FtsH, a membrane-bound ATPase, forms a complex in the cytoplasmic membrane of Escherichia coli.

The FtsH (HflB) protein of Escherichia coli is integrated into the membrane with two N-terminally located transmembrane segments, while its large cytoplasmic domain is homologous to the AAA family of ATPases. The previous studies on dominant negative ftsH mutants raised a possibility that FtsH functions in multimeric states. We found that FtsH was eluted at fractions corresponding to a larger molecular weight than expected from monomeric structure in size-exclusion chromatography. Moreover, treatment of membranes or their detergent extracts with a cross-linker, dithiobis(succinimidyl propionate), yielded cross-linked products of FtsH. To dissect possible FtsH complex, we constructed an FtsH derivative with c-Myc epitope at its C terminus (FtsH-His6-Myc). When membranes prepared from cells in which FtsH-His6-Myc was overproduced together with the normal FtsH were treated with the cross-linker, intact FtsH and in vitro degradation products of FtsH-His6-Myc without the tag were cross-linked with the tagged FtsH protein. Co-immunoprecipitation experiments confirmed the interaction between the FtsH molecules. To identify regions of FtsH required or sufficient for this interaction, we constructed chimeric proteins between FtsH and EnvZ, a protein with a similar topological arrangement, by exchanging their corresponding domains. We found that only the FtsH-EnvZ hybrid protein with an FtsH-derived membrane anchoring domain and an EnvZ-derived cytoplasmic domain caused a dominant ftsH phenotype and was cross-linked with FtsH. We suggest that the N-terminal transmembrane region of FtsH mediates directly the interaction between the FtsH subunits.

Escherichia coli FtsH (HflB) protein belongs to a novel ATPase family whose members are widely found among eukaryotic and prokaryotic organisms (1). They all have one or two copies of the conserved regions of about 200 amino acid residues that include a set of ATP binding consensus motifs (2). They are suggested to be involved in diverse cellular functions such as regulation of cell cycle, vesicular transport in protein secretion, biogenesis of organelles, nuclear division, regulation of transcription, and protein degradation (2). This protein family is called AAA (ATPases associated with a variety of cellular activities) (3). However, their modes of involvement in the above mentioned cellular processes are mostly unclear. Even ATPase activities have been demonstrated only for a few of them (4 -6). Their localizations in the cell are also diverse; some are bound to the plasma or the organella membrane, but many others are soluble proteins (2).
We previously showed that mutational impairments of the ftsH gene of E. coli caused an Std phenotype in which a normally cytoplasmic reporter PhoA 1 domain of a model membrane protein (SecY-PhoA) was exported to the periplasmic space (7,8). Since the Std phenotype signifies insufficient anchoring of the transmembrane segment that precedes the reporter domain, we suggested that FtsH is involved in the process of protein assembly into the membrane. We also found that a decreased cellular content of the FtsH protein resulted in a strong Std phenotype and an impaired translocation of some secreted proteins (Sec phenotype) (7). Therefore, FtsH might have a role in protein export as well. Additionally, we found that the expression of C-terminally truncated forms of FtsH or ATP binding site mutants of FtsH from a plasmid caused the Std and Sec phenotypes dominantly (8). The existence of dominant negative alleles of ftsH raises a possibility that FtsH may function in multimeric states.
This study was aimed at clarifying the quaternary structure of FtsH in the cell. We showed that FtsH in the wild-type cells exists as a complex. Co-immunoprecipitation and cross-linking experiments using a Myc epitope/His 6 -tagged FtsH revealed that the FtsH molecules interact with each other. A series of chimeric proteins between FtsH and EnvZ were constructed, and cross-linking experiments using them showed that the FtsH-FtsH association required the N-terminal membrane association region but not the cytoplasmic domain.
