Dissecting the Role of a Conserved Motif (the Second Region of Homology) in the AAA Family of ATPases

Escherichia coli FtsH is an ATP-dependent protease that belongs to the AAA protein family. The second region of homology (SRH) is a highly conserved motif among AAA family members and distinguishes these proteins in part from the wider family of Walker-type ATPases. Despite its conservation across the AAA family of proteins, very little is known concerning the function of the SRH. To address this question, we introduced point mutations systematically into the SRH of FtsH and studied the activities of the mutant proteins. Highly conserved amino acid residues within the SRH were found to be critical for the function of FtsH, with mutations at these positions leading to decreased or abolished ATPase activity. The effects of the mutations on the protease activity of FtsH correlated strikingly with their effects on the ATPase activity. The ATPase-deficient SRH mutants underwent an ATP-induced conformational change similar to wild type FtsH, suggesting an important role for the SRH in ATP hydrolysis but not ATP binding. Analysis of the data in the light of the crystal structure of the hexamerization domain ofN-ethylmaleimide-sensitive fusion protein suggests a plausible mechanism of ATP hydrolysis by the AAA ATPases, which invokes an intermolecular catalytic role for the SRH.

The AAA protein family (for ATPases associated with diverse cellular activities) is a distinct subfamily of the Walker-type ATPases that has been defined on the basis of amino acid sequence homology (for reviews see Refs. [1][2][3]. Walker-type ATPases have two consensus motifs, Walker A and B. In addition to these consensus motifs, AAA proteins have another highly conserved amino acid sequence within their ATPase domain that has been termed the second region of homology (SRH) 1 (4). A 200 -250-amino acid residue sequence that en-compasses the Walker A and B motifs and the SRH is named the AAA module, and proteins that contain one or two copies of this module are classified into the AAA protein family. Thus, the SRH is the defining feature that distinguishes the AAA family from other Walker-type ATPases.
FtsH of Escherichia coli is the first identified prokaryotic member of the AAA family (5,6). FtsH is a membrane-bound ATP-dependent protease with two N-terminal transmembrane segments ( Fig. 1) (7,8). It possesses one copy of the AAA module and at its C terminus possesses a Zn 2ϩ -binding motif, which is thought to be the catalytic center for proteolysis. Several substrates of FtsH have been identified, including cytoplasmic proteins, such as the heat shock transcription factor 32 (8,9), LpxC (10), SsrA-tagged proteins (11), CII (12), CIII (13), and Xis (14), as well as integral membrane proteins such as SecY (15,16), subunit a of the proton ATPase F O sector (17), and YccA (18). Although FtsH is essential for cell growth (5,19), cells lacking ftsH can grow if the sfhC mutation coexists (10,20). Interestingly, several findings have also raised the possibility that FtsH has a chaperone-like activity. Some ftsH mutations cause abnormal protein translocation and defects in protein export. In particular, it has been shown that the PhoA moiety of SecY-PhoA fusions, in which PhoA is attached to the C-terminal cytoplasmic region of SecY, is abnormally translocated (Std phenotype) in ftsH mutant strains (19,21). These phenotypes are differentially suppressed by overproduction of other chaperones (22). The data also indicated that FtsH binds to denatured but not intact PhoA polypeptide and that this binding does not result in proteolysis of PhoA (23).
The AAA family has been growing rapidly for the past several years with more than 200 family members identified to date. Members of this family are spread among various species in all kingdoms, archaea, bacteria, and eukaryotes and participate in various cellular activities. The AAA family can be divided into at least six subfamilies: metalloproteases including FtsH, subunits of the 26 S proteasome, proteins involved in vesicle-mediated secretion, homotypic fusion, peroxisome biogenesis, and meiosis/mitochondrial functions. No clue to the common function of AAA proteins, besides ATP binding and/or hydrolysis, has yet emerged.
