- Email: [email protected]
Protein translocation into and across the bacterial plasma membrane and the plant thylakoid membrane
REVIEWS
TIBS 24 – JANUARY 1999
Protein translocation into and across the bacterial plasma membrane and the plant thylakoid membrane Ross E. Dalbey and Colin Robinson Over the past decade, some familiar themes have emerged on how proteins are inserted into or translocated across the plant chloroplast thylakoid membrane and bacterial inner membranes. In the SecA and signal recognition particle (SRP) pathways, nucleotides and soluble factors are used to translocate proteins across the membrane bilayer in the unfolded state. However, the DpH-dependent pathway in thylakoids uses a radically different mechanism: transport of proteins across the membrane is driven by the transmembrane pH gradient, and neither stromal factors nor nucleotide triphosphates are needed. In addition, this pathway, which requires the membrane-bound protein Hcf106, appears to translocate proteins in a tightly folded form. Recently, a similar pathway has been shown to operate in eubacteria, and several of its components have been identified. GRAM-NEGATIVE EUBACTERIA and plant chloroplasts have much in common. We have known for a long time that chloroplasts evolved from endosymbiotic, cyanobacteria-like cells, and the chloroplast still exhibits many prokaryotic features. The chloroplast contains its own genetic system, and the gene structure and protein-synthesizing machinery of chloroplasts are similar to those of prokaryotes in many respects. However, the chloroplast genome now encodes only a small proportion of the chloroplast proteins, the vast majority of genes having been transferred to the nucleus since the initial endosymbiotic event. Chloroplast biogenesis therefore requires the import of hundreds of proteins into the organelle. In structural terms, there are again clear similarities. Gram-negative eubacteria and chloroplasts are enclosed by two membranes and, in both cases, two aqueous spaces are formed: the periplasm and cytoplasm in eubacteria; and the intermembrane space and stroma in chloroplasts. However, the chloroplast contains in addition the thylakoid R. E. Dalbey is at the Dept of Chemistry, Ohio State University, Columbus, OH 43210, USA; and C. Robinson is at the Dept of Biological Sciences, University of Warwick, Coventry, UK CV4 7AL.
membrane, which in turn encloses a further aqueous phase, the thylakoid lumen. In terms of protein biogenesis, eubacteria and chloroplasts differ in many respects because protein traffic in eubacteria is directed outwards towards the periplasm/medium whereas proteins are imported into chloroplasts. In recent years, however, it has become clear that many proteins are targeted into or across the thylakoid membrane by mechanisms that strongly resemble well-studied processes in Escherichia coli. The learning process has not been unidirectional, and studies on thylakoid protein transport have led to the identification of a unique type of protein translocase that operates in a wide variety of eubacteria. Here, we focus on these similarities and attempt to summarize the salient features of the increasingly complex web of protein-sorting processes in these organisms.
The basic targeting pathways In eubacteria, export across the inner membrane requires an N-terminal signal peptide (also known as a leader peptide), which is removed by either signal peptidase I (also known as leader peptidase) or signal peptidase II (also known as lipoprotein signal peptidase). The leader peptide is characterized by a positively charged N-terminal region (n),
0968–0004/99/$ – See front matter © 1999, Elsevier Science. All rights reserved.
a hydrophobic central domain (h) and a polar C-terminal region (c). Residues at the 23 and 21 positions are necessary for leader-peptidase cleavage1 (Fig. 1a). Most proteins that are imported from the cytoplasm to the chloroplast thylakoid lumen are synthesized with a bipartite presequence2 (Fig. 1b). The N-terminal peptides, which are 40–70 residues long and are enriched in threonine and serine residues, contain the information necessary for import of the protein into the stroma. They are removed by the stromal processing peptidase (SPP). The second signal directs proteins across the thylakoid membrane, after which the signal is removed by a thylakoidal processing peptidase (TPP), which shares significant sequence similarity with the bacterial leader peptidase.
The translocation mechanisms: some familiar themes Studies over the past few years have demonstrated that multiple pathways translocate proteins into or across the bacterial plasma membrane and the plant thylakoid membrane. The known translocation components in eubacteria and chloroplasts are shown in Table 1. Of the four primary pathways identified in chloroplasts (Fig. 2), two appear to be similar, in most respects, to well-studied bacterial targeting mechanisms. Several labs elucidated the bacterial Sec pathway, exploiting genetic and biochemistry techniques. Genetic studies revealed that the cellular machinery includes SecA, SecB, SecD, SecE, SecF, SecG and SecY3,4. Translocation studies of reconstituted systems helped to reveal the biochemical functions of the Sec components5,6. Protein translocation across the inner membrane in E. coli (see Fig. 2a) requires typically, but not always, a molecular chaperone, such as SecB, and a translocase complex that comprises membrane-bound SecYEGDFyajC and the peripheral membrane protein SecA (Ref. 7). SecB binds to preproteins and retards their folding; it also participates in the delivery of the preprotein to the membrane by interacting with the membrane-bound SecA. SecA is required for translocation; it is an ATPase and probably functions as a translocation motor that pushes exported proteins across the membrane. The trimeric SecYEG complex is believed to function as a protein-conducting translocation channel. SecD, SecF and YajC also form a trimeric complex and can associate with SecYEG to form a hexameric complex8.
PII: S0968-0004(98)01333-4
17
REVIEWS
TIBS 24 – JANUARY 1999
(a) Bacterial signal peptides SPase-cleavage site R/K n region
h region
R/K
RR n region
c region
h region
c region
(b) Bipartite thylakoid-lumen-targeting presequences SPP-cleavage site
TPP-cleavage site h region
n region Stroma-targeting peptide
c region Thylakoid signal peptide
RR Stroma-targeting peptide
Thylakoid transfer peptide
Figure 1 Distinct types of targeting signals. (a) Bacterial signal peptides and twin-arginine signal peptides. The twin-arginine motifs target proteins to the Tat pathway. (b) Chloroplast thylakoidlumen bipartite presequences. The N-terminal portion of the chloroplast thylakoid-lumen presequence resembles a stroma-targeting sequence; the C-terminal part contains a hydrophobic sequence that targets the protein to the thylakoid lumen. Of the two thylakoid transfer sequences shown, the upper type targets proteins to the Sec-dependent pathway; the lower type targets proteins to the DpH pathway. The presequence is cleaved twice – first by the stromal processing peptidase (SPP) and then by the thylakoid processing peptidase (TPP; a membrane-bound protease whose catalytic site is exposed to the thylakoid lumen).
Table 1. Proteins implicated in translocation in bacteria and chloroplast Function
Bacteria
Chloroplast
Targeting
SecB Ffh
cpSRP54
Translocation
SecA SecY SecE SecG SecD SecF YajC TatE(Ybec) TatA(YigT) TatB
SecA SecY
Hcf106
TatC TatD ? Ftsy Signal-peptide cleavage
SPase I SPase II
TPP SPP
Abbreviations used: SPP, stromal processing peptidase; TPP, thylakoid processing peptidase.
