IlDepartment of Biochemistry, University of Aberdeen, AB9 IAS, Scotland, United Kingdom, and ... tionally identical to stroma-targeting pre-sequences) followed.
THEJOURNAL OF BIOLOGICAL CHEMISTRY Q 1991 by The American Society for Biochemistry and Moleculal’ Biology, Inc
Val. 266, No. 35, Issue of December 15, pp. 23606-23610.1991 Printed in U.S.A.
Transport of Proteins into Chloroplasts DELINEATION OF ENVELOPE“TRANSIT” AND THYLAKOID “TRANSFER SIGNALS WITHIN THE PRESEQUENCES OF THREE IMPORTED THYLAKOID LUMEN PROTEINS* (Received for publication, July 17,1991)
Diane C. Bassham$, Dieter Bartlingg, Ruth M. Mould$, Bryan Dunbarll, Peter WeisbeekII , Reinhold G. HerrmannQ,and Colin Robinson$** From the $Department of Biological Sciences, University of Warwick, Coventry, CV4 7AL, United Kingdom, the SBotanisches Znstitut der Ludwig-Marimilian-Universitat,Menzingerstrasse 67, 8000 Munchen 19, Federal Republic of Germany, the IlDepartment of Biochemistry, University of Aberdeen, A B 9 I A S , Scotland, United Kingdom, and the IlDepartment of Molecular Cell Biology and Institute of Molecular Biology, University of Utrecht, Padualaan8, 3584 CH Utrecht, The Netherlands
The targeting of cytosolically synthesized proteins they are essentially devoid of similarity at either the primary into the thylakoid lumen is mediated by an aminoter- or secondary structure level (von Heijne et al., 1989). Indeed, minal pre-sequence consisting of an “envelope transit” von Heijne and Nishizawa (1991) have proposed thatan and a “thylakoid transfer” signal in tandem. We have essential feature of stroma-targeting signals is the ability to investigated the structural characteristics of several assume random-coil conformations in order to interact with thylakoid transfer signals by determining the inter- “chaperone” molecules involved in the protein translocation mediate sites at which the stromal processing peptidase system in the chloroplast envelope. cleaves to remove the transit sequences. Using this After import, precursors of stromal proteins are converted approach we have found that the thylakoid transfer to the mature sizes by a metal-dependent stromal processing signals of Silene pratensis plastocyanin, 23-kDa oxygen-evolving complex protein from wheat, and 33-kDa peptidase (SPP)’ which is highly specific for imported preoxygen-evolving complex protein from wheat, are 25, cursors (Robinson and Ellis, 1984). As with the targeting 39, and 48 residues in length, respectively. All of the signals, however, the absence of sequence similarity in the transfer signals contain hydrophobic core sequences vicinity of the cleavage sites has precluded a detailed underand a “-3,-1”motif reminiscent of those found in signal standing of the mechanism involved. Chloroplast biogenesisinvolves the import of numerous sequences, but the amino-terminal regions of the transfer signals of the 23- and 33-kDa proteins are both proteins intothe stromal phase, but asubset of these proteins longer and more highly charged. The net charge of undergoes further targeting into the internal thylakoid neteach amino-terminal region of the transfer sequences work. Nuclear-encoded thylakoid lumen proteins, such as is +1, including the amino-terminal amino group. In plastocyanin (PC’) and the 33-, 23-, and 16-kDa proteins each case, the stromal processing peptidase cleaves (33K, 23K, 16 K’) of the oxygen-evolving complex, are imimmediately after a positively charged residue, but ported by a complex pathway which can be divided into two otherwise the cleavage sites exhibit no common ele- phases. Initially, these proteins are imported into the stroma ments of either primary or secondary structure. and processed to intermediate forms by SPP, after which the intermediates are transportedacross the thylakoid membrane by a separate transportsystem. During or shortly after transport into the lumen, cleavage to the mature sizes is carried The biogenesis of the chloroplast involves protein targeting out by a thylakoidal processing peptidase, TPP’ (Hageman et on a large scale, since most chloroplast proteins are synthe- al., 1986; James et al., 1989; Kirwin et al., 1989). In keeping sized in the cytosol and post-translationally imported. The with this two-step import pathway, thylakoid lumen proteins targeting mechanisms involved appear to be both efficient are synthesized with bipartite pre-sequences consisting of an and accurate in that thereis little evidence of significant mis- amino-terminal domain or “transit” sequence (which is funcsorting of chloroplast proteins. However, the molecular basis tionally identical