| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 279, Issue 27, 28817-28825, July 2, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
¶¶||**
From the
Department of Molecular Biology and Biochemistry and the **Department of Chemistry, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada
Received for publication, March 24, 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-helix (8, 9). This domain acts as an autoinhibitory domain in the soluble form of the enzyme (10, 11). It is a mixed structure in the lipid-free enzyme, but upon membrane binding it adopts a purely -helical conformation (12). How does partitioning into a membrane bilayer relieve the inhibition at the active site? As part of our effort to answer this question we have examined changes in the quaternary interactions of the CCT dimer upon interaction with lipid vesicles.
Early work established a homodimer structure for CCT by chemical cross-linking (13) and by gel filtration and sedimentation (14). Craig et al. (8) provided the first hint that the N-terminal two-thirds of the protein are sufficient for dimerization. After proteolytic digestion, which removed the more sensitive C-terminal domains, a species double the mass of the major N-terminal fragment could be detected on SDS-gels. The first x-ray structure of a cytidylyltransferase (glycerolphosphate cytidylyltransferase; GCT), which is homologous to the catalytic domain (amino acids 78–210) of CCT, showed that this fold interacts as a homodimer and gave insight into the dimer interface (15). In particular, the Arg within one of three signature motifs for the cytidylyltransferase family, 139RYVDEVV145, is found at the dimer interface in GCT. A CCT truncated at residue 236, containing the N-terminal domain plus the entire catalytic domain, migrates as a dimer on gels after chemical cross-linking and in sedimentation analysis (11). This suggests that the dimerization of CCT does not require the C-terminal region. There are no data from these papers specifying that the catalytic domain alone mediates dimerization. Other domains may participate in addition to the catalytic domain. For purposes of this work, the domains of CCT are designated domain N (1–72), domain C (73–236), domain M (237–300), and domain P (301–367) (Fig. 1). The boundaries are approximate.
|
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-Chymotrypsin, PMSF, and phosphoglycerate mutase were from Sigma. Glutaraldehyde was from BDH. DSP and BS3 were from Pierce. [14C]NEM was from PerkinElmer Life Sciences. Alexa Fluor 532 succinimidyl ester, Prolong antifade, and Oregon Green- or Texas Red-conjugated secondary antibodies were from Molecular Probes. Lipids were from Avanti or Northern Lipids. Restriction enzymes, Vent polymerase, and T4 ligase were from New England Biolabs. Oligonucleotides were from Invitrogen. Antiserum against the N-terminal 15 residues of rat CCT and against a 33-residue segment of domain M were obtained as described previously (16, 17). Antibody against amino acids 164–176 was generously provided by A. J. Ryan and R. K. Mallampalli (Iowa). Anti-hemagglutinin antibody was from Covance. The yeast two-hybrid vector pACT was from Clontech, and pBTM116 (18) was a gift from Dr. Charles Boone (Toronto). SYPRO Orange protein gel stain was from Amersham Biosciences. Microcon-30 filters were from Amicon.
Yeast Two-hybrid Construction and Analysis—Constructs were prepared which linked the LexA DNA binding domain (DBD) or the Gal4 activation domain (AD) to the N termini of full-length CCT or CCT fragments (amino acids 1–236, domains N + C; 73–239, domain C; 237–367, domains M + P). For this purpose we used the yeast expression vectors pBTM116 containing full-length LexA (18), and pACT2 containing residues 768–881 of Gal4 and a nuclear targeting signal upstream of the Gal4 domain (24). All CCT sequences were inserted in-frame into the BamHI or BamHI/SalI sites in the linker regions of these vectors. We engineered stop codons and BamHI or SalI sites flanking the CCT sequences by PCR mutagenesis, using WT rat CCT in pAX142 (16) as the template DNA. For CCT(1–367) and CCT(1–236) the 5'-primer was 5'-AT AGG ATC CCT TAT AAC ATG1 GAT2 GCA3 CAG4 AGT5 TCA6 GCT7 AAA8 GTC9-3' (BamHI site in italics), adding a linker encoding RYN before the first CCT codon (CCT codons indicated with superscript). The 3'-primer for CCT(1–236) was 5'-ATA GGA TCC TCAstop GTT236 GAT235 AAA234 GCT233 GAC232 ATT231 GAG230 CTC229-3'. The 3'-primer for CCT(1–367) was ATA GGA TCC TCAstop GTC367 CTC366 TTC365 ATC364 CTC363 GCT362 GAT361 GTC360. These oligonucleotides introduced stop codons after codon 236 or 367 followed by the BamHI site. The CCT(237–367) construct was generated with the same 3'-primer as used to generate CCT(1–367), and the 5'-primer was 5'-AT AGG ATC CGT TAT AAC ATG GAA237 AAG238 AAA239 TAC240 CAC241 TTG242 CAA243 GAA244-3', which added RYNM to the linker sequence just before CCT codon 237. CCT(73–239) was generated using a 5'-primer, 5'-AT AGG ATC CGT TAT AAC ATG TGT73 GAG74 CGG75 CCT76 GTG77 AGA78 GTT79-3'. The 3' primer was 5'-T CGT CGA CTAstop TTT239 CTT238 TTC237 GTT236 G-3', which introduced a SalI site flanking the stop codon following codon 239. The correct sequences were checked by inserting the amplified DNA into the BamHI or BamHI/SalI sites of pBS KS+ and sequencing both strands. For the pBTM and pACT CCT1–367 constructs a 1.1-kb 3'-EcoRI fragment was replaced with an EcoRI fragment from a previously sequenced full-length CCT so that only the 5'-end of the new construct required sequencing.
