Regulation of IS1 transposition by the insA gene product

J. Mol. Biol. (1989) 208, 567-574

Regulation

of IS1 Transposition

by the insA Gene Product

Chiyoko Machida and Yasunori Machida? Department of Biology, Faculty of Science Nagoya University, Chikusa-ku Nagoya 464, Japan) (Received 16 January

1989)

The IS2 element contains two adjacent genes called insA and insB, both required for IS2 transposition and ISZ-mediated plasmid cointegration. These two genes are transcribed polycistronically from the promoter in the left terminal inverted repeat of IS1 (insL). We constructed overexpression systems of these genes with the tat promoter, which are regulated by an exogenous inducer, isopropyl-fl-n-thiogalactopyranoside (IPTG). Then we have examined, under various conditions of induction with IPTG, how overexpression of these genes affects IS1 transposition, using an assay based on plasmid cointegration. When the insA and insB genes were organized identically to the wild-type IS1 genes and simultaneously expressed using low concentrations of IPTG, activity of a mutant IS1 in cis was restored, but not in trans. Higher IPTG concentrations resulted in lower transposition activity. Expression in trans of insd and insB results in a 50 to loo-fold reduction of the frequency of cointegration mediated by wild-type ISl. Such a reduction is also observed when only the insA gene is overexpressed in trans. Overexpression of either mutant insA or insB does not affect the cointegration event. Tests with the insA-la& fusion gene showed that the InsA product inhibits the expression of IS1 genes directed by its own promoter in insL. These results suggest that’ the InsA product regulates IS1 transposition by inhibiting expression of IS1 transposition genes in addition to acting as part of a transposase complex. 1. Introduction Transposition is thought to be a major source of genetic variability in a wide variety of organisms, and sometimes causes mutations of genes. In bacterial cells, transposition is strictly regulated to very low levels (10d4 to 10e9 per division cycle; Kleckner, 1981) since a high level of transposition may be lethal. The transposable element generally encodes functional protein(s) required for its transposition, which in some cases is referred to as some transposable transposase. In addition, elements also have a gene (or a sequence) for a negative regulator that interferes with their ability to transpose (Gill et al., 1979; Chou et al., 1979; Johnson et al., 1982; Tsberg et al., 1982; Simons & Kleckner, 1983). Such a self-regulation may be one of the mechanisms that keep the frequency of t’ransposition low. Some of the negative regulators inhibit the expression of transposase genes. Transposon Tn3 encodes a repressor (also known as a resolvase) that inhibits expression of the transposase gene by binding to the promoter region (Gill et al., 1979; Chou et al., 1979; Grindley et al., 1982). The ISlOR (right) element, flanking TnlO also has a mechanism

to regulate expression of its transposase gene by an anti-sense RNA (Simons & Kleckner, 1983). IS50R, flanking transposon Tn5, encodes both an inhibitor of transposition and transposase (Johnson et al., 1982; Isberg et al., 1982). In contrast to the Tn3 and IS10 repressors, the ISSOR inhibitor does not repress expression of the transposase gene (Johnson et al., 1982; Isberg et al., 1982). Insertion sequence ISI, at 768 bpf in length, is the smallest’ active transposable element, known in prokaryotes (Ohtsubo & Ohtsubo, 1978). It is thought to contain two genes, insA and insB (91 and 125 amino acids, respectively), both of which are required for ISI-mediated plasmid cointegration (see Fig. 1) (Ohtsubo et al., 1981; Machida et al., 1982; Y. Machida et al., 1984). Since base substitutions in either coding sequence cause fluctuation in the length of the duplicated target sequence at an insertion site, both may be involved in the cleavage reaction at the transposition target (Machida & Machida, 1987). We refer to both ind and insB as $ Abbreviations used: bp, base-pair(s); TPTG. isopropyl-b-n-thiogalactopyranoside. Symbols used for representing fusion genes: “:“, transcriptional fusion; “;“, sequences joined as in wild-type ISI: “-“. translational fusion; prime (‘), 3’- or 5’-truncated gene; cat, chloramphenicol acetyltransferase gene, kb. lo3 bases.