18
SecD and SecF are membrane proteins, and possess large periplasmic domains that help to maintain the proton motive force across the membrane, to stabilize the membrane-inserted form of SecA and to promote the release of exported proteins into the periplasm. A Sec-dependent pathway also operates in chloroplasts (Fig. 2b); a SecA homolog is involved in thylakoid protein transport9–11, and a thylakoid SecY homolog has been cloned – although a direct role for this protein has not been established12. The other Sec proteins remain to be identified in plants, and it is still unclear whether stromal chaperones bind to proteins imported into the stroma from the cytosol and prevent them from aggregating and misfolding. Nevertheless, so far, the two types of Sec system appear to be very similar in most respects. Walter and Blobel13 elucidated the role of signal recognition particle (SRP) in export of proteins to the ER over 15 years ago. The mammalian SRP is
composed of six subunits and one RNA component. A hint that SRP might also be involved in targeting in the bacterial system was the discovery in eubacteria of ffh and ftsY genes, which are homologous to the genes that encode the 54-kDa SRP subunit and the SRP receptor, respectively, in eukaryotes, although homologs of the other gene products have not been found. The role of Ffh and FtsY in bacterial protein translocation was controversial initially, however, because genetic screens did not identify a role for these proteins in protein targeting14. Support for their role in vivo came first from Phillips and Silhavy15, who showed that depletion of Ffh led to defects in the export of certain proteins. There is now agreement in the bacterial field that Ffh and FtsY play a critical role in targeting and therefore translocation of a subset of membrane proteins16. The majority of periplasmic and outer-membrane-protein precursors, by contrast, can be exported even under conditions where Ffh is limiting within the cell. Recent studies have shown that bacterial SRP transfers some substrates to the SecYEG translocase17. In fact, several integral membrane proteins that insert in a Ffh-dependent manner may insert in a SecY-dependent manner. In 1994, Li and co-workers18 showed that chloroplasts also contain a homolog, cpSRP54, of the SRP 54-kD subunit (Fig. 2b). The cpSRP54 protein seems to play a similar functional role: its immunodepletion inhibited insertion of a multispan membrane protein [the lightharvesting chlorophyll-binding protein (LHCP) of photosystem II] into thylakoids18. Moreover, insertion of LHCP into thylakoid membranes is ten times more efficient in the presence of GTP than it is in the presence of other nucleotides19, and is inhibited by non-hydrolyzable GTP analogs. Interestingly, the cpSRP54 subunit associates with a novel 43-kDa component, cpSRP43 (Ref. 20); there is no evidence, so far, that any RNA molecule is present. To date, an FtsY homolog has not been discovered in chloroplast thylakoids. One major question concerns the determinant, within the exported or membrane protein, that enables interaction with cpSRP54 and targeting to the membrane. This determinant might in fact be the hydrophobicity of the protein chain. High et al.21 showed, using nascent-chain crosslinking, that cpSRP54 interacts with precursor proteins that have particularly hydrophobic signals. They suggested that the SRP-dependent targeting system is important for routing integral
REVIEWS
TIBS 24 – JANUARY 1999 membrane proteins to the thylakoid membrane. This is consistent with the observation that the SRP-dependent pathway seems to be critical only for membrane proteins.
(a)
C D F
A novel type of protein translocase in thylakoids and eubacteria In some respects, the field of thylakoid protein targeting has followed in the footsteps of bacterial studies in that similar pathways have emerged. A prime example is the Sec pathway used for a subset of thylakoid lumen proteins. However, in one critical respect, ideas have flowed in the other direction, following the finding that some luminal proteins are transported by a mechanism that could hardly be more different from the Sec-type mechanism. Substrates for this pathway, such as the 23-kDa and 16-kDa photosystem-II proteins (known as 23K and 16K, respectively) require neither stromal factors nor nucleoside triphosphates for transport across the thylakoid membrane but are instead totally reliant on the pH difference across the thylakoid membrane (DpH)22–26. Both Sec- and ∆pH-dependent substrates have signaltype targeting signals in which the three typical domains can be discerned: a charged N-terminal domain (N domain); a hydrophobic core region (H domain); and a more polar C domain that ends with a signal-peptidase-cleavage motif, usually Ala-X-Ala. Remarkably, however, these presequences are able to direct highly accurate, pathway-specific targeting, and there is little evidence of cross-talk between pathways25,27. This finding prompted a more rigorous examination of the targeting signals; Chaddock et al.28 showed that a crucial determinant in transfer peptides for the DpH pathway is the presence of an essential twinarginine motif immediately before the H domain. Other factors must, however, be involved: a twin-arginine motif alone is unable to divert Sec substrates onto this pathway. Recently, Settles and co-workers29 identified a plant protein, Hcf106, that is critical for the DpH pathway, following the isolation of a maize mutant in which this pathway is defective30. The Sec and SRP-dependent pathways are unaffected in this mutant, which suggests that Hcf106 is a specific and essential element of the elusive DpH-driven translocation system. The cloning of the hcf106 gene has provided major insights into both the origins and likely mechanism of this system. Homologs of hcf106 are present in a wide range of bacterial genomes as
?
Tat
Periplasm
Y EG FtsY
A
PMF
P SR P GT
PMF ATP
Cytoplasm Spontaneous
Ffh
B
(b)
Hcf106
?
Thylakoid lumen
Y A
SRP GTP
ATP
∆pH
Stroma Spontaneous
SRP 54
Figure 2 Protein export and membrane insertion pathways in eubacteria (a) and plant chloroplast thylakoids (b). Proteins are exported by different pathways. Some of these pathways require Sec proteins; others require Tat proteins. Classical signal peptides target proteins to the Sec pathway; twin-arginine signal peptides target proteins into the Tat pathway. The Tat pathway is DpH-dependent in chloroplasts, but the energetics remain to be determined in eubacteria. Insertion of integral membrane proteins occurs either by the signal recognition particle (SRP)-dependent pathway or by an apparently spontaneous pathway. At least one SRP substrate is inserted through SecYEG, although it is not known whether all SRP substrates follow this rule. The SRP-dependent pathway might also require an unidentified membrane component or might not require a protein-export apparatus. Blocks indicate hydrophobic regions; shaded blocks indicate hydrophobic regions that are signal peptides that are cleaved by signal peptidase (in Escherichia coli) or thylakoid processing peptidase (TPP; in chloroplasts). A, SecA; B, SecB; C, YajC; D–G, SecD–SecG; PMF, proton motive force; Y, SecY.
unassigned reading frames; this provides clear indications that a similar system might operate in prokaryotes. Berks31 has pointed out that periplasmic proteins that contain any of a range of redox cofactors invariably bear signal peptides that possess twin-arginine motifs. This has promoted speculation that the latter might be recognized by a system akin to the thylakoidal DpH system. Very recently, this scenario has been proved to be correct, and the first details of a Sec-independent bacterial export system have emerged. Two hcf106 homologs are found in the E. coli genome; one
of these genes (originally designated ybeC ) appears to be monocistronic, whereas the other (originally yigT ) appears to be part of an operon. The original yigT gene comprises two separate genes, which in this operon are followed by two further genes, yigU and yigW (see Fig. 3). Sargent et al.32 have probed the functions of these genes by using in-frame deletions and found that disruption of either yigT (the first gene shown in Fig. 3) or ybeC caused a marked inhibition of the export of at least five periplasmic proteins that contain twin-arginine presequences. In a double-deletion strain,
19
REVIEWS yigQRS
TIBS 24 – JANUARY 1999
(yigT ) tatA
(yigT ) tatB
(yigU ) tatC
(yigW ) tatD
rfaH
(ybeC) tatE
Figure 3 The Tat operon in Escherichia coli. The operon comprises four genes. The tatA gene is homologous to the maize hcf106 gene, and a second Hcf106 homolog is encoded by the unlinked tatE gene. Note that TatA, TatB, TatC and TatE have been shown to be involved in Sec-independent protein export but that the role of the TatD protein has yet to be established. Previous designations for reading frames are shown in parentheses. The yigQRS gene is the final ORF in a cluster of genes involved in ubiquinone biosynthesis; rfaH encodes a transcriptional regulator.
clarified experimentally but there seems little doubt at present that this system achieves the remarkable feat of transporting large globular proteins across coupled biological membranes. Teter and Theg40 have shown that the thylakoidal DpH is not compromised when the ∆pH-driven translocase is at full capacity; hence, the challenge for the future is to determine how an array of proteins can be squeezed through these membranes while protons are excluded so effectively.
The spontaneous pathway for membrane protein insertion: more variations on a theme export is totally defective, which implied that these gene products play a central, and perhaps overlapping, role. The observation that Sec-dependent export is unaffected in these strains confirms the existence of a separate export pathway. Members of this group of genes have been designated tat genes (for twin-arginine translocation), and the hcf106 homologs are tatA, tatB and tatE (see Fig. 3). A separate study has analyzed the function of this operon. Weiner et al.33 have identified an E. coli mutant in which this export process is defective, and the mutation lies in the yigT region; the authors concluded that an hcf106 homolog had been disrupted in this strain. However, it is now clear that this mutation lies instead in the second gene of this operon. This study thus effectively demonstrated a critical role for TatB. Finally, TatC (the product of the gene formerly known as yigU ) was identified as a fourth component of this export pathway, following the demonstration that disruption of the tatC gene leads to a complete block in the export of a range of cofactor-containing proteins that bear twin-arginine targeting signals34.
