Proc. Nail. Acad. Sci. USA Vol. 82, pp. 8389-8393, December 1985 Biochemistry Structural and functional organization of the colicin El operon ( expression/induction/SOS system/and-sense RNA/molecular cloning) NAHID S. WALEH AND PAUL H. JOHNSON Molecular Biology Program, SRI International, Menlo Park, CA 94025 Communicated by Robert L. Sinsheimer, August 16, 1985

ABSTRACT We analyzed the structural and functional caused by the lys gene product is biochemically distinct from relationships among independently cloned segments of the cell death caused by the action of either exogenous or ColEl region that regulates and codes for colicin El endogenous colicin on the nonimmune bacterial membrane. (cea), immunity (imm), and the mitomycin C-induced lethality The results presented in this report demonstrate that the function (lys). On the basis of physiological properties, restric- colicin El operon represents a novel gene arrangement in tion endonuclease mapping, and DNA sequence analysis, the . The discontinuity of the colicin operon and following recombinant were determined to represent the overlapping nature of individual transcriptional units and three major functional classes: pNP12 (cea+, imm+, lys+), regulatory sequences suggest an unusual economy ofgenetic pNP4 (cea+, imm+, lys-), and pNP6 (cea+, imm-, lys-). Our information that might have several important consequences results have established the order, boundaries, and relative for the control of gene expression in this system. orientation of the three structural , the location of the promoter region for imm gene transcription, and the predicted MATERIALS AND METHODS amino acid sequences of the imm and lys gene products. Hydropathicity analysis suggests that both proteins have hy- Strains and Media. A colicin-producing derivative strain of drophobic amino acid segments characteristic of membrane- JC411 (7) was the source of ColEl plasmid. E. coli K-12, associated proteins. A model for the structure and expression strain 294 (endoP hsdR- hsdM' thi-) (16), was the host for of the colicin El operon is proposed in which the cea and lys pBR322 and the recombinant plasmids. Strain CL142 (K-12 genes are expressed from a single inducible promoter that is Row) was used as the indicator strain that was sensitive to all controlled by the lexA repressor in response to the SOS system colicins (17). of Escherichia coli. The imm gene lies between the cea and lys Preparation of Sheared ColEl Fragments. Plasmid DNA genes and is expressed by transcription in the opposite direction was purified as described (18). Approximately 140 ,ug of from a promoter located within the lys gene. This arrangement open-circular ColEl DNA in 0.3 M sodium acetate was of structural genes indicates that the transcriptional units for sheared at 3800 rpm for 20 min at 00C, using the microattach- all three genes overlap. We suggest that the formation of ment of a DuPont Omnimixer. The sheared DNA was anti-sense RNA may be an important element in the coordinate purified by velocity centrifugation using a 5-20% linear regulation of gene expression in this system. sucrose gradient. Colicin Tests. The enhanced production of colicin by Colicin El can be induced by treating cells containing transformants was determined by the procedure ofMales and plasmid ColEl with agents that damage DNA or interfere Stocker (19). Colicin was released from induced cells by with DNA synthesis (1). Previous work has demonstrated using chloroform, according to the procedure of Suit et al. that colicin induction is one of the SOS responses (2-4). The (15). SOS repressor, lexA protein, is also the colicin El gene repressor (5). The colicin El gene (cea) has been sequenced RESULTS (6), and its protein product has been purified (7). Colicin El has been characterized