t Author to whom all correspondence should be addressed. 0022-2836/89/160.567-08

$03.00/O

567

0

1989 Academic

Press Limited

C. Machida and Y. Machida

568

IS1 transposase genes, although the mechanism of synthesis of the active IS1 transposase is only poorly understood (see below). The putative IS1 transposase preferentially acts on the IS1 element present in the same plasmid as the transposaseproducing ISI (that is, the activity is cis-acting) (Machida et al., 1982). A similar cis-acting property of transposase function was also observed in ISIO, IS903 and IS50 (Kleckner, 1981). The terminal inverted repeats, each consisting of approximately 35 bp (insL and insR, see Fig. 1) are cointegration also essential for IS1 -mediated (Machida et al., 1982). The terminal repeats of IS1 additional function of transcriptional have promoter (C. Machida et al., 1984). The insA and in& genes seem to be transcribed polycistronically from the promoter overlapping with insL (pinsL) The promoter in insR (pinsR) is used for synthesizing RNA being oriented oppositely to the insA and insB genes. InsA protein is in fact detectable in Escherichia coli cells when a strong promoter is linked to its upstream region (Zerbib et al., 1987), whereas the InsB protein has not yet been shown to be present. Sekine & Ohtsubo (1989) proposed that the reading frameshift of some of the polycistronic mRNA from IS1 may occur during translation which could generate the fusion of InsA and InsB in frame. The fusion protein could act as the active IS1 transposase. If this is the case, at least two kinds of proteins could be synthesized from ISl: InsA protein, and InsA-InsB fusion protein, the presence of which has not yet been substantiated. We examined in vivo functions of the insA and insB gene products using overproduction systems of these genes. Since E. coli K-12 contains ten copies of IS1 sequence in its chromosome DNA (Nyman et al., 1981), it is difficult to examine in vivo functions of IS1 genes manipulated in vitro. Overexpression systems are more powerful approaches for understanding functions of such multicopy genes.

(c) Plasmid construction DNA fragments containing the IS1 sequence were isolated from pJL3 (Machida et al., 1983) or pYM plasmids (Machida el al., 1982) by digesting with suitable restriction enzymes. The la& fusion genes listed in Fig. 2 were inserted into suitable restriction sites of pMC1403 generating pCMlO1, pCM104, pCM5001, pCM5QO4 and pCM5002. Nucleotide sequencing confirmed that the IS1 genes were joined to the la& sequence in frame. pCM2004, pCM2001, pCM2002 and pCM2001-3 were constructed by inserting the overproducing IS1 genes listed in Fig. 2 between the EcoRI cleavage site and the BamHI cleavage site of pBR322. pCF144 and pCF103 were constructed by deleting the DNA region containing ISI01, which is activated by transposon gamma-delta, from pMZ71 and pYM103, respectively (Machida et al., 1982; Ishizaki 6 Ohtsubo, 1985). pCF103AB and pCF103A were constructed by inserting the ptac : insA : insB fragment and the ptac: insA fragment into the pCF103 plasmid. respectively (Fig. 3). To construct pBVl104, the BamHI-PvuII DNA fragment (115 bp) containing the IS1 sequence from positions 1 to 75 and the sequence outside the IS1 was isolated from pJL3-49 (Machida et al., 1983). To construct pBVl105, the HaeIIT DNA fragment (243 bp) containing the ISI sequence from positions 559 to 768 and the sequence outside IS1 was purified from pJL3-1 (C. Machida et al., 1984). Each fragment was inserted into pBV234-Cm’ together with the DNA fragment containing only the coding sequence of the cat’-ZacZ’ (Matsumoto et al., 1988) to generate transcriptionally fused genes. pBVllO3 was constructed by inserting only the DNA fragment containing the cat’-la&’ coding sequence. Details of the procedure will be sent on request.

(d) Assay of P-galactosidase activity E. co& JM109 or NK5830 cells harbouring pCM and/or pBV plasmids were grown in 1xA medium (Miller, 1972) supplemented with 05% (w/v) Casamino acids, O-2% (w/v) glucose, 1 pg thiamine/ml, 1 mM-MgSO,, with or without IPTG. Cells were harvested by centrifugation. The reaction and measurement of /3-galactosidase activity were carried out as described by Miller (1972).