Transport of folded proteins by the Tat apparatus? One of the canonical rules of protein transport is that proteins must be transported in an unfolded state. For example, Randall and Hardy35 showed that tight folding of the preprotein of maltosebinding protein is incompatible with translocation across the membrane by the Sec pathway. Other data showed that slowing down the rate of protein synthesis, by addition of chloramphenicol, suppresses mutations that render the bacterial Sec machinery less efficient. This suggested that there is a kinetic competition between folding and export of proteins, which led to the view that Sec-dependent translocation requires the domain to be translocated in an
20
unfolded state. It does, however, seem that the Sec system is not as strict as was once believed; Mizushima and coworkers36 showed that a mutant of proOmpA that possesses an intramolecular disulfide bridge can be translocated in the presence of a proton motive force. The emerging evidence suggests that the DpH-dependent and Tat systems in thylakoids and eubacteria operate by a completely different principle. Studies on the thylakoidal system, using proteasesensitivity as an assay for the folding state of a luminal protein, showed that the extrinsic 23-kDa protein of photosystem II is in a tightly folded state in the stroma prior to translocation across the thylakoid membrane37. This observation and others38 raised the possibility that the system transports fully folded proteins, and such an idea is strongly supported by the studies on eubacteria. The known substrates for this export pathway bind a variety of complex redox cofactors, including FeS and molybdopterin centers, all of which are believed to be inserted in the cytoplasm31. This would inevitably necessitate folding of the mature protein. This idea is strongly supported by the finding that most of these proteins accumulate as active cytoplasmic forms in the E. coli mutants in which the tat genes are disrupted, which clearly indicates that cofactor insertion has taken place32–34. However, the story gets even more interesting because there is also evidence that export of some proteins can be blocked when cofactor insertion is prevented31,39; this observation raises the fascinating possibility that the system requires substrate proteins to be folded, at least in some cases, and that a proofreading mechanism actively prevents export until cofactor insertion has taken place. Such a proofreading mechanism would of course be logical (to avoid export of useless proteins) but unique among known protein-translocation systems. Many of these points need to be
The availability of bacterial strains carrying temperature-sensitive SecA and SecY mutants has allowed us to show that some proteins insert across the membrane independently of the Sec translocase. The M13-virus coat protein is such a protein41. It probably inserts into the membrane by using a spontaneous mechanism, given that it can insert into protein-free liposomes42. The protein is synthesized as a precursor, termed procoat, that has a signal peptide (Fig. 2a). Procoat is targeted to the membrane through electrostatic interactions between the positively charged residues at the termini of the protein and acidic phospholipids in the membrane bilayer. The insertion mechanism seems to be unique. The protein inserts into the membrane by a mechanism that involves both the hydrophobic leader peptide and the membrane-anchor domain43. Translocation of the acidic residues in the central loop is driven across the membrane by the proton motive force. After translocation, the leader peptide is removed by leader peptidase. What is the driving force for translocating the central acidic region across the membrane to the periplasm? The main driving force seems to be hydrophobic forces, although the proton motive force stimulates procoat insertion when negatively charged residues are present in the central region, but is not required when no charges are present44. The Pf3-virus coat protein also probably inserts into the bacterial membrane by a spontaneous mechanism. It is 44 residues long and contains one transmembrane segment, and its N-terminus is exposed to the periplasm. The protein inserts into the membrane under conditions in which the Sec machinery is impaired. Strikingly, the Pf3 coat protein moves its N-terminal domain across the membrane only in the presence of the proton motive force and when negatively charged residues are present45.
REVIEWS
TIBS 24 – JANUARY 1999 Current data are also consistent with a spontaneous mechanism for integration of several thylakoid proteins. Three single-span thylakoid membrane proteins – CFoII, and the W and X subunits of photosystem II (PsbX and PsbW) – all have cleavable signal-type peptides yet are inserted in vitro into thylakoids in the absence of stromal factors, nucleoside triphosphates or the thylakoid ∆pH46–48. Integration into thylakoids is unaffected even by high levels of proteases, which suggests that a protein component does not catalyze translocation – although this has yet to be proved unambiguously. Like the M13 procoat, these thylakoid protein precursors all contain short translocated domains that are followed by a membrane anchor, and PsbW insertion involves a loop intermediate that includes the signal peptide and transmembrane-spanning region (Fig. 2b)49. However, some differences are also apparent: these proteins require neither a DpH nor a transmembrane potential for insertion despite the presence of acidic translocated domains that strongly resemble that of procoat. In the case of the thylakoid proteins, we can trace the development of the signal peptide, thanks to the numerous genomesequencing projects carried out to date. Genes that encode CFoII and PsbX are found in cyanobacteria and the plastid genomes of several eukaryotic algae and, strikingly, these homologs do not have signal-type presequences. It is therefore likely that these peptides were acquired only after the transfer of these genes to the nucleus: these signal peptides therefore differ from true Sec-type signal peptides in both function and origin.
Other pathways? In eubacteria, translocation of Nterminal domains probably involves a novel Sec-independent mechanism: insertion occurs under conditions where the Sec machinery is impaired50. It will be interesting to see whether a Secindependent pathway is used to translocate N-terminal tail domains of chloroplast thylakoid proteins. Although N-terminal translocation has not been studied in plant chloroplast, it has been studied in the mitochondrial system. Like chloroplasts, mitochondria are believed to be derived from a bacterial ancestor. Intriguingly, the bacterial-type Sec machinery is absent in the yeast mitochondrial genome51. However, there is reason to believe that mitochondria, eubacteria and, possibly, chloroplasts share a common pathway. In mitochondria,
Oxa1 is required for insertion of innermembrane proteins from the matrix52,53. This is significant to the bacterial field because the sorting of a subset of innermembrane proteins occurs from the matrix by a process that is analogous to that found in eubacteria. These are the similarities: (1) certain N-terminal tails and loops of mitochondrial inner-membrane proteins require the transmembrane electrical gradient for insertion; (2) the hydrophobic transmembrane segments are required for membrane insertion; (3) the N-terminal tails (which are exposed to the intramitochondrial space) are enriched in acidic residues; (4) the introduction of positively charged residues blocks Nterminal translocation from the matrix; and (5) mitochondria do not require the Sec proteins for insertion (mitochondria do not contain SecA, SecY or SecE). The E. coli homolog of Oxa1 is YidC. In chloroplasts, the homolog is Albino3.
Conclusions and future perspectives Studies over the past few years have shown that translocation pathways are conserved in eubacteria and chloroplast thylakoids, and translocate proteins across the membrane in different ways. The Sec translocase promotes the export of proteins in unfolded states, which is reminiscent of protein export across the ER membrane and the mitochondrial and chloroplast envelopes. By contrast, translocation by the DpH pathway in thylakoids, and the Tat pathway in eubacteria, seems to occur in a folded state and, in some cases, translocation requires the binding of redox cofactors to the protein prior to the translocation step. Translocation of folded proteins also occurs in peroxisomes. These differences in export mechanisms necessitate translocation mechanisms that are fundamentally different and hence require multiple pathways. The SRP and the spontaneous pathways seem to play an important role in targeting and translocation of integral membrane proteins. However, the SecAand pH-dependent pathways play a major role in the transport of luminal proteins across the membrane. In eubacteria, the SRP pathway and the SecA pathway might both use the SecYEG translocase to promote insertion of polar regions. In the thylakoid system, this is still an open question. In the plant field, labs are racing to identify SecE and other bacterial homologs in thylakoids. How Ffh is targeted to the membrane is not known, because the Ffh receptor has not been identified in thylakoids.
This is a very exciting time in the bacterial and the plant thylakoid fields. With the cross fertilization between the two fields, progress will most certainly be more rapid in the future. In the next century, the general themes for protein export will be elucidated in far greater detail, and the problem of how proteins are translocated into or across membranes in an unfolded or folded state will be closer to being solved.
Acknowledgements R. E. D. would like to dedicate this manuscript to the memory of Tracy A. Schuenemann. We apologize to those workers who have not been mentioned because of space restraints.