as a protein that forms ion channels Molecular Cloning of the Colicin Gene Region. Sheared in the bacterial membrane, leading to cell death (8, 9). fragments of ColEl DNA were cloned into plasmid pBR322 The size and location of the colicin El immunity gene using the poly(dG-dC) tailing procedure (20). Twenty (6.8%) (imm) have not been well-documented. On the basis of of 368 randomly selected tetracycline-resistant (Tcr) ampicil- transposon mutagenesis, a region of the ColEl plasmid lin-sensitive (Aps) transformant colonies were found to pro- adjacent to the colicin El gene has been proposed for the duce colicin (Col+). Transformant strains were designated immunity function (10-12). Transcriptional studies and DNA NP1-NP20; the corresponding plasmids were designated sequence analysis indicate that an open reading frame exists pNPl-pNP20. in which the transcriptional direction is opposite that of the Physical Mapping. Restriction endonucleases Pst I, EcoRI, colicin El gene (13). The immunity protein has not been and Sma I were used to construct maps of ColEl-derived isolated, however, and there is no direct evidence that the fragment inserts carried by Col' recombinant plasmids pNP1- presumed polypeptide product of this DNA sequence is pNP20. Fig. 1 presents a summary of these results in which responsible for immunity to colicin El. the individual physical maps are oriented to the known Upon induction, death of the bacteriocinogenic cells oc- organization and direction oftranscription ofthe colicin gene, curs concomitantly with the overproduction of colicin. Cell the proposed immunity region, and the origin of ColEl death occurs in the absence of active colicin molecules (14). replication (11). Recombinant plasmids are ordered by in- A gene (here designated lys) that is close to or overlaps the creasing length of the Pst I fragment that contains the immunity region ofColEl plasmid has been postulated for the COOH-terminal coding region ofcolicin and extends through mitomycin C-induced lethality function (15). Cell death the proposed immunity region toward the ColEl replication origin. Arrows indicate the orientation of the inserts to the The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" Abbreviations: cea, colicin gene; imm, immunity gene; lys, lysis in accordance with 18 U.S.C. §1734 solely to indicate this fact. gene. 8389 Downloaded by guest on September 25, 2021 8390 Biochemistry: Waleh and Johnson Proc. Natl. Acad. Sci. USA 82 (1985)

Plasmid Physical Map NP12, NP6, and NP4 are presented in Fig. 2 and Table 1. Plasmids pNP12 and pNP6 contain fragments of ColEl DNA Cea 1mm that extend, respectively, 580 and 50 base pairs beyond the 600bp ' COOH-terminal end of the colicin gene. Strains carrying Col El these two plasmids, however, exhibit similar sensitivity ' PstI Pst I patterns to mitomycin C. A decrease by a factor of 10 in the pNP6 ~~~~~~~~~~Ori pNP77 viability of both cultures is observed within 2 hr after the pNP1 _----- addition of the inducing agent (Fig. 2). pNP13 --_ __ _i_(_ ) C induction, cells carrying plasmids with pNP 8 s-___ Upon mitomycin pNP 4 a functional lys gene are killed and colicin is released into the pNP14 H medium (15). Moreover, plasmid-carrying cells lacking a pNP12 (.) functional lys gene but protected from the action of endog- pNP16 ' .