To test interactions we used yeast strain Y274 containing the promoter sequence specific for LexA upstream of His-3 and LacZ, and lacking leucine and tryptophan synthases for selection of pBTM116 and pACT2 transformants. Yeast cells were made competent by modification of a lithium acetate procedure (24) and were transformed using 33% polyethylene glycol and a 15-min, 42 °C heat shock (24). Transformants were selected by culturing in leucine- and tryptophan-deficient medium at 30 °C for 48 h. To quantify the levels of LacZ expression, yeast cell lysates were prepared from mid-log phase cultures (A600 nm = 1.0). Cells from 3-ml cultures were collected by centrifugation at 13,000 rpm for 1 min and were suspended in 160 µl of Z buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1 mM MgSO4) with 50 mM 
-mercaptoethanol. 40 µl of sample was diluted to 1 ml with 0.15 M NaCl, 3.7% formaldehyde, and the A600 nm was determined. The remainder was solubilized by adding sarkosyl and toluene to final concentrations of 0.2%, and 3.8%, respectively. The sample was vortexed and centrifuged at 13,000 rpm for 10 s. The solvent was evaporated from the supernatant, and 100 µl was mixed with 650 µl of Z buffer containing 0.67 mg/ml o-nitrophenyl -D-galactopyranoside to initiate the -galactosidase reaction. The reaction was quenched at various times with 250 µl of 1 M Na2CO3 to a final concentration of 250 mM. The absorbance at 420 nm was determined. The units of -galactosidase activity were calculated as shown in Equation 1.
 | (Eq. 1) |
Isolation of Yeast Spheroplasts and Nuclei and Analysis of Expression Levels of CCT Constructs—Cell walls were digested by zymolyase in the presence of 1 M sorbitol, essentially as described previously (25). Preparation of spheroplasts was checked by microscopy. Nuclei were isolated from Dounce-homogenized spheroplasts by differential sedimentation in 18% Ficoll (25). The nuclear prep was stained with DAPI and examined by microscopy using a Zeiss LSM-410 confocal microscope to assess the enrichment with nuclei. Immunoblotting was performed as described previously (26) except that that transfer buffer was 39 mM glycine, 48 mM Tris, 0.0375% SDS, pH 8.7. For immunofluorescence analysis of CCT localization, spheroplasts were absorbed onto 0.1% polylysine-coated slides. The cells were fixed using sequential methanol and acetone baths at -20 °C, for 6 min and 30 s, respectively. After blocking with 2% bovine serum albumin in phosphate-buffered saline for 1 h, anti-hemagglutinin or anti-domain M antibodies were added in blocking buffer using a dilution of 1:100 for incubation over-night at 20 °C. After five washes, the secondary antibodies conjugated to Oregon Green or Texas Red were applied at 1:100 dilution in blocking buffer. The slides were prepared for viewing with Prolong antifade. The cells were viewed with a Zeiss LSM-410 microscope (26).
