Characterization of the uup Locus and Its Role in Transposon Excisions and Tandem Repeat Deletions in Escherichia coli

Abstract

Null mutations in the Escherichia coli uup locus (at 21.8 min) serve to increase the frequency of RecA-independent precise excision of transposable elements such as Tn10 and to reduce the plaque size of bacteriophage Mu (Uup⁻ phenotype). By the combined approaches of physical mapping of the mutations, complementation analyses, and protein overexpression from cloned gene fragments, we have demonstrated in this study that the Uup⁻ phenotype is the consequence of the absence of expression of the downstream gene (uup) of a two-gene operon, caused either directly by insertions in uup or indirectly by the polar effect of insertions in the upstream gene (ycbY). The promoter for uup was mapped upstream of ycbY by primer extension analysis on cellular RNA, and assays of reporter gene expression indicated that it is a moderately active, constitutive promoter. The uup mutations were also shown to increase, in a RecA-independent manner, the frequencies of nearly precise excision of Tn10 derivatives and of the deletion of one copy of a chromosomal tandem repeat, suggesting the existence of a shared step or intermediate in the pathways of these latter events and that of precise excision. Finally, we found that mutations that increase the frequency of precise excision of Tn10 are divisible into two categories, depending upon whether they did (uup, ssb, polA, and topA) or did not (mutHLS, dam, and uvrD) also increase precise excision frequency of the mini-Tn10 derivatives. It is suggested that the differential response of mini-Tn10 and Tn10 to the second category of mutations is related to the presence, respectively, of perfect and of imperfect terminal inverted repeats in them.

One of the features of mutations generated by the insertion of transposable elements is their ability to undergo true reversion, at characteristically low frequencies, by precise excision between the pair of host sequence-derived direct repeats that flank each insertion. Unlike other properties associated with transposons, precise excision is mediated by host-encoded functions and does not depend on the transposase encoded within each element (10, 13). In studies with the tetracycline resistance element Tn10 (13, 25, 26), Kleckner and coworkers have identified precise excision as one of three related genetic rearrangements, the other two being nearly precise excision (in which a deletion event between two repeats internal to Tn10 results in excision of all but 50 bp of the element, so that the target gene remains nonfunctional but there is relief of polarity on the expression of downstream genes in the operon) and precise excision of the 50-bp remnant of nearly precise excision. All three rearrangements are RecA independent and fall into the category of illegitimate recombination events.

The mechanism by which precise excisions occur is not known, nor is it clear what, if any, are the other non-transposon-related mutations, resulting from illegitimate recombination events in bacteria, that are mechanistically related to precise excision. Foster et al. (13) had provided early evidence that, whereas precise excision and nearly precise excision of Tn10 may occur by very closely related pathways, precise excision of the 50-bp remnant appears to occur by a different mechanism. One model has been that precise excision occurs by a RecA-independent replication slippage event across the pair of direct repeats (of host-derived sequence) flanking the insertion and that the inverted repeats at the ends of the element facilitate the process (10, 13, 48). The inverted repeats may, for example, participate in formation of intrastrand stem-loop structure(s), although alternative structures involving interactions between the inverted repeats as duplex DNA have not been excluded (10, 13, 40, 48). The former possibility is supported by the findings that the frequencies of precise and nearly precise excision are increased under conditions where the single-stranded template (which would more readily be able to form the stem-loop structure) is expected to be abundant, such as in the presence of an M13 ori sequence on the template (7) or during Tra-dependent synthesis of single-stranded DNA during conjugal transfer of an F′ plasmid (30, 49). An in vitro model that mimics precise excision and that is mediated by replication slippage has also been reported (6). Finally, a separate phenomenon of UV-induced transposon precise excision that appears to require functions encoded by the SOS regulon has also been described (22, 34).

Mutations (designated tex, for transposon excisions) in several host genes that increase the frequency of precise excisions have been identified (16, 25, 26, 38). One such locus is uup (16, 38), which maps at 21 min on the Escherichia coli chromosome and increases the precise excision frequency of both transposons Tn5 and Tn10. Mutants in uup also exhibit a reduction in lytic growth of Mu bacteriophage. In an earlier study (38), we had described the isolation of several independent insertion mutations in the uup locus. Molecular cloning, complementation analysis, and nucleotide sequence determination of the gene identified by one of the disruptions had indicated that the Uup protein is cytosolic and belongs to the superfamily of ATP-binding cassette domain proteins (23).

In the present study, we have investigated the organization and regulation of the uup locus. Our results indicate that uup is part of a complex operon and that it is situated downstream of a conserved gene of apparently unrelated function (ycbY) with which it is cotranscribed from a moderately active, constitutive promoter. We also report that uup mutations increase the frequency of two other RecA-independent recombination events, namely, nearly precise excisions and deletions of one copy of a chromosomal tandem repeat (tandem repeat deletion). Finally, our results permit classification of the tex mutations into two categories, depending upon their differential effects on Tn10 derivatives with imperfect versus perfect terminal inverted repeats.

MATERIALS AND METHODS

Growth media and conditions.

The defined and nutrient media were, respectively, minimal A medium (supplemented with glucose or other indicated C source and the appropriate auxotrophic requirements) and Luria-Bertani medium (31). Concentrations of antibiotics and 5-bromo-4-chloro-3-indolyl-β-d-galactoside (X-Gal) used were as earlier described (37, 38).

Bacterial strains and plasmids.

The genotypes of E. coli strains used in this study are listed in Table 1. Plasmids were constructed from the following vectors: the high-copy-number derivatives pBluescript II KS (pBKS; Stratagene, La Jolla, Calif.) and pET21b (Novagen, Madison, Wis.); a medium-copy-number pSC101 derivative, pCL1920 (21); and a very-low-copy-number IncW derivative, pMU2385 (51), carrying the lacZ reporter gene for promoter cloning experiments. The extent of uup locus carried on each of the plasmids is depicted in Fig. 1.