--- enous are not pNP17 (4) colicin by the immunity function killed, and the pNP99() colicin is not released into the medium. The results presented pNP 5 (+) in Table 1 indicate that most of the colicin synthesized by pNPl1 ( pNP155 , ------NP12 cells is released, whereas colicin made by NP6 cells is pNP10 >___ + released only when the cells are disrupted with chloroform. pNP 2 - --- These results, together with the observation that NP12 cells pNP19 (__- pNP 3 are immune to colicin El and yet are killed upon induction pNP20 with mitomycin C, indicate that pNP12 carries a functional lys gene. On the basis of a similar argument, we conclude that pNP6 lacks a functional lys gene and that the observed FIG. 1. Physical maps ofColEl-derived fragments. Wavy arrows represent the direction of transcription of the colicin and immunity lethality is due to the intracellular toxicity of colicin in the genes. Arrowheads indicate the orientation ofthe inserts with respect absence of a functional immunity gene. Plasmids pNP1, to the direction oftranscription from the A-lactamase promoter ofthe pNP7, and pNP8, which also lack the immunity function, cloning vector pBR322. Ori represents the replication origin of confer upon their host the same properties with respect to ColEL. Heavy lines indicate the Pst I fragments containing the mitomycin C-induced lethality and colicin release as does carboxyl-terminal coding region of the colicin gene. The presence pNP6 (data not shown). and absence of immunity function is indicated by (+) and (-), respectively. bp, Base pairs. Plasmid pNP4 contains a region that extends 400 base pairs beyond the COOH-terminal end of the colicin gene. Strain NP4, in contrast to NP6 and NP12, shows only a partial loss direction of transcription from the p-lactamase promoter of ofcell viability upon mitomycin C induction. Phenotypically, the pBR322 cloning vector. The Col+ fragments were insert- ed in both orientations with approximately equal frequency. NP4 is imm'. However, when the 600-base-pair Pst I Plasmid pNP18 was not mapped because it produced ambig- fragment of pNP4 (Fig. 1) is isolated and recloned into uous results upon double digestion with endonucleases Pst I and EcoR. 100 Functional Characterization of Transformants. To distin- guish between transformant strains that were immune to 50 colicin El and those that had lost their outer membrane btuB protein and had therefore become resistant to all E colicins, we tested strains NPl-NP20 for immunity to colicins El-E7. The btuB protein is essential as a receptor-binding site for E colicins, vitamin B-12, and phage BF23 (21-23). The immu- 10 nity properties of the transformed strains are represented in Fig. 1 by a (+) for recombinant plasmids that were immune 0.50 to colicin El and a (-) for those that were not. -a U) Fifteen ofthe transformants were immune to colicin El, as Un demonstrated by their lack of susceptibility to colicin El and -j Jr their sensitivity to all other E colicins. Strains NP6 and NP7, > although colicinogenic, lacked colicin El immunity. During W their initial testing with all of the E-producing strains, strains 0.10 NP1, NP8, and NP13 showed either no zone of inhibition or a faint narrow zone of inhibition. These strains, however, 0.05 were sensitive to non-E colicins such as group B colicins (24). To establish the immunity property of these presumed btuB mutant strains, the plasmid DNA was isolated from each ofthe transformants described above and was used to transform E. coli 294. Several colonies from each transformation plate were 0.01 picked and tested for immunity to colicin El and other E colicins. In each case, 20-60%o ofthe colonies were sensitive to all E colicins. These clones maintained colicinogenicity and 0.0 0.5 1.0 1.5 2.0 2.5 tetracycline resistance. The remaining clones were resistant to TIME AFTER INDUCTION (hours) all E colicins. Strains NP1, NP6, NP7, NP8, and NP13 grew media. FIG. 2. Survival of cells containing various ColEl-derived plas- poorly in both solid and liquid mids after treatment with mitomycin C. Cells were grown in L broth To correlate the structure of various regions of the colicin at 370C to mid-logarithmic phase. Mitomycin C at 2 ,Ag/ml was added gene cluster with specific functional properties, we analyzed