THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1986 by The American Society of Biological Chemists, Inc

Vol. 261, No. 10, Issue of April 5, pp. 4600-4606,19S6 Printed in U.S.A.

Expression of the Proteus mirabilis Lipoprotein Gene in Escherichia coli EXISTENCE OF TANDEM PROMOTERS*

(Received for publication, July 22, 1985)

Gee Ching and Masayori Inouyeg From the Department of BwchemistT, State University of New York at Stony Brook, Stony Brook, New York 11794

The lpp gene from Proteus mirabilis was cloned onto pBR322 and expressed in Escherichia coli. The P. mirabilis lpp gene is unique in that it has two tandem promoters transcribing two mRNAs that differ in length by approximately 70 nucleotides at their 5’- ends. The two mRNAs thus encode the identical lipo- protein. The P. mirabilis prolipoprotein has a 19- amino acid signal peptide and a 59-amino acid lipopro- tein sequence. In spite of the substantial differences in the amino acid sequence from the E. coli prolipopro- tein, the P . mirabilis prolipoprotein is normally mod- ified and processed in E. coli, and the resultant lipo- protein is assembled.in the E. coli outer membrane as is the E. coli lipoprotein.

The lipoprotein is a major outer membrane protein in various enteric bacteria (1; see also review Ref. 2 ) . The lpp genes of Escherichia coli (3), Serratia marcescens (4), Erwinia amylouora (5), and Morganella morganii (6) have been cloned, and their nucleotide sequences have been determined. In all these bacteria, the lipoprotein is synthesized as a precursor prolipoprotein containing a 20-amino acid-long signal peptide and a 58-amino acid-long mature sequence. The cleavage of the signal peptide requires a glyceride modification of the cysteine residue at the cleavage site (7, 8). This cleavage can be specifically inhibited by an antibiotic, globomycin (9, 10). One third of the lipoprotein molecules are known to be covalently linked to the peptidoglycan layer (2).

Proteus mirabilis is distantly related to E. coli. Its DNA has a low G + C content (39%) and shares little homology (6%) with E. coli DNA (11,12). We have previously shown that the lipoprotein from P. mirabilis was very weakly cross-reactive with anti-E. coli lipoprotein serum (1) and that the level of lipoprotein in P. mirabilis was considerably less than that of E. coli (13). Recently, we cloned the P. mirabilis lpp gene into pBR322 and determined the nucleotide sequence of its coding region (1.4). The coding sequence shows that the P. mirabilis prolipoprotein consists of a 19-amino acid signal peptide and a 59-amino acid lipoprotein. Its amino acid sequence is sub- stantially different from those of the other enterobacterial lipoproteins. Therefore, it is of great interest to study the expression of the P. mirabilis lpp gene in E. coli.

In this paper, we attempted to express the P. mirabilis lpp

*.This work was supported by Grant GM19043 from the National Institute General Medical Sciences and &ant NP387I from the American Cancer Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement’’ in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom correspondence should be addressed.

gene in E. coli and to identify the P. mirabilis lipoprotein promoters. We found that in spite of the substantial structural differences from the E. coli prolipoprotein, the P. mirabilis lpp gene was expressed in E. coli and the P. mirabilis lipopro- tein was normally assembled in the E. coli outer membrane. We also found that the P. mirabilis lpp gene had two tandem promoters, both of which were functional in E. coli.

EXPERIMENTAL PROCEDURES

Materials-Restriction enzymes were purchased from Bethesda Research Laboratories and New England Biolabs. S1 nuclease, nu- clease-free bovine serum albumin, and Klenow enzyme were obtained from Bethesda Research Laboratories and T4 polynucleotide kinase and E. coli RNA polymerase from Boehringer Mannheim. Rifampicin was purchased from Sigma. Globomycin was obtained from Dr. M. Arai (Sankyo Pharmaceutical Co., Japan).