Construction of the ftsH-his 6 -myc Plasmids-pSTD101 carrying ftsHhis 6 -myc was constructed as follows. pSTD40 in which a mutant ftsH gene (the ftsH40 allele) was placed under the lac promoter/operator was described previously (7). 2 The 2.7 kb EcoRI-PstI fragment of pSTD40 was blunt-ended by treatment with T4 polymerase and cloned into SmaI site of a pBlueScript SK(Ϫ) (Stratagene) derived vector, pTYE007, which carried a sequence encoding a bipartite His 6 /c-Myc tag * This work was supported by grants from the Ministry of Education, Science, and Culture, Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Constructions of Hybrid Genes between ftsH and envZ-pSTD117 that carried an envZЈ-ЈftsH hybrid gene was constructed by site-directed mutagenesis as follows. First, a 0.8-kb XbaI-EcoRV fragment of pAT2005S (15) carrying the envZ gene was ligated with pSTD113 that had been digested with BamHI, blunt-ended by treatment with T4 polymerase, and then digested with XbaI. Then, the region encoding the membrane anchoring domain (from the amino terminus to the 179th amino acid residue) of EnvZ and the region encoding the cytoplasmic domain (from the 121th amino acid residue of FtsH to the carboxyl terminus) of FtsH-His 6 -Myc were fused in frame according to the method of Kunkel et al. (14) using a mutagenic primer (5Ј-TAGG-CGGGGCGTGGCTGTTTATTCGTCAAATGCAGGGCGGCGGTGG-3Ј). pSTD122 that carried the ftsHЈ-ЈenvZ hybrid gene was constructed similarly. An about 2-kb HpaI-NruI fragment of pAT2005S was ligated with pSTD113 that had been digested with SmaI and EcoRV, and the region encoding the membrane anchoring domain (from the amino terminus to the 120th amino acid residue) of FtsH and the region encoding the cytoplasmic domain (from the 180th amino acid residue to the carboxyl terminus) of EnvZ were fused in frame using a mutagenic primer (5Ј-TTGGTGTCTGGATCTTCTTCATGCGTATCCAGAACCGA-CCGTTGGT-3Ј). pTYE030 was constructed as follows. The envZ open reading frame was amplified by polymerase chain reaction with primers of 5Ј-GCTCTAGAATAAGGAGGCTCTAAAGCATGAGGC-3Ј and 5Ј-CGGGATCCCCCTTCTTTTGTCGTGCC-3Ј. The amplified fragment was subcloned into pBluescript SK(Ϫ) using an XbaI site and a BamHI site introduced by the polymerase chain reaction. While a central part of the insert (a 0.98-kb MunI-BglII fragment) was replaced by that of pAT2005S, the remaining part of it was confirmed by sequencing. Low copy number plasmids carrying envZЈ-ЈftsH (pSTD119), ftsH-envZ-his 6 myc (pSTD125), or envZ (pSTD124) were constructed by inserting a 2.2-kb XbaI-SmaI fragment of pSTD117, a 1.6-kb XbaI-EcoRI fragment of pSTD122, or a 1.7-kb XbaI-HpaI fragment of pAT2005S into the multicloning region of pMW119, respectively.
Fractionation of Membrane Proteins by Size-exclusion Chromatography-Cells of AD202 were grown to a mid-log phase in L medium, collected, and washed with buffer C (50 mM Hepes-KOH, pH 7.0, 50 mM KCl, 1 mM dithiothreitol, 20% glycerol) (15). Total membrane fraction was prepared by disruption of cells by sonication followed by ultracentrifugation essentially as described previously (9). Membranes were suspended in buffer C and solubilized with OG in the presence of E. coli phospholipids as described previously (16). After removal of insoluble materials by centrifugation, proteins were mixed with molecular-size standards (obtained from Bio-Rad), loaded to Superose 6 column, and developed with 50 mM, 150 mM NaCl, 1.25% OG, 10% glycerol. Proteins in each fraction were precipitated with 5% trichloroacetic acid, separated by 15% acrylamide, 0.12% N,NЈ-methylenebisacrylamide polyacrylamide gel electrophoresis (9) and subjected to immunoblotting with anti-FtsH (17) or anti-SecY (18).
Pulse-Chase Experiments and Immunoprecipitation of Denatured Proteins-Cells were grown in M9 medium supplemented with 18 amino acids (20 g/ml) other than Met and Cys, thiamine (2 g/ml), 0.4% glucose, and appropriate antibiotics. After 10 min of induction with isopropyl-1-thio-␤-D-galactopyranoside (1 mM) and cAMP (5 mM), cells were pulse-labeled for 30 s with about 0.37 MBq/ml [ 35 S]methionine followed by chase with 200 g/ml of nonradioactive L-methionine. 100 l of samples were removed at intervals and mixed with an equal volume of 10% trichloroacetic acid. Immunoprecipitation with anti-FtsH (17) or anti-c-Myc (Ab-1) (Oncogene Science, Inc) was carried out as described previously (7). Proteins were separated by 10% polyacrylamide gel electrophoresis (19).
