The C-terminal Domain of Escherichia coli Trigger Factor Represents the Central Module of Its Chaperone Activity *□S
Received for publication, May 30, 2006, and in revised form, August 7, 2006 Published, JBC Papers in Press, August 22, 2006, DOI 10.1074/jbc.M605164200
Frieder Merz, Anja Hoffmann, Anna Rutkowska, Beate Zachmann-Brand, Bernd Bukau, and Elke Deuerling1 From the Zentrum fu¨r Molekulare Biologie der Universita¨t Heidelberg Im Neuenheimer Feld 282, Universita¨t Heidelberg, 69120 Heidelberg, Germany
In bacteria, ribosome-bound Trigger Factor assists the fold- ing of newly synthesized proteins. The N-terminal domain (N) of Trigger Factor mediates ribosome binding, whereas the mid- dle domain (P) harbors peptidyl-prolyl isomerase activity. The function of the C-terminal domain (C) has remained enigmatic due to structural instability in isolation. Here, we have charac- terized a stabilized version of the C domain (CS), designed on the basis of the recently solved atomic structure of Trigger Factor.
Strikingly, only the isolated CSdomain or domain combinations thereof (NCS, PCS) revealed substantial chaperone activityin vitroandin vivo. Furthermore, to disrupt the C domain without affecting the overall Trigger Factor structure, we generated a mutant (⌬53) by deletion of the C-terminal 53 amino acid resi- dues. This truncation caused the complete loss of the chaperone activity of Trigger Factorin vitroand severely impaired its func- tionin vivo. Therefore, we conclude that the chaperone activity of Trigger Factor critically depends on its C-terminal domain as the central structural chaperone module. Intriguingly, a struc- turally similar module is found in the periplasmic chaperone SurA and in MPN555, a protein of unknown function. We spec- ulate that this conserved module can exist solely or in combina- tion with additional domains to fulfill diverse chaperone func- tions in the cell.
A complex network of molecular chaperones controls the folding of newly synthesized proteins in the cytosol. In bacteria, the Trigger Factor (TF),2because of its location on the ribo- some is the first chaperone to assist the folding of newly syn- thesized proteins. This first encounter is succeeded by the cyto- solic Hsp70 and Hsp60 chaperone systems consisting of DnaK, DnaJ, GrpE and GroEL, GroES, respectively (1–3). While the
GroEL system is essential for cell viability at all conditions, cells tolerate the absence of TF or DnaK at growth temperatures between 20 and 37 °C. However, the simultaneous deletion of the TF-encoding gene tig and the dnaK gene provokes the aggregation of several hundred cytosolic protein species and the loss of cell viability at temperatures above 30 °C (4 – 6).
TF is a very abundant protein (50Min the cell cytosol) (5) and is present in a 2- to 3-fold excess over ribosomes. It associ- ates in a 1:1 stoichiometry with ribosomes (7, 8), whereas uncomplexed TF is in an equilibrium between a monomeric and dimeric state (9).In vivo, ribosome-associated TF assists the folding of newly synthesized proteins by binding to nascent polypeptides when they emerge from the ribosomal exit tunnel (4, 5, 8, 10 –12).In vitro, TF chaperone activity can be moni- tored by its ability to prevent the aggregation and to promote the refolding of chemically denatured GAPDH (glyceralde- hyde-3-phosphate dehydrogenase) (13, 14). Moreover, TF effi- ciently stimulates the refolding of denatured RNase T1in vitro (15), where a slowtranstocisisomerization of prolyl residues is the rate-limiting step (16). The high catalytic efficiency of TF in RNase T1 refolding results from two combined activities: the binding of TF as a chaperone and its catalytic activity as a pep- tidyl-prolylcis/transisomerase (PPIase) (17).
Escherichia coliTF consists of 432 amino acids and was bio- chemically characterized by limited proteolysis to be a three- domain protein (18 –21). The N-terminal fragment (aa 1–144) contains a stable domain (N, aa 1–118) that is essential and sufficient for the ribosomal attachment of TF (18). The second domain (P, aa 145–247) displays PPIase activity and shows homology to the PPIase family of FK506-binding proteins (15, 20). The largest domain, which shows no sequence homology to any other known protein family, is formed by the C-terminal segment of TF (C, aa 248 – 432). No precise function had been assigned to this isolated domain so far, in contrast to the N and P domains. Previous in vitro studies showed that only full- length TF, but none of the isolated domains, displayed chaper- one activity toward denatured GAPDH and unfolded RNase T1 (18, 21–23).
The recent crystal structures of E. coli TF revealed that this chaperone has an unusual extended shape (11, 24) and its three domains structurally do not align in a linear manner (Fig. 1). The C-terminal domain of TF builds the center of the molecule with two protruding extensions forming the
“arms,” whereas the N-terminal and the catalytic PPIase domains localize to opposite distal ends of the protein.
Importantly, the structure implies that the C-terminal domain of TF is stabilized by a long N-terminal linker region
*This work was supported by DFG Grant SFB638 (to B. B. and E. D.), by the Human Frontiers in Science Program (to E. D.), a Heisenberg Fellowship of the DFG (to E. D.), a Fonds der Chemischen Industrie grant (to B. B.), and by a Boehringer Ingelheim Fonds scholarship (to A. H.). The costs of publica- tion of this article were defrayed in part by the payment of page charges.
This article must therefore be hereby marked “advertisement” in accord- ance with 18 U.S.C. Section 1734 solely to indicate this fact.
□S The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1 and S2.
1To whom correspondence should be addressed. Tel.: 49-6221-546870; Fax:
49-6221-545894; E-mail: e.deuerling@zmbh.uni-heidelberg.de.