References 1 von Heijne, G. (1983) Eur. J. Biochem. 133, 17–21 2 Robinson, C. and Mant, A. (1997) Trends Plant Sci. 2, 431–437 3 Schatz, P. J. and Beckwith, J. (1990) Annu. Rev. Genet. 24, 215–248 4 Ito, K. (1996) Genes Cells 1, 337–346 5 Driessen, A. J. M. et al. (1991) Methods Cell Biol. 34, 147–165 6 Mizushima, S., Tokuda, H. and Matsuyama, S-I. (1991) Methods Cell Biol. 34, 107–146 7 Wickner, W. and Leonard, M. R. (1996) J. Biol. Chem. 271, 29514–29516 8 Duong, F. and Wickner, W. (1997) J. Biol. Chem. 16, 2756–2768 9 Yuan, J., Henry, R., McCaffery, M. and Cline, K. (1994) Science 266, 796–798 10 Nakai, M. et al. (1994) J. Biol. Chem. 269, 31338–31341 11 Voelker, R., Mendel-Hartvig, J. and Barkan, A. (1997) Genetics 145, 467–478 12 Laidler, V. et al. (1995) J. Biol. Chem. 270, 17664–17667 13 Walter, P. and Blobel, G. (1981) J. Cell Biol. 91, 557–561 14 Bassford, P. et al. (1991) Cell 65, 367–368 15 Phillips, G. J. and Silhavy, T. J. (1992) Nature 359, 744–746 16 De Gier, J. W. L., Valent, Q. A., von Heijne, G. and Luirink, J. (1997) FEBS Lett. 408, 1–4 17 Valent, Q. A. et al. (1998) EMBO J. 17, 2504–2512 18 Li, X. et al. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 3789–3793 19 Hoffman, N. E. and Franklin, A. E. (1994) Plant Physiol. 105, 295–304 20 Schuenemann, D. et al. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 10312–10316 21 High, S. et al. (1997) J. Biol. Chem. 272, 11622–11628 22 Mould, R. M. and Robinson, C. (1991) J. Biol. Chem. 266, 12189–12193 23 Cline, K., Ettinger, W. F. and Theg, S. M. (1992) J. Biol. Chem. 267, 2688–2696 24 Klosgen, R. B., Brock, I. W., Herrmann, R. G. and Robinson, C. (1992) Plant Mol. Biol. 18, 1031–1034 25 Henry, R., Kapazoglou, A., McCaffery, M. and Cline, K. (1994) J. Biol. Chem. 269, 10189–10192 26 Brock, I. W., Mills, J. D., Robinson, D. and Robinson, C. (1995) J. Biol. Chem. 270, 1657–1662 27 Robinson, C. et al. (1994) EMBO J. 13, 279–285 28 Chaddock, A. et al. (1995) EMBO J. 14, 2715–2722
21
REVIEWS
TIBS 24 – JANUARY 1999
29 Settles, A. M. et al. (1997) Science 278, 1467–1470 30 Voelker, R. and Barkan, A. (1995) EMBO J. 14, 3905–3914 31 Berks, B. C. (1996) Mol. Microbiol. 22, 393–404 32 Sargent, F. et al. (1998) EMBO J. 17, 3640–3650 33 Weiner, J. H. et al. (1998) Cell 93, 93–101 34 Bogsch, E. G. et al. (1998) J. Biol. Chem. 273, 18003–18006 35 Randall, L. L. and Hardy, S. J. S. (1986) Cell 46, 921–928 36 Uchida, K., Mori, H. and Mizushima, S. (1995) J. Biol. Chem. 270, 30862–30868 37 Creighton, A. M. et al. (1995) J. Biol. Chem. 270, 1663–1669
38 Clark, S. A. and Theg, S. M. (1997) Mol. Biol. Cell. 8, 923–934 39 Santini, C-L. et al. (1998) EMBO J. 17, 101–112 40 Teter, S. A. and Theg, S. M. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 1590–1594 41 Wolfe, P. B., Rice, M. and Wickner, W. (1985) J. Biol. Chem. 260, 1836–1841 42 Geller, B. L. and Wickner, W. (1985) J. Biol. Chem. 260, 13281–13285 43 Kuhn, A., Kreil, G. and Wickner, W. (1986) EMBO J. 5, 3681–3685 44 Cao, G., Kuhn, A. and Dalbey, R. E. (1995) EMBO J. 14, 866–875 45 Kiefer, D., Hu, X., Dalbey, R. and Kuhn, A. (1997) EMBO J. 16, 2197–2204
Dealing with energy demand: the AMP-activated protein kinase Bruce E. Kemp, Ken I. Mitchelhill, David Stapleton, Belinda J. Michell, Zhi-Ping Chen and Lee A. Witters The AMP-activated protein kinase (AMPK) is a member of a metabolitesensing protein kinase family that is found in all eukaryotes. AMPK activity is regulated by vigorous exercise, nutrient starvation and ischemia/ hypoxia, and modulates many aspects of mammalian cell metabolism. The AMPK yeast homolog, Snf1p, plays a major role in adaption to glucose deprivation. In mammals, AMPK also has diverse roles that extend from energy metabolism through to transcriptional control. THE AMP-ACTIVATED protein kinase (AMPK) should not be confused with the better-known cAMP-dependent protein kinase (PKA) that transduces cAMP signals within cells. The discovery of AMPK was a consequence of studies by several laboratories. Berg and co-workers first detected a protein kinase activity associated with HMG-CoA reductase (reviewed in Ref. 1) that was later found to be activated by AMP (Ref. 2). Parallel studies by Kim and co-workers identified an acetylCoA-carboxylase kinase with related properties (reviewed in Ref. 1) but it was Hardie and co-workers3 who showed B. E. Kemp, K. I. Mitchelhill, D. Stapleton, B. J. Michell and Z-P. Chen are at the St Vincent’s Institute of Medical Research, 41 Victoria Parade, Fitzroy, Victoria 3065, Australia; L. A. Witters is in the EndocrineMetabolism Division, Depts of Medicine and Biochemistry, Dartmouth Medical School, Hanover, NH 03755, USA. Email: [email protected]
22
that the acetyl-CoA-carboxylase kinase and HMG-CoA-reductase kinase are one and the same. This, along with the observation that AMPK phosphorylates glycogen synthase and hormone-sensitive lipase, firmly established it as a multisubstrate enzyme. Because AMPK inhibits enzymes involved in glycogen, fatty acid and cholesterol synthesis, it was considered to be primarily a cellular fuel gauge that recognizes ATP depletion and limits further ATP utilization by anabolic pathways4. However, AMPK not only inhibits anabolic pathways but also initiates a series of compensatory changes that maintain cellular ATP levels. In rat skeletal and cardiac muscle, AMPK plays an important role in accelerating fatty acid oxidation and, possibly through increased glucose uptake and glycolysis, in meeting the energy demands of mechanical activity (Fig. 1). AMPK is a member of a larger metabolite-sensing protein-kinase family.
46 Michl, D. et al. (1994) EMBO J. 13, 1310–1317 47 Lorkovic, Z. J. et al. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 8930–8934 48 Kim, S. J., Robinson, C. and Mant, A. (1998) FEBS Lett. 424, 105–108 49 Thompson, S. J., Kim, S. J. and Robinson, C. (1998) J. Biol. Chem. 273, 18979–18983 50 Dalbey, R. E., Kuhn, A. and von Heijne, G. (1995) Trends Cell Biol. 5, 380–383 51 Glick, B. S. and von Heijne, G. (1996) Protein Sci. 5, 2651–2652 52 Bauer, M. et al. (1994) Mol. Gen. Genet. 245, 272–278 53 Hell, K. et al. (1997) FEBS Lett. 418, 367–370
It is an abg heterotrimer5. The yeast enzyme (the Snf1p protein kinase complex), which is essential for release from glucose repression, is also thought to be a trimer10. Stapleton and colleagues8 identified two isoforms of the AMPK catalytic a subunit, a1 and a2, which are encoded by different genes; a1 corresponds to the isoform purified from liver, for which there was the first unequivocal protein sequence5,8. The mammalian b subunits, b1 and b2, are homologous to yeast Sip1p, Sip2p and Gal83p, whereas the g subunits g1, g2 and g3 are homologous to Snf4p1,6,7. Snf1p phosphorylates yeast acetyl-CoA carboxylase; this indicates that there is a functional relationship between the two kinases5,9. For simplicity, we use AMPK a1 to refer to the heterotrimeric holoenzyme (a1, b1, g1); the catalytic subunit alone is referred to as the AMPK a1 subunit. For other isoforms, the heterotrimer is given in full – for example, we refer to the skeletal muscle isoform as AMPK (a2b2g1).
Structures of AMP-activated protein kinase and Snf1p-complex subunits Our understanding of the structure– function relationships of AMPK proteins has come from studies both of the mammalian enzyme and of yeast Snf1p. The kinase AMPK a contains two functional regions: an N-terminal catalytic core; and a C-terminal tail that is responsible for autoregulation and targeting to other subunits (Fig. 2). In yeast, two-hybrid analysis suggests that Snf4p binds Snf1p in part of the C-terminal regulatory domain (RD) and that the Sip1p/Sip2p/Gal83p family bind to a different part of the RD; Snf4p also binds to a conserved ASC sequence (association with SNF1 complex) at the C-terminus of members of the Sip1p/Sip2p/Gal83p family10. Within their catalytic cores, a1 and a2 share 90% identity; their catalytic cores also share 64% identity with that of Snf1p.
0968–0004/99/$ – See front matter © 1999, Elsevier Science. All rights reserved.