There appears to be a confusing contrast between the extreme sequence conservation within the module and the diversity of function of the AAA proteins. There must be a very important reason for this high conservation of sequence both among proteins with a common function in evolutionarily distant species, as well as among AAA proteins with a variety of functions within the cell of a given species. Elucidation of the basic function of the module represents an exciting challenge, and it must provide a clue to understanding the common function of the AAA proteins.
Because the SRH is a part of the ATPase domain, it is assumed to have some relation to the ATPase activity. However, there have as yet been no reports of an experimental investigation into the function of the SRH. To elucidate SRH function, we introduced a series of mutations into the SRH sequence of E. coli FtsH and studied their consequences both in vivo and in vitro. The results indicate that the SRH is important for the ATPase activity.
Plasmids-pUC118 carries the gene for ampicillin resistance (amp R ). pET29b is a plasmid vector of the pET system carrying the T7 promoter (Novagen). pKY297 carries a truncated ftsH gene that lacks the sequence encoding the two transmembrane segments and the Walker motifs (21). pSTD401 carries the wild type ftsH gene with its own Shine-Dalgarno sequence under the control of the lac promoter (19). pSTD41 carries the K201N mutant ftsH gene (21). pQM21 carries the H421Y mutant ftsH gene (26). pIFH100 is the plasmid vector used in this study and was constructed as follows. The NdeI-BamHI fragment of pET29b was deleted to eliminate the S⅐Tag sequence, and the NsiI short fragment of the resulting plasmid was replaced with the BspHI amp R -containing fragment of pUC118 by blunt end ligation after treatment with Klenow fragment. pIFH108 is the wild type FtsH expressing plasmid constructed by inserting the EcoRI-HindIII ftsH fragment of pSTD401 into pIFH100. We checked the whole sequence of the cloned ftsH gene on pIFH108 and found a silent mutation of Ala 163 (GCA to GCG) in the ftsH coding region.
Construction of Mutant ftsH Plasmids-All mutant ftsH plasmids were constructed by replacing a segment of ftsH on pIFH108 with a mutated fragment. The numbering of amino acid residues of FtsH was according to Wang et al. (27). The mutants are as follows: 1) K201N (AAA to AAC): the KpnI-PstI segment of pIFH108 was replaced with the corresponding segment of pSTD41. 2) V296A (GTT to GCA), T300A (ACT to GCT), N301A (AAC to GCA), P303A (CCG to GCG), D304A (GAC to GCC), D307A (GAC to GCG), D307N (GAC to AAC), D307E (GAC to GAG), L310A (CTG to GCG), R312A (CGT to GCT), R312L (CGT to CTT), R312K (CGT to AAA), G314A (GGC to GCA), R315A (CGT to GCA), R315L (CGT to CTT), and R315K (CGT to AAA): mutagenized ftsH fragments were produced by sequential polymerase chain reaction steps using synthetic primers (28). The KpnI-NcoI segment of pIFH108 was replaced with each mutated fragment. 3) E418Q (GAA to CAG): the mutated fragment was produced by sequential polymerase chain reaction steps. The fragment digested with MluI and Bsu36I was cloned into pKY297. The KpnI-PstI fragment was then subcloned into pIFH108. 4) H421Y (CAT to TAT): the KpnI-PstI segment of pQM21 was subcloned into pIFH108. We checked the whole sequence of ftsH to confirm that no other mutations had been introduced adventitiously.
Purification of FtsH-AR5088 cells carrying pIFH108 were grown in 2 liters of L medium containing ampicillin (100 g/ml) at 37°C to a cell density corresponding to 50 Klett units. At this point, expression of FtsH was induced by the addition of IPTG (1 mM) followed by growth for a further 3 h.
Pooled Superose 6 fractions were applied to a MonoP fractionation column with buffers A and B. The peak fraction of FtsH was dialyzed against dialysis buffer (20 mM monoethanolamine HCl, pH 9.0, 5 mM magnesium acetate, 10% glycerol, 0.5% Nonidet P-40) and used for in vitro assays.
Assay for in Vitro Degradation of 32 -The proteolytic activity of purified wild type and mutant FtsH proteins was assayed as described (8).