S%221/F’lacZq (hsdM+ hsdR- recA leuB6 lacy trpE5 lpp-/F’lacZq lac+ Bacterial Strains, Phage, and Plasmids-P. mirabilis (151, E. coli

pro+) (16), E. coli JA221 (hsdM+ hsdR- recA leuB6 lacy trpE5 lpp+/ F’lacP lac+ pro’) (16), and M. morganii (6) were used. The following plasmids were used: pGC-01 carrying the M. morganii lpp gene (6) and pYM140 carrying the E. coli lpp gene (17). The construction of the P. mirabilis lpp-carrying plasmid, pGC9-25, and the transforma- tion of pGC9-25 into E. coli SB221/F’lncIq were described previously

Protein Labeling Experiments-E. coli SB221/F‘lacIq carrying pGC9-25 was grown at 37 “C in M9 medium supplemented with 0.4% glucose, 2 pg of thiamine/ml, 200 pg of MgSO,. 7HzO/ml, and 50 pg each of tryptophan, leucine, and ampicillin/ml. At a cell density of approximately 2.5 X IO8 cells/ml, globomycin was added to a final concentration of 100 pg/ml. After 20 min incubation, 200 pCi of [2,3,4,5-3H]arginine (43.5 Ci/mmol; Amersham Corp.) was added to 10 ml of the culture, and the mixture was incubated for 30 s. The labeling was terminated by addition of 10 ml of ice-cold stop solution (40 mM sodium phosphate (pH 7.1), 0.8% formaldehyde, and 2 mg of arginine/ml). The procedures for the membrane preparation, immu- noprecipitation with anti-E. coli lipoprotein serum, and SDS-poly- acrylamide gel electrophoresis were described previously (6).

Sodium lauroyl sarcosinate solubilization (18) was used to separate the outer and cytoplasmic membrane protein fractions. Total mem- brane was prepared from a’ 10-ml culture of E. coli SB221/F’lacZq carrying pGC9-25 that had been labeled with [3H]arginine for 15 min in the absence of globomycin; this fraction was resuspended in 100 pl of 10 mM sodium phosphate buffer (pH 7.0). Fifty pl of the membrane suspension was mixed with 150 pl of water and 200 pl 1% sodium sarcosinate. The mixture was kept at room temperature for 30 min and centrifuged at 105,000 X g for 30 min at 4 “C. The pelleted outer membranes were washed with 10 mM sodium phosphate buffer (pH 7.0), and the supernatant fluid was precipitated with 400 pl of cold 10% trichloroacetic acid. After 60 min incubation in an ice bath, the trichloroacetic acid precipitate (cytoplasmic membrane protein) was collected by centrifugation and washed twice with methanol/ether ( M I .

To isolate the peptidoglycan-lipoprotein complex, purified pepti-

The abbreviations used are: SDS, sodium dodecyl sulfate; bp, base pairs; kb, kilobase pairs; Hepes, 4-(2-hydroxyethyl)-l-piperazineeth- anesulfonic acid; MOPS, 4-morpholinepropanesulfonic acid.

( 1 4 ~

4600

Proteus mirabilis Lipoprotein Gene 4601

doglycan was obtained by the method of Braun and Sieglin (19). The purified peptidoglycan was treated with 20 pg of lysozyme in 20 p1 of 10 mM sodium phosphate buffer (pH 7.0) at 37 "C for 3 h.

SI Mapping and Northern Blot Hybridization-To prepare total cellular RNA, a 40-ml cell culture was grown in L broth (ampicillin was added to a concentration of 50 pg/ml for the E. coli strain harboring plasmid pGC9-25) at 37 "C to a Klett-Summerson color- imeter reading of 110-150 units. The cells were collected by centrif- ugation and lysed with 3 ml of hot lysis solution (10 mM Tris-HC1 (pH 8.0), 1 mM EDTA, 350 mM NaCI, 2% SDS, 7 M urea) (17). The lysate was extracted three times with 3 ml of chloroform/phenol (1:l) and three times with 3 ml of chloroform. Three volumes of 100% ethanol and one-tenth volume of 3 M sodium acetate (pH 5.2) were then added to the lysate to precipitate RNA. The RNA precipitate was dissolved in 200 pl of TE buffer (10 mM Tris-HC1 (pH 7.5), 1 mM EDTA). Three volumes of ethanol and one-tenth volume of 3 M sodium acetate (pH 5.2) were added to the solution to precipitate RNA. This procedure was repeated three times, and the final RNA sample was stored in ethanol at -20 "C.

