Identification of a unique specificity determinant of the colicin E3 immunity protein

Gene, 107 (1991) 133-138 0 1991 Elsevier Science Publishers B.V. All rights reserved. 0378-l 119/91/$03.50

133

GENE 06099

Identification

of a unique specificity determinant of the colicin E3 immunity protein

(Recombinant DNA; nuclease; inhibitor; bacteriocin; ColE3; ColE6; plasmid; molecular recognition)

Haruhiko Masaki,

Akiko Akutsu *, Takeshi Uozumi and Takahisa Ohta

Department of Biotechnology, Faculty of Agriculture. The University ofTokyo, Bunkyo-ku, Tokyo 113 (Japan) Received by A. Nakazawa: 21 January 1991 Revised/Accepted: 23 July/31 July 1991 Received at publishers: 15 August 1991

SUMMARY

Plasmid immunity to a nuclease-type colicin is defined by the specific binding of an immunity (or inhibitor) protein, Imm, to the C-terminal nuclease domain, T2A, of the colicin molecule. Whereas most regions of colicin operons exhibit extensive sequence identity, the small plasmid region encoding T2A and Imm is exceptionally varied. Since immunity is essential for the survival of the potentially lethal colicin plasmid (Col), we inferred that T2A and fmm must have co-evolved, retaining their mutual binding specificities. To evaluate this co-evolution model for the col and imm genes of ColE3 and ColE6, we attempted to obtain a stabilized clone from a plasmid which had been destabilized with a non-cognate immunity gene. A hybrid Col, in which the immE3 gene of the ColE3 was replaced with immE6 from ColE6, was lethal to the host cells upon SOS induction. From among this suicidal cell population, we isolated a stabilized, i.e., evolved, clone which produced colicin E3 (E3) stably and exhibited immunity to E3. This change arose from only a single mutation in ImmE6, from Trp48 to Cys, the same residue as in the ImmE3 sequence. In addition, we constructed a series of chimeric genes through homologous recombination between immE3 and immE6. Characterization of these chimeric immunity genes confirmed the above finding that colicins E3 and E6 are mostly distinguished by only CYSTSof the ImmE3 protein.

INTRODUCTION

Among the Co&directed E-group colicins, E2 and E8 act as DNases (Schaller and Nomura, 1976; Toba et al., 1988), Correspondence to: Dr. H. Masaki, Department of Biotechnology, Faculty of Agriculture, The University of Tokyo, Yayoi, Bunkyo-ku, Tokyo 113

(Japan) Tel. (81-3)3812-2111, ext. 3081; Fax@l-3)5684-0387. * Present address: Tsukuba Research Laboratories, Tokuyama Soda Co., Ltd., 40 Wadai, Tsukuba, Ibaraki 300-42 (Japan) Tel. @l-298)64-4023. Abbreviations: aa, amino acid(s); Ap, ampicillin; bp, base pair(s); cat, gene encoding Cm acetyltransferase; Cm, chloramphenicol; Col, colicin [also Co1 plasmid(s plasmid( col, gene encoding Col; DF13, cloacin DF13; E3, colicin E3; E6, colicin E6; Imm, immunity protein(s) which protects the host from the cognate colicin; imm, gene encoding Imm; ImmE3, E3 Imm; ImmM, E6 Imm; LB, Luria-Bertani (medium); nt, nucleotide(s); R, resistance/resistant; SOS, T2A, nuclease domain of colicin; wt, wild type; ::, novel joint (fusion).

and E3 (and probably E6) act as RNases (Ohno and Imahori, 1978 ; Mock and Pugsley, 1982; Akutsu et al., 1989). These activities are exclusively located on their C-terminal ‘T2A’ domains, and the plasmids protect the host cells from their own colicin actions by synthesizing cognate Imm. The immunity of nuclease-type colicins is thus defined by the specific binding of T2A and Imm (Ohno-Iwashita and Imahori, 1980; Jakes, 1982; Pugsley and Oudega, 1987). We have dete~ined the primary structures of several E-group co1 operons (Ma&i and Ohta, 1985; Masaki et al., 1985; Toba et al., 1988; Akutsu et al., 1989). Most operon regions exhibited extensive sequence identity and almost all nonhomologous residues were confined within the limited regions encoding T2A and Imm. These fmdings suggest that the plasmids have recently diverged through alteration in these regions, their immunity specificities changing as a result. Some mutation within either the T2A