Construction of Cys Mutants—WT CCT was engineered with an N-terminal His-tag by PCR amplification using pBSKS(-) WT rat CCT as the template. The forward primer, 5'-CAG CAG GTC TAG ACG CGT AGG ACC ATG GCT AAG CAC CAC CAT CAC CAT CAC ata gaa gga aga TCT GCC ATG1 GAT2 GCA3 CAG4 AGT5 TC-3' introduced the His-tag (bold), a Factor Xa site (lowercase), and an optimal translation initiation sequence (underlined). Superscripts denote CCT codons. T3 was used as the reverse primer. The PCR product was inserted into pBSKS(-) at the XbaI and SalI sites. For expression in COS cells it was inserted in pAX142 at the MluI and SalI sites.
The C139S was engineered by PCR using as template a His-tagged WT CCT in pBSKS(-) with a deleted SalI site. The forward primer was the His-tag-generating primer (see above), and the reverse primer introduced the mutation (underlined) at codon 139 followed by the AccI site (italics): 5'-CAC145 CTC144 GTC143 TAC142 GTA141 TCT140 AGA139 ATG138 CTG137 CAC136 CGC135 GTC134-3'. The PCR product was inserted into pBS WT CCT with a deleted SalI site using MluI and AccI. For expression in COS cells it was inserted in pAX142 at the MluI and EcoRV sites.
To construct CCT with serines at positions 354 and 359, WT CCT in pBSKS(-) was PCR amplified with the forward primer: 5'-CCA348 GCA349 AGC350 TTA351 TCC352 AGA353 TCT354 AAG355 GCT356 GTG357 ACT358 AGT359 GAC360 ATC361 AGC362 GAG363-3', to introduce Cys to Ser mutations (underlined) as well as three new restriction sites: HindIII, BglII, and SpeI (italics). T3 was used as the reverse primer. His-tagged WT CCT was amplified using 5'-GA CTG CCG CGG GTT53 GAC54 TTT55 AGT56 AAG57-3' as a forward primer and 5'-CCT353 AGA352 TAA351 GCT350 TGC349 TGG348 GGA347 GGA346-3' as a reverse primer to create a HindIII site (italics). The HindIII to XhoI fragment of the C354,359S DNA was ligated with the SstI to HindIII fragment of the pBS His-tagged WT CCT amplification. The resulting SstI to XhoI fragment was then inserted into SstI/XhoI-cleaved pBSKS(-) His-tagged WT CCT. For expression in COS cells the construct was inserted into pAX142 using MluI and SalI sites.
Further single cysteine to serine mutations were engineered using the QuikChange PCR mutagenesis materials (Stratagene) following the manufacturer's instructions. The PCR template was His-tagged WT CCT in pBSKS(-). The complementary primer pairs replaced the cysteine codon with a serine codon. The sequences of the mutagenic codons of the forward primers were: C37S, TGT37 to TCT37; C68S, TGC68 to AGC68; C73S, TGT73 to TCT73; C113S, TGC113 to AGC113. The MluI to EcoRV fragments of the single cysteine mutants were each subcloned into MluI/EcoRV-cleaved pAX142-WT CCT. Cysteine-free CCT was engineered by replacing the N-terminal cysteines with serines one after another using the QuikChange PCR mutagenesis kit, with pBS-His-tagged C139S CCT as a template. MluI and EcoRV were used to subclone this into pAX142-HisC354,359S CCT. A single cysteine was engineered back into codon 37 using His-tagged Cys-free CCT in pBSKS(-) as a template. The sequence of the mutagenic codon of the forward primer was TCT37 to TGC37. All mutants were sequenced in pBSKS(-) to ensure correct sequence.
Purification of Untagged CCT and Activity Assay—The isoform of rat CCT was expressed in Trichoplusia ni cells using the baculovirus system (19). It was purified by the method of Friesen et al. (11) with modifications (4). The purified CCT was stored in Buffer A (10 mM Tris, pH 7.4, 0.15 M NaCl, 1 mM EDTA, 2 mM dithiothreitol) at -80 °C. The CCT activity assay was carried out as described previously (20) in the presence of 0.1 mM PC/oleic acid (1:1)-sonicated vesicles.