2. Materials and Methods (a) Bacteria and plasmids E. coli JM109 (Yanish-Perron et al.. 1985) was obtained from Takara-Shuzo (Kyoto, Japan). E. coli NK5830 nalR rifR/F'laciQ L8 proAB+) (recA56 arg- 1acproXIII (Morisato et al., 1983) was kindly provided by Dr N. Kleckner. The plasmids pJL3 (Machida et al., 1983), pJL3-49 (Machida et al., 1983) and pJL3-1 (C. Machida et al., 1984) were described previously. pCM plasmids listed in Fig. 2 were derived from pBR322 (Bolivar et al., 1977) or pMC1403 (Casadaban et aZ., 1980). The pCF plasmids, which were used for cointegration assay, were derivatives of pSClO1 (Machida et al., 1982). The pBV234-Cm’ plasmid was a derivative of RlOO (Ohtsubo et al., 1982). (b) Enzymes and chemicals Enzymes were purchased from Toyobo (Osaka, Japan) and New England BioLabs. IPTG was obtained from Wako Pure Chemical Industries (Osaka, Japan).

(e) Analysis of frequency of plasmid cointegration The frequency of formation of cointegrabes was determined by a fluctuation test as described (Machida et al., 1982). E. coli JM109 or NK5830 cells carrying pCF plasmid and ColEl. or pCM plasmid were grown at 25°C in L-broth overnight with or without lo-’ to 10e3 M-IP’I’G. Then 0.1 ml of a 1: lo6 dilution was inoculated into each of 20 test tubes, each containing 5 ml of L-broth supplemented with or without IPTG, and the cells were grown at 25% for 48 h. Next, @l ml of culture was plated on agar plates containing tetracycline (Tc) incubated at 42°C for 24 h. Tc-resistant colonies were counted. To determine the 2 types of cointegrates in Fig. 3, we collected a number of independent colonies (2 to 40 colonies), which were both temperatureand tetracycline-resistant, so that each colony was from an independent culture. We then examined cointegrate structure by size and cleavage analysis with restriction enzymes such as BstEII. ThrlllI and S&II (Machida et al., 1982).

Regulatory

hsL

insA (91 AA)

l

I

/ :L

I &

insB (125AA)

56

328

376

750

piisL

function

insR

mm &L :

768

pins*

PPPUY 40-44 e;PPP

of the IS1 gene

pCMlO1

569

pinsL InsA’ 1acZ’ p=yam 208 I msi3’

msA pCMl04

pCM5001

”

f%$@-J

i ptac

5b0 20

pm5004

733

prac:msAkd

-1

208

+,.$m

I

reported (Machida et al., 1982). The numbers indicate nucleotide positions on the IS1 sequence co-ordinate (Ohtsubo & Ohtsubo, 1978).

of chime&

pCM2004

/nsB’ _<, _ . DT))),.%m’ -‘+--J 371 500 msA ins9 P

IS1 genes

Figure 1 shows the genetic organization of the ISI element. For constructing overproduction systems of the ISI genes, we applied the maximizing expression procedure (Guarente et al., 1980) using translational fusion of the truncated 1ac.Z gene (la&‘). As shown in Figure 2, we first joined, in frame, the la& gene to the middle region of the insA gene (pinsL ; insA’-la&‘). Subsequently, these chimeric genes were transcriptionally fused to the tat promoter (ptac) (Amann et al., 1983) to generate ptac : insA’-la& and ptac : insA ; insB’lacz’ .

When the activity of /I-galactosidase was assayed in cells harbouring la& fusion genes, approximately 100 units of activity were detected in cells having pinsL ; insA’-la&‘, suggesting that the insA region is expressed at a detectable level in cells (Table I, line 2). In contrast to the insA fusion gene, practically no activity was detected in cells containing pinsL ; insA ; insB’-1acZ’ (line 3). Even when ptac is linked to upstream from the insA region, only 15 units were detected in the cells contain-

ptac:ms&/acZ’

prac:msA;/nsB

:,I

2:

insA pCM2001

ptac:!nsA

B

500

2;

ins9 pCM2002

prac:jnsB

7 41

:a ms A-3

pCM2001- 3

Ouw

2; dcletlo” 177-196

3. Results (a) Construction

KM5002

prac:msA;msBlacZ’