TABLE 1.

List of E. coli strains^a

Strain	Genotypeb	Source or reference
BL21(DE3)	hsdS gal (λDE3)	47
JJC520	deo lacZDR624lacY1 Cmr	3
KM22	Δ(recC ptr recB recD)::(Plac-bet exo Kanr)	33
MC4100	Δ(argF-lac)U169 rpsL150 relA1 araD139 flbB5301 deoC1 ptsF25	Lab stock
MG1655	Wild type	CGSCc
GJ1885d	ara zbh-900::Tn10dKan(Ts)1 lacZ4525::Tn10dKan	38
GJ1886	GJ1885 uup-351::Tn10dTet1	38
GJ1887	GJ1885 uup-352::Tn10dTet2	38
GJ1888	GJ1885 uup-353::Tn10dTet2	38
GJ1934	MC4100 recA56 srl-300::Tn10 zgb-910::Tn10dCm	This work
GJ2240	KM22 ΔycbY::Tet	This work
GJ2241	MC4100 lacIq L8 lacZ+ zai-911::Tn10dCm	This work
GJ2242	MC4100 (λcI857)	This work
GJ2243	GJ1885 topA72::Tn10dTet2	This work
GJ2255	GJ1885 ΔycbY::Tet	This work
GJ2256	MG1655 lacZDR624 Cmr	This work
GJ2258	GJ1885 recA56 srl-300::Tn10	This work
GJ2259e	GJ1886 recA56 srl-300::Tn10 zgb-910::Tn10dCm	This work
GJ2260e	GJ1887 recA56 srl-300::Tn10 zgb-910::Tn10dCm	This work
GJ2261e	GJ1888 recA56 srl-300::Tn10 zgb-910::Tn10dCm	This work
GJ2265	GJ1885 lacZ+	This work
GJ2268	GJ2265 lacZ2900::Tn10	This work
GJ2269	GJ2265 lacZ4526::Tn10dTet2	This work
GJ2272f	GJ2265 uup-351::Tn10dTet1 lacI3098::Tn10Kan lacZ2900::Tn10	This work
GJ2273f	GJ2265 uup-351::Tn10dTet1 lacI3098::Tn10Kan lacZ4526::Tn10dTet2	This work
GJ2274f	GJ2265 ssb-200 zjc-904::Tn10dTet1 lacI3098::Tn10Kan lacZ2900::Tn10	This work
GJ2275f	GJ2265 ssb-200 zjc-904::Tn10dTet1 lacI3098::Tn10Kan lacZ4526::Tn10dTet2	This work
GJ2276	GJ2265 polA12 zih-3166::Tn10Kan lacZ2900::Tn10	This work
GJ2278	GJ2265 polA12 zih-3166::Tn10Kan lacZ4526::Tn10dTet2	This work
GJ2280	GJ2265 ΔuvrD288::Kan lacZ2900::Tn10	This work
GJ2281	GJ2265 ΔuvrD288::Kan lacZ4526::Tn10dTet2	This work
GJ2282	GJ2265 dam::Tn9 lacZ2900::Tn10	This work
GJ2283	GJ2265 dam::Tn9 lacZ4526::Tn10dTet2	This work
GJ2284	GJ2265 mutH471::Kan lacZ2900::Tn10	This work
GJ2285	GJ2265 mutH471::Kan lacZ4526::Tn10dTet2	This work
GJ2288f	GJ2265 mutS::Tn10dTet1 lacI3098::Tn10Kan lacZ2900::Tn10	This work
GJ2289f	GJ2265 mutS::Tn10dTet1 lacI3098::Tn10Kan lacZ4526::Tn10dTet2	This work
GJ2290f	GJ2265 mutL::Tn10 lacI3098::Tn10Kan lacZ2900::Tn10	This work
GJ2291f	GJ2265 mutL::Tn10 lacI3098::Tn10Kan lacZ4526::Tn10dTet2	This work
GJ2292	ara zbh-900::Tn10dKan(Ts)1 lacZDR624 Cmr	This work
GJ2293	GJ2292 recA::Kan	This work
GJ2294	GJ2292 uup-351::Tn10dTet1	This work
GJ2295	GJ2292 uup-351::Tn10dTet1 recA::Kan	This work

^aBL21(DE3) is E. coli B. All other strains are K-12 derivatives. 

^bGenotype designations are as in the work of Berlyn (2). All strains are F⁻. Allele numbers are indicated where they are known. The designations Tn10dTet1 and Tn10dTet2 have been described earlier (38). In the GJ strains listed, the following mutations were transduced from strains described earlier: recA56 and srl-300::Tn10, GJ971 (43); lacI^q L8, NR3835 (45); lacZ2900::Tn10, NK5661 (26); lacI3098::Tn10Kan, CAG18420 (46); ssb-200 and zjc-904::Tn10dTet1, GJ1890 (38); polA12, MM383 (CGSC no. 5022); zih-3166::Tn10Kan, CAG18601 (46); ΔuvrD288::Kan, SK6776 (52); mutH471::Kan, GW3773 (CGSC no. 7306); mutS::Tn10dTet1, CSH115 (31); and mutL::Tn10, ES1484 (CGSC no. 7050). Strains carrying the dam::Tn9 and recA::Kan alleles were obtained from N. Kleckner and R. Jayaraman, respectively. The zgb-910::Tn10dCm and zai-911::Tn10dCm insertions were new alleles obtained in this study (with the aid of the transposon vehicle phage λNK1324 [17]), cotransducible 60 and 80%, respectively, with the recA and lac loci. The lacZ4526::Tn10dTet2 insertion was also obtained in this study. 

^cCGSC, Coli Genetic Stock Center. 