134 or Imm region affecting the binding specificity might have rendered the host cells lethal and consequently evoked complementary mutation(s) within the counterpart protein, giving rise to a new Co1 plasmid with an altered immunity specificity. The new plasmid should kill the parental colicinogenic cells and survive in some cir~umst~~es. We expected that one of these elementary steps in plasmid evolution might be demonstrable under particular experimental conditions.

---__-----

Colicin E6 shows extensive sequence identity to E3 and shares immunity specificity with cloacin DF13 (Akutsu et al., 1989). Sequence comparison of these three bacteriotins led to the prediction that, at most, eight of the 97 aa in T2A should distinguish between ImmE3 and ImmE6 (or ImmDFl3), and nine of the 84 aa in Imm between E3 and E6 (or DF13) (Fig. 1). The E3 and E6 pair is thus promising as starting material for experimental co-evolution to obtain a stable variant from a plasmid which has been artificially destabilized with a noncognate immunity gene.

_____-__- ______________________”____________________________~__ ____________________ .

467

472

.

.

478

.181

463.

ColE3 ColE6

AGGGA---

_ --

__~_

CloDF13

ColE3 ColE6

CloDF13

. ColE3

. . . . . . . . . . . . . . . . . . . . . ...*....

ColE6 CloDF13

T..C.....A..A..C Asp

. . . . . . . . . ..CG.A..C..C...G.......T..A..A..G.....G...G.T.....G..G Glu Al8 Asp

(BarnHI). SauSAI

LYs

Vd

ColE3 ColE6 CloDFt3

TGGGGCTTAAATT

A~GGT~GATAAGACGACCGA snTrpPheAspLysThrThrG1 . . . . . . . . . . . . . ..AA.....

. . . . . . ..T..T..

ColE3 ColE6 CloDFl3

ColE3 ColE6 CloDF13

. . . . . . . .._.................................................................................A..........

.

BCII

.

CAAATTGATAn”PCCGATAATGAGTA~G~CG~ATTATCOTGATGGTGA~~AATA~A~AGGGATGAG~A~TACGGGC~C GlnfleAspIleSerAspAsnGluTyrPhaVelSerPheAspTyrArgAspGlyAspTrp GT...C . . . . ..AGTA.ATT...T..C......A..C....T...C..G......A.C....A.O.....AG......A...........AA...A...... Asn Vsl LysPheAsp Val

Fig. 1. Nt and aa sequences of the T2A-imm regions of ColE3-CA38 (Masaki and Ohta, 1982; 1985), COW-CT14 (Akutsu et al., 1989; Lau and Condie, 1989) and CloDF13 (Van den Elzen et al., 1983). For the ColE3 and CloDF13 sequences, only the residues not identical to those of ColE6 are presented. Possible determinants of the E3 and Eli (or DFl3j immu~ty specificities are boxed in both the T2A and Imm regions. The n~~ng of aa starts at the N-terminal Met for both the colicin and the Imm, though the purified Imm, at least ImmE3 and ImmDF13, lack the N-terminal Met (Mochitate et al., 1981; Van den Elzen et al., 1983). A* adjacent to the Sau3AI site at the end of the E3-T2A region was converted to G, introducing a unique BarnHI site without changing the encoding aa (pSH357, Fig. 2A). The unique CIaI site is at about 230 bp upstream beyond the area described in the figure, and the BclI site is marked at the end of the immE3 gene. Six types of chimeric genes between im&3 and timE6 were constructed through homologous recombination at regions # 1, # 2 and # 3.