at time 0. At the indicated times, samples were removed, diluted, and the recombinant plasmids represented in Fig. 1 for cell plated on L plates supplemented with tetracycline at 25 Ag/ml. viability and colicin production and release as a function of Percentage survival is expressed as the ratio of the number of time after induction with mitomycin C. The results for strains survivors at a given time to the number of viable cells at time 0. Downloaded by guest on September 25, 2021 Biochemistry: Waleh and Johnson Proc. Natl. Acad. Sci. USA 82 (1985) 8391 Table 1. Colicin production and release after induction of quences for the proposed imm and lys gene products. The recombinant plasmids with mitomycin C imm gene is predicted to encode a 113-amino acid protein that is transcribed in a direction opposite that of cea, and the lys Time, pNP12 pNP4 pNP6 gene is predicted to encode a 45-amino acid protein that is hr C T C T C T transcribed in the same direction as the cea gene but is 0 0.1 0.02 0.02 0.1 0.001 0.02 separated from it by the imm gene. 1 1 1 1 1 0.01 0.2 The complete double-stranded DNA sequence of this 2 10 10 1 10 0.02 2 region (630 base pairs) and the proposed amino acid se- 3 20 20 1 20 ND ND quences of the imm and lys gene products are shown in Fig. 4 100 100 2 100 0.2 100 4. Initiation of for the immunity protein begins at Samples from induced cultures were serially diluted in L broth and nucleotide 373 and ends at nucleotide 34, overlapping the were spotted onto a lawn of strain CL142 before and after 2 hr of colicin termination codon. Seven bases to the 5' side of the treatment of the diluted sample with chloroform. Plates contained imm initiation codon is a potential ribosome-binding site streptomycin at 100 ,ug/ml to allow only the growth of the colicin- having the sequence 5' -GAGGT- 3'. The lys protein trans- sensitive indicator strain CL142. Results represent colicin titer lation begins at nucleotide 421 and ends at nucleotide 556. A (x 10-3). The colicin titer is defined as the reciprocal of the last possible ribosome-binding sequence, 5' -TAAGGA- 3', exists dilution giving noticeable clearing of the indicator lawn. ND, not done; C, control; CHCl3-treated. 7 bases to the 5' side of the lys initiation codon. Further T, aspects of the structural and functional properties of this pBR322, only the orientation with the f3-lactamase promoter region are discussed below. in the correct direction for immunity gene transcription from Protein Structure Prediction. The locations of hydrophobic it shows immunity function (H.-I. Oh, personal communica- and hydrophilic regions of both the lys and imm proteins are tion). This suggests that pNP4 has the complete immunity presented in Fig. 5 A and B; the average hydropathicities

gene-coding sequence but does not have its natural transcrip- 10 20 29 46 . Asn Asn Lys Ile Ser Ile Phe Leu tional promoter. Therefore, the intermediate level ofNP4 cell TTGAATTATG A TAT TTA CTC CAC AAT CCC TAAAT TMA TAA GAA ATA ACT ATA rrT TTC viability (Fig. 3) in the absence of the lys gene may be the 5' -AACTTAATAC T ATA MT GAG GTG TTA GGG ATTTA ATT ATT CTT TAT TGA TAT AAA AAG result of reduced immunity expression, possibly as a conse- Col(1-515) - lie Asn Glu Val Leu Gly Ile quence of a reduced transcription level due to the G+C-rich 61 76 91 106 Gly Leu Leu Gly Phe Pro Ile Ala Leu Val Leu Cys Leu Leu Asn Tyr Leu MET region resulting from the cloning procedure used and/or a AGG ATC GTT AGG TTT ACC CTA ACG ATC CTG GTT TGT TTC ATC TAA TAT ATC GTA weaker f-lactamase promoter. Our conclusion that pNP4 5 -TTC TAG CAA TCC AM TGG GAT TGC TAG GAC CAA ACA MG TAG ATT ATA TAG CAT does not produce a functional lys gene product is also 121 136 151 166 Phe Leu Asn Leu Lys Ala Val Pro Ala Asn Asn Leu Asa Lys Arg Arg Glu Phe