ATP Hydrolysis Is Essential for the Function of the Uup ATP-binding Cassette ATPase in Precise Excision of Transposons*

In Escherichia coli K-12, the RecA- and transposase-independent precise excision of transposons is thought to be mediated by the slippage of the DNA polymerase between the two short direct repeats that flank the transposon. Inactivation of the uup gene, encoding an ATP-binding cassette (ABC) ATPase, led to an important increase in the frequency of precise excision of transposons Tn10 and Tn5 and a defective growth of bacteriophage Mu. To provide insight into the mechanism of Uup in transposon excision, we purified this protein, and we demonstrated that it is a cytosolic ABC protein. Purified recombinant Uup binds and hydrolyzes ATP and undergoes a large conformational change in the presence of this nucleotide. This change affects a carboxyl-terminal domain of the protein that displays predicted structural homology with the socalled little finger domain of Y family DNA polymerases. In these enzymes, this domain is involved in DNA binding and in the processivity of replication. We show that Uup binds to DNA and that this binding is in part dependent on its carboxyl-terminal domain. Analysis of Walker motif B mutants suggests that ATP hydrolysis at the two ABC domains is strictly coordinated and is essential for the function of Uup in vivo.

mutations in topA (toposimerase I), dnaA (DNA polymerase I), and dnaBE (helicase and ␣-subunits of DNA polymerase III) also lead to the same phenotype (8,9). All of these mutations affect genes encoding proteins involved in DNA replication or repair, supporting the notion that deletions are the result of errors during replication (1).
Other tex mutations affect the uup gene and cause an increase in the frequency of precise excision of transposons Tn10, mini-Tn10, and Tn5 and a defective growth of bacteriophage Mu (10). Molecular cloning and nucleotide sequence determination of the uup gene suggested that the Uup protein is cytosolic and belongs to the superfamily of ATP-binding cassette (ABC) 3 proteins (7,9). tex mutations affecting transposon precise excision fall into two categories: the first, such as uup, ssb, topA, and polA, increase precise excision of both Tn10 and mini-Tn10; and the second, such as mutHLS, dam, and uvrD, increase precise excision of Tn10 but not of mini-Tn10. The role of Uup in DNA metabolism was never investigated.
ABC proteins constitute one of the largest families of paralogues in sequenced genomes (11). They couple ATP hydrolysis to a wide variety of cellular processes, including transmembrane transport, gene regulation, and DNA repair. All ABC proteins are characterized by the presence of a conserved domain, known as the nucleotide-binding domain (NBD), which contains at least five characteristic sequence motifs (12). In addition to the Walker motifs A and B identified in many ATPase families, NBDs contain the ABC signature motif (motif S), a conserved glutamine loop (Q-loop or Lid), and a conserved histidine loop (H-loop or Switch). In ABC proteins associated with membrane transport, the NBDs bind and hydrolyze ATP and transmit conformational changes to membrane-spanning domains, which typically form a pathway for the transported substrate (12). Genome and phylogenetic analyses have identified three classes of ABC proteins that correlate well to functional categories (13). Classes 1 and 3 are comprised of exporters and importers, respectively. In contrast, class 2 consists of ABC proteins devoid of transmembrane domains or known transmembrane protein partners and composed of two tandemly repeated ABC domains fused together. Within class 2, phylogenetic analyses identified four subfamilies of proteins in which some members have been involved in antibiotic resistance (ARE), transcription or translation regulation (REG and RLI), and DNA repair (UVR) (13,14). Uup belongs to the REG subfamily, which is comprised mainly of eukaryote and prokaryote proteins with unknown function. However, a few are known to participate in gene expression regulation, although their exact mechanism of action and the role of ATP hydrolysis in their function are unknown. Among the five ATPases of this family that have been studied, the best characterized is the yeast protein GCN20. It associates with another protein, GCN1, to stimulate the activity of GCN2, a kinase that phosphorylates the eukaryotic translation initiation factor eIF2, ultimately leading to increased translation of the transcriptional activator GCN4 in amino acid-starved cells (15). In prokaryotes, none of the four proteins have been characterized at the molecular level. The Agrobacterium tumefaciens ChvD protein was found to be inactivated in a mutant selected for the reduced transcription of the virA and virG genes (16). The Caulobacter crescentus HfaC protein was found to be important for the attachment of the holdfast adhesive organelle to the cell (17). In E. coli, inactivation of yheS causes an increase in the periplasmic production of a recombinant Sc-Fv antibody fragment by an unknown mechanism (18). To date, the sole phenotype reported for uup null mutants is an increase in the frequency of precise excision of transposons and of sequences located between two tandemly repeated sequences (7,9,10).