Expression and Purification of His-tagged CCTs—COS-1 cells were transfected for 68 h with the His-tagged CCT cDNA constructs in pAX142 (16). The soluble fraction was prepared as described previously (27), except the homogenization buffer was 20 mM KH2PO4, pH 7.4, containing a full compliment of protease inhibitors. Prior to the high speed centrifugation, binding buffer was added from a 10x stock to a final concentration of 5 mM Na2HPO4, pH 8.0, 0.5 M NaCl, 15 mM imidazole. A 1/8 volume of 50% nickel-agarose beads was incubated with the soluble protein by rotation at 4 °C for 1 h. Beads were washed three times with 50 mM Na2HPO4, pH 8.0, 500 mM NaCl, 25 mM imidazole, 1% Nonidet P-40 and three times with 50 mM Na2HPO4, pH 8.0, 100 mM NaCl, 25 mM imidazole. His-tagged proteins were eluted with 50 mM Na2HPO4, pH 8.0, 100 mM NaCl, 350 mM imidazole. The CCT in the preparations purified by this method represented from 61 to 88% of the total protein in the samples, as assessed by densitometry of silver or SYPRO Orange-stained gels. The activities of the purified CCTs were determined as described above.
-Chymotrypsin Digestion—The reactions were carried out typically in a volume of 30 µl at 37 °C in a shaking water bath for various times using 1–4 µM CCT, a 1:150 weight ratio of chymotrypsin to CCT, and Buffer A with 0.15 mM Triton X-100. The reactions were initiated with chymotrypsin and were stopped with PMSF to a final concentration of 2 mM. A 10 mM PMSF solution in 
90 °C water was prepared from a 0.5 M dimethyl sulfoxide stock immediately before the assay and was maintained at 37 °C. Some of the samples were prequenched with PMSF prior to chymotrypsin treatment. Other digested samples were subsequently reacted with glutaraldehyde within 2 min of adding PMSF (see below). Samples were analyzed by 12% SDS-PAGE.
Cross-linking Reactions—Stocks of CuSO4 and phenanthrolene were mixed 1:3 immediately before initiating the reactions. For cross-linking with Cu(Phe)3 or DSP, the dithiothreitol and Tris were removed from the CCT preparation by dialysis against 20 mM K2HPO4, pH 7.4, 0.1 M NaCl, 0.15 mM Triton X-100 or by Microcon-30 filtration followed by three washes in the above buffer. For other reactions this step was not needed. The reactions were typically in a volume of 30 µl at 37 °C in a shaking water bath. The reaction buffer was 20 mM phosphate, pH 7.4, the Triton concentration was adjusted to be equivalent in all samples (generally 75 µM), and the concentrations of CCT (0.5–8 µM) and phospholipid vesicles were as specified in the figure legends. Small unilamellar vesicles were prepared by sonication as described previously (3). The reactions were initiated with cross-linking reagent using the concentrations and incubation times specified in the figure legends. The glutaraldehyde reaction was quenched with 100 mM ethanolamine; the DSP reaction was quenched with 100 mM ammonium acetate, Cu(Phe)3 was quenched with 10 mM NEM, and BS3 was quenched with 0.1 M glycine, pH 6.5. Some samples were prequenched prior to the addition of cross-linker. Samples were analyzed by 10% SDS-PAGE.
Labeling with [14C]NEM—The reaction with NEM was done in a volume of 20 µl and used 2.5 µM CCT (dithiothreitol-free), 225 µM [14C]NEM (specific activity, 39 Ci/mol) in 20 mM phosphate buffer, pH 7.4, 1 mM EDTA, 0.075 mM Triton X-100, with or without 500 µM PG vesicles. The sample was incubated with shaking for 30 min at 37 °C. The reaction was quenched by boiling for 3 min in Laemmli sample buffer containing 2% -mercaptoethanol. The stained CCT bands were excised from the gel, treated with 50% H2O2 at 70 °C for 17 h, and the radioactivity determined by liquid scintillation counting.