PII: S0968-0004(98)01340-1
TIBS 24 – JANUARY 1999
Protein translocation into and across the bacterial plasma membrane and the plant thylakoid membrane Ross E. Dalbey and Colin Robinson Over the past decade, some familiar themes have emerged on how proteins are inserted into or translocated across the plant chloroplast thylakoid membrane and bacterial inner membranes. In the SecA and signal recognition particle (SRP) pathways, nucleotides and soluble factors are used to translocate proteins across the membrane bilayer in the unfolded state. However, the DpH-dependent pathway in thylakoids uses a radically different mechanism: transport of proteins across the membrane is driven by the transmembrane pH gradient, and neither stromal factors nor nucleotide triphosphates are needed. In addition, this pathway, which requires the membrane-bound protein Hcf106, appears to translocate proteins in a tightly folded form. Recently, a similar pathway has been shown to operate in eubacteria, and several of its components have been identified. GRAM-NEGATIVE EUBACTERIA and plant chloroplasts have much in common. We have known for a long time that chloroplasts evolved from endosymbiotic, cyanobacteria-like cells, and the chloroplast still exhibits many prokaryotic features. The chloroplast contains its own genetic system, and the gene structure and protein-synthesizing machinery of chloroplasts are similar to those of prokaryotes in many respects. However, the chloroplast genome now encodes only a small proportion of the chloroplast proteins, the vast majority of genes having been transferred to the nucleus since the initial endosymbiotic event. Chloroplast biogenesis therefore requires the import of hundreds of proteins into the organelle. In structural terms, there are again clear similarities. Gram-negative eubacteria and chloroplasts are enclosed by two membranes and, in both cases, two aqueous spaces are formed: the periplasm and cytoplasm in eubacteria; and the intermembrane space and stroma in chloroplasts. However, the chloroplast contains in addition the thylakoid R. E. Dalbey is at the Dept of Chemistry, Ohio State University, Columbus, OH 43210, USA; and C. Robinson is at the Dept of Biological Sciences, University of Warwick, Coventry, UK CV4 7AL.
membrane, which in turn encloses a further aqueous phase, the thylakoid lumen. In terms of protein biogenesis, eubacteria and chloroplasts differ in many respects because protein traffic in eubacteria is directed outwards towards the periplasm/medium whereas proteins are imported into chloroplasts. In recent years, however, it has become clear that many proteins are targeted into or across the thylakoid membrane by mechanisms that strongly resemble well-studied processes in Escherichia coli. The learning process has not been unidirectional, and studies on thylakoid protein transport have led to the identification of a unique type of protein translocase that operates in a wide variety of eubacteria. Here, we focus on these similarities and attempt to summarize the salient features of the increasingly complex web of protein-sorting processes in these organisms.
The basic targeting pathways In eubacteria, export across the inner membrane requires an N-terminal signal peptide (also known as a leader peptide), which is removed by either signal peptidase I (also known as leader peptidase) or signal peptidase II (also known as lipoprotein signal peptidase). The leader peptide is characterized by a positively charged N-terminal region (n),
0968–0004/99/$ – See front matter © 1999, Elsevier Science. All rights reserved.
a hydrophobic central domain (h) and a polar C-terminal region (c). Residues at the 23 and 21 positions are necessary for leader-peptidase cleavage1 (Fig. 1a). Most proteins that are imported from the cytoplasm to the chloroplast thylakoid lumen are synthesized with a bipartite presequence2 (Fig. 1b). The N-terminal peptides, which are 40–70 residues long and are enriched in threonine and serine residues, contain the information necessary for import of the protein into the stroma. They are removed by the stromal processing peptidase (SPP). The second signal directs proteins across the thylakoid membrane, after which the signal is removed by a thylakoidal processing peptidase (TPP), which shares significant sequence similarity with the bacterial leader peptidase.
The translocation mechanisms: some familiar themes Studies over the past few years have demonstrated that multiple pathways translocate proteins into or across the bacterial plasma membrane and the plant thylakoid membrane. The known translocation components in eubacteria and chloroplasts are shown in Table 1. Of the four primary pathways identified in chloroplasts (Fig. 2), two appear to be similar, in most respects, to well-studied bacterial targeting mechanisms. Several labs elucidated the bacterial Sec pathway, exploiting genetic and biochemistry techniques. Genetic studies revealed that the cellular machinery includes SecA, SecB, SecD, SecE, SecF, SecG and SecY3,4. Translocation studies of reconstituted systems helped to reveal the biochemical functions of the Sec components5,6. Protein translocation across the inner membrane in E. coli (see Fig. 2a) requires typically, but not always, a molecular chaperone, such as SecB, and a translocase complex that comprises membrane-bound SecYEGDFyajC and the peripheral membrane protein SecA (Ref. 7). SecB binds to preproteins and retards their folding; it also participates in the delivery of the preprotein to the membrane by interacting with the membrane-bound SecA. SecA is required for translocation; it is an ATPase and probably functions as a translocation motor that pushes exported proteins across the membrane. The trimeric SecYEG complex is believed to function as a protein-conducting translocation channel. SecD, SecF and YajC also form a trimeric complex and can associate with SecYEG to form a hexameric complex8.
PII: S0968-0004(98)01333-4
17
REVIEWS
TIBS 24 – JANUARY 1999
(a) Bacterial signal peptides SPase-cleavage site R/K n region
h region
R/K
RR n region
c region
h region
c region
(b) Bipartite thylakoid-lumen-targeting presequences SPP-cleavage site
TPP-cleavage site h region
n region Stroma-targeting peptide
c region Thylakoid signal peptide
RR Stroma-targeting peptide
Thylakoid transfer peptide
Figure 1 Distinct types of targeting signals. (a) Bacterial signal peptides and twin-arginine signal peptides. The twin-arginine motifs target proteins to the Tat pathway. (b) Chloroplast thylakoidlumen bipartite presequences. The N-terminal portion of the chloroplast thylakoid-lumen presequence resembles a stroma-targeting sequence; the C-terminal part contains a hydrophobic sequence that targets the protein to the thylakoid lumen. Of the two thylakoid transfer sequences shown, the upper type targets proteins to the Sec-dependent pathway; the lower type targets proteins to the DpH pathway. The presequence is cleaved twice – first by the stromal processing peptidase (SPP) and then by the thylakoid processing peptidase (TPP; a membrane-bound protease whose catalytic site is exposed to the thylakoid lumen).
Table 1. Proteins implicated in translocation in bacteria and chloroplast Function
Bacteria
Chloroplast
Targeting
SecB Ffh
cpSRP54
Translocation
SecA SecY SecE SecG SecD SecF YajC TatE(Ybec) TatA(YigT) TatB
SecA SecY
Hcf106
TatC TatD ? Ftsy Signal-peptide cleavage
SPase I SPase II
TPP SPP
Abbreviations used: SPP, stromal processing peptidase; TPP, thylakoid processing peptidase.
18
SecD and SecF are membrane proteins, and possess large periplasmic domains that help to maintain the proton motive force across the membrane, to stabilize the membrane-inserted form of SecA and to promote the release of exported proteins into the periplasm. A Sec-dependent pathway also operates in chloroplasts (Fig. 2b); a SecA homolog is involved in thylakoid protein transport9–11, and a thylakoid SecY homolog has been cloned – although a direct role for this protein has not been established12. The other Sec proteins remain to be identified in plants, and it is still unclear whether stromal chaperones bind to proteins imported into the stroma from the cytosol and prevent them from aggregating and misfolding. Nevertheless, so far, the two types of Sec system appear to be very similar in most respects. Walter and Blobel13 elucidated the role of signal recognition particle (SRP) in export of proteins to the ER over 15 years ago. The mammalian SRP is
composed of six subunits and one RNA component. A hint that SRP might also be involved in targeting in the bacterial system was the discovery in eubacteria of ffh and ftsY genes, which are homologous to the genes that encode the 54-kDa SRP subunit and the SRP receptor, respectively, in eukaryotes, although homologs of the other gene products have not been found. The role of Ffh and FtsY in bacterial protein translocation was controversial initially, however, because genetic screens did not identify a role for these proteins in protein targeting14. Support for their role in vivo came first from Phillips and Silhavy15, who showed that depletion of Ffh led to defects in the export of certain proteins. There is now agreement in the bacterial field that Ffh and FtsY play a critical role in targeting and therefore translocation of a subset of membrane proteins16. The majority of periplasmic and outer-membrane-protein precursors, by contrast, can be exported even under conditions where Ffh is limiting within the cell. Recent studies have shown that bacterial SRP transfers some substrates to the SecYEG translocase17. In fact, several integral membrane proteins that insert in a Ffh-dependent manner may insert in a SecY-dependent manner. In 1994, Li and co-workers18 showed that chloroplasts also contain a homolog, cpSRP54, of the SRP 54-kD subunit (Fig. 2b). The cpSRP54 protein seems to play a similar functional role: its immunodepletion inhibited insertion of a multispan membrane protein [the lightharvesting chlorophyll-binding protein (LHCP) of photosystem II] into thylakoids18. Moreover, insertion of LHCP into thylakoid membranes is ten times more efficient in the presence of GTP than it is in the presence of other nucleotides19, and is inhibited by non-hydrolyzable GTP analogs. Interestingly, the cpSRP54 subunit associates with a novel 43-kDa component, cpSRP43 (Ref. 20); there is no evidence, so far, that any RNA molecule is present. To date, an FtsY homolog has not been discovered in chloroplast thylakoids. One major question concerns the determinant, within the exported or membrane protein, that enables interaction with cpSRP54 and targeting to the membrane. This determinant might in fact be the hydrophobicity of the protein chain. High et al.21 showed, using nascent-chain crosslinking, that cpSRP54 interacts with precursor proteins that have particularly hydrophobic signals. They suggested that the SRP-dependent targeting system is important for routing integral
REVIEWS
TIBS 24 – JANUARY 1999 membrane proteins to the thylakoid membrane. This is consistent with the observation that the SRP-dependent pathway seems to be critical only for membrane proteins.