supported by the results shown in Table 1, which indicate that TTT ATC CAA ATT AAA ACG ATG CCC CCG CAA TAA ATC CAA AAA AGA AGA MG CTT the colicin synthesized upon induction appears to remain 5'-AAA TAG GTT TM TTE TGC TAC GGG GGC GTT ATT TAG GT TTT TCT TCT TTC GAA intracellular, because most of it can be detected only when 1 6 21 1 Phe Asp Lys Lys Ile81 Phe Asn Phe Ala 19Ile Tyr Glu Ile Ala Tyr Lys Ser Tyr the cells are disrupted with chloroform. TEm TAG AM GM ATA CTT CM TTT TCG ATA CAT MG TTA TCG TAT AAA ACT TAT Structural Organization. To examine the structure of the 5'A-MA ATC TTT CTT TAT GM GTT AAA AGC TAT GTA TTC MT AGC ATA mTT TGA ATA we 226 241 256 271 proposed imm-lys genes in further detail, analyzed the Pro Cys Leu Ile Thr Ser Ile Val Ile Ala Tyr Leu MET Lys Asp Ser Val Leu DNA sequence of this region. The DNA sequence presented ACC TGT ATC TTA TCA CGA ATA GTG ATA ACG TAT ATC GTA GAA TAG ACT GTG TTC in Fig. 4 is a compilation of various overlapping sequences 5'-TGG ACA TAG MT AGT GCT TAT CAC TAT TGC ATA TAG CAT CET ATC TGA CAC AAG 2 86 3 01 3 16 reported in the literature (2, 6, 13, 25) and partial sequences Phe Tyr Tyr Gly Glu Ser Asn Lys Thr Ile Leu Tyr Ile Tyr Ile Leu Thr Cys CTT TAT TAT GGG MG CGA CM AM ACA ATA TTC CAT ATA TAT ATA TTC ACA CGT of selected recombinants described here (C. Green, personal 5-GM ATA ATA CCC communication). The final sequence compilation includes the TTC GCT GTT TTT TGT TAT AAG GTA TAT ATA TAT MG TGT GCA 331 346 361 383 entire sequence contained in the plasmid pNP12 insert (Fig. 1). Tyr Leu Gly Phe Leu Ile Asn Lys Ile Tyr Tyr Arg Leu Ser MET CAT GTC CGG TTT ATT TTA TM AAA ATA CAT CAT AGA ATT CGA GTA TTAMTTEGG Fig. 3 is a schematic diagram of the possible translation 5' -GTA CAG GCC AAA TM MT ATT TTT TAT GTA GTA TCT TM GCT CAT MATTAAACC products of the imm-lys region (as deduced from an analysis 393 403 413 435 of open reading frames from both DNA strands) and shows AGCGGTATAT AAGAAAAGTA AMTATTCCT AGCTCM TAC TCC mTE TCT AAA AAA CAC CCT 5' -TCGCCATATA TTCTTTTCAT TTTATMGGA TCGAGTT ATG AGG AM AGA TTm TET GTG GGA the location of the boundaries of pNP6, pNP4, and pNP12. MET Arg Lys Arg Phe Phe Val Gly

The correlation of structure-function properties of these 450 465 480 495 plasmids with the translation map suggests only one reason- TAT MG CGC TAT TTG GAG GM CAA CCT ACA GTC CGA TTG ATA TAT GCA CTA CM 5' -ATA TTC GCG ATA MC CTC GTT GTT GGA TGT CAG GCT MC TAT ATA CGT GAT GTT able possibility for the assignment of specific coding se- Ile Phe Ala Ile Asn Leu Leu Val Gly Cys Gln Ala Asn Tyr Ile Arg Asp Val

510 525 540 GTC CCT CCC TGG TAG CGT GGT AGG AGG AGA AGA TTT GAC TGC CCC TAG CGC CAA (Imm) 113 5'-CAG GGA GGG ACC ATC GCA CCA TCC TCC TCT TCT AAA CTG ACG GGG ATC GCG GTT Gln Gly Gly Thr Ile Ala Pro Ser Ser Ser Ser Lys Leu Thr Gly Ile Ala Val

7 7 12 555 568 578 588 598 608 618 GTC' ATC TTTTCTMTT TCCTAGMGA ACTCTAGGM AAAAAGACGC GCATTAGACG ACGMCGTTE 5 -CAG TAG AAAAGATTM AGGATCTECT TGAGATCCTT TTETECTGCG CGTMTCTGC TGCTTGCAM ~- Gln 4 pNP 4 pNP14 pNP12 NP66 GTTTTTTTGG TG 13 5' -CAAAMAACC AC 12 33 45 6Obp (Lys) FIG. 4. DNA sequence for the immunity and lysis gene regions and the predicted polypeptide sequences. The various functional FIG. 3. Schematic representation of all possible open reading sequences are indicated by sequence number as follows: 32-35, frames in the region of plasmid ColEl, 600 base pairs (bp) beyond the overlapping translation termination codons for colicin El and im- COOH-terminal end of the colicin El gene. Horizontal arrows munity; 58-96, rho-dependent termination site (Tj) for transcription represent the direction of translation and numbers indicate the of colicin El; 373, initiation of immunity gene translation; 381-386, number of amino acids in each possible polypeptide. The * shows ribosome-binding site