For S1 mapping (20), a 1.9-kb DdeI fragment isolated from plasmid pGC9-25 was 5'-end-labeled with [y3'P]ATP (3000 Ci/mmol; Amer- sham Corp.) and T4 polynucleotide kinase and cut with HindIII to generate a 520-bp DdeI-Hind11 fragment carrying the entire pro- moter region and part of the structural region of the P. mirabilis lpp gene (from the HindIII site of pBR322 to +193 of P. mirabilis lpp gene; see Figs. 1 and 2). Forty ng (1.2 X lo6 cpm) of this 32P-labeled DNA fragment was ethanol-precipitated together with 90-95 pg of an RNA sample. After lyophilization, the DNA/RNA mixture was dis- solved in 30 pl of hybridization buffer (80% formamide, 0.4 M NaC1, 40 mM Hepes (pH 6.5), and 1 mM EDTA). The solution was heated at 75 "C for 15 min and then rapidly transferred to a 40 "C water bath and incubated for 3 h (in some experiments, a hybridization temperature of 35 or 45 "C was used instead). After incubation, 300 pl of ice-cold S1 buffer (30 mM sodium acetate (pH 4.6), 50 mM NaC1, 1 mM ZnS04, 5% glycerol) and 150 units of S1 nuclease were added. The mixture was incubated at 37 "C for 25 min. The S1 nuclease reaction was stopped by adding 5 pl of 0.25 M EDTA and 10 pg of yeast tRNA, and the mixture was precipitated with 1 ml of 100% ethanol. The resultant precipitate was washed twice with 70% ethanol, lyophilized, and dissolved in 4 pl of the loading buffer (80% formamide, 10 mM NaOH, 1 mM EDTA, 0.025% xylene cyanol, 0.025% bromphenol blue). Two microliters of the sample was then subjected to 8% polyacrylamide gel electrophoresis in 8 M urea. To determine the relative amounts of the bands of interest, the autora- diograms of the gels were scanned with a Shimadzu densitometer.

For the Northern blot experiment (21, 22), a 187-bp RsaI-XbaI DNA fragment (from +78 to +264 of the P. mirabilis lpp gene; see Figs. 1 and 2) isolated from plasmid 'pGC9-25 was nick-translated with [cx-~'P]~ATP (3000 Ci/mmol; Amersham Corp.) and the E. coli DNA polymerase I (23) and used as the probe for hybridization. Twenty pg of the total cellular RNA was incubated at 60 "C for 10 min in 40 pl of MOPS buffer (20 mM MOPS, 5 mM sodium acetate, 1 mM EDTA, (pH 7.0)) containing 50% formamide and 6% formal- dehyde. After the sample was cooled to room temperature, 4 p1 of the loading buffer (50% glycerol, 1 mM EDTA, 0.05% xylene cyanol, 0.05% bromphenol blue) was added, and the sample was subjected to 1.5% agarose gel electrophoresis in MOPS buffer containing 6% formaldehyde at 75 V for 4.5 h. The agarose gel was soaked in 20 X SSC (1 X SSC = 0.15 M NaCl plus 0.015 M sodium citrate) for 30 min and blotted to a nitrocellulose filter for 15 h in 20 X SSC. After being soaked in 20 X SSC for 20 min and baked at 80 "C for 2.5 h, the nitrocellulose filter was hybridized wtih the 32P-labeled DNA probe in the hybridization medium containing 50% formamide and 5 X SSC at 42 "C for 17 h. The filter was washed at 42 "C twice with the hybridization medium for 2 h each and four times with 2 X SSC for 15 min each. The 581-bp HindIII-XbaI fragment (from HindIII site of pBR322 to +264 of the P. mirabilis lpp gene), 520-bp HindIII- DdeI fragment (from Hind111 site of pBR322 to +193), and 187-b~ RsaI-XbaI fragment (+78 to +264) prepared from pGC9-25 were treated as described for the RNA samples and were used as size markers.

Northern hybridization analysis of the total cellular RNA from M. morganii and E. coli JA221 lpp+/F'lacIq was carried out as described above. A 300-bp XbaI-MspI fragment (+24 to +323 of the M. morganii lpp gene (6)) from pGC-01 and a 423-bp XbaI-MspI fragment (+24 to +446 of the E. coli lpp gene (3)) from pYM140 were used as the

hybridization probes for the M. morganii lpp mRNA and E. coli lpp mRNA, respectively.