5.90

20

Figure 1. Genetic organization of the ISI element. Thick arrows represent the position and orientation of insA and insB genes. Striped regions indicate the terminal inverted repeats, insL and insR. The upper wavy line represents mRNA synthesized from the promoter in the in.sL inverted repeat (p&L). The lower wavy line represents RNA having polarity opposite to that of the ISI genes, synthesized from the promoter in the in& inverted repeat (p&a@. Presence of both RNA molecules and their initiation sites were previously

p,nsL;msA;ins&lacZ’

DIzlmpE~LX

mRNA anti-mRNA

ptnsL;tnsA’-IacZ’

ptac:msA-3 5b0

Figure 2. Schematic representation of the plasmid and fusion genes used in this work. Thick arrows or thick lines represent IS1 genes, as in Fig. 1. The numbers under the fusion genes are the nucleotide co-ordinates of the region in ISZ. m, ZacZ’ sequence; w, insL region; m, tat promoter fragment. Symbols used for representing fusion genes (e.g. :, ‘) are defined in the footnote to p. 567.

ing ptac : insA; insB’-lace’, while 1400 units were induced in the presence of 1O-3 M-IPTG in cells containing ptac : insA’-lace’ (lines 5 and 4). The inefficient expression of the insB region may be due to the presence of a transcriptional terminator between the two regions (Hiibner et al., 1987) or to the absence of an efficient ribosome binding site upstream from insB (Ohtsubo et al., 1981). To region efficiently, the ptac express the insB fragment containing the ribosome binding site was joined to the 6 bp upstream region of the initiation (position 376-378) generating codon of insB ptac : insB’-la&‘. When cells harbouring ptac : in&ZacZ’ were grown in the presence of 10m3 M-IPTG, 1900 units of fl-galactosidase were induced (line 6). The la&’ moiety of each chimeric gene was then replaced wit#h a suitable DNA fragment for

Table 2 Table 1 Expression

Plasmid pMC1403 pCMlO1 pCM104 pCM500 1

of /3-galactosidase produced from ptac : insA’1acZ’ in the presence of various concentrations of IPTG in E. coli JM109 Activity

of IS1 gene-la& chimeric genes directed by the insL and tat promoters Activity of /&galactosidass (units) +IPTG (1O-3 M) - IPTG 3

4

117

93 3 1400

pCM5004

2 26 3

pCM5002

16

15 1900

IPTG

(M)

0

Activity of /l-galactcsidase (units)

10-l 10-e

26 84 200

1o-5 10-a 10-s

950 1200 1400

Essentially

the same results were obtained in E. coli NK5830.

570

C. Machida

and Y. Machida (b)

(a)

i

pCFl44 pCFlO3

IS/ or W-3 Tc’

ori

ori

ori

ori

Figure 3. (a) cis assay system. Structures of 2 parental plasmids, pCF plasmids and ColEl, and 2 types of cointegrates between them are schematically shown. pCF144 and pCF103 carry a wild-type IS1 and the mutant, IS1-3 having the deletion mutation in insA, respectively. The filled box indicates overproducing IS1 genes. pCF144 and pCF103 do not contain overproducers. pCF103AB and pCF103A contain the ptac : insA : insR and ptac : ksA overproducarrs, respectively. ISI (or ISI-3) and IS102 are shown by open and hatched boxes, respectively. Cleavage sites for restriction enzymes are indicated. IS1 (or IS1-3)-mediated cointegrates are approximately 0.3 kb smaller than ISZWmediated cointegrates, because IS102 is 0.3 kb larger than ISI. (b) tranv assay system. A pCM plasmid carries t)hr overproducers depicted in Fig. 2. The region of the overproducing IS1 genes is homologous to the internal region of the IS1 element. To avoid homologous recombination between them generating cointegrate structures. all experiments were carried out in NK5830 (wcA) strain. If homologous recombination took place, the ISI sequence was not duplicated at recombination junctions in these cointegrates. Therefore, we distinguished cointegrate formed by IS7 from them by isolating cointegrate molecules and digesting them with suitable restriction enzymes.