ATP-induced Conformational Change of FtsH-Conformational changes of FtsH induced by ATP were analyzed according to the procedures described by Akiyama et al. (23). The reaction was started by the addition of 2.5 g/ml of trypsin. Samples were removed at intervals, and analyzed by 10% SDS-polyacrylamide gel electrophoresis (PAGE) followed by immunoblotting using the anti-FtsH serum and detected by ECL (Amersham Pharmacia Biotech).
Western Blotting-1 ml of culture (Klett unit ϭ 50) was harvested by centrifugation. Pelleted cells were suspended in 100 l of the SDS loading buffer. 20 l of the samples were analyzed by SDS-PAGE followed by immunoblotting using anti-FtsH or anti-32 serum (29) and detected by ECL.

Development of a System That Facilitates Both in Vivo Functional Analysis and Purification of Mutant FtsH Proteins-To
examine the activity of mutant FtsH proteins in vivo and to purify them without contamination from wild type FtsH, it is necessary to use a strain lacking the ftsH gene as a host. Recently, we have identified a suppressor mutation, sfhC, that allows cells lacking ftsH to survive (10,20). To take advantage of the efficient pET expression vectors that utilize the T7 promoter, we constructed a derivative of BL21(DE3) carrying ⌬ftsH and sfhC, AR5088, and cloned the ftsH gene on a derivative of pET29b, pIFH108.
In this system, intact FtsH protein is expressed from pIFH108, and its expression is inducible by IPTG. We found that this system has an additional merit. In the absence of IPTG, expression of FtsH from pIFH108 in AR5088 was leaky such that levels of FtsH comparable with that derived from the wild type chromosomal ftsH gene are produced (see Fig. 2, A and B, lanes 1, 2, and 5). This property allows the in vivo activity of mutant FtsH proteins in AR5088 to be assayed.
In principle, FtsH cannot be induced from pIFH108 in host strains other than DE3 lysogens owing to the absence of T7 RNA polymerase. Indeed as expected, we observed negligible expression of FtsH from pIFH108 in such strains in the absence of IPTG. However, somewhat surprisingly, in the presence of IPTG, the level of expression of FtsH from pIFH108 was comparable with that from the chromosomal ftsH gene (data not shown). Although the mechanism underlying this observation is not understood, this property was exploited in the complementation experiments with the temperature-sensitive mutant strain (see Fig. 3).

Construction of SRH Mutant Plasmids by Site-directed Mutagenesis and Their
Protease Activity in Vivo-Several amino acid residues such as Thr 300 , Asn 301 , Asp 307 , Ala 309 , Arg 312 , and Arg 315 in the SRH of FtsH are very highly conserved among the AAA family proteins (Fig. 1B), whereas others are less well conserved. Changing each residue to Ala, in what is termed "alanine scanning," is a useful strategy for identifying functionally important residues in proteins (30). To determine whether or not the SRH is important for FtsH function and, if so, which amino acid residues in the SRH are important, we introduced Ala substitutions into some conserved and less conserved residues in the SRH.
Plasmids expressing wild type or mutant FtsH were introduced into AR5088, and the levels of 32 , which is one of the known specific substrates of the FtsH protease (8,9), were analyzed by Western blotting (Fig. 2A). In untransformed AR5088 (⌬ftsH) or AR5088 harboring the vector pIFH100, 32 accumulated ( Fig. 2A, lanes 3 and 4). When wild type FtsH was expressed from pIFH108, the accumulation of 32 was not observed (Fig. 2A, lane 5). In the case of mutants of less conserved residues such as V296A, P303A, D304A, and G314A, the accumulation of 32 was not observed ( Fig. 2A, lanes 6, 9, 10,  and 14), indicating that these mutants have almost the same activity as wild type FtsH and that these less conserved residues are not important for the in vivo protease activity of FtsH. On the other hand, with the mutants of the highly conserved residues N301A, D307A, R312A, and R315A, 32 accumulated to the same levels as those in AR5088 harboring the vector ( Fig.  2A, lanes 8, 11, 13, and 15), indicating that these highly conserved residues are essential for the in vivo protease activity of FtsH. In cells expressing the T300A or L310A mutants, the accumulation of 32 was partially suppressed ( Fig. 2A, lanes 7  and 12), suggesting that these mutants have a low protease activity. This implies that Thr 300 and Leu 310 are less important. Overall, these results indicate that the SRH is important for FtsH function and that there is good correlation between the functional importance and the extent of conservation of the amino acid residues in the SRH.