2The abbreviations used are: TF, Trigger Factor; PPIase, peptidyl-prolylcis/
transisomerase; N, N-terminal domain of Trigger Factor; P, PPIase domain of Trigger Factor; C, C-terminal domain of Trigger Factor; Cs, stabilized C-terminal domain of Trigger Factor; GAPDH, glyceraldehyde-3-phos- phate dehydrogenase; ICDH, isocitrate dehydrogenase; IPTG, isopropyl 1-thio--D-galactopyranoside; LB, Luria Bertani; aa, amino acid; BSA, bovine serum albumin.
OCTOBER 20, 2006 •VOLUME 281 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 31963
at Universitaet Heidelberg on November 23, 2006 www.jbc.orgDownloaded from
http://www.jbc.org/cgi/content/full/M605164200/DC1 Supplemental Material can be found at:
Konstanzer Online-Publikations-System (KOPS)
(aa 112–144), providing an explanation for the instability of the isolated C-terminal domain (aa 248 – 432) lacking this linker region (18, 21). Furthermore, the structural insights suggest that the central C-terminal domain may strongly contribute to TF chaperone activity. However, recent analy- ses of TF truncation mutants provide contradicting results:
whereas deletions in the C-domain affected the refolding of GAPDHin vitro, this domain was suggested to be dispensa- ble for TF chaperone activityin vivo(25).
To unravel the function of the C-terminal domain and its potential contribution to TF chaperone activity, we designed several new constructs. These comprised either the stabi- lized C domain, including the N-terminal linker (aa 112–
144) or domain combinations thereof. In addition, we trun- cated the full-length TF by deletion of the 53 C-terminal amino acids. Thein vitroandin vivochaperone activities of these TF variants were characterized in detail by analyzing their capacity to (i) prevent the aggregation and promote the refolding of denatured GAPDH, (ii) refold denatured RNase T1, (iii) form dimers in solution, (iv) cross-link to nascent polypeptides, and (v) complement the synthetic lethality of cells lacking TF and DnaK and prevent the aggregation of cytosolic proteins in these cells.
EXPERIMENTAL PROCEDURES
Strains and Growth Conditions—E. colistrains were deriva- tives of MC4100. Construction of⌬tig⌬dnaKS1 and⌬dnaK strains were described previously (14, 26). Strains were grown in Luria Bertani (LB) medium containing IPTG as indicated and supplemented with ampicillin (100g/ml), spectinomycin (20
g/ml), or kanamycin (40g/ml) where appropriate.
Mutant Construction, Expression, and Protein Purification—
TF variants were generated by PCR. The oligonucleotides 5⬘-ggccggatccatgtatccggaagttgaactgcaggg-3⬘and 5⬘-ccgggg- atcccgcctgctggttcatcagc-3⬘were used to amplify the PCSand CS fragments. The plasmid pDS56-tig-His6(18) served as template for the PCSfragment, whereas pDS56-NC (23) was used as template for construction of the CS variant. The C-terminal truncation of 53 amino acids was created using the primers 5⬘-ggccggatccatgcaagtttcagttgaaacc-3⬘ and 5⬘- ggccggatccttcgtacgcagaagccatctcttcg-3⬘, whereby pDS56- tig-His6 served as template. All PCR products encoded BamH1 sites on the 5⬘ and 3⬘ ends. These were used for insertion into the vector pDS56 to generate the plasmids pDS56-PCS, pDS56-CS, and pDS56-⌬53. The DNA sequence integrity of all mutants was confirmed by sequencing. TF con- structs were expressed inE. coliMC4100⌬tig(4) as recombi- nant proteins with a C-terminal His6tag. Cells were grown in LB medium, and expression was induced with 500MIPTG (isopropyl--D-thiogalactoside). TF and variants were purified as described (23). The protein concentration was determined via Bradford assay (Bio-Rad), while the purity was assessed by both SDS-PAGE and mass spectrometry.
GAPDH Activity Assay—The prevention of aggregation and the refolding of denatured GAPDH were performed as described previously (13, 14). In all measurements we used gua- nidine hydrochloride denatured GAPDH from rabbit muscle (G-2267; Sigma) at a final concentration of 2.5M. The final
concentrations of TF, TF variants, or BSA varied from 1.25 to 20M. Data were fitted to a single-exponential equation (GraFit program). Relative activities of refolded GAPDH were calcu- lated by dividing the respective rate constants by that deter- mined for the same amount of non-denatured GAPDH.
RNase T1 Refolding—The PPIase-dependent refolding of denatured RCM-RNase T1 was performed as previously reported (14, 27). Briefly, RNase T1 fromAspergillus oryzae (R-1003; Sigma) was reduced with dithiothreitol and applied to iodoacetic acid treatment for carboxymethylation. Upon dilu- tion into 1.6M NaCl, RCM-RNase T1 refolding (1M final concentration) was followed by measuring the intrinsic trypto- phan fluorescence at 320 nm in a spectrofluorometer (LS55;
PerkinElmer Life Sciences) in the presence of TF, TF variants, or BSA (0.2– 4M). Refolding rates were obtained by fitting the data to a single-exponential equation using the program GraFit.
Rates of TF variants are expressed relative to wild-type TF.
Glutaraldehyde Cross-linking—TF and TF variants (5 g each) were incubated at different concentrations (0.156, 0.625, 2.5, 10, and 20Mfor TF, PCS, NCS,⌬53; and 0.625, 2.5, 10, 20, and 40Mfor N, P, CS) in 20 mMHepes, pH 7.5, 100 mMNaCl, and 1 mMEDTA for 25 min at 25 °C. Cross-linking was initiated by addition of 0.1% glutaraldehyde (Sigma), and the reaction was quenched after 10 min at 25 °C with 100 mMTris-HCl, pH 7.5, for 15 min at 25 °C. Samples were trichloroacetic acid pre- cipitated (5% trichloroacetic acid, 0.02% NaDOC, 1 h, 4 °C);
proteins were pelleted by centrifugation (30 min, 16,000⫻g, 4 °C) and pellets resuspended in alkaline sample buffer. SDS- PAGE (12% gel) with Coomassie Brilliant Blue staining was sub- sequently performed.