(a)
C D F
A novel type of protein translocase in thylakoids and eubacteria In some respects, the field of thylakoid protein targeting has followed in the footsteps of bacterial studies in that similar pathways have emerged. A prime example is the Sec pathway used for a subset of thylakoid lumen proteins. However, in one critical respect, ideas have flowed in the other direction, following the finding that some luminal proteins are transported by a mechanism that could hardly be more different from the Sec-type mechanism. Substrates for this pathway, such as the 23-kDa and 16-kDa photosystem-II proteins (known as 23K and 16K, respectively) require neither stromal factors nor nucleoside triphosphates for transport across the thylakoid membrane but are instead totally reliant on the pH difference across the thylakoid membrane (DpH)22–26. Both Sec- and ∆pH-dependent substrates have signaltype targeting signals in which the three typical domains can be discerned: a charged N-terminal domain (N domain); a hydrophobic core region (H domain); and a more polar C domain that ends with a signal-peptidase-cleavage motif, usually Ala-X-Ala. Remarkably, however, these presequences are able to direct highly accurate, pathway-specific targeting, and there is little evidence of cross-talk between pathways25,27. This finding prompted a more rigorous examination of the targeting signals; Chaddock et al.28 showed that a crucial determinant in transfer peptides for the DpH pathway is the presence of an essential twinarginine motif immediately before the H domain. Other factors must, however, be involved: a twin-arginine motif alone is unable to divert Sec substrates onto this pathway. Recently, Settles and co-workers29 identified a plant protein, Hcf106, that is critical for the DpH pathway, following the isolation of a maize mutant in which this pathway is defective30. The Sec and SRP-dependent pathways are unaffected in this mutant, which suggests that Hcf106 is a specific and essential element of the elusive DpH-driven translocation system. The cloning of the hcf106 gene has provided major insights into both the origins and likely mechanism of this system. Homologs of hcf106 are present in a wide range of bacterial genomes as
?
Tat
Periplasm
Y EG FtsY
A
PMF
P SR P GT
PMF ATP
Cytoplasm Spontaneous
Ffh
B
(b)
Hcf106
?
Thylakoid lumen
Y A
SRP GTP
ATP
∆pH
Stroma Spontaneous
SRP 54
Figure 2 Protein export and membrane insertion pathways in eubacteria (a) and plant chloroplast thylakoids (b). Proteins are exported by different pathways. Some of these pathways require Sec proteins; others require Tat proteins. Classical signal peptides target proteins to the Sec pathway; twin-arginine signal peptides target proteins into the Tat pathway. The Tat pathway is DpH-dependent in chloroplasts, but the energetics remain to be determined in eubacteria. Insertion of integral membrane proteins occurs either by the signal recognition particle (SRP)-dependent pathway or by an apparently spontaneous pathway. At least one SRP substrate is inserted through SecYEG, although it is not known whether all SRP substrates follow this rule. The SRP-dependent pathway might also require an unidentified membrane component or might not require a protein-export apparatus. Blocks indicate hydrophobic regions; shaded blocks indicate hydrophobic regions that are signal peptides that are cleaved by signal peptidase (in Escherichia coli) or thylakoid processing peptidase (TPP; in chloroplasts). A, SecA; B, SecB; C, YajC; D–G, SecD–SecG; PMF, proton motive force; Y, SecY.
unassigned reading frames; this provides clear indications that a similar system might operate in prokaryotes. Berks31 has pointed out that periplasmic proteins that contain any of a range of redox cofactors invariably bear signal peptides that possess twin-arginine motifs. This has promoted speculation that the latter might be recognized by a system akin to the thylakoidal DpH system. Very recently, this scenario has been proved to be correct, and the first details of a Sec-independent bacterial export system have emerged. Two hcf106 homologs are found in the E. coli genome; one
of these genes (originally designated ybeC ) appears to be monocistronic, whereas the other (originally yigT ) appears to be part of an operon. The original yigT gene comprises two separate genes, which in this operon are followed by two further genes, yigU and yigW (see Fig. 3). Sargent et al.32 have probed the functions of these genes by using in-frame deletions and found that disruption of either yigT (the first gene shown in Fig. 3) or ybeC caused a marked inhibition of the export of at least five periplasmic proteins that contain twin-arginine presequences. In a double-deletion strain,
19
REVIEWS yigQRS
TIBS 24 – JANUARY 1999
(yigT ) tatA
(yigT ) tatB
(yigU ) tatC
(yigW ) tatD
rfaH
(ybeC) tatE
Figure 3 The Tat operon in Escherichia coli. The operon comprises four genes. The tatA gene is homologous to the maize hcf106 gene, and a second Hcf106 homolog is encoded by the unlinked tatE gene. Note that TatA, TatB, TatC and TatE have been shown to be involved in Sec-independent protein export but that the role of the TatD protein has yet to be established. Previous designations for reading frames are shown in parentheses. The yigQRS gene is the final ORF in a cluster of genes involved in ubiquinone biosynthesis; rfaH encodes a transcriptional regulator.
clarified experimentally but there seems little doubt at present that this system achieves the remarkable feat of transporting large globular proteins across coupled biological membranes. Teter and Theg40 have shown that the thylakoidal DpH is not compromised when the ∆pH-driven translocase is at full capacity; hence, the challenge for the future is to determine how an array of proteins can be squeezed through these membranes while protons are excluded so effectively.
The spontaneous pathway for membrane protein insertion: more variations on a theme export is totally defective, which implied that these gene products play a central, and perhaps overlapping, role. The observation that Sec-dependent export is unaffected in these strains confirms the existence of a separate export pathway. Members of this group of genes have been designated tat genes (for twin-arginine translocation), and the hcf106 homologs are tatA, tatB and tatE (see Fig. 3). A separate study has analyzed the function of this operon. Weiner et al.33 have identified an E. coli mutant in which this export process is defective, and the mutation lies in the yigT region; the authors concluded that an hcf106 homolog had been disrupted in this strain. However, it is now clear that this mutation lies instead in the second gene of this operon. This study thus effectively demonstrated a critical role for TatB. Finally, TatC (the product of the gene formerly known as yigU ) was identified as a fourth component of this export pathway, following the demonstration that disruption of the tatC gene leads to a complete block in the export of a range of cofactor-containing proteins that bear twin-arginine targeting signals34.
