found exclusively in the thylakoid fraction, and only as the mature form. With nigericin present, however, some imported protein is found in the stromal fraction and it is notable that the primary stromal form is the full precursor protein. Some mature PsaK is also found in the stroma, together with a further, intermediate-size protein (iPsaK) that may represent a processing intermediate (alternatively, this may result from proteolysis of pPsaK). From these data it is evident that optimal insertion efficiency is dependent on the thylakoidal proton electrochemical gradient, ⌬ H ϩ. The appearance of the full precursor protein, furthermore, suggests that pPsaK is not necessarily processed immediately upon entry into the stroma, and we suggest in addition that at least some molecules may be processed only during the later stages of the insertion pathway, since the appearance of pPsaK most likely stems from an inhibition of insertion.
To obtain further data on the ⌬ H ϩ dependence of insertion we conducted thylakoid insertion assays with PsaK in the presence of nigericin (Fig. 4B). The data show that the presence of nigericin (lanes N) reduces insertion efficiency to a moderate extent when compared with the control samples shown in lanes C (in this experiment by 20 -25%) in both the presence and absence of stromal extract. Taken together, these data indicate that the ⌬ H ϩ stimulates PsaK insertion but is not essential.
PsaK Insertion Does Not Require the Thylakoidal Sec Machinery or Alb3-The data shown in Figs. 3 and 4 exclude the involvement of SRP or NTPs in PsaK insertion but it is equally important to understand whether membrane-bound translocation machinery is involved. The question of Alb3 involvement is critical because the homologous YidC protein is required for the efficient insertion of every E. coli membrane protein tested to date (19), and the SecYEG complex is also a prime candidate since this translocon is also used for some membrane proteins in bacteria (reviewed in Ref. 1). We addressed these possibilities in two ways. Previous studies (31) have shown that the insertion of Lhcb1 is totally inhibited by pretreatment of the thylakoids with trypsin, and the same study showed that translocation of Sec-dependent lumenal proteins is also completely blocked. This technique provides a simple means of destroying both the membrane-bound Sec apparatus and inhibiting integration by the SRP pathway. In previous studies using this approach we have maintained a ⌬ H ϩ by driving the ATP synthase in reverse in the dark (the synthase is highly resistant to trypsin) and we used the same method in this study since insertion of PsaK is stimulated by the ⌬ H ϩ. Fig. 5 shows experiments in which thylakoids were treated with trypsin, washed with buffer containing trypsin inhibitor to remove protease, and then assayed for their ability to import pre-23K (a substrate for the twin-arginine translocation, or Tat pathway), Lhcb1 and PsaK. The pre-23K imports serve as a test for the establishment of the ⌬ H ϩ since transport of this protein into the lumen is completely dependent on the proton gradient (1), and the data show that in the absence of trypsin treatment this protein is indeed imported with high efficiency and processed to the mature size, in total darkness. This observation confirms the presence of a ⌬pH, and import is abolished by trypsin treatment which has been shown previously to inactivate the Tat system (31). Insertion of Lhcb1 is also completely inhibited by this treatment; very little Lhcb1 is found associated with the thylakoids after the incubation (lane T of the "Trypsin" panel) and essentially no resistant degradation product is found after protease treatment (lane Tϩ) and other FIG. 4. Insertion of PsaK is stimulated by the thylakoidal ⌬ H ؉. A, intact pea chloroplasts equivalent to 50 g of chlorophyll were incubated with 12.5 l of in vitro translated pPsaK (lane Tr) in the absence (Control) and presence (ϩ Nigericin) of the proton ionophore nigericin (at 2 M final concentration), for 20 min in the light. After