^dThe zbh-900::Tn10dKan(Ts)1 allele confers kanamycin-resistance at 30 but not at 42°C (37). 

^eGJ2259, GJ2260, and GJ2261 were constructed by P1 transduction of GJ1886, GJ1887, and GJ1888, respectively, with a P1 lysate prepared on GJ1934. The presence of the srl-300::Tn10 allele in these strains has not been verified. 

^fThe lacZ allele in each of these strains was introduced by cotransduction with the linked lacI3098::Tn10Kan marker, with selection for kanamycin resistance at 42°C. 

FIG. 1.

Extents of insert DNA from E. coli uup locus in plasmids used in this study. On top is depicted, to the indicated scale, the position of recognition sites for the enzymes BamHI (B), BclI (Bc), ClaI (C), EcoRI (E), HindIII (H), PstI (P), and SalI (S); for BclI, ClaI, and SalI, only the relevant sites have been marked. Also depicted are the positions of the uup-351, -352, and -353::Tn10dTet insertions (as inverted triangles). Immediately beneath is depicted the alignment of the ycbY, uup, pqiA, and pqiB′ ORFs and the promoters (as hooked arrows) within uup and upstream of ycbY identified, respectively, by Roe's group (18) and in this study. The uup locus sequence is from the work of Blattner et al. (4) and is corrected from that reported earlier (18, 38). The inset shows a map, drawn to one-half scale, of Tn10dTet in the orientation present in each of the three uup insertions, and the solid bar marks the fragment used for radiolabeled probe preparation in the Southern blot of Fig. 2; the inset map is that of Tn10dTet2 present in uup-352 and uup-353, whereas the Tn10dTet1 element in uup-351 lacks the pair of BamHI sites shown (17, 38). Each line aligned beneath the uup locus physical map represents the extent of chromosomal DNA, delimited by the cut sites marked, that has been cloned into a plasmid(s) whose numerical pHYD designation(s) (and vector derivation[s] in parentheses) is indicated alongside. Abbreviations for plasmid vectors: pCL, pCL1920; pET, pET21b; and pMU, pMU2385. The interrupted line segment in the insert of pHYD650 depicts deletion of DNA between the parenthetical BclI sites marked, and the filled rectangle indicates insertion at this site of the 2.7-kb Tet^r-encoding BamHI fragment from Tn10dTet2 (see inset). Digestions at the BclI sites and at the right ClaI site marked were done on DNA prepared from a dam strain.

DNA methods.

The standard protocols of Sambrook et al. (42) were followed for experiments involving recombinant DNA, including plasmid manipulations, gel electrophoresis, transformation, preparation of radiolabeled probes, Southern blot hybridization, and DNA sequence determination on double-stranded plasmid DNA templates. The oligonucleotide primer 5′-TGGTCACCAACGCTTTTCCCGAG-3′, designed so as to read outward from a site immediately internal to the right terminal inverted repeat of Tn10dTet2 (see Fig. 1), was used to determine the junction sequences of insertions generated with this element.

Construction of a chromosomal ycbY deletion-insertion mutant.

A 1.2-kb BclI fragment that spans the promoter and proximal third of the ycbY open reading frame (ORF) was excised from plasmid pHYD633, and in its stead was ligated a 2.7-kb BamHI fragment (derived from Tn10dTet2) comprising the tetA and tetR genes. The resulting plasmid pHYD650 thus carries both a deletion and a tetracycline resistance insertion in ycbY. The ΔycbY::Tet mutation was recombined into the chromosome of the recD strain KM22 as described elsewhere (33), following transformation with 1 μg of a gel-purified 5.3-kb fragment from pHYD650 that carries the mutation and flanking DNA from the ycbY-uup locus. In order to control for the possibility that the mutation might be lethal, the linear transformation with the fragment from pHYD650 was attempted in both KM22 and KM22/pHYD646 (where pHYD646 is expected to provide ycbY⁺ and uup⁺ functions even after disruption of the chromosomal locus); equal numbers of Tet^r transformants were obtained with both recipient strains. A P1 lysate prepared on one KM22 Tet^r transformant, GJ2240, was then used to transduce the lacZ::Tn10dKan strain GJ1885, as well as the pHYD646 derivative of GJ1885, to Tet^r. Once again, Tet^r transductants were obtained in both strains at equal frequencies, and a 100% linkage was observed in GJ1885 between Tet^r and the Uup⁻ phenotype (data not shown). One such GJ1885 derivative was designated GJ2255.

Measuring mutation frequencies.

In strains where reversions to lacZ⁺ were being examined, Lac⁺ papillation tests (37, 38) were employed to obtain rapid and qualitative estimates of mutation frequency. The general procedure for quantitative determination of mutant frequencies in cultures was as described by Fijalkowska and Schaaper (12). Briefly, the mutant frequency was calculated as the ratio of the median number of mutants in a series of (four to eight) cultures to the average number of viable cells per culture. The median frequency was chosen so as to avoid the problem of disproportionate contributions by mutational jackpots in individual cultures. In agreement with earlier reports (12, 13), threefold or greater differences in mutation frequencies between different strains were clearly reproducible in these experiments. In all the experiments involving selection for utilization of lactose, melibiose, or phenyl-β-d-galactoside as C source, appropriate minimal plates prespread with 10⁹ cells of the Δlac strain MC4100 (or its derivatives carrying the plasmid vector pBKS, pCL1920, or pMU2385 for surviving appropriate antibiotic supplementation) as scavenger were used.