In Vitro Transcription-The following fragments were prepared from pGC9-25 and used as the truncated templates in the transcrip- tion reaction: a 372-bp EcoRI-RsaI fragment (-295 to +77 of P. mirabilis Ipp gene), a 245-bp EcoRI-AluI fragment (-295 to -50), and a 127-bp AluI-RsaI fragment (-50 to +77) (see Figs. 1 and 2). The single-round transcription reaction was carried out according to Kajitani and Ishihama (24) with the following modifications. The reaction mixture contained 0.6 pmol of DNA, 0.9 units of E. coli RNA polymerase, 0.15 mM each unlabeled ATP, GTP, and CTP, 0.016 mM unlabeled UTP, 5 pCi of [a-32P]UTP (3000 Ci/mmol; Amersham Corp.), and 1 pg of rifampicin/ml. The same buffer system was used except that it contained 1 M KC1 instead of NaC1. The DNA template and RNA polymerase were preincubated together at 37 "C for 15 min prior to the addition of ribonucleotides and rifampicin. The transcrip- tion reaction was carried out at 37 "C for 10 min. The samples were subjected to electrophoresis on an 8% polyacrylamide, 8 M urea gel. The in vitro transcription products of the XDNA (77-base and 193- base (25)) and the MluI-Hind111 fragment (-96 to +142) of the crp gene (26) were used as size markers.

RESULTS

Cloning and DNA Sequence of the P. mirabilis lpp Gene- We have previously cloned a 1.1-kb fragment carrying the entire P. mirabilis lpp gene into the EcoRI site of pBR322 (14). The resultant plasmid, designatedpGC9-25, is shown in Fig. 1. The nucleotide sequence encompassing the entire P. mirabilis lpp gene, together with the deduced amino acid sequence, is shown in Fig. 2.

We have previously shown that the cloning of the E. coli lpp gene into pBR322 was unsuccessful due to the lethal over- production of the lipoprotein (27). However, the promoters of the P. mirabilis lpp gene appeared to be much weaker in E. coli than that of the E. coli lpp gene. Therefore, we were able to clone the P. mirabilis lpp gene into pBR322.

Expression of the P. mirabilis lpp in E. c o l t I t has previ- ously been demonstrated that E. coli SB221F'lacZq totally lacks the E. coli lipoprotein (6, 16). In order to examine whether the P. mirabilis lipoprotein was expressed in E. coli, E. coli SB221F'lacIq cells harboring pGC9-25 were labeled with [3H]arginine for 30 s in the presence and the absence of globomycin (100 pg/ml). In the absence of globomycin, a protein species (lane 1 in Fig. 3A) that could be immunopre- cipitated with anti-E. coli lipoprotein serum (lane 2 in Fig. 3A) was observed at position a. When globomycin was added, this species disappeared and a new species appeared at posi-

0 0.2 0.4 0.6 0.8 1.0 1.2 I I Ikb

c::z3P3 PZb-----"

p"

HindI l I MspI R s a I Ode1 XbaI Ms I DdeIRsoIRsaI pGC9-25

A l u I AluI I P3 P2 PI

I P3 P2 +d $0.. templates

Probe-1

Probe-2

FIG. 1. Structure of the P. mirabilis lpp gene carrying plas- mid pGC9-25. The 1.1-kb EcoRI fragment carrying the entire Ipp gene (hatched bar) was inserted at the EcoRI site of pBR322. The length and direction of the 1pp transcripts are indicated by open arrows. PI, P2, and P3 indicate the promoters described in the text. The direction of transcription from P3 is indicated by a broken arrow. Probes 1 and 2 used for S1 mapping and Northern blot hybridization, respectively, and DNA fragments used for in vitro transcription are also shown.

4602 Proteus mirabilis Lipoprotein Gene -200

s . , , ~ c - ~ , - , ~ R - , ~ ~ W I A c ~ = A W = ~ - A ~ A c ~ - - -” -

-150 -100 b . b . .

A T A I U W t r m * C G M I * A A U I C r m G T G U A I M ~ ~ ~ ~ A I A ~ ~ A ~ ~ ~ ~ A ~ ~ A I -35

-so b 4 .