the import incubation, the chloroplasts were washed, fractionated, and analyzed by SDS-PAGE and fluorography. Lanes C, total washed chloroplasts; lanes Cϩ, thermolysin-treated chloroplasts; lanes S, stromal extract (prepared, as always, in the presence of 10 mM EDTA, to prevent residual thermolysin activity from degrading any stromal intermediates); lanes T, thylakoid membranes; lanes Tϩ, trypsin-treated thylakoid membranes; pPsaK, precursor of PsaK; iPsaK, an intermediate form of PsaK; PsaK, mature protein; TM1 and TM2, degradation products corresponding to transmembrane spans 1 and 2, respectively. B, isolated pea thylakoids equivalent to 20 g of chlorophyll were incubated with 5 l of in vitro-translated PsaK (lane Tr), in the presence (SE) and absence (HM) of stromal extract, and in the presence (N) and absence (C) of 2 M proton ionophore nigericin, for 20 min in the light. After the insertion incubation, the membranes were washed with HM and TB, before being extracted with urea (upper panel, ϩUrea). After urea extraction, the membranes were digested with trypsin (lower panel, ϩ Trypsin). Urea-resistant mature protein is marked PsaK, while the protease degradation products corresponding to transmembrane spans 1 and 2 are marked TM1 and TM2, respectively. After the experiment the samples were analyzed by SDS-PAGE, and insertion efficiencies (shown below the lower panel, in %) measured by a Phos-phorImager to calculate the amounts of urea-resistant mature PsaK relative to the control sample (HM, C).

FIG. 5. Predigestion of the thylakoid membranes with trypsin does not prevent insertion of PsaK.
Thylakoid membranes were digested with 60 g/ml trypsin (Trypsin) or buffer (Control) and the chloroplast ATPase was subsequently activated to generate a ⌬pH in the dark, as described in detail in Ref. 31. The activated membranes (20 g of chlorophyll) were incubated with 5 l of in vitro translated PsaK, pLhcb1, or p23K (lanes Tr) in the dark for 30 min, with all manipulations being carried out under a dim, green, safe light. After the incubation, the membranes were washed with HM and reisolated. PsaK and Lhcb1 samples were washed further with TB, before being subjected to urea extraction (lanes T) and then trypsin digestion (lanes Tϩ). 23K samples were analyzed directly after washing with HM (lanes T) or after digestion with 0.2 mg/ml thermolysin for 40 min on ice (lanes Tϩ). All samples were analyzed by SDS-PAGE and fluorography. The insertion efficiencies of PsaK (values in %, relative to the control sample) were measured using a PhosphorImager. PsaK, mature protein; TM1 and TM2, trypsin degradation fragments corresponding to transmembrane helices 1 and 2 of PsaK; pLhcb1, precursor of Lhcb1; DP, trypsin degradation fragment of inserted Lhcb1; p23K, precursor of 23K; 23K, mature protein.
experiments (not shown) confirmed that import of a Sec substrate was also blocked. However, the upper panel shows that PsaK still inserts into trypsin-treated thylakoids, which indicates that the Sec system is not required.
The specific question of Alb3 involvement was approached by more direct means. Preincubation of thylakoids with polyclonal anti-Alb3 antibodies almost blocks Lhcb1 insertion without affecting the Tat-or Sec-dependent pathways (13) and the same technique was used to test for its involvement in PsaK insertion. Fig. 6A shows a control assay using Lhcb1, in which insertion was monitored after incubation of the thylakoids with buffer (HM; as a control) with preimmune antibodies or anti-Alb3 antibodies. The preimmune serum causes a slight inhibition of insertion (to 89% of the control value) but the Alb3 antibodies reduce insertion efficiency to 27% of the control value. A similar level of inhibition was observed previously (13). In contrast, Fig. 6B shows that neither the preimmune nor the Alb3 antibodies affect insertion of PsaK into thylakoids, and the levels of urea-resistant protein or TM1 or TM2 degradation products remain undiminished. These data clearly indicate that PsaK is not dependent on Alb3 for insertion.