135 EXPERIMENTAL

that the phenotypic

AND DISCUSSION

transversion (a) Construction of a colE34mmE6 hybrid plasmid The parental ColE3 plasmid used in this experiment was pSH357, which is ApR and carries the wt E3 operon (Masaki and Ohta, 1985), except that the Sau3AI site near the 3’ end of the colE3 gene had been converted to a BamHI site without changing the encoding aa (paper in preparation). In pSH357, the T2A-encoding region and the immE3 gene are roughly subdivided by three unique restriction sites, ClaI, BamHI and BclI (Figs. 1 and 2A). A colE3itnmE6 hybrid plasmid, pSH358, was constructed placing the BamHI-BcZI fragment of pSH357

by reby the

Sau3AI fragment (immE6) of plasmid ColE6, using an recA _ strain as a host in order to repress the colicin expression from the SOS-inducible colicin promoter (Masaki and Ohta, 1985). When pSH358 was transferred to a recA + strain, it made the host cells lethal on either the addition of an SOS inducer or elevation of the growth temperature, although the plasmid could be maintained relatively stably at 30°C in the absence of the inducer. (b) Isolation of a stabilized plasmid E. coli KH1250recA ‘mutD5 (Maruyama et al., 1983) was transformed with pSH358 and then incubated overnight at 37°C in LB containing 50 pg Ap/ml and 50 PM vitamin B12, which protects the cells from attack by the colicin spontaneously produced in the medium (DiMasi et al., 1973). The cells were transferred to the same fresh medium and then incubated another night. The cell suspension was spread on LB agar plates containing Ap, with or without colicin E3. Phenotypically stabilized clones appeared on the plates without colicin, some of which proved to have lost its colicinogeny possibly through the IS1 insertion. From a plate containing colicin, we isolated a clone carrying pEV1, which produced E3 stably and showed immunity to both E3 and E6. (c) Identification of the mutation site(s) of pEV1 To localize the possible mutation site(s) in pEV1, we prepared a series of CmR acceptor plasmids, into which a restriction fragment bearing the mutation(s) can be rescued in one step (Fig. 2A). Phenotypes producing E3 and immune to both E3 and E6 were recovered by the transfer of the BumHI-BcZI fragment to the imm-acceptor plasmid. Then the removal of the HincII-cut-cartridge from the blu gene restored ApR and gave rise to pEV5, which has the same sequence as pSH357 and pSH358 besides this BamHI-BcZI segment. The phenotypes of pEV1 and pEV5 were indistinguishable and the mutation(s) were restricted within the BumHI-BcfI fragment. Sequence analysis (Yanisch-Perron et al., 1985) showed

change

in immE6, TGG

of pEV5 is due to a single (Trp4’) to TGT (Cys). This

indicates that ImmE6 acquired immunity toward E3 only through a single replacement of Trp4’ by Cys. Interestingly, Cys is the 48th residue of ImmE3, and appears only once throughout the E3, E6, ImmE3 and ImmE6 sequences. On the other hand, plasmid pEV5 did not lose the original immunity toward E6, suggesting that Trp4x is not the absolute immunity determinant of ImmE6. (d) Construction Sequence determinants

of chimeric immunity genes

comparison restricted the possible immunity of ImmE3 and ImmE6 to nine aa pairs,

including Cys and Trp at position 48 (Fig. 1) (Akutsu et al., 1989). To evaluate the contributions of these aa to each immunity specificity, the immE3 and immE6 genes were cloned in tandem into pUC18 (Yanisch-Perron et al., 1985) and three immE3: :immE6 and three immE6 ::immE3 chimeric genes were constructed through recAdependent homologous recombination (Fig. 2B). Then, the nine aa possibly involved in immunity determination were divided into four groups by three recombination regions, # 1 to # 3 in Fig. 1 (also Fig. 3). Since the C-terminal 44% regions are identical in ImmE3 and ImmE6, the recombination at site # 3 consequently represents a point mutation and the immE63C gene thus encodes the same aa sequence as that of pEV5 (pAM63C in Fig. 3). (e) Uniqueness of the immunity determinant in ImmE3 The properties of the chimeric genes were consistent with those of pEV5. The results of phenotypic examination (Fig. 3, the cross-streak test) suggest that a single replacement of Trp4’ by Cys in ImmE6 (pAM63C) confers the E3 immunity, the E6 immunity being retained. Coincidentally, replacement of Cys4’ by Trp in ImmE3 (pAM36C) abolishes the E3 immunity and the E6 immunity is not acquired. Thus, Cys 48 should be a unique immunity determinant for ImmE3, while Trp 48 is not the absolute immunity determinant for ImmE6. The uniqueness of Cys4’ in ImmE3 suggests that its counterpart in the E3-T2A region also consists of one or only a few aa. This will be confirmed when a E6 derivative with E3 specificity is obtained based on an alternative experimental co-evolution starting with a colE6-immE3 hybrid plasmid, but we have not succeeded in this type of co-evolution so far. Since an ‘evolved’ colE6 gene would be accompanied by the native immE3 gene, a new plasmid molecule will be, if at all, still sensitive to E6 and will acquire a selective advantage only after complete segregation from its parental plasmid, unlike in the case of the evolution of the immunity gene, in which a stabilizing mutation confers an immediate selective advantage.