for immunity gene; 407-412, ribosome-binding location of all methionine residues. Vertical arrows indicate the site for lysis gene; 421, initiation of lysis gene translation; 440 and boundaries of plasmids pNP6, pNP4, pNP14, and pNP12. Open 475, possible start sites of immunity transcription; 557, lysis gene arrowhead represents the COOH-terminal region and the direction of translation termination; 566-589, rho-independent termination site transcription of cea. (T2) for transcription of colicin-lysis. Downloaded by guest on September 25, 2021 8392 Biochemistry: Waleh and Johnson Proc. Natl. Acad. Sci. USA 82 (1985)

-3.50 . and the location of their associated genetic control elements. A This region is characterized by a high degree of compactness and overlap of functional segments, including (i) overlapping 1-2 L-1 -1.75 . termination codons and transcripts for the imm and cea L-2 II 1-1 genes, (ii) the presence of a rho-dependent transcriptional - >)-of 1-4 0 termination site for the colicin gene within the imm coding t0 sequence at about position 79(5), and (iii) the close proximity 0.00 (47 base pairs) of the amino-terminal coding regions of the imm and lys genes and the likely overlap ofthe transcriptional control sequence for the imm gene and the coding sequence for lys. 1.75 The location of the imm gene promoter within the lys gene-coding sequence is inferred from the results presented in Figs. 1 and 3. Plasmid pNP4 contains the entire coding 3.50 region for imm but lacks its natural promoter, as suggested by -3.50 . the dependence of immunity phenotype on the orientation of the fragment relative to the P-lactamase promoter. However, plasmid pNP14 appears to contain the imm promoter, be- -1.75 . cause immunity expression is observed in the absence of the correct orientation relative to the P-lactamase promoter. Computer analysis of the lys gene-coding sequence, partic- ularly the region defined by the boundaries of pNP4 and 0.00 I pNP14, has revealed the existence of two overlapping can- didate sequences for the imm promoter (results not shown). Both promoters and transcription start sites (positions 440 and 475, Fig. 4) are within the lys gene-coding sequences. 1.75 Analysis has also demonstrated a high degree of sequence homology between the ColEl lys gene and gene H ofplasmid CloDF13. Gene H encodes a protein, located predominantly 3.50 in the membrane, that is involved in bacterial cell lysis and is 1 28 56 84 responsible for a similar mitomycin C-induced lethality in AMINO ACID No. that system (27). Fig. 6 presents a schematic representation ofthe transcrip- FIG. 5. Hydropathicity profiles of the lysis (A) and immunity (B) to a proteins. (Inset) Comparison between the average hydropathicity tional organization of the colicin El operon relative values as a function of segment length for hydrophobic segments of restriction endonuclease cleavage map showing the location the lysis protein (L-1 and L-2) and the immunity protein (I-1, 1-2, I-3, of the major infrequent sites throughout this region. General and I-4) and for segments of membrane (o) and nonmembrane (n) features of the in vivo transcription of the colicin gene region proteins (taken from ref. 26). have been deduced previously by Si mapping and RNA blot hybridization studies (28). calculated for sequential segments of seven amino acids are Although the imm gene transcript has not been identified in plotted as a function of residue position (26). The lys protein vivo, coupled transcription-translation using DNA fragments contains two major nonpolar regions (designated L-1 and from the imm gene region has demonstrated in vitro the L-2), and the imm protein has four major nonpolar regions synthesis of a unique RNA (150 kDa) and protein (13 kDa), (I-1-I-4) distributed throughout the sequence. characterized as products of the imm region (29). We have Kyte and Doolittle (26) have demonstrated that hydropa- also observed the synthesis of a 14-kDa protein specified by thicity analysis can distinguish between polypeptide se- plasmid pNP4, using maxi cells (H.