To provide insight into the mechanism of Uup in transposon excision, we undertook a biochemical and genetic analysis of this protein. In this report, we provide evidence that purified recombinant Uup undergoes a conformational change upon ATP binding. We also demonstrate that recombinant Uup has intrinsic ATPase activity and interacts with DNA in vitro. This binding is in part dependent on the 90-amino acid carboxyl-terminal domain of Uup that shares similarities with little finger domains found in the Y family of DNA polymerases. Analysis of Walker motif B mutants led to the conclusion that ATP hydrolysis at the two ABC domains of the protein is essential for the function of Uup in vivo.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Strain Constructions-Strains are listed in Table 1. All cloning steps were performed with E. coli K-12 strain Top10FЈ (Invitrogen). Strain MG1655 was used for construction of mutant DM1. In strain DM1, the uup gene was replaced by an aphA cassette, encoding a kanamycin resistance determinant, by a PCR-based procedure (19). Complementation studies were achieved in strains GJ1885 and GJ1886 (7) and in derivatives of the latter. Transductions were performed as described previously (20). Overexpression of uup alleles placed under the control of the T7 promoter was achieved in E. coli strain BL21(DE3)(pLysS).
Plasmid Constructions-The plasmids used in this study are listed in Table 2. Oligonucleotides used for uup cloning and mutagenesis are listed in Table 3. Standard DNA methods were performed as described previously (21). PCR amplification was achieved by using the Expand High Fidelity PCR kit (Roche Applied Science) according to the supplier's recommendations. The mutations of the uup gene affecting codons 181 and 465 and the replacement of lysine 551 by a stop codon to construct the carboxyl-terminal domain (CTD) deletion were carried out by site-directed mutagenesis of plasmid pETuup as described in Stratagene's QuikChange protocol. The resulting fusion proteins, His 6 -UupD181N, His 6 -UupD465N, His 6 -UupD181N/D465N, and His 6 -Uup⌬CTD called hereafter Uup1, Uup2, Uup3, and Uup⌬CTD, respectively, contained a 20-amino acid amino-terminal extension (MGSSHHHHHHSSGLVPRGSH). The same point mutations were constructed on a uup gene cloned under the control of araBAD promoter in plasmid pBAD18. Cloned DNA sequences of wild-type and mutants alleles of uup were checked by DNA sequencing (Genome Express, Meylan, France).
Purification of His 6 -Uup and Its Variants-All purification protocols were carried out at 4°C. Wild-type and mutant His 6 -Uup proteins were purified from cells of strain BL21(DE3)(pLysS) carrying the respective alleles on plasmid vector pET15b. After growth, cells were suspended in 20 ml of buffer TNA (10 mM Tris-HCl, pH 8.0, 150 mM NaCl) supplemented with one tablet of Roche Applied Science Complete EDTA-free protease inhibitor mixture. Cell suspension was disrupted in an SLM Aminco French pressure cell press (SLM Instruments, Urbana, IL) at 14,000 p.s.i., and the cell lysates were centrifuged for 5 min at 5,000 ϫ g. After removal of cell debris, cell extracts were centrifuged at 100,000 ϫ g for 1 h at 4°C, and His 6 -Uup and its variants were purified from the supernatants, supplemented with 5 mM imidazole on an Ni-NTA-agarose column (Qiagen), as described in the supplier's handbook. The column was washed with 20 ml of TNA containing 40 mM imidazole, and proteins were eluted in 10 ml of TNA containing 250 mM imidazole, except for Uup⌬CTD, which was eluted at 1 M imidazole. 1-ml fractions were analyzed by SDS-PAGE, and those containing recombinant proteins were pooled and extensively dialyzed against TNA in SlideALyser cassettes (Pierce, 30,000 Da). After concentration with a centrifugal filter (Amicon, 30,000-Da cutoff) to a final volume of 5 ml, proteins were applied onto a Superose 6 10/300 GL column, equilibrated in the same buffer, and fitted on an ⌬KTA device (Amersham Biosciences). The purity of the protein samples was assessed by SDS-PAGE followed by Coomassie Blue staining. Finally the fractions containing pure proteins were combined and concentrated. Protein concentration was colorimetrically determined by using the Bradford protein assay kit (Bio-Rad). ATPase Assay-Hydrolysis of ATP was monitored by assaying the amount of inorganic phosphate liberated from ATP using NaH 2 PO 4 as a standard as described previously (21) with slight modifications. Standard reaction mixtures (250 l) contained 50 mM Tris-HCl, pH 7.5, 200 mM NaCl, 10 mM MgCl 2 , 20% (v/v) glycerol, 0.1 mM EDTA, 2 mM dithiothreitol, and purified protein (5-100 g). The reaction was started by adding ATP (0.25-2 mM) and terminated after various times by the addition of the Malachite Green reagent. Absorbance was measured at 630 nm. Vanadate inhibition was achieved by adding 0.01-10 mM orthovanadate prepared as described previously (22). N-Ethylmaleimide inhibition was achieved by adding 0.01-4 mM N-ethylmaleimide prepared in 100% ethanol.