Cross-linking to Nascent Polypeptides—An E. coli-basedin vitrotranscription/translation system derived from MC4100⌬tig strain was used (4, 6, 28). To arrest nascent isocitrate dehy- drogenase (ICDH) chains at the ribosomes (400 nM) in the presence of TF variants (5M), truncated mRNA was gener- ated by adding an antisense oligonucleotide (40 ng/l; 5⬘- cccccatctcttcacgcagg-3⬘) and RNase H (0.04 units/l) to the transcription/translation system. After 20 min of synthesis at 37 °C, translation was stopped by adding 2 mMchloram- phenicol. Chemical cross-linking was achieved by addition of 2.5 mMdisuccinimidyl suberate and incubation at 25 °C for 30 min. The reaction was quenched with 50 mMTris- HCl, pH 7.5, for 15 min at 25 °C. Ribosome-nascent chain-TF complexes were isolated by sucrose cushion ultra- centrifugation and subsequently co-immunoprecipitated using a TF-specific antiserum. Isolated [35S]methionine-la- beled polypeptides and cross-link products were separated on SDS-PAGE and visualized by autoradiography.
In Vivo Complementation Analysis and Preparation of Aggregates—E. coli⌬tig⌬dnaKS1 strain was transformed with pDS56 plasmids expressing TF or TF variants under the control of an IPTG-regulatable promoter (18, 26). Expression of TF variants in the absence of IPTG was repressed by LacI encoded on the plasmid pZA4. Cells were grown overnight at 30 °C in the absence of IPTG, diluted to concentrations corresponding to 105, 104, 103, 102, or 10 cells/5l, spotted on LB plates con- taining 0, 10, 20, 50, 100, or 250MIPTG, and incubated for 30 h at 30, 34, or 37 °C. Preparation of aggregates was per-
at Universitaet Heidelberg on November 23, 2006 www.jbc.orgDownloaded from
formed according to published procedures (6, 23) from cultures that were grown in LB containing 20MIPTG at 30 °C and harvested at anA600⫽1. Experiments were reproduced at least three times.
RESULTS
Design of Trigger Factor Variants—To investigate the contri- bution of the C-terminal domain to the chaperone activity of TF, we designed several variants (Fig. 1) based on recent crystal structures. The structures suggested that a long N-terminal linker region is important for the three-dimensional organiza- tion of the C-terminal domain (11, 24). By fusing this putative stabilizing linker region (aa 112–144) N-terminal to the C ter- minus (aa 248 – 432) of TF, we generated a stabilized C-termi- nal domain (CS). As controls, we included in our analyses the isolated N-terminal domain (N, aa 1–118) and the PPIase domain (P, aa 145–247), which were characterized earlier (18, 23).
Combinations of the CSdomain with other TF domains were also investigated. We designed a PCStwo-domain fragment comprising amino acids 112– 432 and in addition made use of the NC fragment. This was previously generated by fusing the N-terminal 144 amino acids, including the stabilizing linker region, to the C terminus (aa 247– 432) (23). The NC fragment
is designated hereafter NCSto indi- cate the presence of the linker region (Fig. 1).
We also constructed a C-terminal truncated TF mutant by deleting the 53 C-terminal amino acids (⌬53).
The truncation site was designed based on the crystal structure of TF fromVibrio cholerae(24); this ver- sion lacks part of the second arm, which results in local rearrange- ments within the C-terminal domain but has no structural effect on the N or P domain (Fig. 1). All purified TF variants were soluble and thus amenable to functional analysis (data not shown).
The CSDomain Is Essential for the in Vitro Chaperone Activity of TF on Denatured GAPDH—To character- ize thein vitrochaperone activity of the various TF fragments, we tested their capacity to prevent aggregation of chemically denatured GAPDH (in 3MGndHCl) (13, 14, 23). Denatured GAPDH rapidly aggregated upon a 50-fold dilution in non-denaturing reaction buffer as indicated by the increase in light scattering signal at 620 nm. However, the presence of stoichiometric amounts of wild- type TF efficiently prevented GAPDH aggregation as no increase of signal was observed (Fig. 2A).
When TF variants were added in a 1:1 ratio, only the PCSfrag- ment inhibited the aggregation of denatured GAPDH by⬃50%
(Fig. 2A). However, increasing the concentrations of the TF variants revealed that PCSfully suppressed aggregation at a 4-fold (data not shown) or 8-fold excess (20M, Fig. 2B) over GAPDH. In addition, at an 8-fold excess NCSreduced aggre- gation by⬃60%. Remarkably, even the isolated CSdomain displayed chaperone activity by decreasing GAPDH aggrega- tion by ⬃15% (Fig. 2B). Neither the individual N nor P domain showed any chaperone activity, whereby the isolated N domain aggregated at high concentrations leading to an increase in the light scattering signal (Fig. 2B). Moreover,
⌬53 revealed no activity in the prevention of GAPDH aggre- gation at 20M(Fig. 2B) or 40M(data not shown). Taken together, these data suggest that an intact C-domain is cru- cial for the ability of TF to prevent aggregation.
We next investigated whether the TF fragments are able to promote the refolding of GAPDH into its active state. Dena- tured GAPDH was diluted 50-fold in the presence or absence of TF or TF fragments at different concentrations, and the restored GAPDH activity was determined after 4 h (Fig. 2C).