Because highly conserved residues such as Asp 307 , Arg 312 , and Arg 315 were found to be important for FtsH function by alanine scanning mutagenesis, we constructed further mutants that introduce less drastic substitutions at these positions. Specifically, we changed Asp307 (negatively charged) to Asn (polar but uncharged) or Glu (similarly negatively charged) or Arg 312 and Arg 315 (positively charged) to Leu (bulky and uncharged) or Lys (similarly positively charged). The mutants D307N, R312L, R312K, R315L, and R315K showed no detectable protease activity (Fig. 2B, lanes 7-12). The mutant D307E exhibited a low but nevertheless significant protease activity similar to T300A and L310A. Thus, the negative charge of the Asp 307 side chain seems to be important for FtsH activity. On the other hand, the two conserved Arg residues, Arg 312 and Arg 315 , could not be functionally substituted by Lys even though the positive charge is retained with this substitution.
For comparison, we constructed additional mutants that have mutations outside the SRH. The K201N substitution resides in the Walker A motif, which is believed to be essential for ATPase activity, whereas the E418Q and H421Y mutations are in the Zn 2ϩ -binding motif believed to form the active site of the protease domain. As expected, the mutants E418Q and H421Y had no protease activity (Fig. 2B, lanes 13 and 14). The mutant K201N also showed no detectable protease activity (Fig. 2B,  lane 6), confirming the notion that the protease activity is dependent on the ATPase activity.
Complementation of the Temperature-sensitive ftsH1 Muta- The SRH is defined according to Lenzen et al. (31). The numbers above the sequence indicate the amino acid residues in FtsH and are based on the revised sequence reported by Wang et al. (27). The letters under the sequence indicate the amino acid substitutions constructed in this study. B, amino acid conservation in the SRH. Sequences of the SRH of 54 representative AAA proteins, which are listed in the data base BLOCKS (BL00674D), were analyzed. Conserved amino acids (Ͼ50%) are indicated by filled bars. Bars shorter than 100% are due to the lack of amino acid residues at the corresponding positions in some AAA modules. The numbering is again that of FtsH. The predicted secondary structure elements, based on the crystal structure of D2 of NSF, are underlined (31,32,43).

FIG. 2. In vivo protease activity of wild type and mutant FtsH proteins.
A, Western blotting with anti-FtsH and anti-32 antisera. vector and wild-type represent pIFH100 and pIFH108, respectively. Lanes 6 -15 represent alanine scanning mutants of pIFH108. Strains AR5074 (sfhC) and AR5088 (sfhC ⌬ftsH) are derivatives of BL21(DE3). AR5088 was used as the host for all of the plasmids (lanes 4 -15). Cells were cultured in the absence of IPTG at 37°C. B, same as in A except that lanes 6 -14 represent additional mutants of pIFH108. The amounts of three mutant FtsH proteins carrying substitutions at position 307 (D307A, D307N, and D307E) seem to be significantly lower than those of the others. However, this is due to poor recognition of these mutants by the anti-FtsH serum. 2 Note that the anti-FtsH serum used was raised against the synthetic peptide TNRPDVLDPALLRPGR corresponding to residues 300 -315 of the SRH of FtsH (7). tion-To determine whether or not SRH mutants retain the essential functions of FtsH for cell growth, the ability of the mutants to complement the temperature-sensitive ftsH1 mutation was examined. Wild type and mutant pIFH108 plasmids were introduced into the ftsH1 strain AR754, and the growth of the transformants was examined at 42°C in the presence of IPTG (Fig. 3). Under these conditions, wild type and mutant plasmid encoded FtsH proteins are expressed at levels comparable with that from the chromosomal ftsH gene (data not shown). Wild type FtsH and the mutants T300A and D307E, which exhibit a low in vivo protease activity, complemented ftsH1, whereas the other SRH mutants lacking the protease activity did not. Because the mutants K201N and H421Y also failed to complement ftsH1, it is clear that the ability of FtsH to complement ftsH1 is related to the in vivo protease activity.