Gel Electrophoresis and Western Blot Analysis—Samples were boiled for 3 min in sample buffer prior to separation of components by SDS-PAGE using 10 or 12% acrylamide (21). For Cu(Phe)3 and DSP cross-linked samples, the sample buffer did not contain -mercaptoethanol. SYPRO Orange staining of gels was done as described by the manufacturer, and the gels were visualized on the Typhoon 9410 Variable Mode Imager with a 488 nm laser and a 580 nm filter. Silver staining of gels was done as described previously (22). Immunoblots were done as described (23) with an antibody directed against the N-terminal 15 amino acids (15) or residues 256–288 of domain M (17). Immunoblots with an antibody directed against residues 164–176 of domain C were done at 37 °C with both the primary antibody (anti-domain C), and the secondary antibody, goat anti-rabbit horseradish peroxidase, diluted 1:1,000 in 1% gelatin, 0.1% Tween 20 in Tris-buffered saline.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
7 Å to be cross-linked by the aldehydes. DSP and BS3 contain succinimidyl groups separated by 12 Å that react selectively with lysine amino groups. Cu(Phe)3 selectively oxidizes cysteine sulfhydryls. When lipid-free CCT was incubated with glutaraldehyde or the succinimidyl-based reagents, the 42-kDa monomer species observed on gels was converted to a diffuse molecular set between 80 and 110 kDa (Fig. 2, A and B, lanes 4–6; Fig. 3, A, lanes 2–5, and C, lanes 2–4). Autoxidation during the dialysis step preceding treatment with DSP or Cu(Phe)3 generated species at 84 kDa and 120 kDa in addition to the monomeric band (Fig. 2C, lanes 1–3). After incubation with Cu(Phe)3 the 84 and 120 kDa bands became predominant, and the 42 kDa band disappeared (Fig. 2C, lanes 4–6). The diffuse DSP-linked band at 80–110 kDa has been identified as heterogeneously linked homodimers by two-dimensional electrophoresis (9). The identity of the 120 kDa disulfide-linked species has not been resolved. The predominance of this band varied from experiment to experiment (e.g. see Fig. 4C). In addition, higher order cross-linked oligomers were present in many samples as minor species (Fig. 2).
|
|
|
Lipid Vesicles Decrease Cross-linking Efficiency at the Dimer Interface—When 770 M excess anionic lipid vesicles (small unilamellar vesicles) were added to CCT prior to initiation of cross-linking with the lysine-specific succinimidyl cross-linker BS3, the monomer species was retained, with only 12% conversion to dimer after 10 min compared with 88% conversion to dimer for the vesicle-free CCT (Fig. 3, A and B). With glutaraldehyde as the cross-linker the monomer band persisted in samples containing 400 M excess PG vesicles, with only a weak dimer band at 84 kDa appearing along with higher molecular mass oligomers (diffuse smear above 84 kDa in lanes 6–8 in Fig. 3C). Increasing the PG/CCT ratio from 400 to 4,000 (to generate 1 CCT/vesicle) reduced the proportion of large oligomeric species and generated an increase in the proportion of monomer band at 42 kDa (Fig. 3C, lane 9). This suggests that the higher molecular mass oligomers, prominent only at high CCT/vesicle ratios, are formed by covalent trapping of interdimer collisions on the surface of the vesicles, where the concentration of bound CCT molecules would be increased by 4 orders of magnitude compared with CCT in the aqueous compartment. These results suggested that the PG vesicles were interfering with some aspect of the cross-linking reaction or were altering the structure of the dimer interface.
CCT interacts preferentially with anionic versus zwitterionic vesicles. The effects of zwitterionic PC vesicles compared with PG and two other anionic vesicles, PC/oleic acid (1:1) and PA, on the cross-linking potential of glutaraldehyde, DSP, and Cu(Phe)3 are shown in Fig. 4. Only the anionic vesicles reduced the extent of cross-linking. The anionic phospholipids were more inhibitory than PC/oleic acid vesicles (e.g. compare lanes 4, 5, and 6 of Fig. 4, A and C), in keeping with their higher negative charge density and higher affinity for CCT (2, 3). DSP has lower water solubility than glutaraldehyde or Cu(Phe)3, raising the question of whether membrane sequestration of DSP could account for the reduction in cross-linking with DSP. However, increasing the DSP concentration from 125 to 625 M excess over CCT did not overcome the inhibitory effects of the anionic lipid vesicles on CCT cross-linking (Fig. 4B, lanes 4–6). These data suggested that there are one or more sets of lysines and cysteines in the CCT dimer interface that are altered in positioning or flexibility upon interaction of CCT with anionic membranes.
To exclude trivial explanations of the above results we conducted several other tests. We examined whether the reduced efficiency of disulfide bond formation upon membrane binding could be explained by a loss of cysteine on CCT available for reaction with the cross-linkers. We compared the labeling of cysteines with [14C]NEM in the presence and absence of 100 M excess PG vesicles. The radioactivity associated with CCT was quantitated after excision of the Coomassie-stained band. The labeling of CCT was consistently increased by 16% in the presence of the PG vesicles. The stoichiometry of NEM/CCT was 3.5 ± 0.08 mol/mol in the absence of PG, and 4.0 ± 0.24 mol/mol in the presence of PG (mean ± S.D. of four determinations). These results indicate that the lipid vesicles did not reduce the availability of cysteines for reaction with aqueous reagents. The lack of complete labeling of all 7 cysteines on CCT may mean that 3 of the 7 cysteines are buried. Alternatively, it may be a reflection of the inaccuracy of the protein concentration estimation. We attempted to compare the labeling of lysines with the succinimidyl ester of Alexa Fluor 532 in the presence and absence of PG vesicles. However, the presence of 29 lysines in CCT complicates interpretation of any changes in labeling. Changes in accessibility of only a few sites, which could impact on the cross-linking results, would be hard to detect. In fact the extensive labeling of CCT caused self-quenching of the Alexa Fluor and was not pursued further.