~ ~ ~ I A I r m G T G U A ~ M R A ~ A T A C ~ ~ A I r m G T G U A I r m G T G U A ~ ~ ~ ~ ~ A I M ~ C ~ ~ T4T -

-10 -31 -10

-I +l 4.

+so 4 4 . 4 . 4 .

~ A = A ~ , I A - I C ~ ~ C - ~ ~ G T G U A ~ = M ~ M T ~ R , ~ C - I UatL~.AI.LymIl

b 4 4 . . I G I A = - ~ M T I C I C C ~ , ~ = A , A ~ C ~ ~ ~ ~ C ~ ~ , A W C ~ ~ ~ V ~ ~ L . ~ G l y ~ V ~ l l I ~ ~ N ~ S ~ f f i l ~ L a ~ u N ~ C ~ ~ ~ ~ S ~ r S ~ r S ~ r A ~ ~ ~ ~ C l n ~ m ~ ~ ~ l l e S

+1so

+zoo b 4 . 4 b

C r T ~ ~ r m G T G U A C C ~ r m G T G U A ~ C ~ ~ ~ ~ = M = A ~ - ~ ~ M C ” ~ C ~ r S ~ r A ~ p V ~ U . a A r ~ k u ~ ~ r C l n V ~ ~ C l ~ C l ~ ~ S ~ r S ~ ~ A ~ ~ V ~ C l ~ S ~ r ~ ~ C ~ ~ l ~ L ~ ~ ~

+ n o +3M) b . 4 . . 4

G ~ A ~ C A C C T C - M - ~ ~ I M C C A * G I N ~ I ~ ~ E ~ ~ . * ~ . * ~ ~ ~ C ~ ~ ~ A ~ ~ ~ C I U V ~ ~ ~ ~ ~ ~ I ~ L ~ ~ L ~ ~

+350 4 b . 4 b

c w M A c G ~ M C ~ ~ M ~ M I M I ~ ~ W I A C ~ A ~ G M ~ ~ ~ ~ ~ ~ l A I * v

-00 4450 b b b 4 4 4

C I A ~ - t A ~ A , ~ C ~ A ~ A C ~ C W ~ I M ~ ~ , I C , - C - , -

+Sp b 4

C C “ I A C C M C C C 3 ,

FIG. 2. Nucleotide twquence encompassing the P. mirabilis lpp gene. The translation initiation and termination codons, and the Shine-Dalgarno sequence are boxed. The -35 and -10 regions of the P2 and P1 promoters are underlined and indicated by the numbers -35 and -10, respectively. The putative transcription termination site is indicated by a large arrow and is deduced from the E. coli lpp mRNA (36). The first nucleotide of transcript I determined by S1 mapping (Fig. 4A) is numbered as +l. The transcription initiation sites of transcript I1 described in the tex t are indicated by small arrows. The transcription direction of the P3 promoter is indicated by a broken arrow; the exact location of the transcription initiation site is not known. The amino acid sequence deduced from the nucleo- tide sequence is also shown.

tion b (lane 3 in Fig. 3A), which could also be immunoprecip- itated with anti-E. coli lipoprotein serum (data not shown). By analogy to the effect of globomycin on E. coli lipoprotein synthesis (9, lo), these results suggest that the protein at position a is the P. mirabilis lipoprotein and the protein at position b is the P. mirabilis prolipoprotein. To examine further if the P. mirabilis lipoprotein and prolipoprotein were modified with glycerol and palmitic acid in E. coli, cells harboring pGC9-25 were labeled with [3H]palmitic acid and [‘H]glycerol in the presence and absence of globomycin. Both P. mirabilis lipoprotein and prolipoprotein were labeled with [3H]palmitic acid and with [3H]glycerol (data not shown). Sarcosyl solubilization (18) indicated that the P. mirabilis lipoprotein produced in E. coli was found exclusively in the outer membrane fraction (lane 2 and Fig. 3B) , as is the case for the E. coli lipoprotein (28). The P. mirabilis lipoprotein was also found to be covalently linked to the peptidoglycan in E. coli (lane 1 in Fig. 3C), again like the E. coli protein (29). Based on these results, we conclude that the P. mirabilis prolipoprotein translocates across the E. coli cytoplasmic membrane, is normally modified and processed, and the ma- ture lipoprotein is properly assembled in the E. coli outer membrane.