Limited Proteolysis by Trypsin-Digestion with trypsin was carried out as described previously (21) with slight modifications. 20 g of purified His 6 -Uup were incubated at 0°C in TNA buffer containing 10 mM MgCl 2 with and without 2 mM ATP. The reaction was initiated by the addition of trypsin (Uup:trypsin, 1:1, w/w). After 10 -60 min of incubation, 15-l samples were withdrawn and mixed immediately with concentrated gel loading buffer. Digestion products were separated by SDS-PAGE (15% acrylamide) and visualized by staining with Coomassie Blue.
Gel Mobility Shift DNA Binding Assay-DNA probes were generated by PCR, purified, and 3Ј-end-labeled with phage T4 polynucleotide kinase and [␥-32 P]ATP (370 MBq⅐ml Ϫ1 ). DNA binding experiments were performed in a buffer as described previously (23). 5-20 g of protein were incubated for 20 min at room temperature with [␥-32 P]ATP-labeled DNA (20,000 cpm) and 1 g of poly(dI-dC) (nonspecific competitor) in a final volume of 10 l. Samples were loaded onto polyacrylamide gels, which were then dried, analyzed by autoradiography, and quantitated with a PhosphorImager (Amersham Biosciences).
Complementation Studies-Detection of precise excision of a mini-Tn10 inserted within the lacZ gene was done by a colony papillation assay on MacConkey lactose medium as described previously (9). To study the ability of uup and its Walker motif B variants to complement a null mutant of uup, these genes were placed under the control of the araBAD promoter into vector pBAD18 (24), and these constructs were introduced into the prophage attB site of MG1655 by using the InCh system (25). The attB-inserted genes, whose expression was inducible by L-arabinose at a final concentration of 0.2%, were transduced in strain GJ1886, which carried a uup gene inactivated by the insertion of a defective Tn10 transposon (7). Then the araBAD::aadA7 mutation was transduced from MG1655⌬araBAD⌬lacIZ into these strains to avoid the effects of arabinose utilization on the phenotype of mutants on MacConkey lactose plates.
Sequence Analysis-Sequence similarity searches were performed using PSI-BLAST (26) on a non-redundant version of Swiss-Prot (178,545 sequences) using as query amino acids 511-635 of the E. coli uup sequence. The search was limited to iteration 3 due to the inclusion of nonspecific sequences in subsequent iterations. Results of iteration 3 only gathered ABC proteins and DNA polymerase Y little fingers. Hydrophobic cluster analysis (HCA) was used to predict and compare secondary structures (27,28). In the HCA plots, the sequence is written on a duplicated ␣-helical net in which hydrophobic amino acids (Val, Ile, Leu, Phe, Met, Tyr, and Trp) are contoured. These form clusters, which mainly correspond to regular secondary structures (␣-helices and ␤-strands). Clusters often constitute robust signatures allowing an efficient comparison of related but divergent sequences.
Miscellaneous Methods-Subcellular fractionation, protein solubility tests, SDS-PAGE, and immunoblotting were performed as described previously (21). For raising antibodies directed against His 6 -Uup, two New Zealand White rabbits were immunized by intradermic injection of purified protein according to standard protocols. Antibodies were absorbed on total cellular extracts of strain DM1. Protein identification by matrix-assisted laser desorption ionization time-of-flight mass spectrometry was performed by the Institut Pasteur Proteomic Facility. Protein identification by amino-terminal partial sequencing was performed by the Institut Pasteur Protein Analysis Facility on Applied Biosystems model 473 and 494 sequencers.

TABLE 3 List of oligonucleotides
Restriction sites are underlined, and mutagenized nucleotides are in bold characters.