The presence of TF (2.5–5M) restored GAPDH activity up to
⬃55%, whereas a BSA control had no effect. Higher TF concen- trations inhibited the refolding of GAPDH as previously FIGURE 1.Structure of TF and TF variants.E. coliTrigger Factor (TF) and TF fragments N, P, CS, NCS, and PCS
were modeled according to the structure coordinates of Ref. 11 using the program PyMol and are depicted in ribbon representation. In addition, TF is shown in a space-filling illustration at theleft. The C-terminal truncated
⌬53 mutant was modeled according to the TF structure coordinates from Ref. 24. The N-terminal domain is shown inred, the PPIase is colored ingreen, and the C-terminal domain is displayed in different shades ofblue to emphasize the “body” and the two protruding “arms.”
at Universitaet Heidelberg on November 23, 2006 www.jbc.orgDownloaded from
reported (13, 23). The addition of increasing concentrations of PCS, NCS, and CS led to increasing total yields of refolded GAPDH up to⬃55, 40, and 25%, respectively. However, in con- trast to wild-type TF no substantial inhibition of GAPDH refolding was observed (Fig. 2C). Complementing the results above, N, P, and⌬53 did not demonstrate any detectable refold- ing activity at all concentrations tested.
In summary, all fragments containing the CSdomain (NCS, PCS, and CS) showed significant residual chaperone activity although they differed from wild-type TF (i) with a reduced efficiency, and (ii) in their broad effective concentration range for GAPDH refolding. This suggests differences in the interac- tion of TF, NCS, PCS, and CSwith unfolded protein substrates.
Furthermore,⌬53 lost the ability to prevent the aggregation and to promote the refolding of denatured GAPDHin vitro. We conclude that the CSdomain is essential but not sufficient for the full chaperone activity of TF toward denatured GAPDH in vitro.
The CSDomain Is Crucial for Efficient RNase T1 Refolding by TF—A second characteristicin vitroactivity of TF is its ability to catalyze the refolding of RNase T1. This process is rate-lim- ited by the slowtranstocis isomerization of peptidyl-prolyl bonds at Pro-39 and Pro-55 (16) and can be followed by meas- uring the intrinsic tryptophan fluorescence of RNase T1. Wild- type TF in substoichiometric amounts (0.2M) efficiently cat- alyzed the refolding of RNase T1 (1 M) due to the prolyl isomerase activity of the P domain combined with the capacity to bind substrate as a chaperone (Fig. 3Aand Refs. 14, 17, 21).
Remarkably, PCSdisplayed an extremely high catalytic activity with refolding rates similar to full-length TF. In contrast, none of the other TF variants exhibited an efficient catalytic activity even when concentrations of 4Mwere tested (Fig. 3,AandB).
The⌬53 mutant promoted the refolding of RNase T1 similar to the isolated P domain, suggesting that the requisite comple- mentary chaperone function lies in the C-terminal domain (Fig.
3B). Taken together, these results suggest that the CSdomain is essential and sufficient for the binding of the unfolded RNase T1, subsequently enabling the efficient isomerization of the peptidyl-prolyl bonds via the P domain.
Dimer Formation Is Not a Prerequisite for the in Vitro Chap- erone Function of TF Variants—Although ribosome-bound TF exerts its function in a monomeric state, previous studies sug- gest that thein vitrochaperone function of TF in the absence of ribosomes correlates with dimerization (25, 29, 30). To deter- mine whether the chaperone activity of TF variants is depend- ent on dimer formation, we examined the quaternary struc- tures of the purified TF variants by glutaraldehyde cross-linking (Fig. 4). As previously demonstrated, the appearance of glutar- aldehyde cross-linking products reflects the dimerization of TF (9). In agreement with this, wild-type TF showed concentration- dependent dimer formation, with an aberrant SDS-PAGE migration of ⬃150 kDa. In addition, cross-linking products
⬎170 kDa were observed and most likely represent higher oli- gomeric or aggregated species of TF generated by glutaralde- hyde treatment (9). Although isolated N, P, and CSdomains as well as PCSand⌬53 revealed no or little dimer formation, the two-domain NCSfragment demonstrated considerable dimer- ization in a concentration-dependent manner (Fig. 4). This FIGURE 2.In vitrochaperone activity of TF variants.AandB, aggregation
of chemically denatured GAPDH (2.5M) was monitored as an increase in light-scattering signal at 620 nm. The prevention of GAPDH aggregation was tested in the presence of 2.5M(equimolar concentrations) (A) and 20
M(B) TF or TF variants.C, refolding of denatured GAPDH was monitored by measuring its enzymatic activity 4 h after 50-fold dilution (final concen- tration 2.5M) in the absence or the presence of different concentrations (1.25, 2.5, 5, 10, and 20M) of TF or TF fragments. TF,filled circles; N,open triangles up; P,open triangles down; CS,open circles; NCS,filled triangles up;
PCS,filled triangles down;⌬53,open diamonds; BSA control,crosses.
at Universitaet Heidelberg on November 23, 2006 www.jbc.orgDownloaded from
indicates that the N and CSdomains together contribute signif- icantly to TF dimer assembly.
Importantly, the ability of TF fragments to form dimers did not correlate to their chaperone activity. For example, the PCS fragment was severely impaired in dimer formation and yet showed near wild-type-like activities in GAPDH and RNase T1 refolding (Figs. 2 and 3). Hence, dimer formation is not a pre- requisite forin vitrochaperone activity of the TF variants, sug- gesting that the monomeric conformation represents an active state.