The frequency of precise excision of the kanamycin resistance insertion in lacZ::Tn10dKan strains was measured following selection for Lac⁺ revertants; all complementation experiments involving plasmid-borne genes were done in recA strains. The frequency of nearly precise excision of lacZ::Tn10dKan was measured following selection for polarity relief mutants capable of expressing LacY permease and growth on melibiose as sole C source at 42°C (31), on plates additionally supplemented with isopropyl-β-d-thiogalactopyranoside (IPTG) and X-Gal so that the ratio of white Lac⁻ Mel⁺ colonies (nearly precise-excision mutants) to blue Lac⁺ Mel⁺ colonies (precise-excision mutants) could be determined; typically, this ratio was at least 200:1. All Mel⁺ Lac⁻ colonies tested had also lost the Kan^r marker in lacZ and were capable of reverting in a subsequent step to Lac⁺ (on Lac⁺ papillation medium), indicating that they had uniformly suffered nearly precise excision of the lacZ::Tn10dKan element. The frequency with which the 50-bp remnant of Tn10dKan in lacZ following nearly precise excision undergoes precise excision was measured exactly as for precise excision of lacZ::Tn10dKan itself.

Tandem repeat deletion frequency was measured by selection for Lac⁺ revertants of the strain GJ2292 (or its derivatives). GJ2292 carries a 624-bp in-frame duplication within the lacZ gene (lacZDR₆₂₄) and a chloramphenicol resistance marker gene near lacZ and is similar to strain JJC520, which had been used earlier by Michel and coworkers (3), except that the former is lacY⁺ whereas the latter also carries the lacY1 mutation. GJ2292 was constructed in two steps as follows. (i) A P1 phage lysate prepared on JJC520 was used to infect strain MG1655, and a double selection was imposed for Cm^r and Mel⁺ at 42°C (that is, for the marker closely linked to lacZDR₆₂₄ and for lacY⁺, respectively), followed by screening for Lac⁻ colonies; one transductant so recovered was designated GJ2256. (ii) In the second step, P1 transduction to Cm^r with a lysate prepared on GJ2256 was used to replace the lacZ::Tn10dKan allele in GJ1885 with lacZDR₆₂₄, and the resulting strain was designated GJ2292.

lacI mutation frequencies were determined in strain GJ2241 or its uup derivatives following selection for growth on 0.05% phenyl-β-d-galactoside as sole C source (in the absence of IPTG), as described elsewhere (45). The frequency of occurrence of spontaneous deletions in the attλ-gal locus was measured using a strain (GJ2242) carrying a λcI857 prophage or its uup derivatives, following selection for survivors at 42°C on minimal A-glycerol medium supplemented with 1 mM 2-deoxy-d-galactose.

Isolation of a topA insertion mutant as tex.

Random insertions of transposon Tn10dTet2 were generated in the chromosome of the lacZ::Tn10dKan strain GJ1885 after infection with the transposon vehicle phage λNK1323, as described elsewhere (17). Tet^r colonies were screened on Luria-Bertani–lactose–X-Gal agar plates by the method earlier described (38), for clones that exhibited increased Lac⁺ papillation frequency following precise excision of the Tn10dKan insertion in lacZ. One tex mutant so identified also showed extremely poor growth characteristics on both defined and rich media and was designated GJ2243. The Tn10dTet2 insertion was cloned as part of a 12-kb PstI fragment from the chromosome of GJ2243 in the plasmid vector pCL1920, and the resulting plasmid was designated pHYD661. A radiolabeled probe prepared from DNA of plasmid pHYD661 hybridized to the recombinant λ phages 253 and 254 of the ordered Kohara miniset phage library (20, 41). Data from physical mapping of pHYD661 permitted the inference that the Tn10dTet2 insertion in GJ2243 is located in topA (encoding topoisomerase I). Determination of the junction sequence (using pHYD661 as template) confirmed that the insertion had occurred immediately preceding the first base of codon 481 in the 865-residue-long topA ORF. The insertion allele in GJ2243 has been designated topA72::Tn10dTet2.

Other techniques.

Procedures for transduction with P1 phage (15); determination of burst size following phage Mu c(Ts) infection (16); IPTG-mediated overexpression, and identification by gel electrophoresis, of the products of genes cloned into pET plasmid vectors and introduced into the BL21(DE3) strain (47); and measurement of β-galactosidase activities in cultures (31) were as described previously. Strains were made recA following transduction either to Cm^r with a P1 lysate prepared on strain GJ1934 that carries a Tn10dCm insertion 60% linked to the recA locus or to Kan^r with a lysate prepared on a recA::Kan mutant.

RESULTS

Insertions in both uup and ycbY confer a Uup⁻ phenotype.

In an earlier study (38), we had described the isolation and mapping by phage P1 transduction of three independent Tn10dTet insertion mutations (uup-351, −352, and −353) to a single chromosomal locus. The cloning, physical mapping, and sequence analysis of the gene designated uup that had been rendered null by the uup-351::Tn10dTet1 insertion was also described. Plasmids bearing the cloned uup⁺ gene complemented all three mutants, and conversely, a plasmid with the uup-351 insertion complemented none of them.

The physical map and organization of genes in the vicinity of uup deduced from the genome sequence of E. coli (4) are shown in Fig. 1. In this study, we determined, by Southern blot analysis of genomic DNA prepared from the uup-352 and −353 mutants (GJ1887 and GJ1888, respectively), the physical positions of the cognate Tn10dTet insertions in them. For this purpose, a radiolabeled probe prepared from an internal fragment (EcoRI-HindIII) of the tet gene was used, and the data are presented in Fig. 2.

FIG. 2.

Mapping of uup-352 and -353 insertions. Reproduced is the autoradiograph following Southern blot hybridization to electrophoresed DNA from strains GJ1887 (uup-352) and GJ1888 (uup-353) after digestion with EcoRI (E), EcoRI-PstI (EP), or HindIII-PstI (HP) of a radiolabeled probe prepared from the EcoRI-HindIII Tn10dTet fragment described for Fig. 1. At the left are shown the positions of migration of DNA markers of the indicated size in kilobases. Calculated sizes (in kilobases) of the hybridizing fragment for each digest of the uup-352 and uup-353 mutants were, respectively, 3.2 (E), 1.5 (EP), and 10.1 (HP) and 6.0 (E), 4.2 (EP), and 7.2 (HP).