5DM1
Cloning of uup into pET15b and pBAD18 CGGAATTCATATGTCATTAATCAGTATGCATGGCGCAT 3DM1 Cloning of uup into pET15b CCGCTCGAGTCAGCCACCATTTTTTAACGCTTCA 5DM2 Cloning of uup into pBAD18 GGAATTCCGCAGCCTGAAAGGAATAGTAATG 5MUT1 Site-directed mutagenesis: T541A CCGCGCGTGCTGTTGCTTAATGAACCGAC 3MUT1 Site-directed mutagenesis: A541T GTCGGTTCATTAAGCAACAGCACGCGCGG 5MUT2 Site-directed mutagenesis: T1393A CCAAGCAACTTATTGATTCTTAACGAACCGACCAACG 3MUT2 Site-directed mutagenesis: A1393T CGTTGGTCGGTTCGTTAAGAATCAATAAGTTGCTTGG 5CTD Site-directed mutagenesis A1651T GCCGCCGCGGCATAAGCAGAAACTGTAAAAC 3CTD Site-directed mutagenesis T1651A GTTTTACAGTTTCTGCTTATGCCGCGGCGGC grown at different temperature and induction conditions. As can be seen in Fig. 1 lane 5, 0.25 mM IPTG-induced cells synthesized a major protein with an apparent molecular mass of 72 kDa, that was absent in non-induced cells (lane 1) or in control cells harboring pET15b (data not shown). When induced cells were disrupted by sonication, 70% of the 72-kDa protein was found in the supernatant of the cell extract centrifuged at 20,000 ϫ g (Fig. 1, lanes 6 and 7). The apparent molecular mass corresponds well with the data deduced from the nucleotide sequence of His 6 -Uup. Purification of His 6 -Uup was performed as described under "Experimental Procedures." The soluble fraction of broken cells was purified on a nickel-chelating resin. Because few bands were visible on the Coomassie Blue-stained gel after the first purification step ( Fig. 2A), a second purification step on a Superose 6 10/300 GL column was undertaken to improve purification and to estimate the native apparent molecular mass of the His 6 -Uup protein. The protein eluted at an apparent molecular mass of 110 kDa (Fig. 2B). The same molecular mass was deduced from the migration of the protein on a Sephacryl S300 column (data not shown), indicating that the anomalous mass was not due to interactions with the matrix. Treatment of the protein sample with 50 units of DNase I prior to chromatography does not lead to a change in apparent molecular mass. Peak fractions analyzed by SDS-PAGE revealed that His 6 -Uup was pure to near homogeneity (Fig. 2C). A minor band (highlighted by an asterisk) was shown to correspond to a degradation product of His 6 -Uup as determined by mass spectrometry analysis (data not shown). The purification yield was 60 mg⅐liter Ϫ1 . The protein was used to raise specific antibodies and for biochemical characterization.
Uup Is a Cytosolic Protein-The antibodies raised against purified recombinant Uup protein were used to detect a protein that is present in  total extracts of MG1655 and absent in those of the uup deletion strain DM1 by Western blotting. As shown in Fig. 3A, lane 1, a band of 69 kDa molecular mass is present only in total cell extracts of MG1655. We showed that Uup is detected from early exponential phase to stationary phase in similar amounts (data not shown). To determine the cellular location of Uup, equivalent volumes of membrane, ribosomal, and soluble fractions, prepared as described previously (21), were probed by immunoblotting. No band was detected in membrane and ribosomal fractions; however, a strong signal corresponding to the Uup protein was detected in unbroken cells and debris and in soluble fractions (Fig. 3B, lanes 1 and 4), suggesting that Uup is a cytosolic protein. In the soluble fraction, two other bands were immunodetected, and their migration is similar to that of material detected in the uup deletion strain.
Mg-ATP-induced Conformational Changes as Assessed by Limited Trypsinolysis-We have shown that binding of Mg-ATP increased the resistance of purified MalK against limited proteolysis with trypsin (21,29), suggesting that substrate binding is accompanied by a major structural alteration of the protein. To determine whether ATP binding leads to conformational changes in His 6 -Uup, the purified protein was subjected to  trypsin digestion in the absence or in the presence of ATP and subsequently analyzed for the resulting cleavage fragments. In the absence of ATP (Fig.  4A, lanes 1, 3, 5, and 7) limited trypsinolysis of purified His 6 -Uup led to two major degradation products designated fragments A and B (32 and 29 kDa, respectively). Amino-terminal microsequencing demonstrated that they correspond to the two ABC domains of the protein (Fig. 4B). Fragment A starts at the end of the amino-terminal extension introduced by the histidine tag. Fragment B starts at the end of a predicted linker region that joins the two ABC domains of His 6 -Uup (Fig. 4B). Taken together, these results indicate that ABC domains are quite inaccessible to trypsin and probably structured.