To test whether the anionic phospholipids interfered with the chemistry of the cross-linking reactions, we examined their effect on glutaraldehyde or BS3-induced cross-linking of phosphoglycerate mutase, a nonmembrane dimeric protein. Neither PA nor PG vesicles affected the cross-linking of a 30-kDa monomeric species into a 60-kDa dimeric species (Fig. 5), although these vesicles inhibited cross-linking of CCT (Fig. 4 and Fig. 5, lanes 9–11). In Fig. 5B, the BS3 reaction conditions resulted in only partial cross-linking; thus the lack of effect of lipids cannot be explained by excessive cross-linking conditions.
|
28 and 26 kDa, Fig. 6, A and B, lane 4). After digestion this sample was subjected to glutaraldehyde cross-linking (Fig. 6B, lane 5). The intensity of fragments 7 and 8 diminished, and a new band appeared, centered at 53.5 kDa. This species reacts with antibody against the N terminus of CCT as well as with antibody against residues 164–176 of domain C (Fig. 6, B and C). After a 5-min digestion, the major fragments were CCT(1–216) and CCT(1–173) (bands 8 and 11, migrating at 26 and 21 kDa; Fig. 6A, lane 6). Glutaraldehyde treatment of the 5-min digest generated a major band that migrated slightly faster than the cross-linked species in lane 5 and centered around 52 kDa (lane 7). In the 5-min digest (but not the 2-min digest), glutaraldehyde linked some heterogeneous species migrating at 43 kDa, which also reacted with anti-N antibody and anti-domain C (Fig. 6, A–C, lane 7). Fragments 7–12 disappeared to form the 43- and 52-kDa species. Further proof of the identity of the 54- and 52-kDa cross-linked species was obtained by two-dimensional electrophoretic analysis using the cleavable cross-linker, DSP (13). In the second (reducing) dimension the 54-kDa species resolved principally into a 27-kDa species, and the 52-kDa species resolved principally into bands at 27 and 25 kDa (data not shown).
|
The most reasonable hypothesis to explain these results is that the cross-linked species (present only in glutaraldehyde-treated samples) represent heterodimers of N-terminal fragments; CCT(1–225), (1–216), (1–191), (1–173) (20–25.5 kDa each monomer). This cross-linking approach confirmed that the N-terminal region (domain N + C) is sufficient to maintain a dimer interface and suggested that the C-terminal end of domain C is not essential for dimer interaction. Because no fragment corresponding to the catalytic core (residues 75–225) was isolated, this analysis did not resolve whether the catalytic domain alone is sufficient to mediate the interactions of the dimer. Because the C-terminal domain is cleaved rapidly into small fragments we could not use this approach to assess the participation of the C-terminal domain in dimerization.
Yeast Two-hybrid Results Implicate an Important Role for the N-terminal 72 Amino Acids in CCT Dimerization—Interactions between CCT domains were explored using the pBTM116/pACT system. Constructs (see Fig. 1) corresponding to full-length CCT, the first 236 amino acids, the catalytic core (73–239), or the C-terminal domain (237–367) were fused to both the LexA DBD and the Gal4 AD and were expressed in yeast. Positive interactions were observed between the two CCT(1–236) fusion proteins (very strong), CCT(1–236) and CCT(73–239) (medium strength), and the two CCT(73–239) fusion proteins (weak), as assessed by a quantitative -galactosidase assay (Table I). The same pattern of interactions was also observed in three separate transformations using a quick LacZ assay, which scores the development of blue color by spheroplasts incubated with o-nitrophenyl -D-galactopyranoside (data not shown).
|
62 and 35 kDa). Thus the negative two-hybrid results with constructs containing the CCT C-terminal domains are not explained by their lack of nuclear expression or by unbalanced ratios of the AD versus DBD constructs.