136

H

PXHB PXH

b/a

pUC18 ColE3

Q

B

--f=-b COlE6

I jy

(4) T

Fig. 2. Structural design ofplasmids. (A) Isolation of a stable mutant from a destabilized Co1 plasmid, pSH358, and restriction of the mutated segment. Plasmid pSH357 is a pSH350-derivative (Toba et al., 1986) and has the wt colicin E3 operon, except for a unique BamHI site (Fig. 1). Step 1: the BarnHI-BclI fragment of pSH357 carrying immE3 was replaced by the corresponding Sau3AI fragment of ColE6CTl4, giving rise to pSH358. Step 2: a stabilized (evolved) plasmid, pEV1, was isolated from the suicidal plasmid, pSH358. Step 3: the b[u gene was disrupted by inserting the H&II-cut cartridge (Close and Rodriguez, 1982), and the BumHI-lucZ cartridge from pMCl871 (Casadaban et al., 1983) was inserted in-frame near the end of the colE3 gene. Steps 4 and 5: deletion of the EclI segment, and then the ClaI segment gave rise to &n-acceptor and TZA-imm-acceptor plasmids, respectively. Step 6: the BumHI-BclI fragment of pEV1 was rescued with the CmR imm-acceptor plasmid. Step 7: deletion of the Him%-cut fragment restored the blu gene, and pEV5 showed the same phenotype as pEV1. (B) Construction of chimeric genes from immE3 and immE6. Step 1: the immE3 Suu3AI fragment was inserted into the PsrI site ofpUC18 (Yanisch-Perron et al., 1985) after the sites had been flush-ended. Step 2: insertion of the immE6 Suu3AI fragment gave rise to a double-immune plasmid, pAM63. Steps 3 and 4: pAM63 was linearized and introduced into E. coli RR1 (Ma&i et al., 1985), and a series of immE63 chimeric genes was obtained through homologous recombination (Akutsu et al., 1989). Step 5: a control plasmid having the wild-type immE3 gene was constructed by inserting the ColE3-CA38 Suu3AI fragment into the BumHI site ofpUC18. Steps l’-5’: similar procedures, starting with ColE6CT14, gave rise to immE36 chimeric plasmids and a control plasmid having the wt immE6 gene. B, BumHI; Bc, BclI; C, ClaI; H, HincII; P, PstI; X, XbuI. The structures of ColE3-CA38 and ColE6-CT14 were described previously (Masaki and Ohta, 1985; Toba et al., 1986; Akutsu et al., 1989).

137 E3

Plasaid

1m11 protein

CTDBSspot streak

test

immunity

cl-DS.¶ spot strcsk

test

2’8

pOCl8 pAMGO pAH36A pAH36B pAH36C pAff30 pA~B3A pAM63B pAM63C

E6

immunity

2’5

2'5

t

213 210

-29

<2O

26 2'2

2'0 t

<2@

-

t

<20

-

219

;

29 25

t t

;,"

2'4

Fig. 3. Immunities to E3 and E6 conferred by the wt (pAM30, pAM60) and six ehimeric (pAM36A-C, pAM63A-C) plasmids in the absence of the fat inducer. The Imm are schematically represented by arrows; the blackened, open and hatched areas indicate the immE3- and ImmEd specific sequences, and the C-terminal identical sequence of immunity proteins, respectively. For the cross-streak test (Masaki and Ohta, 1985), a plus symbol denotes immunity and a minus symbol sensitivity to each colicin. For the spot test (Ma&i and Ohta, 1985), the apparent titers of the given E3 and E6 samples, which were serially diluted by a factor of two, are presented. A value of < 2* represents complete immunity, and 2” for E3 and 2’s for E6 complete sensitivity in this experiment.