to stroma-targeting pre-sequences) followed for this targeting specificity is presently obscure because the by a thylakoid “transfer” sequence. targeting signals have proved to be difficult to define. All A comparison of several lumenal protein pre-sequences imported chloroplast proteins (to date) are initially synthe- reveals that they have two features in common: the presence sized with amino-terminal pre-sequences, and in a number of of short chain residues (usually alanine) at the -3 and -1 cases such pre-sequences have been shown to contain essential residues, relative to theTPP cleavage site, and ahydrophobic targeting information (van den Broeck et al., 1985; Schreier stretch of residues shortly before the -3, -1 motif (von Heijne et al., 1985; Kavanagh et al., 1988; Lubben et al., 1989; Mead- et al., 1989; Halpin et al., 1989).These features are also shared ows et al., 1989). However, a comparison of the pre-sequences by “signal“ sequences which target proteins into the endoof various stromal proteins hasshown that although they are plasmic reticulum or bacterial periplasm, and it is clear that all positively charged and enriched in hydroxylated residues, these peptide types display functional similarities. Halpin et al. (1989) showed that TPP is a signal-type peptidase capable * This work was supported in part by the Science and Engineering Research Council and by Schering Agrochemicals. The costs of pub- of cleaving signal peptides of both eukaryotic and prokaryotic lication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ** To whom correspondence should be addressed.
’ The abbreviations used are SPP, stromal processing peptidase; TPP, thylakoidal processing peptidase; pre-33K, i33K, 33K and pre23K, i23K, 23K: precursor intermediate and mature forms of the 33and 23-kDa proteins of the oxygen-evolving complex, respectively.
23606
23607
Targeting Signalsfor Thylakoid Lumen Proteins origin, and Seidler and Michel (1990) showed that the carboxyl-terminal residues of the spinach33K pre-sequence can direct exportof 33K into theEscherichia coli periplasm. In spite of the above studies, theoverall features of thylakoid transfer sequences have remained partially obscure, especially with respect to the amino-terminal regions because it is impossible to deduce from cDNA sequence data where the SPP cleavage sites are located within lumenal protein pre-sequences. In the present study,we have used a microsequencing approach to identify theSPP cleavage sites within several lumenal protein pre-sequences, and we describe the structural features of the thylakoid transfer signals in each case. EXPERIMENTALPROCEDURES
Materials-Radioactive amino acids [?H]leucine,['Hllysine, and ['H]phenylalanine were purchasedfromAmershamInternational, United Kingdom (catalog numbers T R K 683, TRK 752, and TRK 648). I n Vitro Translations-Precursor proteins were synthesized by translation of synthetic mRNAs encoding pre-23K, pre-33K, or prePC generated by in uitro transcription using SP6 or T7 RNA polymerase (James et al., 1989; Hageman et al., 1986). Capped transcripts were translated in a wheat germ lysate in thepresence of one labeled amino acid (see figure legends) at 1 pCi/pl of translation mixture. Generation and Edman Degradation of Intermediate Forms-30 pl of translationmixture(containing 10,000-40,000 cpm of labeled precursor) were mixed with 30 pl of partially purified SPP and incubated for 60 min a t 27 "C as described by Robinson and Ellis (1984). The mixture was resolved on a sodium dodecyl sulfate-polyacrylamide gel according totheconditions described in Applied Biosystems User Bulletin no.25. The gel was thenblottedonto Immobilon membrane (Millipore, U. K.) and the band corresponding to the intermediate-sized protein identified by counting slicesfor radioactivity. The band was then excised and placed in the cartridge block of an Applied Biosystems model 470a proteinsequencer equipped with a 120a on-line phenylthiohydantoin analyzer. Fractions from each cycle were counted for "H radioactivity. Expression of a n Artificial i23K"Codon 34 (lysine) of the wheat pre-23K cDNA, p23K-1 (James and Robinson, 1991) was changed from AAG to ATG by site-specific mutagenesis using the Amersham kit and manufacturer's protocol. The cDNA was then digested with BgA, whichcleaves 5' totheintroduced ATG, and EcoRI. This fragment was blunt-ended usingT4 DNA polymerase and ligated into pGem 42 (Promega Biotech)which had been digested with SmaI and EcoRI. The i23K was synthesized by transcription using T 7 RNA polymerase, followed by translation in a wheat-germ lysate.