In the presence of Mg-ATP (Fig. 4A, lanes 2, 4, 6, and 8) three major degradation products were detected. In addition to fragments A and B identified previously, we detected a third fragment (C) with a molecular mass of 31 kDa. The apparition of fragment C on gels was accompanied by a decrease in fragment B staining intensity, suggesting that fragments C and B share the same origin. Indeed amino-terminal sequencing revealed that fragment C has the same amino-terminal sequence as fragment B. The presumed carboxyl-terminal proteolytic cleavage sites are indicated under the diagrammatic sequence of Uup (Fig. 4B). These results suggest that binding of ATP to Uup leads to a conformational change, which hides a trypsin cleavage site that is exposed in the absence of ATP. This site is located between the second ABC domain and the carboxy-terminal tail of Uup. To confirm this result, we attempted to monitor the effects of ATP binding by measuring changes in the intrinsic fluorescence of tryptophan. The absence of observed changes upon ATP addition suggests that the solvent accessibility of none of the 7 tryptophan residues of Uup is modified (data not shown).
His 6 -Uup Displays an ATPase Activity-In the absence of the membrane-integral components, most ABC transporter ATPases exhibit spontaneous ATPase activity (30 -32). To determine whether REG subfamily ABC ATPases display the same property, the recombinant Uup protein was analyzed for its ability to hydrolyze ATP in vitro by monitoring the release of inorganic phosphate from ATP. ATPase activity required a minimum Mg 2ϩ concentration of 1 mM. When purified His 6 -Uup protein was incubated with ATP, release of P i occurred in a timeand protein concentration-dependent manner (Fig. 5, A and B).
To determine the Michaelis-Menten constants of the ATP hydrolyzing activity of His 6 -Uup, ATPase activity was measured over a range of ATP concentrations under initial rate conditions (Fig. 5C). The values of V max and K m were 26 nmol min Ϫ1 mg Ϫ1 and 1.35 mM, respectively (Fig.  5D), corresponding to a turnover number of 0.03 s Ϫ1 . Such a low activity toward ATP has also been measured for the ATPase domains of mouse P-glycoprotein (33,34) and CDR1 from Candida albicans (35).
It is established that several ABC ATPases are inhibited by vanadate. Vanadate is an analogue of inorganic phosphate that can mimic the ␥-phosphate of ATP in the transition state for ATP hydrolysis (36). The His 6 -Uup ATPase activity was completely insensitive to vanadate up to 10 mM (data not shown). In contrast, ATPase activity was strongly inhibited by 0.5 mM N-ethylmaleimide (data not shown), a result that is in line with the presence of a cysteine residue within the carboxylterminal Walker motif A of Uup (Fig. 4B).  amino acids. PSI-BLAST analyses (26) indicated that Uup CTD shared a weak but significant sequence similarity with Y family DNA polymerase little finger (LF) domain. We observed 26% identity over 107 amino acids with LF of DNA polymerase IV from Pseudomonas syringae pv. tomato (E-value 10 Ϫ23 by iteration 3). Similar significant values were obtained with DNA polymerase IV from other species. In Uup CTD, HCA (27,28) detected the presence of vertical or mosaic clusters typical of amphipathic ␤-strands (e.g. LSYKL and FF sequences) and horizontal clusters corresponding to amphipathic ␣-helices (e.g. the LLEDLEAK-LEALQTQV sequence), suggesting an ␣/␤ folding pattern. We found similar hydrophobic clusters in Uup CTD and in DNA polymerase Y (DinB) LFs, suggesting secondary structure conservation (Fig. 6). Moreover many aligned amino acids were identical as shown within black circles in Fig. 6. The crystal structure of the E. coli DNA polymerase Y DinB LF domain, which shares 31% sequence identity with the Pseudomonas sequence, is known (37). This allowed us to associate observed secondary structures with hydrophobic clusters. It is worth noting that the sequence identities observed between Uup CTD and LF domains of DNA polymerase IV from E. coli and Pseudomonas species are much higher than those observed between LF domains from eubacteria and Archaea, which showed structural similarity, despite the apparent lack of sequence similarities (37). LF domains were demonstrated to participate in DNA binding through the curved ␤-sheet formed by the four ␤-strands (38). The LF domain was also co-crystallized with the ␤-clamp subunit of the replication complex and was suggested to be involved in the processivity of these Y DNA polymerases by interacting with DNA and the ␤-clamp (37). The latter interaction is dependent on a ␤-clamp binding motif, corresponding to a hydrophobic cluster LVLGL at the LF extreme carboxyl terminus and lying outside the globular LF core, in an extended conformation (Fig. 6). This motif was not detected in Uup CTD, thus limiting its similarity to the LF core.