(f) Evaluation of the immunity determinants in ImmE3 and ImmE6 In the spot test (Fig. 3), the apparent titers of purified E3 and E6 samples toward cells harboring various imm plasmids were compared quantitatively. Here again, replacement of TrpJ8 by Cys in ImmE6 confers a great extent of the E3 immunity (pAM63~ vs. pAM60), and replacement of CYSTSby Trp in ImmE3 greatly reduces the E3 immunity (pAM36C vs. pAM30). Moreover, pAM63C (ImmE63C) exhibited a lower value toward E3, i.e., higher immunity to E3, than pAM36C (ImmE36C). ImmE63C has Cys4’ as the only ImmE3-speci~c aa, while fmmE36C is the same as ImmE3 except for at position 48 (Trp). This means that CYSTSalone (ImmE63C) contributes much more to the E3 immunity specificity than the eight other ImmE3-specific aa together (ImmE36C). The results of the spot test also suggest that other aa affect each immunity specificity to various extends. Alteration of AspI and Glu2’ of ImmE3 to those of ImmEG seriously reduces the E3 immunity (pAM63B), but unlike in the case of Cys4’, the acquirement of these two residues alone does not confer suf%cient E3 immunity (pAM36B), suggesting that Asp” and/or GluZo are subdeterminants of E3 immunity. In contrast with the case of ImmE3, the determinants of ImmE6 seem to be more dispersed over the sequence and operate well in an accumulative manner. His6 and/or Ile7 are needed but not sufficient by themselves for the full E6 immunity (pAM36A vs. pAM63A). Likewise, Trp48 is operative but not sufficient for the E6 immunity (pAM36C vs. pAM63C).

(g) Conclusions The extensive sequence similarity of most regions of the colicin operons suggests a rather recent divergence of these bacteriocins. A limited number of nonidentical aa is, however, not randomly distributed but confined within a small variable protein region, T2A, as well as in the cognate Imm, both of which exclusively define the immunity specificity of a plasmid. The strict immunity specificity is indispensable for the survival of the host and consequently of the Col. This trait implies strong selective pressure operative not only among Col, but also between the co1 and imm genes within a plasmid. We would like to refer to this special kind of co-evolution as ‘the hand-in-hod evolution’, which m~ntains the mutual binding specificity between colicin and Imm. Our present experiment suggests that a similar process has actually formed a part of the elementary steps in the evolutionary history of ColE plasmids. Furthermore, in this experiment, we directly identified the unique imm~ity determinant of ImmE3 which distinguishes E3 and E6. This finding should be important for further understanding of the structural bases of specific protein-protein interactions. The binding specificities of T2A and Imm must finally be explained in terms of physicochemical parameters. Many kinds of T2A and Imm prepared in the present study are promising for further examinations along these lines.

ACKNOWLEDGEMENTS

We wish to thank H. Nakae and S. Yajima for their helpful cooperation. This work was supported by a Grantin-Aid for Scientific Research from the Ministry of Education, Science and Culture, Japan.