In Vitro Protease Activity-To examine the activities of these mutants in vitro, we purified wild type FtsH and the mutants K201N, T300A, D307N, D307E, R312L, R315L, and H421Y to near homogeneity. All of the mutant proteins purified in this study behaved as high molecular mass complexes upon Superose 6 size fractionation (see "Experimental Procedures") similar to the wild type protein (data not shown), indicating that their oligomeric states are not significantly affected by the mutations. Using these purified FtsH proteins, we tested the in vitro protease activity toward the 32 substrate (Fig. 4). ATPdependent degradation of C-terminally histidine-tagged 32 was observed only with wild type FtsH. Significant proteolysis was not observed with any of the mutant FtsH proteins. The low proteolytic activities of the mutants T300A and D307E observed in vivo were not detected in the less sensitive in vitro assay used. Overall, these in vitro results seem to be consistent with in vivo results described above.
In Vitro NTPase Activity-We measured the NTPase activities of the purified wild type and mutant FtsH proteins (Table  I). Among the four nucleotides, FtsH hydrolyzed ATP and CTP efficiently. This is consistent with the observation that CTP, but not GTP or UTP, can substitute for ATP in the degradation of 32 and SecY (8,16). Interestingly, however, FtsH also hydrolyzed GTP and UTP at appreciable rates. The reason that GTP and UTP do not support the proteolysis reaction is unknown.
The specific activity of wild type FtsH in ATP hydrolysis is ϳ459 nmol/min/mg. As expected, the mutant K201N showed no detectable ATPase activity. This also excludes the possibility of significant contamination of any other ATPases in FtsH preparations purified by the procedures used. Among the SRH mutants, T300A and D307E, which both have low protease activity, have low specific ATPase activities of ϳ70 and ϳ40 nmol/min/mg, respectively. The other SRH mutants D307N, R312L, and R315L, which have lost the protease activity completely, showed no detectable ATPase activity. Thus, it seems that the SRH is important for both the ATPase and the protease activity of FtsH, presumably because these two activities are tightly coupled.
Interestingly, the ATPase activity of the mutant H421Y decreased to about a quarter of that of wild type FtsH (ϳ107 nmol/min/mg), even though this mutation is in the Zn 2ϩ -binding motif essential for the protease activity. The H421Y mutant did not rescue the lethality of the ftsH1 mutation at the nonpermissive temperature (Fig. 3), indicating that the protease activity of FtsH is essential for cell growth regardless of its ATPase activity.
ATP-induced Conformational Change-As some SRH mutants showed defects in nucleotide hydrolysis, it is important to establish whether or not they are defective in nucleotide binding. We have tried unsuccessfully to demonstrate ATP binding by FtsH in gel filtration assays, ATP agarose chromatography assays, and UV cross-linking assays employing radiolabeled ATP. 2 We therefore followed the procedures developed by Akiyama et al. (23), which allowed us to detect conformational changes in FtsH induced by ATP binding.