The fact that, for both mutants, the size of the hybridizing fragment following EcoRI-PstI digestion was approximately 1.8 kb smaller than that following digestion with EcoRI alone permitted the inference that the insertions were situated in the interval between the EcoRI and PstI sites that delimit the major portion of the two ORFs ycbY and uup (Fig. 1), in the common orientation shown. From the size of the EcoRI-PstI hybridizing fragment, the position of each insertion within this interval was calculated and is marked in Fig. 1 (along with that of the uup-351 insertion characterized earlier, for comparison). The results indicated that the uup-352 insertion is located downstream of uup-351 in the uup gene, approximately 0.5 kb from the 3′ end. On the other hand, uup-353 is an insertion in the anonymous ORF ycbY situated immediately upstream of, and in the same orientation as, uup. The Southern hybridization data from the HindIII-PstI chromosomal digests (Fig. 2) were consistent with these conclusions. Thus, insertions in two adjacent ORFs, ycbY and uup, appear to confer a Uup⁻ phenotype.

We also cloned the chromosomal PstI fragment encoding Tet^r from the uup-353 mutant strain GJ1888 into the vector pCL1920, and the resulting plasmid was designated pHYD638 (Fig. 1). Analysis of restriction digests of pHYD638 (data not shown) provided confirmation for the fact that the Tn10dTet insertion in the plasmid is in ycbY. The exact site of uup-353 mutation in pHYD638 was identified by DNA sequencing across the junction of the Tn10dTet2 insertion, and the data indicate that the insertion has occurred between bases 1 and 2 of codon 417 in the 702-residue-long ycbY ORF.

Plasmid pHYD638 (bearing the uup-353 insertion) failed to complement the Uup⁻ phenotype of the chromosomal uup-351 and −352 insertions in the downstream uup gene (Table 2). Taken together with our earlier result (38) that a plasmid carrying ycbY⁺ and the uup-351 insertion in uup also does not complement the chromosomal insertion (uup-353) in ycbY, we conclude that ycbY and uup constitute a single operon and that the failure of the ycbY insertion to complement mutations in uup is because of a polarity effect associated with the former.

TABLE 2.

Plasmid complementation analysis of uup mutants^a

Plasmid	Normalized precise excision frequency of lacZ::Tn10dKan in derivatives of strain:
	GJ2258 (wild type)	GJ2259 (uup-351)	GJ2260 (uup-352)	GJ2261 (uup-353)
pCL1920	1.0	7.1	9.8	6.6
pHYD638	0.7	NDb	5.5	7.8
pHYD642	0.8	0.8	0.9	0.9
pHYD643	0.8	1.2	1.2	1.0
pMU2385	1.0	6.5	ND	12.0
pHYD640	1.1	6.0	ND	8.3
pHYD646c	0.9	0.6	0.8	0.9
pBKS	1.0	7.4	4.2	6.0
pHYD627	0.05	0.03	0.03	0.06

^aThe median frequency of Lac⁺ revertants per plated cell was measured for each of the strain derivatives carrying the indicated plasmids and is expressed as a relative frequency value normalized to the corresponding vector plasmid control in the wild-type strain GJ2258 (taken as 1.0). The actual median mutant frequencies for the three control strains GJ2258/pCL1920, GJ2258/pMU2385, and GJ2258/pBKS were, respectively, 3.3 × 10⁻⁶, 5 × 10⁻⁶, and 2.1 × 10⁻⁶. 

^bND, not determined. 

^cLac⁺ revertants for the pHYD646 derivatives were selected on minimal medium containing phenyl-β-d-galactoside as sole C source and IPTG. 

That both the ORFs ycbY and uup encode proteins of the expected size was established with the aid of an IPTG-inducible T7 RNA polymerase-based in vivo expression system (Fig. 3). Following induction with IPTG, a pair of closely migrating protein bands of approximately 73,000 in M_r was detected from a template (pHYD653) that included both ORFs (Fig. 3, lane 5), whereas the lower band of this doublet was replaced by one of approximately 47,000 in M_r when a template (pHYD654) with a truncated uup ORF was used (Fig. 3, lane 7). The deduced M_rs of the ycbY and uup gene products are, respectively, 78,854 and 72,066; the deduced M_r of the truncated uup′ product expected to be synthesized from pHYD654 is 45,112.

FIG. 3.

Polypeptides encoded by the ycbY-uup operon. Protein extracts prepared from uninduced (−) and IPTG-induced (+) cultures of BL21(DE3) derivatives carrying the plasmid pET21b, pHYD653, or pHYD654 were subjected to gel electrophoresis on lanes as indicated and visualized by staining with Coomassie blue. Lane 1 represents marker proteins of indicated sizes in kilodaltons. Solid and open arrows identify bands of 73,000 and 47,000, respectively, in M_r that are discussed in the text.

ycbY expression is not required for Uup⁺ phenotype.

In the next set of experiments, we examined whether the Uup⁻ phenotype associated with the ycbY insertion was because of (i) merely a polarity effect of the insertion on expression of the downstream uup gene or (ii) the need for ycbY as well in conferring the Uup⁺ phenotype. For this purpose, we constructed a pair of plasmids (pHYD642 and pHYD643) in which a fragment carrying uup⁺ without an intact ycbY had been cloned in either orientation into a site in the vector pCL1920. Both plasmids were able to complement the transposon precise excision phenotype of the mutant carrying the uup-353 insertion in ycbY (Table 2) as well as its Mu plaque size phenotype, suggesting that ycbY itself is not required for the Uup⁺ function.