partially purified peptidase cleaves these precursors at the correct sites. The SPP Cleavage Site in the Wheat 23K Pre-sequenceWheat pre-23K was synthesized by transcription translation of plasmid p23K-1 as described (James etal., 1989). The precursor was labeled with ["Hllysine and processed to the intermediate form by incubation with SPP. Fig. 1 shows the results of Edman degradation of wheat i23K, in which a single peak of radioactivity is released a t cycle 3. Apart from a lysine 7 residues before the terminal cleavage site, the wheat presequence contains lysine only at residues 34 and 37 (James and Robinson, 1991). Cleavage by SPP must therefore take place immediately after lysine 34, in order for a single lysine residue (lysine 37)to emerge in cycle 3 of the sequencing run. The sequencing data for wheat 23K are corroboratedby the analysis of an artificial i23K construct. The last residue of the envelope transit sequence,lysine 34, was changed to methionine by site-specific mutagenesis of p23K-1. The 5' section of the coding sequencecontaining the initiation codon was then removed, generating a construct in which translation initiates at the introduced methionine codon. Fig. 2 shows that this translation product which, according to the radiosequencing data, should be only 1 residue larger than authentic i23K is very similar in terms of gel mobility to i23K which is generated by SPP. These data provide strong support for the accuracy of the radiosequencing approach. Furthermore,
A
K ~ A Q K N D E A A S D A A V T S R R A A L 3H L
B (Cpm)
"I B
2m
am 160 100
so n
3
1
RESULTS 4
2
7
5
6
9 11 13 15 17 19 8 10 12 14 18 18 20 CYCLE NUMBER
Experimental Strategy-The experimentalapproachinvolved subjecting intermediate-sized forms of lumenal proFIG. 1. Identification of the SPP cleavage sitewithin wheat teins to Edman degradation in order to identify the aminopre-23K. A, sequence of residues 34-54 within the pre-sequence. B, terminal residues of the thylakoid transfersequences. In this radiosequencinganalysis of i23K generated by incubation of ['HI procedure, radiolabeled amino acids were incorporated into lysine-labeled pre-23K with SPP. Fractions generated by each cycle theprecursorproteinsduringtranslation,andtheamino of Edman degradation were counted for "H radioactivity. The SPP termini of the thylakoid transfer sequences were deduced from cleavage site in A is denoted by V. the patternsof release of the labeled residues from the intermediate-sized forms during the sequencing reactions. Because 1 2 3 thestromalintermediateformsare difficult to isolate in sufficient quantities from in vitro chloroplast import reactions, theywere instead generatedby incubation with partially pre-23Kb 423K purified SPP. SPP cleaves very specifically in these incubations, with noevidence to dateof incorrect processing of any precursor (Robinson and Ellis 1984; Musgrove et al., 1989). Furthermore, thepolypeptides generated by the incubationof isolated SPP with pre-PC andpre-23K comigrate on sodium FIG.2. Expression of an artificial wheat i23K construct. An dodecyl sulfate-polyacrylamide gels with the stromal interartificial i23K construct encoding wheat 23K plus a truncated, 40mediates generated during import of these proteins into intactresidue pre-sequence was prepared as detailed under "Experimental chloroplasts (Hageman et al., 1986; James et al., 1989). Dif- Procedures" and expressed by in uitro transcription translation. Lune ferences in size of only a few amino acids can be detected I, pre-23K; lane 2, artifical i23K translation product; lane 3, generausing high resolutiongel systems, stronglysuggesting that the tion of authentic i23K by incubation of pre-23K with SPP.