As shown in Fig. 5A, wild type FtsH undergoes a conformational change in the presence of ATP, which is characterized by partial protection of an ϳ33-kDa ATPase domain from trypsin digestion. This conformational change is not detected for the Walker motif mutant K201N, which does not bind ATP (Fig.  5B). We examined three representative SRH mutants, D307N, R312L, and R315L, which showed no detectable ATPase activity as described above, for the ATP-induced conformational change. As shown in Fig. 5 (C-E), the ϳ33-kDa fragment of all 2 K. Karata, and T. Ogura, unpublished results. FIG. 3. Complementation of ftsH1(Ts) by wild type and mutant FtsH proteins. AR754 is a temperature-sensitive mutant of ftsH (24). vector and wild-type represent AR754 harboring pIFH100 and pIFH108, respectively. K201N, T300A, D307N, D307E, R312L, R312K, R315L, R315K, and H421Y indicate AR754 harboring pIFH108 carrying the specified mutations. Cells were incubated on L agar plates with 1 mM IPTG at 42°C.  three mutants was protected from proteolytic degradation by the presence of ATP to a similar extent to that seen for wild type FtsH. Thus, it is most likely that mutations in the SRH do not significantly affect the ability of FtsH to bind ATP, indicating that the SRH plays a catalytic rather than simply a binding role in ATP hydrolysis. There is no obvious difference in the pattern of fragments generated by trypsin digestion of wild type and mutant FtsH proteins, suggesting that the amino acid alterations in these mutants do not significantly affect the overall structure of FtsH.

DISCUSSION
In this paper, we have investigated the function of the SRH in FtsH. The results indicate that the highly conserved residues in the SRH are important for its ATPase activity. Among the pool of SRH mutants examined, there is clear correlation between the ATPase and protease activities. It is most likely that mutations in the SRH affect the ATPase activity directly with the effects on the protease activity of FtsH being an indirect consequence of the tight coupling of the two activities. The consequences of mutations in the SRH match those seen for Walker motif mutations such as K201N.
Crystal Structure of the AAA Module-Recently, the crystal structure of the hexamerization domain of N-ethylmaleimidesensitive fusion protein (NSF-D2) in complex with ATP or its analog AMPPNP has been solved, providing the first high resolution structural data on an AAA module (31,32).
NSF contains two AAA modules, D1 and D2, that are different in function and sequence. D1, whose sequence closely matches the consensus sequence for the AAA module, catalyzes ATP hydrolysis (33), whereas D2, whose sequence is only moderately homologous to the consensus sequence of the AAA module, binds ATP and mediates hexamerization of NSF but displays no significant ATPase activity (33)(34)(35)(36). Thus, although the structure of D2 reveals the general fold of the AAA module and the determinants of nucleotide binding and hexamerization, the precise mechanism of ATP hydrolysis and ATP-induced conformational change have yet to be elucidated. To interpret the data from the site-directed mutagenesis experiments on the SRH of FtsH, we have aligned the sequence of the AAA module of FtsH with those of other AAA modules including D2 and examined these alignments in the context of the crystal structure of NSF-D2 (Fig. 6).
Functions of the SRH in ATP Hydrolysis-The SRH corresponds to the segment connecting strands ␤4 and ␤5 of the ␤-sheet in the N-terminal nucleotide-binding domain of NSF-D2 (Fig. 6, A, B, and E). There is a conserved Asn or Ser residue just downstream of strand ␤4, and the Ser residue in the D2 crystal structure interacts with a water molecule that itself interacts with the ␥-phosphate of ATP ( Fig. 6C and Refs. 31 and 32). This conserved Asn/Ser may be involved in hydrolysis of ATP in FtsH and other AAA domains that possess ATPase activity. The observation that the Asn residue at this position in FtsH, Asn 301 , is important for the protease activity ( Fig. 2A) is consistent with this hypothesis. However, the rest of the SRH is remote from the ATP-binding pocket and the main body of the SRH cannot interact with ATP directly (Fig. 6,  A and B). Our results indicate that the highly conserved residues in this region, Asp 307 , Arg 312 , and Arg 315 , are important for the ATPase activity but not for ATP binding (Table I and Fig. 5).