The uup-353 insertion is situated approximately two-thirds into ycbY, and it is possible that the resulting truncated protein was necessary and sufficient, along with uup⁺, for the Uup⁺ phenotype. To exclude this possibility, we constructed (as described above) a chromosomal ycbY deletion-insertion mutant GJ2255, in which a Tet^r cassette had replaced 1.2 kb of sequence encompassing the promoter and proximal one-third of the ycbY gene (Fig. 1). Strain GJ2255 was Uup⁻, and it too was complemented to Uup⁺ by either of the plasmids pHYD642 and pHYD643 described above which expressed uup⁺ alone without ycbY (data not shown). The results therefore established that the ycbY insertions confer a Uup⁻ phenotype only because of their polar effect on expression of the downstream uup gene.

As described below, the physiologically relevant promoter for chromosomal uup expression is that situated upstream of ycbY. We therefore believe that the expression of uup⁺ from the pair of complementing plasmids pHYD642 and -643 is directed from promoters situated in the vector. Likewise, positive complementation (observed by us earlier [38]) of uup mutations, by a multicopy plasmid-borne fragment which includes uup⁺ and all but the first codon of ycbY but not bearing the upstream promoter region, is because of read-through into uup from a promoter in the vector or of a fortuitous weak internal promoter in ycbY that is phenotypically relevant only in the multicopy state. The same fragment, when borne on a single-copy-number plasmid (pHYD640), failed to complement the uup-353 mutant strain, whereas a larger fragment that included the promoter (pHYD646) successfully complemented the mutant (Table 2).

Characterization of the ycbY−uup operon promoter.

We cloned several fragments from the ycbY-uup region upstream of the lacZ reporter gene in a single-copy-number plasmid, in order to examine promoter activities and regulation in vivo. Each of the fragments extended either upstream or downstream of the EcoRI site that cleaves at codon 2 of ycbY. Two upstream fragments (in plasmids pHYD632 and pHYD647) that extended respectively up to a ClaI site (0.29 kb) and an EcoRI site (1.2 kb) exhibited promoter activities of comparable strength (Table 3), suggesting that the promoter for the ycbY-uup operon is situated downstream of the ClaI site. Experiments of primer-extension analysis on total cellular RNA from uup⁺ strains (data not shown) were also consistent with the existence of a single transcription start site 172 bases upstream of the ycbY ORF.

TABLE 3.

Promoter activity in fragments from uup locus^a

Plasmid	Enzyme sp act
pMU2385	3.3
pHYD630	4.4
pHYD631	30
pHYD632	114 (105)
pHYD640	35 (472)
pHYD647	144 (130)

^aThe specific activity of β-galactosidase was determined for transformants of strain MC4100 carrying the lacZ promoter-cloning vector pMU2385 or its derivatives, and values are given in Miller units (31). Values in parentheses indicate the specific activity for the indicated strains in cultures supplemented with 0.5 mM paraquat. 

The promoter activity identified in either of the plasmids pHYD632 and pHYD647 above was not subject to regulation in vivo by any of the following agents or mutations tested: growth rate, pH, SOS response, oxidative stress, recA, rpoS, hns, oxyR, or soxR; neither was it subject to autoregulation by the products of ycbY and uup themselves (data not shown). Promoter-lac expression was unaffected in a strain carrying multiple copies of the same region on another compatible plasmid (pHYD627 or pHYD628), suggesting that a titratable positive or negative regulatory factor did not exist (data not shown). A plasmid (pHYD631) bearing the 1.2-kb EcoRI fragment in inverted orientation (relative to pHYD647) upstream of the lacZ reporter gene also exhibited a weak and constitutive promoter activity (Table 3), which we believe represents the promoter for the divergently transcribed ORF ycbX upstream of ycbY.

Two downstream fragments from the EcoRI site at the start of the ycbY ORF were also cloned into the lacZ reporter gene plasmid. A 2.3-kb fragment encompassing all of ycbY and the 5′ end of uup exhibited negligible promoter activity (pHYD630 [Table 3]), consistent with other results above showing that the ycbY and uup genes constitute a single unit of transcription. A 5.7-kb fragment that extended further downstream displayed promoter activity that was inducible by paraquat (pHYD640 [Table 3]); this observation is in accord with the findings of Koh and Roe (18, 19) that a paraquat-inducible promoter for the downstream pqi-5 gene in this complex operon is situated within the uup ORF (Fig. 1).

Effect of uup on other types of genetic rearrangements.

We had earlier shown that the frequency of occurrence of spontaneous point mutations to rifampin resistance or nalidixic acid resistance is not altered in uup mutant strains (38). We have now examined the effect of uup on several other types of spontaneous genetic rearrangements that fall under the broad category of RecA-independent recombination events, using the assays described above. We observed that the frequencies of nearly precise excision of Tn10dKan, as well as of tandem repeat deletion in lacZ, were elevated approximately four- to ninefold in both the recA⁺ uup and recA uup strains (Table 4). On the other hand, the uup mutations did not affect the frequencies of occurrence of (i) spontaneous mutations in lacI (roughly 70% of which are caused by insertion or deletion of a 4-bp sequence at a site where this sequence is present in three tandem repeats in the lacI⁺ gene [11, 45]), (ii) deletions in the attλ-gal locus, or (iii) precise excision of the 50-bp remnant following nearly precise excision of lacZ::Tn10dKan (data not shown). The last finding is consistent with an earlier report that precise excision of the Tn10 remnant might occur by a mechanism which is different from that mediating precise excision or nearly precise excision (13).