-
5
Targeting Signals
23608
A
Thylakoid Proteins Lumen for
-
R ~ A ~ G V D A G A R I T C S L Q S D I R E V A S
3
1 2
6 4
9
7 8
8
A
K v A S L K D V G V V V A A T A A
11 1 3 1 6 1 7 1 9 2 1 23 1012141818#)PP24
3Hp)w (apm) so0
L
C ml OD0
100 100
a, 0
1 2 9 4 6 8 7
8 9 101112131416 CYCLE NUMBER
FIG. 3. The SPP cleavage site within wheat pre-33K. A , sequence of residues 31-54 in the wheat 33K pre-sequence. B , Edman degradation of ['H]leucine-labeled i33K generated by SPP. C, Edman degradation of ['H]phenylalanine-labeled i33K. Fractions from each cycle of the Edman analysis were counted for 'H radioactivity. The S P P cleavage site in A is denoted by V.
other work' has shown that the artificial i23K is efficiently imported by isolated thylakoids and processed to the mature size, indicating that the identified thylakoid transfersequence is functional. The SPP Cleavage Site in the Wheat 33K Pre-seqmnceWheat pre-33K was synthesized by transcription translation of p33K-2 in the presence of ["Hllysine or ["Hlphenylalanine, and processed to i33K using SPP as described by James etal. (1989). Edman degradation of the leucine-labeled intermediate (Fig. 3B) releasesasingle peak of leucine at cycle 14, whereas phenylalanine (Fig. 3C is released at cycle 2. Comparison with the cDNA sequence (Fig. 3A) shows that SPP therefore cleaves afterarginine31inthe 79-residue presequence. No other cleavage sites compatible with these data are present in the33K pre-sequence (Meadows et al., 1991). The SPP Cleavage Site in the Silene pratensis Plastocyanin Pre-sequence-iPC was generated by cleavage of pre-PC by SPP as described (Hageman etal., 1986). Edman degradation of lysine-labeled iPC (Fig. 4B) releasesasingleresidue at cycle 4, indicating that cleavage takes place after lysine 41 in the pre-sequence(Fig. 4A).This experimentwas repeated and thesameresultobtained(datanotshown)and a similar analysis was carried out using leucine-labeled iPC (Fig. 4C). In this case, thesamplecontained less than the required
' R. M. Mould and C. Robinson, manuscript in preparation.
FIG. 4. The SPP cleavage site within Silene pratensis prePC. A , sequence of residues 41-56 of the pre-sequence. 8,radiosequencing analysis of ['HH]lysine-labeled iPC generated by SPP cleavage of pre-plastocyanin. C, as in B , but using ['H]leucine-labeled iPC. Fractions generated from each cycleof Edman degradation were counted for 'H radioactivity. The SPP cleavage site in A is denoted by V.
amount of radioactivity, but a small peakof label is neverthelessreleased at cycle 3, which would alsoagreewith SPP cleavage taking place after lysine 41 in the pre-sequence (Smeekenset al., 1985a). Thesedataareconsistentwith previous work (Hageman et al., 1986; Smeekens et al., 1986) which has shown that the SPP cleavage site is shortly after methionine 31. Cleavage Fidelity of SPP-The main aims of this studywere to use SPP as an experimental tool in order to characterize the thylakoid transfersequences of the precursors described. However, the results also serve to demonstrate thevery high degree of reaction specificity exhibited by this enzyme. It is particularlyevidentinthe sequencing runsin which the backgroundradioactivity was low (e.g. Figs. 1 and 3) that processing takes place onlyat a single definedsite. It is clearly of considerable interest to determine the basis for this reaction specificity, since this is presently obscure. DISCUSSION
Fig. 5 summarizes some of the salient features of the thylakoid transfer sequences delineated in this study; these features are also compared with a consensus signal peptide. A comprehensive analysis of prokaryotic and eukaryoticsignal peptides has shown that they areusually about 20 residues in length and that they share three key features: a positively
Targeting Signals for Thylakoid Lumen Proteins
23609