intact insA and insB genes, which resulted in producing ptac : insA ; insB, ptac : insA and ptac : insB (Fig. 2). ptac : insA-3, having a 20 bp deletion in insA, was also constructed by replacement with the DNA fragment from the mutant ISl-3 (Machida et al., 1982) (Fig. 2). Table 2 shows that the level of expression of ptac : insA’-la& increased with the concentration of IPTG. Thus, expression of the ptac-linked IS1 genes can be regulated with IPTG. reconstructing

(b) Effects of overexpression of insA and insB genes on plasmid cointegration mediated by IS1 present in cis or trans (i) Cointegration system used Figure 3 shows structures of plasmids used for examining the ability of IS1 (or ISZ-3) to form cointegrates. pCF144 and pCF103 contain a copy of wild-type IS1 and mutant TSI (ISI-3), respectively. In addition to the IS1 element, pCF plasmids contain another IS element, IS102. The plasmid containing ISI (or ISl-3) and IS102 integrates into ColEl co-existing inside E. coli cells, giving rise to two types of cointegrated plasmid schematically represented in Figure 3(a) and (b): one mediated by IS2 (or ISI-3) and the other mediated by IS102 (Machida et al., 1982). Because DNA replication of pCF plasmids is temperature-sensitive and these plasmids carry the tetracycline resistance gene

(Tc’), cells carrying cointegrates were able to be selected in the presence of tetracycline at 42°C from a population of cells carrying their parental plasmids and ColEl (or pCM plasmid). Note that, the cointegrates have a characteristic structure, a duplication of the IS element that mediates the cointegration event in a direct orientation at the cointegration junctions. Thus, the two types of cointegrates can be distinguished by size and structural analyses of cointegrate DNA molecules (see the legend tlo Fig. 3). An advantage of this genetic system is that the frequency of cointegration mediated by t,hese IS elements can be simply determined by a Luria-Delbriick fluctuation test (Luria & Delbriick, 1943; Machida et al., 1982). Since cointegration mediated by IS102 is expected to be independent from that mediated by ISI. IS202 can be used as an internal reference for the frequency of ISl-mediated assaying cointegration. As shown in Table 3 (lines 1 to 3). pCF144 formed cointegrates with ColEl, which were exclusively those mediated by ISl. However, pCF103 having a deletion mutation in IS1 formed cointegrates mediated by the mutant IS1 (ISI-3) at, 50 to loo-fold reduced frequency (Table 3, lines 4 to 6). Structural analyses of the cointegrates showed that the frequency of cointegration mediated by 1X1-3 was slightly lower than that of cointegration mediated by the reference IS element, IS102.

Regulatory function

of the IS1 gene

571

Table 3. Effects of cis overexpression of insA and insB genes on plasmid cointegration (I&-3) and IS102

Plasmid pCF144

pCFlO3

pCF103AB

IS1 IS1

ISI-

IS-3

Overproducer

IPTG

No

No

ptac : insA ; insB

(M)

0 lo-’ 10-X 0 lo-’ 10-S

0 lo-’ 10-C 10-5 10-d 10-a

pCFlO3A

ISl-3

ptac : insA

Table 3 (lines 1 to 6) also shows that addition of IPTG did not affect the cointegration event mediated by each IS element. (ii) EJfects of overexpression of insA and insB genes on plasmid cointegration mediated by mutant IS1 present in cis We inserted the overproducer ptac : insA ; insB or ptac: insA into the pCF103 plasmid generating pCF103AB and pCF103A (Fig. 3(a)). The frequency of cointegration between these plasmids and ColEl was examined in the presence of IPTG at various concentrations. Table 3 (lines 7 to 18) summarizes the frequencies of cointegration mediated by ISl-3 and ISI&?. Table 3 (lines 7 to 12) showed that pCFlO3AB containing ptac : insA ; insB formed cointegrates at 10 to 20-fold higher frequencies in the presence of lo-’ M to 1O-5 M-IPTG than pCF103. Even without IPTG, the frequency was ten times higher than that of cointegration between pCFlO3 and ColEl, possibly due to the leakiness of ZacI repressor. Further addition of IPTG up to 10e3 M resulted in decreased cointegration frequency (lines 11 and 12). The frequency of ISImediated cointegration was higher than that of TSlOZ-mediated cointegration since all the cointegrates obtained in the presence of 10m3 M-IPTG were those mediated by ISl-3 but not by TS102. When cells harbouring pCFlO3A and ColEl were incubated in the presence of IPTG at various concentrations, the frequency of cointegration mediated by N-3 did not increase and was always lower than that of cointegration mediated by IS102 (lines 13 to 18). These results indicate that the