-I. Oh, personal commu- quences that span membranes and sequences that pass nication). On the basis of the location of the imm gene through the interior of norqmembrane proteins. From their transcription initiation site proposed here and a transcript analysis of nine membrane proteins and nine soluble proteins size of =460 nucleotides (29), we calculate that a possible of known they demonstrated a substantial differ- imm gene transcription termination site is located at T3 (Fig. structure, the ence in average hydropathicity (AH _ 0.5) for segments of 6) within the colicin gene and overlaps by =90 nucleotides various lengths for these two classes of proteins. colicin gene transcript that terminates within the imm gene at as Fig. 5 (Inset) compares the average hydropathicity (H) T1. In an in vitro system using pNP4 (but not pNP6) values for the six major hydrophobic amino acid segments of template, we have observed the synthesis of an RNA ap- imm with those obtained by Kyte and proximately this size (I. Sohel, personal communication). the lys and proteins the of Doolittle (26) for segments of a set of well-characterized However, computer analysis of the region in vicinity membrane and soluble proteins (upper and lower curves, T3 shows only weak dyad symmetry, which is not charac- teristic of known termination sequences (unpublished data). respectively). The lys protein segments L-1 and L-2 and imm El protein segments I-1, I-2, and I-4 have H values of 2.4, 2.2, The unusual structural organization of the colicin These values are to or operon demonstrated here may have several possible conse- 2.2, 2.4, and 1.9, respectively. equal of in greater than those for a typical membrane peptide sequence quences for the coordinate regulation gene expression imm protein segment this system. The transcription map presented in Fig. 6 shows of corresponding segment length. Only of the exhibit H of a soluble- that three major segments operon possible 1-3 has an value (1.0) less than that typical the 3' ends of protein segment. transcriptional overlap; these occurbetween (i) the cea and imm transcripts, (ii) the entire imm transcript and its anti-sense strand formed as part of the cea-lys transcript DISCUSSION terminating at T2, and (iii) the proposed 5' end of the imm The DNA sequence of the imm-lys region presented in Fig. transcript and the region ofthe cea-lys transcript correspond- 4 shows the proposed coding sequences for these two genes ing to the amino-terminal end of the lys gene. Downloaded by guest on September 25, 2021 Biochemistry: Waleh and Johnson Proc. Natl. Acad. Sci. USA 82 (1985) 8393

160160bpbp p_ T3 (?) 13 i

Colicin El

11 12 T2

a: E- - - - :3 -' 0 { ° 0 ,0 _C > V) u_ _ LZ> .0 en h0 .Vu _, D- ui X CLu. CL W < Z

The imm gene is constitutively expressed in cells that are 4. Tessman, E. S. & Peterson, P. K. (1980) J. Bacteriol. 143, not induced for colicin synthesis. imm transcription could 1307-1317. affect cea transcription in the region of overlap and lead to a 5. Ebina, Y., Takahara, Y., Kishi, F. & Nakazawa, A. (1983) J. reduced rate of synthesis, to premature termination, or to an Biol. Chem. 258, 13258-13261. inhibition oftranslation ofthe carboxyl-terminal ofthe 6. Yamada, M., Ebina, Y., Miyata, T., Nakazawa, T. & region Nakazawa, A. (1982) Proc. Natl. Acad. Sci. USA 79, colicin El protein. In this regard, we note that the hydro- 2827-2831. phobic carboxyl-terminal domain of colicin is responsible for 7. Schwartz, S. A. & Helinski, D. R. (1971) J. Biol. Chem. 246, its biological properties that lead to membrane depolarization 6318-6327. and cell death (8). The inverted orientation of the imm gene 8. Davidson, V. L., Cramer, W. A., Bishop, L. J. & Brunden, may therefore be a means of providing an additional control, K. R. (1984) J. Biol. Chem. 259, 594-600. for example, on the low-level expression of this cytotoxic 9. Yamada, M. & Nakazawa, A. (1984) Eur. J. Biochem. 