TABLE 4.

uup effects on nearly precise excisions and tandem repeat deletions^a

uup genotype	Normalized frequency in strain
	Nearly precise excision		Tandem repeat deletion
	recA+	recA	recA+	recA
Haploid uup+	1.0	1.4	1.0	1.3
uup-351	4.2	9.2	4.2	8.3
Haploid uup+ (with pBKS vector)	1.1	1.2	NDb	1.5
Multicopy uup+	0.02	0.03	ND	0.03

^aThe median frequency of mutants obtained per plated cell was measured for each of the strains and is expressed as a relative frequency value normalized to that in the haploid uup⁺ recA⁺ derivatives, GJ1885 and GJ2292, for nearly precise excisions and tandem repeat deletions, respectively (each taken as 1.0). The actual median mutant frequencies for the two strains were 4.7 × 10⁻⁴ and 3.3 × 10⁻⁵, respectively. The other strains used for the nearly precise excision studies were GJ1886 (uup-351 recA⁺), GJ2258 (uup⁺ recA), and GJ2259 (uup-351 recA), and those for the tandem repeat deletion studies were GJ2294 (uup-351 recA⁺), GJ2293 (uup⁺ recA), and GJ2295 (uup-351 recA). Multicopy uup⁺ derivatives carried the plasmid pHYD627. 

^bND, not determined. 

uup⁺ gene dosage effect on genetic rearrangements.

While undertaking the complementation experiments, we observed that strains carrying the multicopy uup⁺ plasmid pHYD627 exhibited lower frequencies of precise excision (approximately 20- to 30-fold) than did the isogenic strain GJ2258 that was haploid uup⁺ (Table 2). The frequencies of nearly precise excision and of tandem repeat deletion were also reduced around 30- to 50-fold in the multicopy uup⁺ strains bearing plasmid pHYD627, compared to the values in strains carrying the vector pBKS (Table 4). Furthermore, the plaque size of Mu c(Ts) when plated on the multicopy uup⁺-bearing strain was significantly larger than that on the haploid uup⁺ strains (data not shown). The relevance of these observations to our understanding of the possible physiological role and mechanism of uup function is discussed below.

tex mutations fall into two categories.

In the course of these studies, we have also found that the various tex mutations described so far can be classified into two categories (Table 5). An example of the first category is uup, which increases the precise excision frequency of both Tn10 (16) (Table 5) and the mini-Tn10 derivatives such as Tn10dKan (38) (Table 2), Tn10dTet (Table 5), and Tn10dCm (data not shown). Other examples of the first category include mutations in ssb (38) (Table 5) and polA (27) (Table 5), the genes encoding the single-stranded DNA binding protein SSB and DNA polymerase I, respectively.

TABLE 5.

tex effects on precise excisions of Tn10 and mini-Tn10

Strain paira	Relevant genotype	Normalized precise excision frequency of strain typeb:
		lacZ::Tn10	lacZ::Tn10dTet2
GJ2268; GJ2269	Wild type	1.0	1.0
GJ2272; GJ2273	uup-351	12.6	13.3
GJ2274; GJ2275	ssb-200	16.2	20.0
GJ2276; GJ2278c	polA12	5.1	9.6
GJ2280; GJ2281	ΔuvrD288::Kan	13.6	0.8
GJ2282; GJ2283	dam::Tn9	5.5	1.7
GJ2284; GJ2285	mutH471::Kan	3.0	0.9
GJ2288; GJ2289	mutS::Tn10	5.4	1.0
GJ2290; GJ2291	mutL::Tn10	8.3	1.2

^aThe first and second strain numbers in each pair correspond, respectively, to the Tn10 and Tn10dTet2 derivatives. 

^bThe median frequency of Lac⁺ revertants per plated cell was measured for each strain and is expressed as a relative frequency value normalized to that in the appropriate wild-type control strains GJ2268 and GJ2269 (taken as 1.0). The actual median mutant frequencies for these two strains were 1.5 × 10⁻⁹ and 2 × 10⁻⁹, respectively. 

^cBecause polA12 is a Ts allele (52), cultures of GJ2276 and GJ2278 were grown in glucose minimal medium at 42°C prior to plating for selection of Lac⁺ revertants. 

Mutations in topA, the gene encoding topoisomerase I, also belong to this first category, because they have earlier been shown to increase precise excision of Tn10 (25), and we found in this study that a new topA insertion (topA72::Tn10dTet2), obtained as described above, increased the median frequency of precise excision of lacZ::Tn10dKan 20-fold in the strain GJ2243 (1.1 × 10⁻⁴ per cell, compared to 5 × 10⁻⁶ in the control strain GJ1885). Strains with topA null mutations are known to accumulate suppressors in loci encoding the gyrase subunits, in order to compensate for the excessive supercoiling of DNA in these strains (9, 36, 39); the following features of topA72, however, lead us to suggest that the observed tex effect is caused by topoisomerase I deficiency itself rather than by the additional suppressor mutation(s). (i) In P1 transduction experiments employing the topA72 strain GJ2243 as donor, no Tet^r transductants could be recovered in either the MC4100 or MG1655 strain backgrounds, whereas slow-growing Tet^r colonies were obtained at the normal expected frequency in strain GJ1885 after 2 days' incubation. This result suggested that different E. coli strains differ in their ability to tolerate topA disruption, some being killed (as has also been reported earlier [9]) and others exhibiting behavior analogous to Salmonella enterica serovar Typhimurium in that they are sick yet retain viability (39). (ii) When the selection for topA72 transductants of GJ1885 (which carries lacZ::Tn10dKan) was undertaken on Lac⁺ papillation medium, the vast majority of the slow-growing colonies exhibited a clear hyperpapillation phenotype, suggesting that the increase in precise excision frequency is a trait that accompanies inheritance of the topA mutation itself.