provisional at present because with this precursor we have not observed an imported, stromal intermediate form which can be tested for comigration on gels with the intermediate E.T. IUSTSCFLHQSTARLMSARPAPAVGRTQLFWCKV + ++ + form generated by isolated SPP. It is therefore possible that, T.T A Q K ? I D E M S D A A W T S ~ S P A M in the intact chloroplast, pre-33K is processed in the stroma both at the site identified in this study and at an additional site, and that the thylakoid transfer sequence is in fact shorter Wheat 33K (Meadows et al., 1991) than 48 residues. The overall structure of the pre-sequence, + + + + however, suggests that the thylakoid transfer signal is indeed E.T. msLQAMTvl4P~IGGRAssARPssHvAR~ - + -++ - + particularly long. Envelope transit sequences rarely contain T . T A P G V D A G A R I T C S L Q S D 1 R E S K C A L ) W G ~ S G A T A negatively charged residues (von Heijne et al., 1989) and most, if not all, of the 4 negatively charged residues in the33K presequence intermediate region are therefore likely to reside in Silene Plastocyanin (Smeekenset al., 1985) the thylakoid transfer signal. KO and Cashmore (1989) have + + + + + shown that the carboxyl-terminal 27 residues of the ArabidopE . T . NATVTSSMVAIPSFAGLKASSTTRMTVKVAVATPRMSIKV sis 33K pre-sequence are sufficient for transport across the +thylakoid membrane, if a stroma-targeting pre-sequence is T.T A S I J C D Y G W V ~ G N A U A used to first direct transport into chloroplasts. may This imply FIG. 5. Structural features of thylakoid transfer seauences that the thylakoid transfer sequence of this precursor is indeed and signal sequences. All of the thilakoid transfer signac (T.T.) shorter than 48 residues, or alternatively may mean that, in contain a hydrophobiccoreregion (underlined) and short chain this case,a truncatedtargeting signal is still capable of residues,often alanine, at the -3 and -1 positions ( A X A m o t i f ) . More variable are the overall lengths of the thylakoid transfer do- functioning efficiently. In addition to delineating the transit and transfer signals mains, and the distribution of positively and negatively charged residuesin the amino-terminal regions. The figure also shows the within these pre-sequences, the microsequencing data also structures of the envelope transit sequences (E.T.) and the sites of provide data on the cleavage sites recognized by SPP. An cleavage by SPP (denoted by '7). identical sequence VCKJA, is at the cleavage site in wheat pre-23K and probably also in spinach pre-23K, and in precharged amino-terminalregion (net chargeusually +1 or +2), 33K and pre-plastocyanin itis interesting to note thatcleava hydrophobic core region, and a more polar carboxyl-terminal age also occurs between a positively charged residue and an region in which the -3 and -1 residues are short chain amino alanine (VARJA and SIKJA, respectively). These data raise acids (von Heijne, 1985, 1986). As noted previously (von the possibility that the presence of a positively charged residue Heijne et al., 1989; Halpin et al., 1989), the hydrophobic core at the cleavage site is important for the selection of these and short chain -3, -1 residues are also found in thylakoid sites by SPP. However, itshould be notedthatthepretransfer sequences. sequences of stromal proteins also usually contain positively In general, the amino-terminalregion of the plastocyanin charged residues near the SPP cleavage sites, but only rarely transfer sequenceresembles signal sequences interms of at position -1. For example, the SPP cleavage site in the length and net charge(+1if the amino-terminal amino group Silene ferredoxin precursor is in the sequence GRVTAMJA, is included).However, the 23K transfer sequence is markedly and in the precursor of peaRubisco smallsubunit is different in two respects. First, thesequence is abouttwice as GGRVKCJM(Smeekens et al., 198513; Cashmore, 1983). long as typical signal sequences, which are on average 18-20 Thus, it may be coincidental that, in the four sites deduced residues in eukaryotes and21-23 residues in prokaryotes(von in this study,cleavage takes place immediately after lysine or Heijne, 1985). Most of the observed differences in length are arginine in each case. Another possibility is that precursors accounted for by significantly longer amino-terminal sections of thylakoid lumen and stromal proteins are processed by of the transfer sequence. The second difference concerns