Cross-linking of Membrane Proteins with DSP-Total membranes were prepared as above. For solubilization, total membranes were treated with OG as described above except that Tris was not included and that pH of Hepes-KOH was 7.5 instead of 7.0. For cross-linking, either total membranes or their OG extracts were treated with 0.25 mg/ml DSP, a membrane permeable cross-linker, at 4°C for 1 h, and the reaction was terminated by the addition of 0.2 M ammonium acetate followed by incubation at 4°C for 10 min. Control samples received 0.2 M ammonium acetate prior to the addition of DSP. Samples were adjusted to 1% SDS, incubated at 37°C for 10 min and subjected to immunoprecipitation as described above. Precipitated proteins were dissolved in SDS sample buffer (20) without 2-mercaptoethanol at 37°C for 10 min before electrophoresis. For cleavage of the cross-linker, 10% 2-mercaptoethanol was included in SDS sample buffer.
Trypsin Digestion and Immunoblotting-Cells were grown in peptone medium supplemented with appropriate antibiotics; rapidly chilled by mixing with NaN 3 (0.02%), chloramphenicol (100 g/ml), and a small piece of ice; and disrupted by lysozyme freezing-thawing (7). The cell lysates were treated with trypsin as described previously (7). Proteins were separated by 10% SDS-polyacrylamide gel electrophoresis and analyzed by immunoblotting with anti-PhoA (obtained from 5 Prime 3 3 Prime, Inc.), anti-FtsH or anti-EnvZ as described previously (7).

Size-exclusion Chromatography of the FtsH Protein-Our
previous findings that ftsH can be mutated to dominant negative with respect to the Std and Sec phenotypes (8) suggested that FtsH may function in multimeric states. To directly examine higher order structures of FtsH, we solubilized the cytoplasmic membrane with OG and subjected the solubilized proteins to size-exclusion chromatography using Superose 6. FtsH was eluted with a peak at fractions 45-47 that corresponded to a molecular mass of about 280 kDa (Fig. 1, A and B), while its monomeric molecular mass should be 71 kDa. Although the value of 280 kDa determined by the calibration using soluble proteins should not be regarded as accurate, FtsH was eluted far earlier than SecY, a major part of which was eluted at the position of about 50 kDa (Fig. 1, A and B). This form of SecY could either be a monomer (the molecular mass is 49 kDa) or in a form of SecY-SecE-SecG complex of estimated molecular mass of about 74 kDa.
Cross-linking of FtsH in Membranes and in Detergent Extracts-We addressed the subunit structure of FtsH by crosslinking experiments. The membranes prepared from wild-type cells were treated with DSP and analyzed by SDS-polyacrylamide gel electrophoresis and immunoblotting with anti-FtsH. Treatment with DSP yielded products with molecular masses of about 240 and 140 kDa ( Fig. 2A, lane 4) that were not observed without DSP treatment (lane 3) or after cleavage of the cross-linker with 2-mercaptoethanol (lane 2). Cross-linked products of FtsH were also generated when solubilized membrane proteins were treated with DSP (Fig. 2B, lane 4). Under the latter condition, however, the intensity of the 240-kDa species was much less than when intact membranes were cross-linked.
Cross-linking between the FtsH Proteins with and without an Epitope Tag-To dissect the putative FtsH complex, we constructed an FtsH derivative with two tandemly located molec-ular tags, oligohistidine residues (His 6 ), and a c-Myc-derived epitope at the C terminus (Fig. 3). The FtsH-His 6 -Myc protein can specifically be isolated, and detected by nickel-nitrilotriacetic acid-agarose and anti-Myc antibodies, respectively. Cells carrying pSTD101 (ftsH-his 6 -myc) were pulse-labeled, and labeled proteins were first solubilized in SDS and then precipi-tated with anti-FtsH or anti-Myc antibodies. 4 Anti-FtsH serum brought down two species of proteins (Fig. 4, lane 2). The upper band represented the tagged FtsH, since it was precipitated by anti-Myc antibodies as well (Fig. 2, lane 5). The lower band represented the normal FtsH, since it comigrated with the chromosomally encoded FtsH (lane 1) and did not cross-react with anti-Myc (lanes 4 and 5). FtsH-His 6 -Myc was stable in vivo; no degradation was observed during a 16-min chase period examined (lanes 2, 3, 5, and 6). The FtsH-His 6 -Myc protein was functional, since pSTD120 (a low copy plasmid carrying ftsH-his 6 -myc) complemented the temperature-sensitive ftsH1 mutation (20). It did not interfere with the cell growth. When pulse-labeled cells were disrupted by sonication and fractionated, most of FtsH-His 6 -Myc, like normal FtsH, was recovered in the membrane fraction (data not shown). We found that a fraction of FtsH-His 6 -Myc was cleaved in vitro by unknown proteases to a product (FtsHЈ) slightly smaller than the authentic FtsH during the process of membrane preparation (see Figs. 5 and 6). The cleavage seemed to occur around the junction between FtsH and the His 6 -Myc tag, since FtsHЈ lost the Myc epitope (Fig. 5B, lane 5).