The C-terminal Transmembrane Span of PsaK Can Insert Independently-The above data show that PsaK inserts by a relatively simple mechanism that does not rely on any known translocation apparatus, including Alb3. This type of mechanism is highly unusual and we have sought to obtain further details on the overall insertion mechanism. One possibility, proposed for many membrane proteins (1), is that the two transmembrane spanning regions may form a "helical hairpin" that is able to insert with high efficiency due to the simultaneous partitioning of two hydrophobic regions. Several membrane proteins are known to form loop intermediates in which the loop region is on the trans side of the membrane, and similar principles may operate for those proteins, such as PsaK, where the loop remains on the cis side. We tested whether a single span of PsaK can insert independently into the thylakoid membrane, by simply using the full precursor protein instead of the mature PsaK construct, under conditions where cleavage by SPP is prevented. As explained above, the N terminus of mature PsaK lies in the lumen, hence cleavage of the large and highly charged presequence would appear to be essential before this N-terminal region can translocate across the membrane. We therefore carried out thylakoid import assays under two conditions: in the complete absence of stromal extract (using thylakoids that had been thoroughly washed to remove residual SPP), and after protease treating the thylakoids (to destroy any SPP on the membrane surface). Mature size PsaK was used as a control since this protein inserts under both sets of conditions as shown above.
The data (Fig. 7) show that the mature size PsaK (PsaK panel) behaves, as in experiments shown above, and the TM1 and TM2 fragments again appear with a labeling ratio which was calculated to be close to 3:1. Insertion occurs with both the washed and protease-treated thylakoids. However, very different results are obtained when the full precursor protein is used (pPsaK panel). The upper degradation fragment (TM2) is again observed after insertion into either washed or protease-treated thylakoids but the lower band (TM1) is now completely absent. A low-intensity smear of label is present below the TM2 band, presumably due to degradation of non-inserted PsaK regions, but no band is present in the TM1 region. We conclude from this result that the N-terminal transmembrane span is indeed unable to insert when the presequence is present, as predicted above, but the clear presence of TM2 is strong evidence that this region is able to insert independently under these conditions. These data also serve to reinforce the efficacy of the in vitro assay because they provide a third line of evidence that the bands denoted TM1 and TM2 do indeed represent inserted transmembrane regions; the N-terminal hydrophobic region is clearly highly susceptible to proteolysis (or is simply removed when the thylakoids are washed after the insertion reaction) when not inserted in the thylakoid membrane. DISCUSSION Several thylakoid membrane proteins have been previously analyzed in terms of insertion mechanism, and in this respect they fall into two broad categories. Lhcb1 follows a complex FIG. 6. Anti-Alb3 antibodies inhibit the insertion of Lhcb1 but not PsaK. Isolated pea thylakoids were preincubated with anti-Alb3 antibodies (Alb3), preimmune serum (PI), or import buffer (HM) for 2 h on ice, as detailed in Ref. 15. After this period the thylakoids were incubated with pLhcb1 or PsaK as indicated and analyzed by protease treatment as described in the legend to Fig. 3.   FIG. 7. Independent insertion of the C-terminal transmembrane span of pPsaK. The full precursor of PsaK (pPsaK) or the mature-size construct (PsaK) were incubated either with thylakoids that had been washed 3 times in HM buffer to remove stromal extract (denoted as "Thyl") or with thylakoids that had been treated with 0.15 mg/ml proteinase K for 30 min on ice and then washed 3 times with HM buffer (denoted as "PK-Thyl"). After incubation, samples of the thylakoids were analyzed directly (Ϫ) or after trypsin treatment as in previous figures (ϩ). Lanes Tr, translation products. pathway involving the input of numerous factors, both in the stroma and at the membrane surface, while several signal peptide-bearing proteins use an apparently simpler insertion mechanism that does not rely on any of the known protein machinery, although the issue of Alb3 involvement has yet to be addressed. PsaK is unlike any of the above proteins in that both the N and C termini are transported to the lumen, the protein is not synthesized with a signal-type peptide and it is not a member of the LHC superfamily of proteins.