To determine whether Uup was directly involved in precise excision by interacting with DNA, we tested the ability of purified His 6 -Uup to bind to DNA in vitro. Purified His 6 -Uup was used in gel mobility shift DNA binding assays with several [␥-32 P]dATP-labeled DNA fragments. We analyzed the binding of His 6 -Uup to a PCR-amplified cat gene. When 5-20 g of His 6 -Uup were incubated with the probe (20,000 cpm), we observed that His 6 -Uup can form a complex with the probe in a dose-dependent fashion (Fig. 7, lanes 2-4). To determine whether His 6 -Uup can recognize a specific site within the cat sequence, we generated by PCR three non-overlapping fragments of the cat gene: cat1, cat2, and cat3. We showed that His 6 -Uup forms a complex with these three fragments (data not shown). No significant sequence similarity was detected between these fragments, suggesting that His 6 -Uup presents an affinity for DNA with no sequence specificity. We also found that His 6 -Uup was unable to bind single-and double-stranded DNA probes with a size smaller than 220 bases (data not shown). These DNA binding properties led us to reinvestigate the enzymatic properties of Uup. No change in K m and V max and in vanadate resistance was observed in the presence of DNA (data not shown).
To determine whether Uup CTD shares more than structural similarity with the little finger domain, we tested its role in DNA binding. We constructed an allele of uup deleted from its last 515 nucleotides by site-directed mutagenesis of pETuup. Purified Uup⌬CTD hydrolyzed ATP with a V max of 109 nmol min Ϫ1 mg Ϫ1 and a K m of 11.2 mM, thereby suggesting that the ABC domains of the truncated protein were correctly folded. Uup⌬CTD was tested for its ability to bind to DNA in vitro. As shown in Fig. 7, lanes 5-7, Uup⌬CTD presented a significant loss of DNA binding activity. Quantitation of bands corresponding to unbound DNA showed that 20 g of Uup⌬CTD shifted 3.07 Ϯ 0.34-fold less DNA than the same amounts of His 6 -Uup. This result suggests that the CTD participates in DNA binding with another domain of Uup.
Walker Motif B Mutant Proteins Are Unable to Hydrolyze ATP in Vitro and Cause an Increase in Precise Excision of Tn10-The uup gene carried by the pETuup plasmid was mutagenized by PCR to express and purify three mutated versions of the Uup protein. The highly conserved Asp residue in the Walker motif B was chosen because several reports highlighted the importance of this residue in the catalytic cycle of ABC ATPases (39 -41). Mutant proteins Uup1 (D181N), Uup2 (D465N) and Uup3 (D181N/D465N) were purified as described under "Experimental Procedures" and were found to display the same apparent molecular mass of 110 kDa as His 6 -Uup. The purification yields were 60 mg⅐liter Ϫ1 for Uup2 and 15 mg⅐liter Ϫ1 for each of Uup1 and Uup3. However, and unlike His 6 -Uup, Uup1, Uup2, and Uup3 were unable to hydrolyze ATP in vitro (data not shown).
The uup variants carrying mutations of Asp-181 (uup1) or Asp-465 (uup2) to asparagine and the double mutant (uup3) were studied in vivo with regard to function in a colony papillation assay (7). In this assay, GJ1885 cells harboring a wild-type uup gene display one or two papilla in 50% of the colonies, whereas 100% colonies of GJ1886 cells carrying a uup null mutation show about 20 red papilla (Fig. 8A). Wild-type and mutant uup alleles inserted into the bacteriophage attB site of strain GJ1886 were tested for their ability to complement the uup null mutation in this strain (Fig. 8B). In the absence of arabinose, strains expressing wild-type and mutant alleles were unable to complement this mutation. In the presence of arabinose, strains expressing the wild type uup gene displayed a rate of transposon excision similar to that of GJ1885, indicating an efficient complementation. In contrast, strains expressing Walker motif B mutant alleles still displayed a high rate of transposon excision. Expression and cellular localization of attB-inserted wild-type and mutant proteins were comparable with that of wild-type strain MG1655 as verified by immunoblotting using anti-Uup antibodies (data not shown). These results demonstrate that mutant uup alleles are unable to complement the interrupted uup gene. Thus, both residues Asp-181 and Asp-465 must play critical roles as mutations are defective in the known function of Uup. Taken together, these results strongly suggest that the ATPase activity of Uup is essential for its function.