Cells of CU141(FЈlacI q ) carrying both the ftsH-his 6 -myc plasmid (pSTD113) and the ftsH plasmid (pSTD401) were induced and pulse-labeled, and total membrane fractions were prepared. To minimize possible artificial effects resulting from overaccumulation of plasmid-encoded proteins, their synthesis was induced only for a short period (10 min) before pulse labeling in this and the following experiments. Membranes were treated with DSP, solubilized with SDS, and subjected to immunoprecipitation using anti-Myc or anti-FtsH antibodies. Samples were analyzed by SDS-polyacrylamide gel electrophoresis without (Fig. 5A) or following (Fig. 5B) cleavage of the cross-linker by 2-mercaptoethanol. Treatment of the membranes with DSP yielded high molecular weight cross-linked products that were immunoprecipitated with anti-FtsH (Fig.  5A, lane 1). Such cross-linked products were not detected when the cross-linker had been quenched by ammonium acetate (lane 2). When DSP was cleaved by 2-mercaptoethanol before electrophoresis, FtsH and FtsHЈ were recovered with anti-Myc antibodies (Fig. 5B, lane 1), whereas they were never recovered with anti-Myc without cross-linking (lane 3). The identities of FtsH and FtsHЈ were confirmed by recovery of these proteins by the second immunoprecipitation with anti-FtsH serum (Fig.  5C). These results suggested that more than two molecules of FtsH form a complex.
Coimmunoprecipitation of FtsH with FtsH-His 6 -Myc-We carried out immunoprecipitation under nondenaturing condi- tions (Fig. 6). Membrane fraction was prepared from FtsH-His 6 -Myc overproducing cells that had been pulse-labeled for 5 min and solubilized with OG, and proteins were immunoprecipitated with anti-Myc or anti-FtsH antibodies. Anti-FtsH precipitated the tagged FtsH, intact FtsH, and FtsHЈ (lane 1), whereas normal serum did not (lane 5). Anti-Myc antibodies also precipitated all of these proteins (lane 3). Inclusion of the FtsH peptide (lane 2) or the Myc peptide (lane 4) during immunoprecipitation abolished the precipitation of all of these proteins. When the anti-Myc-precipitates were dissociated with SDS and subjected to reaction with anti-FtsH serum, all three proteins were precipitated, confirming their identities (data not shown). In contrast, only FtsH-His 6 -Myc was recovered when the membranes were first solubilized in SDS and then subjected to immunoprecipitation with anti-Myc antibodies (see Fig. 5B, lane 5).
These results show that FtsH and FtsHЈ were co-precipitated with the epitope-tagged FtsH. No other proteins were appreciably co-precipitated with anti-Myc antibodies. FtsH and FtsHЈ were also co-purified with FtsH-His 6 -Myc by nickel-nitrilotriacetic acid-agarose affinity column chromatography. 5 Cross-linking (Fig. 2) and co-immunoprecipitation (Fig. 6) after solubilization preclude the possibility that the cross-linking of these proteins in the membrane was caused by artificial proximity resulting from their overaccumulation in the membrane.