The data from this study all point to a strictly SRP-independent insertion mechanism. Stromal extract contains essentially all of the SRP but is not required at any stage, and insertion does not depend at all on NTP hydrolysis. Both of these factors are critical for Lhcb1 insertion. FtsY involvement can also be ruled out since this factor hydrolyzes GTP during its operating mechanism. We cannot rule out the possibility that other, as yet unidentified soluble factors may assist PsaK insertion, and the slight inhibitory effect of apyrase does raise the possibility that proteins in the wheat germ translation system (e.g. chaperones) may aid insertion, but the data nevertheless indicate that the insertion of PsaK is fundamentally different from that of Lhcb1. It is as yet unclear why Lhcb1 is so dependent on SRP activity whereas other thylakoid membrane proteins studied to date are not.
We also find no evidence for the involvement of membranebound translocation machinery in PsaK insertion. Several studies have shown that trypsin treatment blocks the translocation of lumenal Sec substrates, very strongly suggesting that the Sec apparatus is inactivated. This treatment slightly inhibits PsaK insertion (as indeed it does for PsbY (24)) but the effect is not marked and we conclude that the Sec translocon does not play a major role in this pathway. We also find no evidence for Alb3 involvement, since antibodies raised against Alb3 severely inhibit Lhcb1 insertion yet have no effect on the insertion of PsaK. On the basis of these data alone we cannot exclude the possibility that PsaK may interact with Alb3 in a manner which is not affected by the antibodies used in this study. However, other ongoing studies in this laboratory (not shown) have demonstrated that Alb3 is completely degraded when thylakoids are treated with trypsin under the conditions used in this study (e.g. in Figs. 5 and 7) and, since this treatment barely affects PsaK insertion, we conclude that Alb3 is not required for PsaK biogenesis. This finding is significant because the homologous YidC protein plays a central role in the insertion of at least one SRP/Sec-independent membrane protein in E. coli (19). This factor has come to be regarded as a novel form of translocase in its own right, since the related Oxa1 protein also plays an important role in membrane protein biogenesis in yeast (17,18) and there is no evidence as yet for additional subunits in the mitrochondrial Oxa1p complex. Our data thus indicate that PsaK is unique (to date) because it is the only Alb3/YidC-independent membrane protein among those analyzed in bacteria and plant thylakoids.
In general, the topology adopted by PsaK is consistent with the "positive-inside" rule (30) which states that positivelycharged residues are found more frequently on the cis side of the membrane. The stroma-exposed loop region contains 4 basic residues whereas the C-terminal lumenal tail contains only one and the N-terminal region does not contain any (24). However, it is presently unclear why insertion is stimulated to some extent by the thylakoidal ⌬ H ϩ. In bacteria and mitochondria, the insertion of many membrane proteins is stimulated by the proton motive force (1,17,18) and the same applies to the Alb3-dependent insertion of Lhcb1 in thylakoids (9,11,12). However, these effects probably reflect the harnessing of the ⌬ H ϩ by the translocation machinery (the Sec and/or Oxa1-type apparatus) and these factors are not required for PsaK inser-tion. Possibly, there is a mildly stimulatory electrophoretic effect on the translocation of the C-terminal region of PsaK, which contains a single acidic residue. However, this point remains to be investigated in detail.
The Sec/Alb3-independent nature of the insertion mechanism raises the strong possibility that insertion of PsaK occurs spontaneously upon reaching the thylakoid membrane, and such mechanisms have been postulated before on theoretical grounds and following studies on membrane-interactive peptides and toxins (reviewed in Ref. 32). However, further work is required to address this possibility, and one argument against this possibility is that PsaK may then be able to interact with other membranes (e.g. the envelope) in the absence of a specific and dedicated targeting system. Possibly, PsaK is predisposed to insert only into thylakoid-type lipids, and the thylakoid membrane is indeed very unusual in terms of lipid composition, being composed primarily of galactolipids which are chemically very different to phospholipids (reviewed in Ref. 33). Further work is certainly required to determine whether such a lipidbased sorting process operates for PsaK insertion, or whether novel forms of translocation apparatus are involved.