REFERENCES Akutsu, A., Masaki, H. and Ohta, T.: Molecular structure and immunity specificity of colicin E6, an evolutionary intermediate between E-group colicins and cloacin DF13. J. Bacterial. 171 (1989) 6430-6436. Casadaban, M.J., Martinez-Arias, A., Shapira, SK. and Chou, J.: j-Galactosidase gene fusions for analyzing gene expression in Escherichia co/i and yeast. Methods Enzymol. 100 (1983) 293-308. Close,T.J. and Rodriguez, R.L.: Construction and characterization ofthe chIoramphenicol-resistance gene cartridge: a new approach to the transcriptional mapping of extrachromosomal elements. Gene 20 (1982) 305-316. Cole, S.T., Saint-Joanis, B. and Pugsley, A.P.: Molecular characterization of the colicin E2 operon and identification of its products. Mol. Gen. Genet. 198 (1985) 465-472. DiMasi, D.R., White, J.C., Schnaitman, CA. and Bradbeer, C.: Transport of vitamin B12 in Escherid& co& common receptor sites for vitamin B12 and the E ‘colicins on the outer membrane of the cell envelope. J. Bacterial. 115 (1973) 506-513.

138 Jakes, K.S.: The mechanism

of action ofcolicin

DF13. In: Cohen, P. and Van Heyningen, of Toxins

and Viruses.

Elsevier,

Amsteram,

Lau, P.C.K. and Condie, J.A.: Nucleotide E6 and E9 operon: presence in the ColE9-j plasmid. Maruyama, (mutD5)

E2, colicin E3 and cloacin S. (Eds.), Molecular

1982, pp. 13 l-167.

sequences

of a degenerate

from the colicin ES,

transposon-like

Mol. Gen. Genet.

structure

T., Maki, H. and Sekiguchi,

and recessive

(dnaQ49)

M.: A dominant

and the inhibitor

protein

region encoding

of colicin E3-CA38.

the active fragment

FEBS Lett. 149 (1982)

129-132. Masaki,

H. and Ohta, T.: Colicin E3 and its immunity

genes. J. Mol. Biol.

182 (1985) 217-227. Masaki,

H., Toba,

ColE2-P9 Mochitate, immunity

immunity protein

Natl. Acad.

T.: Structure

gene. Nucleic

of the

Acids Res. 13 (1985) 1623-1635.

K. and Imahori, (B subunit)

and expression

K.: Amino

of colicin

acid sequence

E3. J. Biochem.

1609-1618. Mock, M. and Pugsley, A.P.: The BtuB group Co1 plasmids between the colicins they encode. J. Bacterial.

of

89 (1981)

of the functional

84

loci in

of their proteoly-

19 (1980) 652-659.

B.: Methods

for studying

K.G. (Ed.), Plasmids:

colicins

A Practical

and their Approach.

1987, pp. 105-161.

K. and Nomura,

M.: Colicin E2 is a DNA endonuclease.

Proc.

Sci. USA 73 (1976) 3989-3993.

Toba, M., Masaki,

H. and Ohta, T.: Primary

structures

of the ColE2-P9

and ColE3-CA38 lysis genes. J. Biochem. 99 (1986) 591-596. Toba, M., Masaki, H. and Ohta, T.: Colicin E8, a DNase which indicates an evolutionary

M. and Ohta,

K., Suzuki,

In: Hardy,

J. Biochem.

by the characterization

Biochemistry

A.P. and Oudega,

IRL Press, Oxford, Schaller,

H. and Ohta, T.: A plasmid

K.: Assignment

colicin E2 and E3 molecules

plasmids.

of Escherichiu coli. J. Mol.

K.: Colicin E3 is an endonuclease.

(1978) 1637-1640. Ohno-Iwashita, Y. and Imahori,

Pugsley,

Biol. 167 (1983) 757-771. Masaki,

Ohno, S. and Imahori,

tic fragments.

217 (1989) 269-227.

M., Horiuchi,

mutator

Action

relationship

between colicins E2 and E3. J. Bacterial.

170 (1988) 3237-3242. Van den Elzen, P.J.M., Walters, H.J.J.: Molecular bacteriocin

structure

protein

H.H.B.,

Veltkamp,

and function

of plasmid

E. and Nijkamp,

of the bacteriocin

CloDFl3.

Nucleic

gene and

Acids

Res.

11

(1983) 2465-2477. and homology

150 (1982) 1069-1076.

Yanisch-Perron, cloning Ml3mpl8

C., Vieira, J. and Messing,

vectors

and

and pUC19

host

strains:

vectors,

J.: Improved

nucleotide

Ml3

sequences

Gene 33 (1985) 103-119.

phage of the