NSF forms a hexamer (31,32,37). FtsH also forms a ringlike homo-oligomer similar to the NSF hexamer (12). Indeed, such structures may be a general feature of AAA proteins, because several other AAA proteins also form similar hexameric rings (38 -43). The 26 S proteasome contains six distinct but homologous AAA ATPases, and they are included in the base of the regulatory particle of the proteasome (44). Thus, it is reasonable to assume that they form a hetero-hexameric ring similar to the homo-hexamer of NSF-D2.
The SRH of NSF-D2 forms a part of the interface between neighboring protomers in the hexamer (31,32). It seems likely therefore that the SRH will be important for protomer-protomer interaction in FtsH. This type of intersubunit interaction may be required for its ATPase activity and indeed accumulating evidence suggests that the ATPase activity of several AAA proteins including NSF (34,35), p97 (39), and Vps4p (45), is greatly enhanced by oligomer formation. The N-terminal transmembrane and periplasmic domains of FtsH are important for its stable oligomerization (46,47). The deletion of the periplasmic domain lowers the ATPase activity of FtsH 5-fold (47). We have also observed very low or almost no ATPase activity with truncated forms of FtsH corresponding to the ATPase domain. Because these fragments behave as monomers, 3 it is likely that the function of the SRH in ATP hydrolysis is dependent on the oligomeric state of FtsH.
The sequence conservation in the SRH of D2 is low as shown in Fig. 6E. The SRH sequence of D2 is 653 TTSRKDVLQEMEM-LNAFS 670 , whereas the consensus SRH sequence in the AAA family proteins is X(T/S)(N/S)XXXXXDXAXXRXXRX(D/E) (see also Fig. 1B). Only the first two of the highly conserved "signature" residues are conserved in D2. The alignments presented in Fig. 6E suggest that Arg 315 of FtsH corresponds to 3 T. Inagawa, K. Karata, and T. Ogura, unpublished results. Ala 668 of NSF-D2. In a simple modeling experiment, this alanine residue was replaced with an arginine residue. The possible conformations of the arginine side chain were then ex-plored to investigate whether Arg 315 could interact with the ATP ligand and account for its key role in the ATPase activity. Firstly the preferred arginine side chain rotamers were ex-FIG. 6. Modeling of the AAA module. A and B, orthogonal views of the main chain topology and the bound AMPPNP ligand in the crystal structure of the NSF-D2 domain. One molecule of the hexamer is shown. The adenine nucleotide (ball-and-stick) sits in the cavity between a classical ␣/␤ nucleotide-binding domain and a distal helical domain. The mode of binding of the triphosphate species of the ligand is similar to that in other proteins that contain the Walker motifs A (blue) and B (green). The A motif gives rise to the P-loop, which forms a series of contacts to the ␤ and ␥ phosphate groups. The sequence making up the SRH motif is red. C, contacts made by the SRH of NSF-D2 with AMPPNP. AMPPNP is bound to the protomer whose SRH (SRH-1) is red. A second SRH (SRH-2), which is blue, is that of a neighboring protomer. Sw I represents switch I. Wat and Mg 2ϩ represent a water molecule and a Mg 2ϩ ion bound to the ␤ and ␥ phosphates of AMPPNP, respectively. D, modeling of the contacts of the SRH of FtsH with AMPPNP. Ser 655 , Ala 668 , and Ser 670 of NSF-D2 were replaced by the corresponding Asn 301 , Arg 315 , and Asp 317 of FtsH, respectively (see the sequence alignment shown in E). Arg finger? represents a putative arginine finger of the SRH (SRH-2) of the neighboring protomer (for detail see "Discussion"). The replacements were made using simple modeling procedures implemented in the molecular graphics program QUANTA. Panels A-D were produced using the program MOLSCRIPT (56). E, sequence alignment of the AAA module. Sequences of the AAA module of several AAA proteins were aligned using the program CLUSTAL W (57). NSF-D2 and NSF-D1 represent the D2 and D1 domains of Chinese hamster ovary NSF (58). FtsH represents E. coli FtsH (6,27). Cdc48p, Sug1p, and Vps4p represent Saccharomyces cerevisiae Cdc48p (59), Sug1p (4), and Vps4p (60), respectively. The secondary structure is assigned on the basis of the NSF-D2 crystal structure (31,32). plored. A promising conformation was chosen and small manual adjustments to the side chain torsion angles subsequently brought the guanidino moiety of the arginine into close proximity to the ␥-phosphate group of the AMPPNP ligand that is bound by the neighboring subunit in the hexamer (Fig. 6D). This suggests the possibility of an intersubunit catalytic function. This situation would be reminiscent of the interactions of the so-called "arginine fingers" of GTPase-activating proteins with their respective small GTP-binding proteins Ras and Rho. The arginine finger greatly enhances GTP hydrolysis by stabilizing developing charge formed in the transition state of the reaction (for a review see Ref. 48). The effects of the R315L mutation reported here indicate that Arg 315 could fulfill this role in FtsH. The possibility that the Arg residue corresponding to Arg 315 interacts with the ␥-phosphate group of ATP bound to the neighboring subunit has independently been proposed for the AAA and related ATPases, collectively designated the AAA ϩ family, simply from multiple sequence alignment and structural analysis (49,50). Our present data have provided the first experimental evidence to support these consistent models and have shown that the Arg residue is involved in ATP hydrolysis.