the different forms of SPP; we have no evidence that this is the distribution of charged residues in the23K thylakoid transfer case, but it should be emphasized that SPP has yet to be sequences. The amino-terminal region is highly charged and purified to homogeneity and may therefore exist in several is notable for the presence of three negatively charged resiforms. dues. Again, the net charge of the transfer sequence is +1 Further analysis of the SPP cleavage sites indicates that including the amino-terminal amino group. The functional thereis no significantprimary sequence homology which significance of the highly charged amino-terminal section has might serve asa recognition signal. Furthermore, no common yet to be investigated, but thisregion may be responsible for secondary structures are reliably predicted in these regions. maintaining the otherwise hydrophobic transfer signal in a The molecular basis for the high degree of SPP reaction soluble, accessible configuration. specificity thereforeremains obscure, and clearlyrequires Preliminary evidence' suggests that SPP cleaves after lysine 41 in the spinach 23K pre-sequence, within a similar further attentionusing new approaches. motif (VCKJA) to that present at the wheat 23K cleavage REFERENCES site. Thiswould give a thylakoid transfer signalof 40 residues, Cashmore, A. R. (1983)in Genetic Engineering of Plants (Kosuge, T., 1residue longer than the corresponding wheat transfer signal. Meredith, C. P., and Hollaender, A. eds) pp. 29-38, Plenum PubAccording to the positionof the SPP cleavage site in wheat lishing Co., New York pre-33K, the thylakoid transfersequence of this precursor is Hageman, J., Robinson, C., Smeekens, S., and Weisbeek, P. (1986) even longer (48 residues) andmore highly charged, with four Nature 324,567-569 positive and four negative charges within the amino-terminal Halpin, C., Elderfield,P. D., James, H. E., Zimmermann,R., Dunbar, B. and Robinson, C. (1989) EMEO J . 8 , 3917-3921 region prior to the hydrophobic core region (again giving a net charge of +l).However, these data must be regarded as James, H. E., and Robinson, C. (1991) Plant Mol. Bid. 17,179-182 James, H. E., Bartling, D., Musgrove, J. E., Kirwin, P. M., Herrmann, R. G., and Robinson, C. (1989) J . Bid. Chem. 264, 19573-19576 E. Wachter,R. G. Herrmann, and C. Robinson, unpublished data. Jansen, T. Rother, Z., Steppuhn, J., Reinke, H.,Bayreuther, K. Wheat 23K (James and Robinson, 1991)
+
"
+
+
+
23610
Targeting Signals for ThylakoidProteins Lumen
Jansson, C., Anderson, B., and Herrmann, R.G. (1987)FEBS Lett. 216,234-240 Kavanagh, T. A., Jefferson, R.A., and Bevan, M.W. (1988)Mol. Gen. Genet. 216,38-43 Kirwin, P. M., Meadows, J. W., Shackleton, J. B., Musgrove, J. E., Elderfield, P. D., Hay, N. A., and Robinson, C. (1989)EMBO J. 8, 2251-2255 KO,K., and Cashmore, A. R. (1989)EMBO J. 8, 3187-3192 Laemmli, U. K. (1970)Nature 227,680-685 Lubben, T., Gatenby, A. A., Ahlquist, R., and Keegstra, K. (1989) Plant Mol. Biol. 12, 13-19 Meadows, J. W., Hulford, A., Shackleton, J. B., and Robinson, C. (1989)FEBS Lett. 253, 244-246 Meadows, J. W., Hulford, A., Raines, C. A., and Robinson, C. (1991) Plant Mol. Biol. 16,1085-1087 Musgrove, J. E., Elderfield, P. D., and Robinson, C. (1989)Plant PhySiOl. 90,1616-1621
Robinson, C., and Ellis, R. J. (1984)Eur. J. Biochem. 142.337-342 Schreier, P. H., Seftor, E. A., Schell, J., and Bohnert, H.. J. (1985) EMBO J. 4.25-31 Seidler, A., and Michel, H. (1990)EMBO J. 9, 1743-1748 Smeekens, S., deGroot, M., van Binsbergen, J., and Weisbeek, P. (1985a)Nature 317,456-458 Smeekens, S., van Binsbergen, J., and Weisbeek, P. (1985b)Nucleic Acids Res. 13,3179-3194 Smeekens, S., Bauerle, C., Hageman, J., Keegstra, K., and Weisbeek, P. (1986)Cell 46, 365-375 Van den Broeck, G., Timko, M. P., Kausch, A. P., Cashmore, A. R., Van Montagu, M., and Herrera-Estrella, L. (1985)Nature 313, 358-363 Von Heijne, G. (1985)J . Mol. Biol. 184, 99-105 von Heijne, G. (1986)Nucleic Acids Res. 14,4683-4690 von Heijne, G., and Nishizawa, K. (1991)FEBS Lett. 278, 1-3 von Heijne, G., Steppuhn, J., and Herrmann, R.G. (1989)Eur. J. Biochem. 180,535-545