|
has 7 cysteines (Fig. 1). We replaced each with serine by PCR mutagenesis, individually, except for the 2 at the C terminus, which were replaced as a pair. Each mutant was expressed in COS cells with an N-terminal His-tag and was purified by nickel affinity chromatography. The activity was not affected by the His-tag (specific activities: WT CCT, 10.2 ± 1.4 units/µg; His-tagged CCT, 11.3 ± 2.2 units/µg). All mutants were catalytically active, and all but the cysteine-free CCT had specific activities within the range of the WT CCT (Fig. 8). The expression and concentration of the purified preparation of the cysteine-less CCT were the lowest, which contributed to inaccuracy in the protein determination used to calculate the specific activity.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
is altered upon membrane binding. Phospholipid vesicles that promote CCT binding greatly reduce the efficiency of cross-linking by glutaraldehyde, DSP, BS3, and Cu(Phe)3. Because cross-linking is a nonreversible modification, the lack of cross-linked dimer species suggests a profound alteration at the dimer interface, where the contact points available for reaction in the soluble form must be distanced or rigidified so as to forge contacts only as a rare event. We conducted several controls to exclude trivial explanations of our observations. We showed that our cross-linking reactions covalently captured the noncovalent interactions of two monomers, rather than random collisions (Fig. 2). We showed that the phospholipids do not interfere with the chemistry of the glutaraldehyde or succinimidyl-based (BS3) cross-linking reaction (Fig. 5). Lipid vesicles did not appear to reduce the accessibility of cysteines for reactivity with Cu(Phe)3. Moreover, the phospholipid specificity for CCT binding was mimicked in the phospholipid specificity for the inhibition of the cross-linking reactions (Fig. 4). Lastly, the effect of lipids on cross-linking was seen for both water-soluble and less soluble cross-linking reagents, ruling out effects of membranes cause by sequestration of the cross-linking reagents. The dimer interface of CCT within domain C, by analogy to the GCT structure, is extensive and has many hydrophobic contacts. It is possible that membrane binding destabilizes some of these interactions but compensates for this by forming favorable hydrophobic contacts between the membrane-binding amphipathic helix and the lipid bilayer. Does membrane binding dissociate the dimer to monomers? There are several arguments against this idea. 1) The constitutively active CCT(1–236) is a dimer (11), suggesting that dissociation of the dimer is not needed to make an active enzyme. 2) The interactions at the CCT dimer interface are very strong. We have observed traces of dimer band in reducing SDS-PAGE, and the dimer appears to be resistant to urea concentrations as high as 2 M.3 In another project we have found evidence that full-length CCT dimers engage anionic membrane vesicles in a cross-bridging mode that leads to vesicle aggregation.4 Thus, instead of full dissociation to monomers, membrane binding likely changes the position and/or flexibility of regions involved in dimerization.
Some of the cross-linking reactions also generated higher oligomers of CCT, suggesting the capacity for higher order complexes, whose functional significance is not known. The formation of these oligomers was not blocked by CCT binding to lipid vesicles. Rather, it was enhanced when the CCT/vesicle ratio was relatively high (e.g. 10–20 CCT/vesicle) but was reduced when the CCT/vesicle ratio was 1. The simplest explanation is that these cross-linked oligomers represent trapped encounters between two or more CCT dimers on the vesicle surface.
Although our data suggest that at least some dimer contacts are altered during the process of membrane binding, they do not indicate whether or not this change in the protein quaternary structure is involved in catalytic activation. The cross-linking reactions are inhibitory to enzyme activity (13; and other data not shown), but in the case of glutaraldehyde, DSP, or BS3, this could be merely because of chemical modification of the protein, rather than a constrained dimer interface. We found that the copper (without phenanthrolene and without cross-linking of the CCT) is a weak inhibitor of CCT activity, thus the inhibition of CCT activity by Cu(Phe)3 is not only related to disulfide bond formation. This key question will be approached in future studies by assessing the effect of disulfide constraints engineered at specific sites, once the dimer interface is mapped more accurately.
The Dimer Interface Involves Domains C and N—The identities of the major cross-linked protease fragments were derived from their reactivities with antibodies against the N terminus and domain C, their apparent mass on SDS-gels, and from the disappearance of the chymotryptic fragments. The major cross-linked species are composed of domain N + most of domain C. Friesen et al. (11) showed previously that CCT(1–236) interacts as a homodimer. Our data confirm this result within the context of the full-length enzyme and suggest that dimerization does not require amino acids 174–236 because removal of this region by proteolysis did not disrupt the dimer interaction.