Frequency of cointegration per division cycle ( x 109)

630 520 450 8.4

mediated by IS1

Number of examined cointegrates mediated by IS1 (ISI-3) IS102

40

0

10 10

11 12

1

0 0 10 8

2

16

97 244

12 12

0 0

119 114

5 5

0 0 0

;

0

3

33 15

0

6.4

I

11

lo-’ 10-C 10-5 1o-4 10-a

1.8

1 1

3 2

0 0 0

2 5 2

2.3 3.3 7.0 1.3

ptac :insA ;insB construct in pCFl03AB served a transposase-like function, thereby restoring activity of ISl-3 on the same plasmid. The activity was not restored, however, by the InsA product only. (iii) Effects of overexpression of insA and insB genes on plasmid cointegration mediated by IS1 present in trans We examined the effect of overexpression of IS1 genes when the overproducers were present’ in trans. To this end, pCM plasmids having ptac-linked IS1 genes (Fig. 3(b)) were introduced into E. coli cells harbouring pCF103 containing ISl-3. Table 4 (lines 1 to 3) shows that the frequency of cointegration between pCF103 and pCM2004 having the ptac : insA ; insB overproducer did not increase. Therefore, the transposase function encoded by the ptac: insA ; insB in the pCM2004 could not act on the ISl-3 element in other plasmid genomes, although it did act on ISl-3 in the same genome, as described above. Table 4 (lines 4 to 9) also shows that when ptac : insA ; insB was present in trans, cointegration mediated by wild-type IS1 to form cointegrates was inhibited. 4s the concentration of IPTG was increased, the frequency of ISl-mediated cointegration decreased. When pCM2001 having only ptac : insA was of cointegration also present, the frequency decreased (lines 10 to 12). On the other hand, in the presence of pCM2001-3 having ptac : insA -3, cointegration took place at a level similar to that between pCF144 and ColEl (line 13). Overexpression of insB did not affect the cointegrate formatlion (line 14). Note that these frequencies were several-fold higher

572

C. Machida and Y. Machida

Table 4 Effects of trans overexpression of insA and insB genes on plasmid mediated by IS1 (ISl-3) and IS102

Plasmid

Overproducer

pCF103

ptac : insA ; insB

0 lo-’ 1o-3

pCFl44

ptac : inaA ; imB

0 lo-’ 1o-6 1o-5 10-h 1o-3

pCFl44

ptac : irwA

0 1w7 1o-3

pCF144 pCF144

ptac : in.sA-3 ptae : insB

10-3 lo-’

IPTG

Frequency of cointegration per division cycle ( x 109)

(M)

Number of examined cointegrates mediated ‘v ISI (W-3) IS102

5.3 r5.0 3.4

1 1 0

2 2 2

30 25 20 20 5.0 2.5

10 10 10 10 2 1

0 0 0 0 2 5

35 20 1.6

8 10 1

0 0 2

11 10

0 0

200 120

The experiments were carried out in E. coli NK5830, in which background recombination is lower than that in JM109, although both are recA mutants.

than those of cointegration without IPTG in the presence of ptac : insA ; insB or ptac : insA (Table 4, lines 4 and lo), indicating that strong expression of the IS1 genes with IPTG is not required for the reduction of the cointegration frequency. In contrast to ISl, the cointegration ability of IS102 was not affected by overexpression of IS1 genes (Table 4). The inhibition was, therefore, specifically observed in the ISl-mediated cointegration. Only the insA region was responsible for the inhibition, not the insB region. (c) InsA

protein is a negative regulator for expression of the IS1 genes

the

The results in the preceding section showed the inhibitory function of insA product. Subsequently, we examined whether products of the IS1 genes