The CSDomain Shows No Efficient Cross-linking to Nascent Polypeptide Chains—We next sought to investigate the ability of the TF fragments to interact with nascent polypeptide chains. We generated arrested [35S]Met-labeled nascent chains derived from the natural TF substrate ICDH (Fig. 5) in anin vitrotranscription/translation system (6) prepared fromE. coli cells lacking TF. This enabled the controlled addition of TF
variants prior to translation (at 5 M final concentration to saturate ribosomes) and the subsequent cross-linking of TF to nascent polypeptides via the bifunctional chemical cross-linker disuccinimidyl suberate. In the presence of wild-type TF, two major cross-linking products of⬃70 and 90 kDa were obtained (Fig. 5,lane 2) as previously reported (22). In addition, all frag- ments containing the N-terminal ribosome-binding domain (N, NCS, ⌬53) efficiently cross-linked to nascent ICDH polypeptides (Fig. 5,lanes 5,14,20). As anticipated, no cross- linking to nascent polypeptides was observed for the P domain (6, 23), whereas the PCSand CSfragments showed faint cross- linking products; these are likely weak interactions between these non-ribosomal TF fragments and nascent ICDH medi- FIGURE 3.RNase T1 refolding by TF variants.Refolding of denatured RNase
T1 (1Mfinal concentration) was monitored by an increase of intrinsic tryp- tophan fluorescence at 320 nm after excitation at 268 nm.A, RNase T1 refold- ing in the absence or presence of 0.2MTF or TF variants.B, relative refolding rates of denatured RNase T1 obtained by titration of TF and TF fragments (0, 0.2, 0.4, 1, and 4M). TF,filled circles; N,open triangles up; P,open triangles down; CS,open circles; NCS,filled triangles up; PC,filled triangles down;⌬53, open diamonds; BSA control,crosses.
FIGURE 4.Dimerization of TF variants.Increasing concentrations (indicated bytriangular gradient symbols) of purified TF or TF variants (0.156, 0.625, 2.5, 10, and 20Mfor TF, PCS, NCS,⌬53 and 0.625, 2.5, 10, 20, and 40Mfor N, P, CS) were cross-linked with 0.1% glutaraldehyde, trichloroacetic acid, precipi- tated, and separated by 12% SDS-PAGE. Equal amounts of protein were loaded on each lane (5g). The first lane of each TF variant was not treated with glutaraldehyde. Cross-linked dimers are marked bystars. Higher molec- ular mass cross-links represent higher oligomeric or aggregated species (see
“Results”).
FIGURE 5.Interaction of TF and TF fragments with nascent polypeptides.
Ribosome-arrested35S-labeled nascent polypeptides of ICDH (aa 1–177) were generated in anin vitrotranscription/translation system supplemented with TF or TF fragments. After synthesis, an aliquot was taken (S) and the residual was subjected to chemical cross-linking by the addition of disuccin- imidyl suberate. After disuccinimidyl suberate treatment ribosome-nascent chain complexes were purified by sucrose cushion centrifugation (X) and sub- sequently co-immunoprecipitated where indicated (IP) to identify cross-link products of TF and TF variants to ICDH (177 aa, indicated bystars). Additional weak cross-linked bands that appear in the background are of unknown identity.
at Universitaet Heidelberg on November 23, 2006 www.jbc.orgDownloaded from
ated by the CSdomain. This assumption is supported by the presence of similar faint cross-linking products for wild-type TF when nascent ICDH chains were released from ribosomes by puromycin treatment prior to cross-linker addition (data not shown). Taken together, although PCS and CS are active as chaperones in vitro, an efficient interaction with nascent polypeptides requires the ribosome-binding domain (N), a finding that is in agreement with previous results (18, 23).
The CSDomain Is Important for TF Chaperone Function in Vivo—To examine thein vivofunctionality of TF variants, we used anE. coliMC4100⌬tig⌬dnaKS1 strain lacking TF and
DnaK. These cells are viable at tem- peratures up to 30 °C due to an unknown suppressor mutation (S1) but cannot grow at higher tempera- tures. This characteristic provided a means to test thein vivofunctional- ity of the TF variants (26). The
⌬tig⌬dnaK S1 strain was trans- formed with plasmids expressing TF variants under the control of an IPTG-inducible promoter. In liquid culture the expression levels of each TF variant were similar at each par- ticular IPTG concentration tested (see supplemental Fig. S1). Freshly generated transformants and con- trol cells (MC4100 wild-type and MC4100⌬dnaK) were grown over- night without IPTG at 30 °C, spot- ted in serial dilutions on LB plates containing different concentrations of IPTG, and incubated for 30 h at 30, 34, and 37 °C (Fig. 6A).
⌬tig⌬dnaK S1 cells containing the vector plasmid grew exclusively at 30 °C, whereas cells expressing wild-type TF grew in an IPTG-de- pendent manner up to 37 °C, reflecting the growth limit of cells lacking DnaK (Fig. 6A) (4, 31). The TF-complemented ⌬tig⌬dnaK S1 strain grew at 34 °C without the addition of inducer due to the leak- iness of the IPTG-controlled pro- moter. High overexpression of TF resulted in loss of cell viability for unknown reasons as reported ear- lier (23). Under stringent conditions at 37 °C (Fig. 6A,lower panel), only NCSand PCS in addition to wild- type TF were able to complement the growth deficiency of⌬tig⌬dnaK S1 cells. Remarkably, NCScomple- mented the growth in a similar manner to wild-type TF, whereas the PCSfragment lacking the ribo- some-binding domain was clearly less efficient. At lower temperature (Fig. 6A, 34 °C, middle panel), high overexpression of the N domain or the CSdomain (100 and 250M IPTG) also slightly enhanced cell viability.