Transport of folded proteins by the Tat apparatus? One of the canonical rules of protein transport is that proteins must be transported in an unfolded state. For example, Randall and Hardy35 showed that tight folding of the preprotein of maltosebinding protein is incompatible with translocation across the membrane by the Sec pathway. Other data showed that slowing down the rate of protein synthesis, by addition of chloramphenicol, suppresses mutations that render the bacterial Sec machinery less efficient. This suggested that there is a kinetic competition between folding and export of proteins, which led to the view that Sec-dependent translocation requires the domain to be translocated in an
20
unfolded state. It does, however, seem that the Sec system is not as strict as was once believed; Mizushima and coworkers36 showed that a mutant of proOmpA that possesses an intramolecular disulfide bridge can be translocated in the presence of a proton motive force. The emerging evidence suggests that the DpH-dependent and Tat systems in thylakoids and eubacteria operate by a completely different principle. Studies on the thylakoidal system, using proteasesensitivity as an assay for the folding state of a luminal protein, showed that the extrinsic 23-kDa protein of photosystem II is in a tightly folded state in the stroma prior to translocation across the thylakoid membrane37. This observation and others38 raised the possibility that the system transports fully folded proteins, and such an idea is strongly supported by the studies on eubacteria. The known substrates for this export pathway bind a variety of complex redox cofactors, including FeS and molybdopterin centers, all of which are believed to be inserted in the cytoplasm31. This would inevitably necessitate folding of the mature protein. This idea is strongly supported by the finding that most of these proteins accumulate as active cytoplasmic forms in the E. coli mutants in which the tat genes are disrupted, which clearly indicates that cofactor insertion has taken place32–34. However, the story gets even more interesting because there is also evidence that export of some proteins can be blocked when cofactor insertion is prevented31,39; this observation raises the fascinating possibility that the system requires substrate proteins to be folded, at least in some cases, and that a proofreading mechanism actively prevents export until cofactor insertion has taken place. Such a proofreading mechanism would of course be logical (to avoid export of useless proteins) but unique among known protein-translocation systems. Many of these points need to be
The availability of bacterial strains carrying temperature-sensitive SecA and SecY mutants has allowed us to show that some proteins insert across the membrane independently of the Sec translocase. The M13-virus coat protein is such a protein41. It probably inserts into the membrane by using a spontaneous mechanism, given that it can insert into protein-free liposomes42. The protein is synthesized as a precursor, termed procoat, that has a signal peptide (Fig. 2a). Procoat is targeted to the membrane through electrostatic interactions between the positively charged residues at the termini of the protein and acidic phospholipids in the membrane bilayer. The insertion mechanism seems to be unique. The protein inserts into the membrane by a mechanism that involves both the hydrophobic leader peptide and the membrane-anchor domain43. Translocation of the acidic residues in the central loop is driven across the membrane by the proton motive force. After translocation, the leader peptide is removed by leader peptidase. What is the driving force for translocating the central acidic region across the membrane to the periplasm? The main driving force seems to be hydrophobic forces, although the proton motive force stimulates procoat insertion when negatively charged residues are present in the central region, but is not required when no charges are present44. The Pf3-virus coat protein also probably inserts into the bacterial membrane by a spontaneous mechanism. It is 44 residues long and contains one transmembrane segment, and its N-terminus is exposed to the periplasm. The protein inserts into the membrane under conditions in which the Sec machinery is impaired. Strikingly, the Pf3 coat protein moves its N-terminal domain across the membrane only in the presence of the proton motive force and when negatively charged residues are present45.
REVIEWS
TIBS 24 – JANUARY 1999 Current data are also consistent with a spontaneous mechanism for integration of several thylakoid proteins. Three single-span thylakoid membrane proteins – CFoII, and the W and X subunits of photosystem II (PsbX and PsbW) – all have cleavable signal-type peptides yet are inserted in vitro into thylakoids in the absence of stromal factors, nucleoside triphosphates or the thylakoid ∆pH46–48. Integration into thylakoids is unaffected even by high levels of proteases, which suggests that a protein component does not catalyze translocation – although this has yet to be proved unambiguously. Like the M13 procoat, these thylakoid protein precursors all contain short translocated domains that are followed by a membrane anchor, and PsbW insertion involves a loop intermediate that includes the signal peptide and transmembrane-spanning region (Fig. 2b)49. However, some differences are also apparent: these proteins require neither a DpH nor a transmembrane potential for insertion despite the presence of acidic translocated domains that strongly resemble that of procoat. In the case of the thylakoid proteins, we can trace the development of the signal peptide, thanks to the numerous genomesequencing projects carried out to date. Genes that encode CFoII and PsbX are found in cyanobacteria and the plastid genomes of several eukaryotic algae and, strikingly, these homologs do not have signal-type presequences. It is therefore likely that these peptides were acquired only after the transfer of these genes to the nucleus: these signal peptides therefore differ from true Sec-type signal peptides in both function and origin.
Other pathways? In eubacteria, translocation of Nterminal domains probably involves a novel Sec-independent mechanism: insertion occurs under conditions where the Sec machinery is impaired50. It will be interesting to see whether a Secindependent pathway is used to translocate N-terminal tail domains of chloroplast thylakoid proteins. Although N-terminal translocation has not been studied in plant chloroplast, it has been studied in the mitochondrial system. Like chloroplasts, mitochondria are believed to be derived from a bacterial ancestor. Intriguingly, the bacterial-type Sec machinery is absent in the yeast mitochondrial genome51. However, there is reason to believe that mitochondria, eubacteria and, possibly, chloroplasts share a common pathway. In mitochondria,
Oxa1 is required for insertion of innermembrane proteins from the matrix52,53. This is significant to the bacterial field because the sorting of a subset of innermembrane proteins occurs from the matrix by a process that is analogous to that found in eubacteria. These are the similarities: (1) certain N-terminal tails and loops of mitochondrial inner-membrane proteins require the transmembrane electrical gradient for insertion; (2) the hydrophobic transmembrane segments are required for membrane insertion; (3) the N-terminal tails (which are exposed to the intramitochondrial space) are enriched in acidic residues; (4) the introduction of positively charged residues blocks Nterminal translocation from the matrix; and (5) mitochondria do not require the Sec proteins for insertion (mitochondria do not contain SecA, SecY or SecE). The E. coli homolog of Oxa1 is YidC. In chloroplasts, the homolog is Albino3.
Conclusions and future perspectives Studies over the past few years have shown that translocation pathways are conserved in eubacteria and chloroplast thylakoids, and translocate proteins across the membrane in different ways. The Sec translocase promotes the export of proteins in unfolded states, which is reminiscent of protein export across the ER membrane and the mitochondrial and chloroplast envelopes. By contrast, translocation by the DpH pathway in thylakoids, and the Tat pathway in eubacteria, seems to occur in a folded state and, in some cases, translocation requires the binding of redox cofactors to the protein prior to the translocation step. Translocation of folded proteins also occurs in peroxisomes. These differences in export mechanisms necessitate translocation mechanisms that are fundamentally different and hence require multiple pathways. The SRP and the spontaneous pathways seem to play an important role in targeting and translocation of integral membrane proteins. However, the SecAand pH-dependent pathways play a major role in the transport of luminal proteins across the membrane. In eubacteria, the SRP pathway and the SecA pathway might both use the SecYEG translocase to promote insertion of polar regions. In the thylakoid system, this is still an open question. In the plant field, labs are racing to identify SecE and other bacterial homologs in thylakoids. How Ffh is targeted to the membrane is not known, because the Ffh receptor has not been identified in thylakoids.
This is a very exciting time in the bacterial and the plant thylakoid fields. With the cross fertilization between the two fields, progress will most certainly be more rapid in the future. In the next century, the general themes for protein export will be elucidated in far greater detail, and the problem of how proteins are translocated into or across membranes in an unfolded or folded state will be closer to being solved.
Acknowledgements R. E. D. would like to dedicate this manuscript to the memory of Tracy A. Schuenemann. We apologize to those workers who have not been mentioned because of space restraints.