Identification of the FtsH-FtsH Interaction Region Using
Chimeras between ftsH and envZ-We then examined the roles of the two regions, the membrane-associated N-terminal region and the cytoplasmic C-terminal region, in the FtsH-FtsH interaction. We previously showed that an N-terminal fragment of FtsH caused a dominant Std effect. Thus, the N-terminal region of FtsH may be important for the subunit interaction of FtsH. To examine this possibility, we constructed chimeric genes between ftsH-his 6 -myc and envZ. The EnvZ protein is an E. coli inner membrane protein with FtsH-like topology (21). We constructed two kinds of chimeric genes encoding FtsHЈ-ЈEnvZ and EnvZЈ-ЈFtsH-His 6 -Myc (Fig. 3). The FtsHЈ-ЈEnvZ chimeric protein consists of the FtsH-derived transmembrane domain and the EnvZ-derived cytoplasmic domain, whereas EnvZЈ-ЈFtsH-His 6 -Myc has the EnvZ membrane domain followed by the tagged FtsH cytoplasmic domain.
These chimeric genes did not complement the ftsH1 mutation, indicating that both the membrane-bound and the cytoplasmic regions of FtsH are important for the FtsH functions.
Cell fractionation experiments showed that these hybrid proteins are membrane-associated (data not shown).
We then examined whether the chimeric proteins cause a dominant Std phenotype (see Introduction). As the high level overexpression of these proteins from the plasmids used in the cross-linking experiments was found to be deleterious to cells, the fusion genes were recloned into a low copy number vector that is also compatible with the plasmid (pKY221) carrying the reporter secY-phoA C6 gene. Extracts of cells expressing either the chimeras, FtsH, or envZ, in addition to SecY-PhoA C6 fusion, were treated with trypsin and analyzed by immunoblotting with anti-PhoA antibodies (Fig. 7). The PhoA domain of SecY-PhoA C6 from the cells expressing FtsHЈ-ЈEnvZ resisted trypsin (Fig. 7A, lanes 7 and 8), indicating that it was exported to the periplasmic space. On the other hand, expression of the other three proteins, EnvZЈ-ЈFtsH-His 6 -Myc (lanes 1 and 2), FtsH (lanes 3 and 4), or EnvZ (lanes 5 and 6) did not cause the Std phenotype. All of the above proteins evidently accumulated in the cells as shown by Western blotting with anti-FtsH or anti-EnvZ (Fig. 7B). These results suggested that among the above proteins, only FtsHЈ-ЈEnvZ could interact with the chromosomally-encoded FtsH to interfere with its function.
Cells overexpressing FtsH and either FtsHЈ-ЈEnvZ, EnvZЈ-ЈFtsH-His 6 -Myc, or EnvZ were pulse-labeled, and membranes were treated with DSP. Cross-linked products were examined by immunoprecipitation. Fig. 8A shows results of an experiment with FtsHЈ-ЈEnvZ. The anti-FtsH serum used in this study had been directed against a sequence in the cytoplasmic domain of FtsH (17). Thus, without cross-linking, the FtsHЈ-ЈEnvZ protein was immunoprecipitated with anti-EnvZ serum but not with anti-FtsH serum (lanes 5 and 6). The anti-EnvZ antibodies did not cross-react with FtsH (lane 6). After cross-5 Y. Akiyama, unpublished results.

FIG. 4. Synthesis and stability of FtsH-His 6 -Myc.
Cells of AD21/ pSTD101 (ftsH-his 6 -myc) were grown in minimal medium and pulselabeled with [ 35 S]methionine for 30 s before (lanes 1 and 4) or after (lanes 2 and 5) a 10-min induction with 1 mM isopropyl-1-thio-␤-Dgalactopyranoside and 5 mM cAMP. After pulse labeling, induced cells were chased in the presence of unlabeled methionine for 16 min (lanes  3 and 6). Proteins were precipitated with trichloroacetic acid, subjected to immunoprecipitation with anti- FtsH (lanes 1-3) or anti-Myc (lanes 4 -6), and analyzed by SDS-polyacrylamide gel electrophoresis. Quenching of DSP before cross-linking abolished these crossreactions (lanes 3 and 4). On the other hand, EnvZЈ-ЈFtsH-His 6 -Myc was not cross-linked with FtsH, since FtsH was not precipitated with anti-Myc antibodies even after DSP treatment (Fig. 8B). As expected, no cross-linking was observed between FtsH and EnvZ (Fig. 8C). These results confirmed that FtsHЈ-ЈEnvZ can interact with FtsH but EnvZЈ-ЈFtsH-His 6 -Myc cannot. The interaction between FtsH molecules is likely to be mediated by its membrane-associated region.