The GCT dimer consists of a minimal catalytic domain (15). Is the analogous domain in CCT (domain C) an independent folding unit that mediates dimerization without participation of other domains? The yeast two-hybrid interaction scores indicated very strong interactions only for constructs containing domain N. This implies that the catalytic core alone has much weaker dimerization potential than the combined domain N + C. This could be caused by stabilization by the N-terminal 72 amino acids of a dimer interface housed solely within the catalytic core, or the N-terminal domain may directly mediate dimer interactions. If the latter case were correct, the intermediate interaction score when CCT(1–236) was matched with the catalytic core would suggest that domain N of one monomer interacts with domain C of its partner. If the interaction of domain N were only with its partner domain N, the interaction score of domain N +C versus domain C would have been as low as domain C versus domain C.
The two-hybrid probe did not reveal interactions between domain M and any other CCT domain. Moreover, full-length CCT did not show evidence of dimer interactions in the two-hybrid analysis. One possible explanation is that the C-terminal domains M and/or P of CCT interfere with the surfaces of LexA DBD or Gal4 AD for interactions with promoter DNA or with basal transcription factors. Another possibility is that LexA-hybrid proteins containing the C-terminal domains of CCT may form very strong homodimers upon emergence from the polyribosome, precluding interaction with the Gal4 hybrids, and vice versa. A third explanation is that the presence of domain M traps the hybrid protein on the membrane, which prevents productive interaction with promoters and transcription factors. Lastly, the membrane binding of hybrids containing domain M might weaken the dimer interface in domains N/C. This idea goes along with the results from cross-linking that suggest a potential weakening of the dimer interface by binding lipids.
The strongest indication of the participation of domain N in dimerization emerged from the study of single cysteine to serine replacements at all 7 native cysteines. We anticipated that the Cys-139 pair would be identified from this screen as the disulfide bonding pair because our GCT-based model shows these residues are near the dimer interface in domain C. To our surprise this analysis identified the Cys-37 pair as the disulfide bridge that covalently links the two monomers under oxidizing conditions. Furthermore, disulfide bond formation at this position was inhibited by the binding of CCT to membranes, thus identifying one subunit contact site that is modulated by membrane binding. Future dissection of the contact points of the dimer, the identity of those sites that are modulated by membrane binding, and the consequences on enzyme activation will shed light on the mechanism of lipid-triggered conformational changes in CCT. Our successful engineering of a cysteine-free CCT will facilitate progress toward these goals.
|
FOOTNOTES |
|---|
These authors contributed equally to this work.
¶ Present address: Dept. of Molecular Cell and Development Biology, University of California Los Angeles, Los Angeles, CA 90095-1606.
|| Present address: Zhanchunyuan 4-6-60, Haidian, Beijing 100083, People's Republic of China.
To whom correspondence should be addressed: Dept. of Molecular Biology and Biochemistry, Simon Fraser University, 8888 University Dr., Burnaby, BC V5A 1S6, Canada. Tel.: 604-291-3709; Fax: 604-291-5583; E-mail: cornell{at}sfu.ca.
1 The abbreviations used are: CCT, CTP:phosphocholine cytidylyltransferase; AD, activation domain; BS3, bis(sulfosuccinimidyl)suberate; Cu(Phe)3, copper phenanthrolene; DBD, DNA binding domain; DSP, dithiobis(succinimidyl propionate); GCT, glycerolphosphate cytidylyltransferase; NEM, N-ethylmaleimide; PA, phosphatidic acid; PC, phosphatidylcholine; PG, phosphatidylglycerol; PMSF, phenylmethylsulfonyl fluoride; WT, wild-type.
2 M. Bogan, G. Agnes, and R. B. Cornell, unpublished data.
3 J. E Johnson and R. B. Cornell, unpublished data.
4 S. Taneva, P. Patty, B. Frisken, and R. B. Cornell, unpublished data.
|
ACKNOWLEDGMENTS |
|---|
, and Dr. Joanne Johnson for valuable discussions. |
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
, samples without PG; 
, samples with PG. C, 0.64 µM CCT was preincubated for 5 min without (lanes 1–4) or with 250 µM PG vesicles (lanes 5–8) or 2.5 mM PG vesicles (lane 9). Reactions were initiated with 1 mM glutaraldehyde and were quenched at the indicated times with 0.1 M ethanolamine. Samples were electrophoresed and stained as in 