Moreover, the⌬53 variant supported growth of⌬tig⌬dnaKS1 cells at 34 °C, albeit less efficiently than wild-type TF as judged by the number of colony-forming units. Importantly, the observed complementation of the TF variants expressed in
⌬tig⌬dnaKS1 cells resulted from a residualin vivoTF activity rather than from an indirect effect of the unknown suppressor mutation (S1); all fragments that supported the growth of
⌬tig⌬dnaKS1 cells also enabled the simultaneous deletion of FIGURE 6.Analyses of growth and protein aggregation of⌬tig⌬dnaKS1 cells expressing TF variants.A,
growth analysis of wild-type MC4100,⌬dnaK, and⌬tig⌬dnaKS1 cells expressing TF and TF fragments at different temperatures in the presence of varying amounts of IPTG. Cells were grown overnight at 30 °C and after dilution (corresponding to 105, 104, 103, 102, or 10 cells/spot) spotted on LB plates (containing 0 –250M IPTG) and incubated 30 h at the indicated temperatures.B, for aggregation analysis cells were grown at 30 °C in LB medium in the presence of 20MIPTG. At log-phase (A600⫽1), cells were harvested and aggregates were isolated and analyzed by SDS-PAGE and Coomassie Brilliant Blue staining.
at Universitaet Heidelberg on November 23, 2006 www.jbc.orgDownloaded from
thetiganddnaKgenes at 30 °C in MC4100 wild-type cells (data not shown).
Furthermore, we investigated thein vivochaperone func- tion of TF variants by comparing their capacity to prevent the aggregation of cytosolic proteins in⌬tig⌬dnaKS1 cells (Fig. 6B). The analysis was performed under growth condi- tions in which all cells expressing different TF variants were viable and comparable in growth (30 °C, 20MIPTG). Minor protein aggregation was detected in cells lacking DnaK com- pared with wild-type cells (6, 26), whereas pronounced aggregation of cytosolic proteins was found in⌬tig⌬dnaKS1 cells carrying the vector plasmid (Fig. 6B, lanes 1–3). In accordance with earlier findings, the majority of the aggre- gation-prone cytosolic proteins had a molecular mass larger than 40 kDa (6, 26). Expression of wild-type TF in⌬tig⌬dnaK S1 cells efficiently reduced protein aggregation to a level comparable with⌬dnaKcells. Moreover, the NCS, PCS, and CSfragments decreased protein aggregation in⌬tig⌬dnaK S1 cells, albeit to a different extent and less efficiently than wild-type TF. In contrast, the expression of N, P, or⌬53 was ineffective or even enhanced the protein aggregation. These findings suggest that TF fragments lacking an intact CS domain lose their ability to prevent protein aggregation in vivo.
DISCUSSION
The function of the C-domain of TF has remained enigmatic for many years as it was unstable when expressed in isolation (18, 21). We have shown here that the addition of the N-termi- nal linker region stabilizes the C-terminal domain (CS) and alle- viates this problem. In addition, we truncated the C terminus by 53 amino acids (⌬53) to investigate the consequences on TF functionality. We thoroughly tested CS, domain combinations
thereof, and⌬53 for both theirin vitrochaperone activity and theirin vivo functionality. These analyses revealed that the C-terminal domain constitutes the central chaperone module of TF.
In vitro, we found that the CS domain possesses a chaperone activ- ity on its own. Although this activity was lower than that of wild-type TF, the CS domain displayed a similar functionality to prevent aggregation and even promote the refolding of denatured GAPDH into its active state (Fig. 2). Importantly, while none of the other individual domains of TF (N, P) exhibited any chaperone activity, the two-domain fragments (NCSand PCS) showed a conspicu- ously enhanced chaperone activity as compared with the CS domain alone. Thus, although the N and P domains contribute, the CSis essen- tial for the chaperone activity of TF.
This conclusion is supported by the finding that truncation of the C terminus (⌬53) resulted in the complete loss of TFin vitrochaperone activity, evident in the inability of⌬53 to prevent aggregation or promote GAPDH refolding.
The predominant role of the CSdomain in TF chaperone activity is also apparent in thein vitroRNase T1 refolding anal- yses. Efficient refolding of denatured RNase T1 requires both the PPIase activity of the P domain and the chaperone activity for tight substrate binding (17). The two-domain fragment PCS revealed a wild-type-like TF activity in RNase T1 refolding, whereas only a low level of refolding was observed for the iso- lated P domain. Thus, the major chaperone activity required for efficient RNase T1 refolding can be assigned to the CSdomain.
This is further supported by the observation that the ⌬53 mutant and the isolated P domain have similarly low refolding activities. Deletion of the C-terminal 53 amino acid residues is expected to disturb the structural arrangement of the C-do- main alone (Fig. 1) while preserving the fold of the N and P domains. TheV. choleraestructure (24) of a similar truncated TF reveals that the partial deletion of the second “arm” causes a collapse of the C-terminal domain, such that the first “arm” has a closer proximity to the N-terminal domain and a loop of the C-domain inserts into the P domain. This insertion in the P domain does not affect its activity, as⌬53 demonstrated a PPI- ase activity in RNase T1 refolding comparable with the isolated P domain (Fig. 3). Thus, it is unlikely that the disturbance of the C-domain causes severe functional defects in the P or N domain, and the observed chaperone deficiency of⌬53 is pre- dominantly due to the loss of the C-terminal domain activity.