There are some monomeric AAA proteins that exhibit low but significant ATPase activity. This is not necessarily inconsistent with the hypothesis discussed above, because these proteins may form oligomers only transiently under the ATPase assay conditions. Vps4p undergoes concentration-dependent oligomerization, which is accompanied by a dramatic increase in ATPase activity (45).
The model depicted in Fig. 6D does not explain the roles of the other highly conserved residues such as Asp 307 and Arg 312 in the SRH of FtsH. Further work is needed to account for their strong conservation and their crucial role in the ATPase activity.
There has been only one report concerning SRH mutations to date. Shirahama et al. (51) isolated yeast mutants that induce autophagy even in the presence of nutrients and found that one class of mutants has a mutation in the vps4/csc1 gene. The mutation was identified as E291K, which corresponds to Asp 317 of FtsH (Fig. 6E). Complete loss of the vps4/csc1 gene does not induce autophagy in rich media, and an additional mutation (K179A) in Walker motif A also abolished the mutant phenotype. These results suggest a gain-of-function allele in the mutant. The mutant protein might be defective in ATP hydrolysis but not in ATP binding in a similar manner to the SRH mutants of FtsH.
Additional Functions of the SRH-In small G proteins such as Ras and Rho and motor proteins such as myosins and kinesins, there are regions referred to as "switches" (for a review see Ref. 52). These regions undergo dramatic conformational changes upon GTP or ATP hydrolysis and have been postulated to transmit signals through interactions with other proteins. The position of one of these switches (switch I) corresponds to the SRH of the AAA module (Fig. 6, C and D). The ATPase activity of several helicases has been demonstrated to be greatly enhanced by DNA or RNA binding. In the case of DEXX box helicases, motif III, whose position in the crystal structure also corresponds to the SRH of the AAA module, has been implicated as one of the sites for DNA binding (53). In this respect, it should be noted that a proteasomal AAA ATPase, Sug1p, has been shown to have a 3Ј-5Ј DNA helicase activity (54). It has also been demonstrated that its ATPase activity is stimulated by specific RNAs (55). We speculate that the SRH of the AAA module functions similarly as a switch (Fig. 6, C and  D). In FtsH, the SRH might interact with the protease domain. The low observed ATPase activity of the mutant H421Y (Table   I) might be due to an alteration in the interaction between the AAA module and the protease domain mediated through the SRH. This interaction may be crucial in the coupling of the ATPase and protease activities. The SRH of Vps4p/Csc1p might interact with a molecule essential for autophagy, and the gain-of-function mutation E291K might affect this interaction. To elucidate the mechanism of ATP hydrolysis in AAA proteins and in particular to determine the role of the SRH in this function, it will be necessary to extend crystallographic investigations and determine the structures of ATP-and ADP-bound forms of an enzymatically active representative of the family.