References 1 von Heijne, G. (1983) Eur. J. Biochem. 133, 17–21 2 Robinson, C. and Mant, A. (1997) Trends Plant Sci. 2, 431–437 3 Schatz, P. J. and Beckwith, J. (1990) Annu. Rev. Genet. 24, 215–248 4 Ito, K. (1996) Genes Cells 1, 337–346 5 Driessen, A. J. M. et al. (1991) Methods Cell Biol. 34, 147–165 6 Mizushima, S., Tokuda, H. and Matsuyama, S-I. (1991) Methods Cell Biol. 34, 107–146 7 Wickner, W. and Leonard, M. R. (1996) J. Biol. Chem. 271, 29514–29516 8 Duong, F. and Wickner, W. (1997) J. Biol. Chem. 16, 2756–2768 9 Yuan, J., Henry, R., McCaffery, M. and Cline, K. (1994) Science 266, 796–798 10 Nakai, M. et al. (1994) J. Biol. Chem. 269, 31338–31341 11 Voelker, R., Mendel-Hartvig, J. and Barkan, A. (1997) Genetics 145, 467–478 12 Laidler, V. et al. (1995) J. Biol. Chem. 270, 17664–17667 13 Walter, P. and Blobel, G. (1981) J. Cell Biol. 91, 557–561 14 Bassford, P. et al. (1991) Cell 65, 367–368 15 Phillips, G. J. and Silhavy, T. J. (1992) Nature 359, 744–746 16 De Gier, J. W. L., Valent, Q. A., von Heijne, G. and Luirink, J. (1997) FEBS Lett. 408, 1–4 17 Valent, Q. A. et al. (1998) EMBO J. 17, 2504–2512 18 Li, X. et al. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 3789–3793 19 Hoffman, N. E. and Franklin, A. E. (1994) Plant Physiol. 105, 295–304 20 Schuenemann, D. et al. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 10312–10316 21 High, S. et al. (1997) J. Biol. Chem. 272, 11622–11628 22 Mould, R. M. and Robinson, C. (1991) J. Biol. Chem. 266, 12189–12193 23 Cline, K., Ettinger, W. F. and Theg, S. M. (1992) J. Biol. Chem. 267, 2688–2696 24 Klosgen, R. B., Brock, I. W., Herrmann, R. G. and Robinson, C. (1992) Plant Mol. Biol. 18, 1031–1034 25 Henry, R., Kapazoglou, A., McCaffery, M. and Cline, K. (1994) J. Biol. Chem. 269, 10189–10192 26 Brock, I. W., Mills, J. D., Robinson, D. and Robinson, C. (1995) J. Biol. Chem. 270, 1657–1662 27 Robinson, C. et al. (1994) EMBO J. 13, 279–285 28 Chaddock, A. et al. (1995) EMBO J. 14, 2715–2722
21
REVIEWS
TIBS 24 – JANUARY 1999
29 Settles, A. M. et al. (1997) Science 278, 1467–1470 30 Voelker, R. and Barkan, A. (1995) EMBO J. 14, 3905–3914 31 Berks, B. C. (1996) Mol. Microbiol. 22, 393–404 32 Sargent, F. et al. (1998) EMBO J. 17, 3640–3650 33 Weiner, J. H. et al. (1998) Cell 93, 93–101 34 Bogsch, E. G. et al. (1998) J. Biol. Chem. 273, 18003–18006 35 Randall, L. L. and Hardy, S. J. S. (1986) Cell 46, 921–928 36 Uchida, K., Mori, H. and Mizushima, S. (1995) J. Biol. Chem. 270, 30862–30868 37 Creighton, A. M. et al. (1995) J. Biol. Chem. 270, 1663–1669
38 Clark, S. A. and Theg, S. M. (1997) Mol. Biol. Cell. 8, 923–934 39 Santini, C-L. et al. (1998) EMBO J. 17, 101–112 40 Teter, S. A. and Theg, S. M. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 1590–1594 41 Wolfe, P. B., Rice, M. and Wickner, W. (1985) J. Biol. Chem. 260, 1836–1841 42 Geller, B. L. and Wickner, W. (1985) J. Biol. Chem. 260, 13281–13285 43 Kuhn, A., Kreil, G. and Wickner, W. (1986) EMBO J. 5, 3681–3685 44 Cao, G., Kuhn, A. and Dalbey, R. E. (1995) EMBO J. 14, 866–875 45 Kiefer, D., Hu, X., Dalbey, R. and Kuhn, A. (1997) EMBO J. 16, 2197–2204
Dealing with energy demand: the AMP-activated protein kinase Bruce E. Kemp, Ken I. Mitchelhill, David Stapleton, Belinda J. Michell, Zhi-Ping Chen and Lee A. Witters The AMP-activated protein kinase (AMPK) is a member of a metabolitesensing protein kinase family that is found in all eukaryotes. AMPK activity is regulated by vigorous exercise, nutrient starvation and ischemia/ hypoxia, and modulates many aspects of mammalian cell metabolism. The AMPK yeast homolog, Snf1p, plays a major role in adaption to glucose deprivation. In mammals, AMPK also has diverse roles that extend from energy metabolism through to transcriptional control. THE AMP-ACTIVATED protein kinase (AMPK) should not be confused with the better-known cAMP-dependent protein kinase (PKA) that transduces cAMP signals within cells. The discovery of AMPK was a consequence of studies by several laboratories. Berg and co-workers first detected a protein kinase activity associated with HMG-CoA reductase (reviewed in Ref. 1) that was later found to be activated by AMP (Ref. 2). Parallel studies by Kim and co-workers identified an acetylCoA-carboxylase kinase with related properties (reviewed in Ref. 1) but it was Hardie and co-workers3 who showed B. E. Kemp, K. I. Mitchelhill, D. Stapleton, B. J. Michell and Z-P. Chen are at the St Vincent’s Institute of Medical Research, 41 Victoria Parade, Fitzroy, Victoria 3065, Australia; L. A. Witters is in the EndocrineMetabolism Division, Depts of Medicine and Biochemistry, Dartmouth Medical School, Hanover, NH 03755, USA. Email: [email protected]
22
that the acetyl-CoA-carboxylase kinase and HMG-CoA-reductase kinase are one and the same. This, along with the observation that AMPK phosphorylates glycogen synthase and hormone-sensitive lipase, firmly established it as a multisubstrate enzyme. Because AMPK inhibits enzymes involved in glycogen, fatty acid and cholesterol synthesis, it was considered to be primarily a cellular fuel gauge that recognizes ATP depletion and limits further ATP utilization by anabolic pathways4. However, AMPK not only inhibits anabolic pathways but also initiates a series of compensatory changes that maintain cellular ATP levels. In rat skeletal and cardiac muscle, AMPK plays an important role in accelerating fatty acid oxidation and, possibly through increased glucose uptake and glycolysis, in meeting the energy demands of mechanical activity (Fig. 1). AMPK is a member of a larger metabolite-sensing protein-kinase family.
46 Michl, D. et al. (1994) EMBO J. 13, 1310–1317 47 Lorkovic, Z. J. et al. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 8930–8934 48 Kim, S. J., Robinson, C. and Mant, A. (1998) FEBS Lett. 424, 105–108 49 Thompson, S. J., Kim, S. J. and Robinson, C. (1998) J. Biol. Chem. 273, 18979–18983 50 Dalbey, R. E., Kuhn, A. and von Heijne, G. (1995) Trends Cell Biol. 5, 380–383 51 Glick, B. S. and von Heijne, G. (1996) Protein Sci. 5, 2651–2652 52 Bauer, M. et al. (1994) Mol. Gen. Genet. 245, 272–278 53 Hell, K. et al. (1997) FEBS Lett. 418, 367–370
It is an abg heterotrimer5. The yeast enzyme (the Snf1p protein kinase complex), which is essential for release from glucose repression, is also thought to be a trimer10. Stapleton and colleagues8 identified two isoforms of the AMPK catalytic a subunit, a1 and a2, which are encoded by different genes; a1 corresponds to the isoform purified from liver, for which there was the first unequivocal protein sequence5,8. The mammalian b subunits, b1 and b2, are homologous to yeast Sip1p, Sip2p and Gal83p, whereas the g subunits g1, g2 and g3 are homologous to Snf4p1,6,7. Snf1p phosphorylates yeast acetyl-CoA carboxylase; this indicates that there is a functional relationship between the two kinases5,9. For simplicity, we use AMPK a1 to refer to the heterotrimeric holoenzyme (a1, b1, g1); the catalytic subunit alone is referred to as the AMPK a1 subunit. For other isoforms, the heterotrimer is given in full – for example, we refer to the skeletal muscle isoform as AMPK (a2b2g1).
Structures of AMP-activated protein kinase and Snf1p-complex subunits Our understanding of the structure– function relationships of AMPK proteins has come from studies both of the mammalian enzyme and of yeast Snf1p. The kinase AMPK a contains two functional regions: an N-terminal catalytic core; and a C-terminal tail that is responsible for autoregulation and targeting to other subunits (Fig. 2). In yeast, two-hybrid analysis suggests that Snf4p binds Snf1p in part of the C-terminal regulatory domain (RD) and that the Sip1p/Sip2p/Gal83p family bind to a different part of the RD; Snf4p also binds to a conserved ASC sequence (association with SNF1 complex) at the C-terminus of members of the Sip1p/Sip2p/Gal83p family10. Within their catalytic cores, a1 and a2 share 90% identity; their catalytic cores also share 64% identity with that of Snf1p.
0968–0004/99/$ – See front matter © 1999, Elsevier Science. All rights reserved.
PII: S0968-0004(98)01340-1