eointegration

level of homologous

affect their own expression. The short DNA fragment containing pinsL (nucleotide positions 1 to 75 of 1st) or pinsR (nucleotide positions 768 to 559 of ISl) was joined to the coding region of the cat’-la& chimeric gene (Matsumoto et al., 1988). The fused gene was inserted into pBV234-Cm’, a derivative of RlOO which is compatible with pCM plasmids in E. coli cells, generating pBVl104 and pBV1105. pBVllO4 and pBV1105 were introduced into E. coli JMlOQ cells and also JMlOQ cells harbouring pCM plasmids with the ptuc-linked ISI genes (see Table 5). Plasmid pBVllO3, having only the cut’-la&’ coding sequence, was introduced. These E. coli cells were grown in the presence or absence of IPTG, and /?-galactosidase activity was measured. As shown in Table 5 (lines 2 and 7), approximately 350 and 150 units were detected in cells harbouring pBV1104 with pinsl, and

Table 5 Effects of the overexpression of insA and insB genes on their own expression _ Ijga&osidase

(units) +IPTG (1O-3

Plasmid (promoter)

pCM plasmid

pBVllO3

(no promoter)

No plasmid

19

26

pBV1104 pBVllO4 pBVllO4 pBVl104 pBVllO4

(pinsL) (pin&) (pinA) (pinsL) (p&L)

No plasmid pCM2004 pCM2001 pCM2002 pCM2001-3

360 114 114 362 338

338 22 24 278 367

pBVl105 pBV1105 pBVl105 pBVllO5 pBV1105

(pinsR) (pinaR) (p&R) (pin.~R) (pinsR)

No plasmid pCM2004 pCM2001 pCM2002 pCM2001-3

164 134 79 151 128

145 122 97 154 134

M)

Regulatory function pBV1105 with pinsR, respectively. When cells harbouring and with pBVllO4 pCM2904 ptac: insA :insB were grown in the presence of IPTG, /3-galactosidase activity was reduced 15-fold (line 3). Only one third of this activity was observed even in the absence of IPTG. The same level of reduction in fi-galactosidase activity was found in cells having ptac: insA (line 4). On the other hand, overexpression of the insB gene and the insAgene did not affect the level of /.I-galactosidase (lines 5 and 6). Neither products from ptac: insA ;insB nor ptac : insA inhibited the expression of cat’-la&’ directed by the pinsR (lines 8 and 9). These results clearly show that product(s) from the insA region, which was overproduced, specifically inhibited the gene expression directed by the insL promoter of ISl, although the products of insAand .insB regions did not.

4. Discussion Our present study shows that InsA protein represses ISl-mediated plasmid cointegration and expression of IS1 genes. The IS1 ability was inhibited when the insA gene or two IS1 genes were overexpressed in both cis and bans positions. When the expression of both IS1 genes is induced by IPTG at concentrations higher than 10e4 M, the efficiency of complementation of the mutant IS1 decreased even when the overproducer and the mutant IS1 were present in the same plasmid (cis) (Table 3). Although this might be due to inhibition by both overproduced InsA and InsB proteins, it is more likely due to overproduction of the InsA function as product, which has an inhibitory described below. When insA and insB genes were simultaneously overexpressed in trans, only the inhibitory effect was observed (Table 4): the ability of the insA mutant of IS1 was not restored and the level of cointegration mediated by wild-type IS1 was also markedly decreased. Such inhibition observed when only insA was overexpressed in trans. The InsB product was not required for inhibition. Introduction of a deletion mutation in insA abolished the inhibition. These results indicate that the InsA product is responsible for inhibiting the ISl-mediated cointegration. We propose that the InsA protein alone represses IS1 transposition. Our results also show that overexpression of insA results in reduced expression of IS1 genes directed by the insL promoter (Table 5). This reduction did not require the insB region. The overexpression of insA hardly affected the expression level directed by the insR promoter. Thus, the InsA product seems to specifically repress the gene expression directed by the insL promoter. Since the InsA protein binds the insL inverted repeat of IS1 in vitro (Zerbib et al., 1987), binding of the InsA protein to the insL region probably prevents the expression of IS1 genes. The InsA protein could its own expression by negatively regulate competing with RNA polymerase for the insL region, and thereby repress the transposition ability