DISCUSSION
Here we report the first biochemical characterization of a bacterial REG subfamily ABC protein. Members of this subfamily display a large functional diversity in the living kingdom. About 800 members were identified in protein databases, 4 but only a few were characterized, at least summarily. The yeast protein GCN20 and its human orthologue ABC50 have been implicated in the control of mRNA translation (42). However, the information collected on prokaryote proteins, which has been summarized in the Introduction, is quite descriptive and does not afford a clue for their actual physiological role and mechanism of action.
It has been known since 1983 that mutations in the uup gene induce a large increase in transposon Tn10 precise excision in E. coli and also a less important increase in excision of one copy of two directly repeated sequences (7,9,10). To gain information on the mechanism of Uup in transposon excision, we undertook a biochemical and genetic analysis of this protein. Our data show that Uup, when expressed at physiological levels, is localized exclusively in the cytoplasm of bacteria. This evidence considerably weakens the hypothesis that REG subfamily proteins may constitute transporters by interacting with unidentified transmembrane proteins (43,44). This soluble protein was detected in similar amounts during the E. coli cell cycle, suggesting that it is a quite stable protein or that it is expressed constitutively.
The apparent native molecular mass of Uup was 110 kDa. The expected masses for a monomer and a dimer are 72 and 144 kDa, respectively. This result could be interpreted by assuming that His 6 -Uup is monomeric with an asymmetric shape. Alternatively the anomalous molecular mass could be due to the binding of a yet unidentified molecule.
The purified His 6 -Uup protein is able to hydrolyze ATP in vitro with a K m of 1.35 mM and a turnover number of 0.03 s Ϫ1 . This is a real turnover number because mutations in Walker motif B eliminated it. Similar affinity constants were measured for isolated NBDs of ABC transporters (30,31,45). It was reported that ATPase activity of purified NBDs of ABC transporters can be stimulated by the presence of interacting partners, such as transmembrane domains and substrates (33). Although we were unable to detect interacting partners of Uup by coimmunoprecipitation, co-purification experiments, and using a twohybrid system, 5 the hypothesis that His 6 -Uup ATPase activity could be stimulated by an interaction with another yet unidentified protein cannot be excluded. The ATPase activity was found to be insensitive to vanadate even at high concentration. It is known that isolated ATPases of ABC transporters are not inhibited by vanadate (31,46) because the vanadate-ADP complex is unstable in the absence of transmembrane domains (47). The lack of inhibition of His 6 -Uup by vanadate might be due to the absence of an essential interacting partner that does not co-purify with the protein. Alternatively vanadate resistance could be an intrinsic property of REG subfamily ATPases because the ATPase activ-ity of purified YbiT and YjjK proteins of E. coli is also unaffected by vanadate. 5 The roles of the two ATPase sites present in REG subfamily proteins is not known. To address this question, we characterized mutant proteins affected by changes in the Walker motif B highly conserved Asp residue. This residue was shown to coordinate the Mg 2ϩ ion in the catalytic site of ABC ATPases (48,49). Our results indicate that mutants affected in each of these sites display no measurable ATPase activity. This suggests that ATP hydrolysis at these two sites is tightly coordinated either by influencing each other by allosteric interactions or because they interact together to form a composite active site as it was deduced from the crystal structures of MalK, Rad50, and BtuC ABC ATPases (50 -52). This observation is consistent with data obtained on P-glycoprotein where the two NBDs cannot function independently as catalytic sites in the intact molecule (53). Moreover the fact that the phenotype of mutants displaying no ATPase activity in vitro is similar to that of the uup null mutant further indicates that ATP hydrolysis is required for the function of Uup.
In ABC transporters, several lines of evidence emphasize the key role of protein conformational change in the mechanism by which ATP hydrolysis is coupled to transport (for a review, see Ref. 12). We demonstrated that the susceptibility of His 6 -Uup to trypsin was different in the presence and in the absence of Mg-ATP. Changes in susceptibility to proteases upon the addition of nucleotides have been interpreted as evidence for nucleotide-mediated changes in protein structures (29,54,55). Our results are consistent with the view that Uup undergoes an ATP-dependent conformational change that protects a region located in the CTD of the protein. As it was suggested for other ABC ATPases (12), we hypothesize that these conformational changes might be transmitted to CTD to modulate the interaction of the protein with Uup substrates or partners.