The second category of tex mutations consists of those which were earlier known (25, 26) to increase the precise excision frequency of Tn10 and confirmed to be so in this study (Table 5) but which do not affect the precise excision frequency of the mini-Tn10 derivatives (Table 5). Included in this category are mutH, mutL, mutS (encoding the MutHLS proteins involved in methyl-directed mismatch repair), dam (encoding DNA adenine methylase), and uvrD (encoding DNA helicase II). The mechanistic implications of this bipartite classification of the tex mutations are discussed below.

DISCUSSION

Operonic arrangement at the E. coli uup locus.

The following lines of evidence obtained in this study establish that uup is cotranscribed with another ORF (ycbY) from a promoter situated upstream of ycbY. (i) An insertion mutation in chromosomal ycbY confers a Uup⁻ phenotype, because of a polarity effect on uup expression. (ii) Likewise, a plasmid (pHYD638) carrying the ycbY-uup locus with the insertion in ycbY failed to complement chromosomal uup mutations. (iii) Only one promoter capable of transcribing uup⁺ in the haploid state was identified in experiments involving either genetic complementation or lacZ reporter gene expression, and this promoter is situated upstream of ycbY. (iv) Finally, the coordinate production of proteins corresponding to both the ycbY and uup ORFs in the T7 RNA polymerase-based expression system is consistent with the notion that they constitute a single transcriptional unit. The ycbY-uup genes are also part of a more complex operon that includes downstream genes such as pqiA and pqiB for which additional promoters exist that are embedded in the uup ORF (Fig. 1).

Although uup mutants exhibit phenotypes related to transposon excisions, tandem repeat deletions, and phage Mu growth, the normal physiological function of the Uup gene product remains unknown. The fact that ycbY does not share the known uup mutant phenotypes suggests that the two genes perform unrelated functions, notwithstanding their organization together in a single operon. Analysis of the database of microbial genome sequences (obtained from the website of The Institute for Genomic Research at http://www.tigr.org) reveals that the orthologs of the two genes are placed together in an operon in Yersinia pestis, S. enterica serovar Typhi, and S. enterica serovar Paratyphi A; on the other hand, the genes are still clustered but separated by about 240 bp in Vibrio cholerae and Shewanella putrefaciens, whereas they are widely separated on the chromosomes of Actinobacillus actinomycetemcomitans, Haemophilus influenzae, and Pseudomonas aeruginosa. A similar dichotomy, with conservation of gene order in closely related genera and dispersal in the more distant ones, has been reported recently for the complex operon that includes the fpg and mutY genes in E. coli (14).

A shared step in pathways of transposon excisions and tandem repeat deletions?

We found that uup mutations increase the frequency of three different RecA-independent genetic rearrangements, namely, transposon precise excisions and nearly precise excisions and deletions of a tandem chromosomal repeat. This observation implies the existence of a shared step in the pathways of these events which is influenced by the Uup product. That precise excision and nearly precise excision share similar mechanisms has previously also been suggested by Kleckner and coworkers on the basis of other lines of genetic evidence (13, 25, 26).

The exact role played by Uup in these genetic rearrangements is as yet unclear. There is prior evidence to support the notion that each of these three categories of mutations is the consequence of a RecA-independent slippage event (between the pair of direct repeats) during replication, either simple or involving sister-strand chromatid exchange (3, 5, 7, 10, 13, 44). The role of the Uup product (whose deduced sequence suggests that it is cytosolic and belongs to the ATP-binding cassette family of proteins [23]) might then be to actively destabilize the looped and misaligned intermediate that is expected to precede the postulated slippage event. The fact that multicopy uup⁺ strains exhibit a still lower frequency of transposon excisions and tandem repeat deletions in comparison with haploid uup⁺ strains suggests that the Uup-sensitive intermediate contributes to the rearrangement events even in the latter. Likewise, the data for Mu phage growth in uup, haploid uup⁺, and multicopy uup⁺ strains (references 16 and 38 and this study) suggest a dose-dependent effect of Uup on burst size of phage-infected cells, but the mechanism is unknown. At the same time, it must be noted that the mechanistic pathways, at least for precise excision and tandem repeat deletion, are not identical; for example, we have found that mutations in rep and priA, genes encoding the DNA helicase Rep and the primosome assembly protein PriA, respectively, do not affect the frequency of precise excision (data not shown), although they are known to increase that of tandem repeat deletion (3, 44).

Two categories of tex mutations.

It has been shown earlier (10, 13) that the frequency of precise excision is determined in part by the length and degree of matching of the inverted repeat sequence at the ends of the transposable element. Other workers have also shown that the occurrence of spontaneous deletions in vivo and in vitro between short stretches of direct repeats is facilitated by the presence of palindromic sequences within the deletion interval (1, 6, 8, 35, 50, 53). These results have led to the hypothesis that an interaction between the inverted repeats represents an intermediate in the precise excision (or deletion) pathway.

Our results indicate that the tex mutations affecting precise excision of Tn10 fall into two categories: the first, such as uup, ssb, topA, and polA, increasing precise excision of both Tn10 and mini-Tn10 and the second, such as mutHLS, dam, and uvrD, increasing precise excision of Tn10 but not of mini-Tn10. It may be noted that (i) a common feature of the genes (in particular, mutH, -L, and -S) of the second category is their involvement in mismatch repair (32) and (ii) a major distinction between Tn10 on the one hand and the various mini-Tn10 derivatives on the other is that the inverted repeats in the former possess several mismatches whereas those in the latter are perfectly matched (17). The present categorization therefore provides additional support for the hypothesis that during precise excision the inverted repeats interact as intrastrand stem-loop or snapback structures (rather than as a pair of duplex DNA stems), because it explains why a functional MutHLS system reduces precise excision only of elements in which the palindromes are imperfect. The snapback model may also explain the broad-specificity tex nature of topA and ssb mutations, the former because of increased supercoiling that would favour cruciform extrusion (28) and the latter because of the heightened permissiveness for interactions between single-stranded regions of DNA in the mutant strains (24, 29).