140, protein in uninduced cells. 249-255. Read-through of the rho-dependent terminator (T1) results 10. Dougan, G. & Sherratt, D. (1977) Mol. Gen. Genet. 151, in the transcription of the lys gene as well as the synthesis of 151-160. 11. Dougan, G., Saul, M., Warren, G. & Sherratt, D. (1978) Mol. the complete anti-sense strand of the imm mRNA. It is Gen. Genet. 158, 325-327. possible that, after full induction of cea gene transcription, 12. Inselburg, J. (1977) J. Bacteriol. 129, 482-491. imm gene expression is reduced because of the formation of 13. Oka, A., Nomura, N., Morita, M., Sugisaki, H., Sugimoto, K. a hybrid RNA and the inhibition of imm translation. & Takanami, M. (1979) Mol. Gen. Genet. 172, 151-159. A third region of transcriptional overlap potentially exists 14. Shafferman, A., Flashner, Y. & Cohen, S. (1979) Mol. Gen. between the 5' end of the imm transcript and a region of the Genet. 176, 139-146. lys transcript extending from the 5' side of the ribosome- 15. Suit, J. L., Dan, M. L. J., Sabik, J. F., Labarre, R. & Luria, binding site through the sequence coding for the amino- S. E. (1983) Proc. Natl. Acad. Sci. USA 80, 579-583. terminal domain of the lys protein. In principle, either lys or 16. Bochner, B. R., Huang, H., Schieven, G. L. & Ames, B. N. imm (1980) J. Bacteriol. 143, 926-933. expression could be inhibited at the level of translation 17. Ozeki, H., Stocker, B. A. D. & Smith, S. M. (1962) J. Gen. by the formation of an RNA hybrid in this region. Microbiol. 28, 671-687. Although the overlapping nature of transcriptional ele- 18. Kelly, S. V. & Johnson, P. H. (1980) Anal. Biochem. 107, ments implied by the organization of genes in the colicin El 362-368. operon may indicate the existence of a novel regulatory 19. Males, B. M. & Stocker, B. A. D. (1982) J. Gen. Microbiol. mechanism, further analysis is necessary to establish a role 128, 95-106. for anti-sense RNA formation in gene expression in this 20. Villa-Komaroff, L., Efstratiadis, A., Broome, S., Lomedico, system. P., Tizard, R., Naker, S. P., Chick, W. L. & Gilbert, W. (1978) Proc. Natl. Acad. Sci. USA 75, 3727-3731. We thank Drs. Bruce Stocker, Jack Lebowitz, and John Nijkamp, 21. Fredericque, P. (1949) C. R. Seances Soc. Biol. Paris 143, and our laboratory colleagues for their valuable suggestions. This 1011-1013. research was supported by SRI Research and Development Project 22. DiMasi, D. R., White, J. C., Schnaitman, C. A. & Bradbeer, 376D32 FYL and the BIONET National Computer Resource for C. (1973) J. Bacteriol. 115, 506-513. Molecular Biology, National Institutes of Health (Grant 1 U41 23. Davies, J. K. & Reeves, P. (1975) J. Bacteriol. 123, 102-117. RR01685-01). 24. Davies, J. K. & Reeves, P. (1975) J. Bacteriol. 123, 96-101. 25. Patient, R. K. (1979) Nucleic Acids Res. 6, 2647-2665. 1. Ozeki, H., Stocker, B. A. D. & de Margerie, H. (1959) Nature 26. Kyte, J. & Doolittle, R. F. (1982) J. Mol. Biol. 157, 105-132. (London) 184, 337-339. 27. Hakkaart, M. J. J., Veltkamp, E. & Nijkamp, H. J. (1981) 2. Ebina, Y., Fishi, F., Miki, T., Kagamiyama, H., Nakazawa, T. Mol. Gen. Genet. 183, 318-325. & Nakazawa, A. (1981) Gene 15, 119-126. 28. Ebina, Y. & Nakazawa, A. (1983) J. Biol. Chem. 258, 3. Ebina, Y., Fishi, F. & Nakazawa, A. (1982) J. Bacteriol. 150, 7072-7078. 1479-1481. 29. Chen, H.-Z. & Zubay, G. (1983) J. Bacteriol. 154, 650-655. Downloaded by guest on September 25, 2021