Our data demonstrate that His 6 -Uup displays a general, non-sequence-specific DNA binding activity, suggesting that Uup might interact functionally with DNA. Such a property is shared with UvrA, a subunit of the bacterial ABC nucleotide excision repair complex, whose zinc finger domains have been implicated in DNA binding (56). Other ABC mechanoenzymes, such as the structural maintenance of chromosomes (SMC) proteins, also interact with DNA, probably due to their flexible hinge domain (57). We detected neither zinc-binding domains nor sequences similar to hinge domains in Uup, but a carboxyl-terminal extension was shown to display sequence and secondary structure similarities with LF domains of translesional DNA polymerase Y. Although little finger domains display little sequence conservation, they share the same conserved secondary structure (␤-␣-␤-␤-␣-␤) (37). The same structure is predicted for Uup CTD. In Sulfolobus solfataricus DNA polymerase Y (DpoIV), this domain was shown to participate in DNA binding along with the Thumb domain (38). The LF core fits the DNA FIGURE 8. ATPase activity is essential for the function of Uup in vivo. Strains were cultured overnight in LB medium supplemented with 0.4% arabinose or with no arabinose. A, control strains GJ1885 (wildtype uup) and GJ1886 (interrupted uup) were grown in LB and spread on lactose MacConkey plates. B, derivatives of GJ1886 inactivated for araBAD operon and expressing different alleles of uup (uupϩ, uup1, uup2, or uup3) inserted in the chromosomal prophage attB site under the control of the araBAD promoter (BAD). Arabinose-induced cells were spread on MacConkey agar plates containing 0.4% lactose plus 0.4% of arabinose (top pictures). Uninduced cells were spread on MacConkey medium containing 0.4% lactose (bottom pictures). After 4 days of incubation at 37°C, precise excision of Tn10 was monitored by counting the number of red papillae inside white colonies.
major groove through a curved ␤-sheet. The decreased DNA binding activity of the Uup mutant lacking CTD suggests that this domain participates in DNA binding like its homologue in Y family DNA polymerases. It is tempting to speculate that basic residues included within predicted ␤-strands of Uup CTD play a critical role in DNA binding. Because Uup⌬CTD still binds DNA, we hypothesize that the highly charged LD of Uup also could be involved in DNA binding. Site-directed mutagenesis experiments are in progress to test this hypothesis.
In yeast, it was shown that the essential protein RLI, a class 2 ABC ATPase, is involved in mRNA translation initiation and in ribosome biogenesis (58). RLI shares the same general architecture as Uup, i.e. two NBDs on the same polypeptide chain separated by a LD and ending with a CTD. However, RLI and Uup display only 22% sequence identity confined into the ABC domains. The structure of RLI has been determined recently (59). In this structure, NBDs adopt a head-to-tail configuration. An unusual hinge made of a portion of the LD and of the CTD intimately interacts along the NBD1/NBD2 interface, suggesting that the hinge is a key component for the orientation of the two NBDs. A similar situation might be expected for Uup in which LD and CTD might be tightly associated and involved in DNA binding.
How can an ABC ATPase modulate transposon precise excision? This excision process is in fact an illegitimate recombination event that occurs when the replication fork stalls (1). The presence of the inverted repeats constituted by the two insertion sequence elements of transposons might act as a block to DNA replication (for a review, see Ref. 1). On the lagging strand, this would favor the dissociation and realignment of the nascent strand with the second copy of the direct repeats flanking the transposon. Resuming replication at the mispaired position will result in transposon precise excision. A speculative hypothesis involves the Uup protein in an ATP-dependent mechanism that impedes stalling of the replication fork or that permits the replication to restart quickly after stalling. Uup may act directly by interacting with DNA, thus tending to displace intermediate secondary structures that cause polymerase stalling during replication and subsequent deletions.
We have detected more than 100 Uup orthologues in protein databases. 4 In addition to ABC domains, these proteins share highly conserved LDs and CTDs. Uup homologues are found in proteobacteria, and even the highly reduced genome of the aphid endosymbiont Buchnera aphidicola carries a copy of Uup, suggesting that the protein performs important functions in bacteria.