DISCUSSION
FtsH has been implicated to have diverse cellular activities. We suggested previously that FtsH is involved in integration/ assembly of proteins through and/or into the membrane (7,8). It was also found recently that FtsH is involved in rapid degradation of at least three short-lived proteins, cII gene product of phage (22), the heat shock sigma factor, RpoH ( 32 ) (6,23), and uncomplexed forms of SecY (18). From an in vitro study using purified FtsH and RpoH, FtsH was suggested to have a proteolytic activity (6). Yta10, a mitochondrial inner membrane protein, which is closely related to FtsH, was also suggested to participate in degradation of abnormal proteins in the mitochondrial matrix space (24,25). How can these diverse apparent functions of FtsH be reconciled? The E. coli ClpA and ClpX proteins, regulatory subunits of the Clp protease and distantly related to FtsH, do not have any proteolytic activity themselves. Instead, they are proposed to target substrate proteins for ATP-dependent degradation (26 -28). It was also shown that ClpA functions as a molecular chaperone in replication of P1 plasmid or in in vitro protein folding reactions (28,29). Similarly, the AAA family includes some of regulatory ATPase subunits of proteasomes. They have been proposed to function in presentation of substrate proteins to the protease subunits, the process in which energy of ATP hydrolysis is somehow used (30). FtsH may be a multifunctional protein that exerts chaperone-like activities in the assembly or translocation of some cell surface proteins and degradation of some unstable proteins.
Oligomeric structure seems to be a common feature among the above mentioned ATPase subunits as well as some other members of the AAA family. For example, (N-ethylmaleimidesensitive factor) functions as a homotrimer that interacts with many other proteins including SNAPs and SNAREs during process of vesicular transport in eukaryotic cells (31). p97 has also been proposed to be a homohexamer, although its function is not known (5). We have shown here that FtsH is in a complex that includes more than one molecules of FtsH. FtsH remains in high molecular mass state after solubilization in nonionic detergent. The solubilized FtsH could be cross-linked to form oligomeric structure and could be co-immunoprecipitated with the epitope-tagged version of FtsH. It is possible that the FtsH molecules are directly interacting with themselves.  9 and 10 of A and lane 5 of B) were grown in peptone medium containing 1 mM isopropyl-1-thio-␤-D-galactopyranoside and appropriate antibiotics. A, cells were disrupted by lysozyme freezing-thawing and treated with 50 g/ml trypsin as indicated. After separation by 10% polyacrylamide gel electrophoresis, proteins were visualized by anti-PhoA immunoblotting. PhoA* indicates the trypsin-resistant PhoA moiety that is expected if it is exported to the periplasmic space. B, cultures were directly mixed with trichloroacetic acid, and total proteins were separated by 10% polyacrylamide gel and visualized by immunoblotting using antisera against FtsH (upper part) or EnvZ (lower part).
The FtsHЈ-ЈEnvZ chimeric protein is cross-linkable with FtsH and causes dominant Std phenotype. We suggest that the dominant phenotype is at least partly a result of the formation of a nonfunctional FtsH complex containing wild-type and mutant molecules. The results with the hybrid proteins suggested that possible interaction between the FtsH molecules is mediated by direct association of their transmembrane regions. Several examples have been reported for inter-or intramolecular association of transmembrane segments (32)(33)(34). The ftsH101 mutation causes a change of Val 32 to Met in the periplasmic region of FtsH (7). It did not affect the interaction between the FtsH molecules, 5 implicating that the membrane domain is not only important for the oligomerization but may itself have some role in the FtsH functions.
It is not known how many FtsH molecules are present in the FtsH complex and whether any other proteins are associated with it. The major cross-linked products of 140 and 240 kDa might represent dimer and tetramer of FtsH. In addition, no other major proteins were found in the preparation of FtsH that was purified from overproducing strains (6). These results, however, do not exclude the possibility that the physiological complex of FtsH contains additional components. Preliminarily, two proteins of 27 and 16 kDa were found to co-immunoprecipitate with anti-FtsH antibodies. 5 The 27-kDa protein was co-immunoprecipitated even after treatment of the membrane with urea.
Elucidation of the complete structure of the FtsH complex awaits purification of the physiological complex from wild-type cells. The present results showing that FtsH molecules can associate with each other even when they are exclusively overproduced (Figs. 5 and 6) will provide an important guidance for further biochemical characterization of this intriguing membrane protein.