In vivo, all TF variants containing the N-domain necessary for ribosome association (N, NCS, and⌬53) compensated to some degree for the loss of TF in⌬tig⌬dnaKS1 cells, in line with previous findings (23, 32). Surprisingly, we found that PCS FIGURE 7.The C terminus of TF is a putative chaperone module.TF, SurA, and MPN555 share the same
structural features (11, 35, 36) in their putative chaperone active domain, composed of a single helix (green), a two-helix arm (red), and a three-helix arm (blue).
at Universitaet Heidelberg on November 23, 2006 www.jbc.orgDownloaded from
and, to a minor extent, CSwere also able to support growth of cells lacking TF and DnaK at 34 °C (Fig. 6A). These fragments cannot associate with ribosomes (data not shown) and did not interact efficiently with nascent chains as shown by the cross- linking experiments (Fig. 5). How can these results be explained in light of earlier data (10) showing that thein vivoactivity of full-length TF critically depends on ribosome association? Per- haps the weak interaction of the PCSand CSfragments with nascent polypeptides (Fig. 5) might be sufficient to promote the growth of⌬tig⌬dnaKcells. Another plausible explanation is that these fragments reveal a cytosolic chaperone activity dis- tinct from the activity of wild-type TF. Such a modified chap- erone activity might be even more apparent due to the mono- meric state of these fragments and/or the absence of domains that could interfere with accessibility of substrate binding sites in the C domain. This hypothesis is supported by the observa- tion that, whereas PCSand wild-type TF display similarin vitro chaperone functionality, PCSdid not demonstrate wild-type- like inhibition on GAPDH refolding at high concentrations.
This suggests differences between PCS and TF in substrate interaction (Fig. 2C). Finally, another explanation is that TFper sehas a cytosolic chaperone activity that may be more evident in the PCSand CSfragments due to their inability to associate with ribosomes. However, in contrast to PCSand CS, only wild- type TF and the NCSfragment fully complemented the growth defects of⌬tig⌬dnaKS1 cells up to 37 °C, which underscores the importance of the ribosome association of TF for itsin vivo function.
Notably, the ability of TF fragments to complement the growth of⌬tig⌬dnaKS1 cells did not necessarily correlate with their capacity to prevent the aggregation of cytosolic proteins (Fig. 6,AandB). In particular, the⌬53 mutant showed comple- mentation up to 34 °C but failed to prevent protein aggregation in vivo. This is consistent with thein vitroanalyses showing that the destruction of the C terminus in⌬53 causes the loss of chaperone activity. How can the ⌬53 variant promote the growth of ⌬tig⌬dnaK S1 cells without preventing protein aggregation? Recentin vitrostudies showed that TF in complex with the ribosome functions as a protective shield to prevent untimely degradation of unfolded nascent polypeptides by pro- teases (33, 34). It is tempting to speculate that the⌬53 mutant, which has no detectable chaperone activity but still efficiently associated with ribosomes and nascent polypeptides (Fig. 5), may still provide protection at the ribosome and thereby com- plement the loss of TF to some degree. To test this hypothesis, we analyzed the capacity of ⌬53 to protect nascent ICDH polypeptide chains against degradation by proteinase Kin vitro (supplemental Fig. S2). Indeed, the proteolytic digestion of nas- cent ICDH polypeptides was retarded in the presence of⌬53.
However, we also observed degradation of⌬53 itself during proteinase K treatment. Thus, we are not able to unambigu- ously assign a shielding function for⌬53, as part of the protec- tive effect might be due to a substrate competition for protein- ase K.
Interestingly, NCS revealed a shielding activity similar to wild-type TF, whereas PCSdid not (33) (supplemental Fig. S2).
Although thein vitrochaperone activity of NCSis less than that of PCS, this protective ability provides an explanation for how
NCS can both fully support ⌬tig⌬dnaK S1 cell growth and reducein vivoprotein aggregation. In general, all TF fragments containing the CS domain reduced the amount of protein aggregation in⌬tig⌬dnaKS1 cells (NCS, PCS, and to a lesser extent CS). Taken together, these results indicate that the CS domain is the central module of TF chaperone activityin vitro andin vivo.
Interestingly, two other proteins,E. coli SurA (35) andM.
pneumoniae MPN555 (36), possess domains with structural homology to the C-terminal domain of TF (11, 36) despite no relevant similarity at the amino acid level (Fig. 7). MPN555 protein is a single-domain protein with unknown function.
SurA is a periplasmic chaperone dedicated to the folding of outer membrane porins in Gram-negative bacteria with two PPIase domains in addition to the TF CS-like domain. A SurA fragment lacking these two PPIase domains was found to be sufficient for SurA chaperone functionin vitroandin vivo(37).
The structural similarities of the C-terminal domain of TF to SurA and MPN555, together with the finding that this domain exists as isolated protein (MPN555) or can act in isolation (CSof TF and SurA mutant lacking the PPIase domains), suggest that it represents a chaperone module on its own. The activity of this chaperone module can be modified by the addition of other domains to fulfill the appropriate function in the periplasm, as evident for SurA, or at the ribosome, in the case of TF.
Acknowledgments—We thank members of the Deuerling and Bukau laboratories, in particular Renee D. Wegrzyn and Heather Sadlish, for comments on the manuscript and discussions. We thank Nenad Ban, Timm Maier, and Lars Ferbitz for their valuable expertise, Christian Graf and Fernanda Rodriguez for analyses by mass spectrometry, and Steffen Preissler and Sebastian Specht for technical assistance.
REFERENCES
1. Hartl, F. U., and Hayer-Hartl, M. (2002)Science295,1852–1858 2. Bukau, B., Deuerling, E., Pfund, C., and Craig, E. A. (2000)Cell 101,
119 –122
3. Wegrzyn, R. D., and Deuerling, E. (2005)Cell Mol. Life Sci.62,2727–2738 4. Deuerling, E., Schulze-Specking, A., Tomoyasu, T., Mogk, A., and Bukau,
B. (1999)Nature400,693– 696
5. Teter, S. A., Houry, W. A., Ang, D., Tradler, T., Rockabrand, D., Fischer, G., Blum, P., Georgopoulos, C., and Hartl, F. U. (1999)Cell97,755–765 6. Deuerling, E., Patzelt, H., Vorderwu¨lbecke, S., Rauch, T., Kramer, G.,
Schaffitzel, E., Mogk, A., Schulze-Specking, A., Langen, H., and Bukau, B.