573

of the IS1 gene

of ISl. Although recognition sites for the IS1 transposase are not yet known, the insA product could also interfere with the action of the transposase on the inverted repeat. Our in viwo results show that overproduction of insA hardly affected the level of gene expression directed by the insR promoter (Table 5). This suggests that the product of insA has no or very weak affinity to the insR repeat of ISl. Zerbib et al. (1987), however, showed that the InsA protein binds both the insR and insL inverted repeats in vitro. The difference in affinity of the InsA product to the inverted repeats could reflect the divergence of their sequences. The inconsistency between in vitro binding of InsA protein and our in vivo result may arise from the different experimental systems, and needs to be investigated further. Complementation experiments in the present study show that when both insA and insB genes are simultaneously overexpressed, the ability of the mutant IS1 present in the same plasmid as the overproducer (cis) is restored, whereas this mutant IS1 is not complemented by expression of only insA (Table 3). Since the insB region is also essential for ISl-mediated plasmid cointegration (Y. Machida et al., 1984), this finding may indicate that the insB region is not expressed from the ISl-3 sequence because of a polar effect of the deletion. Alternatively, the overproduced InsA protein may not form active transposase with the InsB protein even if synthesized from the mutant IS1 sequence. Both genes may be co-ordinately expressed from the IS1 sequence to synthesize the active transposase (Machida & Machida, 1987). Most recently, Sekine & Ohtsubo (1989) proposed that the shift in reading frame occurs during translation of a polycistronic messenger RNA covering insA, insB and the intercistronic region in such a way that insA and insB regions fuse in frame, generating the active IS1 transposase. Our complementation results are consistent with their proposal. We hope to gain further understanding of the mechanisms by which the IS1 transposase and the InsA repressor are synthesized and how the IS1 transposition can be controlled by these proteins. We thank Dr D. Davison for helpful discussions and critical reading of the manuscript and Dr I. Takebe for his encouragement. We also thank Drs E. Ohtsubo and H. Ohtsubo for providing the pBV234 plasmid and useful discussions and Dr N. Kleckner for providing E. coli NK5830. This work was supported in part’ by grants from the Ministry of Education, Science and Culture, Japan, and a grant to Y.M. from the Naito Foundation.

References Amann, E., Brosius, J. & Ptashne, M. (1983). Gene, 25, 167-178. Bolivar, F., Rodriguez, R. L., Green, P. J., Betlach, M. C., Heyneker, H. L., Boyer, H. W., Crosa, J. H. 8z Falkow, S. (1977). Gene, 2, 95-113. Casadaban, M. J., Chou, J. &, Cohen, S. N. (1980). J. Baeteriol. 143, 971-980.

574

C. Machida

Chou,
and Y. Machida (1982). Proc. Nat. Acad. Ski., U.S.A. 79, 277-291. Machida, Y., Machida, C. & Ohtsubo, E. (1984). J. Mol. Biol. 177, 229-245. Matsumoto, S., Takebe, I. & Machida, Y. (1988). Gene. 66, 19-29. Miller, J. H. (1972). Editor of Experiments in Molecula? Genetics, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Morisato, D., Way, J. C.. Kim, H.-J. & Kleckner, N. (1983). Cell, 32, 799-807. Nyman, K., Nakamura, K., Ohtsubo, H. b Ohtsubo. E. (1981). Nature (London), 289, 609-612. Ohtsubo, H. & Ohtsubo, E. (1978). Proc. Nat. Acad. Sci., U.S.A. 75, 615-619. Ohtsubo. H., Nyman, K., Doroszkiewicz, W. & Ohtsubo. E. (1981). Nature (London), 292, 6409643. Ohtsubo, H., Vassino, B., Ryder, T. & Ohtsubo, E. (1982). Gene, 20, 2455254. Sekine, Y. & Ohtsubo, E. (1989). Proc. XI&. Acad. Sci., U.S.A. 86, in the press. Simons, R. W. & Kleckner. N. (1983). Cell. 34, 683691. Yanish-Perron, C., Vieira, J. & Messing, ,J. (1985). Gene. 33, 1033119. Zerbib, D., Jakowec, M.. Prentki, P., Galas, 1). J. & Chandler, M. (1987). EMBO J. 6, 3163-3169.

by K. Matsubara