(2003)Mol. Microbiol.47,1317–1328
7. Lill, R., Crooke, E., Guthrie, B., and Wickner, W. (1988) Cell 54, 1013–1018
8. Hesterkamp, T., Hauser, S., Lutcke, H., and Bukau, B. (1996)Proc. Natl.
Acad. Sci. U. S. A.93,4437– 4441
9. Patzelt, H., Kramer, G., Rauch, T., Scho¨nfeld, H. J., Bukau, B., and Deuer- ling, E. (2002)Biol. Chem.383,1611–1619
10. Kramer, G., Rauch, T., Rist, W., Vorderwu¨lbecke, S., Patzelt, H., Schulze- Specking, A., Ban, N., Deuerling, E., and Bukau, B. (2002)Nature419, 171–174
11. Ferbitz, L., Maier, T., Patzelt, H., Bukau, B., Deuerling, E., and Ban, N.
(2004)Nature431,590 –596
12. Valent, Q. A., Kendall, D. A., High, S., Kusters, R., Oudega, B., and Luirink, J. (1995)EMBO J.14,5494 –5505
13. Huang, G. C., Li, Z. Y., Zhou, J. M., and Fischer, G. (2000)Protein Sci.9, 1254 –1261
14. Kramer, G., Patzelt, H., Rauch, T., Kurz, T. A., Vorderwulbecke, S., Bukau,
at Universitaet Heidelberg on November 23, 2006 www.jbc.orgDownloaded from
B., and Deuerling, E. (2004)J. Biol. Chem.279,14165–14170
15. Stoller, G., Ruecknagel, K. P., Nierhaus, K. H., Schmid, F. X., Fischer, G., and Rahfeld, J.-U. (1995)EMBO J.14,4939 – 4948
16. Schmid, F. X. (1993)Annu. Rev. Biophys. Biomol. Struct.22,123–143 17. Scholz, C., Stoller, G., Zarnt, T., Fischer, G., and Schmid, F. X. (1997)
EMBO J.16,54 –58
18. Hesterkamp, T., Deuerling, E., and Bukau, B. (1997)J. Biol. Chem.272, 21865–21871
19. Stoller, G., Tradler, T., Rucknagel, K. P., Rahfeld, J.-U., and Fischer, G.
(1996)FEBS Lett.384,117–122
20. Hesterkamp, T., and Bukau, B. (1996)FEBS Lett.385,67–71
21. Zarnt, T., Tradler, T., Stoller, G., Scholz, C., Schmid, F. X., and Fischer, G.
(1997)J. Mol. Biol.271,827– 837
22. Patzelt, H., Rudiger, S., Brehmer, D., Kramer, G., Vorderwulbecke, S., Schaffitzel, E., Waitz, A., Hesterkamp, T., Dong, L., Schneider-Mergener, J., Bukau, B., and Deuerling, E. (2001)Proc. Natl. Acad. Sci. U. S. A.98, 14244 –14249
23. Kramer, G., Rutkowska, A., Wegrzyn, R. D., Patzelt, H., Kurz, T. A., Merz, F., Rauch, T., Vorderwu¨lbecke, S., Deuerling, E., and Bukau, B. (2004)J.
Bacteriol.186,3777–3784
24. Ludlam, A. V., Moore, B. A., and Xu, Z. (2004)Proc. Natl. Acad. Sci.
U. S. A.101,13436 –13441
25. Zeng, L. L., Yu, L., Li, Z. Y., Perrett, S., and Zhou, J. M. (2005)Biochimie
(Paris)
26. Vorderwulbecke, S., Kramer, G., Merz, F., Kurz, T. A., Rauch, T., Zach- mann-Brand, B., Bukau, B., and Deuerling, E. (2004)FEBS Lett. 559, 181–187
27. Mu¨cke, M., and Schmid, F. X. (1992)Biochemistry31,7848 –7854 28. Schaffitzel, E., Ru¨diger, S., Bukau, B., and Deuerling, E. (2001)Biol. Chem.
382,1235–1243
29. Liu, C. P., and Zhou, J. M. (2004)Biochem. Biophys. Res. Commun.313, 509 –515
30. Liu, C. P., Perrett, S., and Zhou, J. M. (2005) J. Biol. Chem. 280, 13315–13320
31. Bukau, B., and Walker, G. C. (1989)J. Bacteriol.171,2337–2346 32. Genevaux, P., Keppel, F., Schwager, F., Langendijk-Genevaux, P. S., Hartl,
F. U., and Georgopoulos, C. (2004)EMBO Rep.5,195–200
33. Hoffmann, A., Merz, F., Rutkowska, A., Zachmann-Brand, B., Deuerling, E., and Bukau, B. (2006)J. Biol. Chem.281,6539 – 6545
34. Tomic, S., Johnson, A. E., Hartl, F. U., and Etchells, S. A. (2006)FEBS Lett.
580,72–76
35. Bitto, E., and McKay, D. B. (2002)Structure (Camb.)10,1489 –1498 36. Schulze-Gahmen, U., Aono, S., Chen, S., Yokota, H., Kim, R., and Kim,
S. H. (2005)Acta Crystallogr. Sect. D Biol. Crystallogr.61,1343–1347 37. Behrens, S., Maier, R., de Cock, H., Schmid, F. X., and Gross, C. A. (2001)
EMBO J.20,285–294
at Universitaet Heidelberg on November 23, 2006 www.jbc.orgDownloaded from