SI Mapping and Northern Blot Hybridization-To deter- mine the site of transcription initiation for the P. mirabilis lpp gene, total cellular RNA from E. coli SB221F’lrrcP car-

A B C

FIG. 3. Characterization of the P. mirabilis lipoprotein pro- duced in E. coli carrying pCC9-26. A, SDS-polyacrylamide gel electrophoresis of the total membrane fraction prepared from cells labeled with 13H]arginine for 30 s. A portion of the solubilized membrane fraction was also treated with anti-E. coli lipoprotein serum. Lane I , the membrane fraction from cells labeled in the absence of globomycin; lane 2, the immunoprecipitate from the mem- brane used in lone I ; lane 3, the membrane fraction from cells labeled in the presence of globomycin. E , SDS-polyacrylamide gel electro- phoresis of the sarcosinate-treated t o t a l membrane fraction from cells labeled with (3H]arginine for 15 min in the absence of globomycin. Lane I , the cytoplasmic membrane fraction; lone 2. the outer mem- brane fraction. C, SDS-polyacrylamide gel electrophoresis of the lipoprotein-peptidoglycan complex fraction from cells labeled with [3H]arginine for 15 min in the absence of globomycin. hne I , the lysozyme-treated peptidoglycan; lone 2, the untreated peptidoglycan; lone 3, the control membrane fraction labeled with [‘Hjarginine in the absence of globomycin. Arrows with letters a and b indicate the positions of the P. mirabilis lipoprotein and prolipoprotein. respec- tively.

ryingpGC9-25 and from P. rnirabilis were used for S1 mapping as described under “Experimental Procedures.” After S1 digestion, RNA preparations from both bacterial strains showed two new unique bands at positions I and 11, in addition to the protected full-length probe (Fig. 4B) . The intensities of these two bands increased as the concentration of RNA was increased (data not shown), and total cellular RNA from E. coli JA221 Ipp’F’lrrcP (lane 5, Fig. 4 8 ) and yeast tRNA ( d a t a not shown) did not give these hybridizable bands. These results demonstrate that bands I and I1 are specific to the P. mirabilis lpp gene. Both bands I and I1 consist of a few individual bands. These multiple bind patterns were observed under different hybridization temperatures and times, S1 concentrations, and digestion temperatures ( d a t a not shown). It is not certain at present if such multiple-band patterns are due to the method used for S1 mapping (30) or due to multiple initiation sites in the lpp promoters. The most intense band for RNA transcript I was found to correspond to the C at position +1 (Fig. 4A); thus, it was concluded that transcript I was initiated with the G at position +1 (see Fig. 2). Similarly, it was concluded that the longer RNA transcript I1 was initiated at two sites, the A at position -69 and the A at position -71 (Fig. 4A and Fig. 2).

Transcript I1 was present seven times and five times 88 much as transcript I in P. mirabilis and E. coli SB221/F’hP carrying plasmid pGC9-25, respectively (lanes 1 and 2, Fig. 4B) . When the hybridization temperature was changed from 35 to 45 “C, the ratios of transcript I1 to transcript I decreased to 4.7 and 1.5 for P. mirabilis and E. coli, respectively (compare lane I with 3, and lane 2 with 4, Fig 4B, respectively). Assum- ing that transcript I1 is initiated at position -69, transcript I1 is longer than transcript I by 69 nucleotides, and 52 out of these 69 nucleotides are A or T (75%). The apparent decrease in the ratios of transcript I1 to transcript I at higher hybridi- zation temperatures may therefore be due to the higher A + T content that lowers the melting temperature of the hybrid between the lpp DNA probe and RNA transcript 11.

Proteus mirabilis Lipoprotein Gene 4603

A 1 2

w-

5’ \ A A

+ I C

$ / A

B t 1 2 3 4 5 6

G A/ G”

5’ : 2 FIG. 4. Analysis of the 5’-ends of the P. mirabilis lpp RNA

by S1 mapping. Total cellular RNAs from E. coli SB221/F’lacIq containing pGC9-25 and from P. rnirabilis were prepared and used for S1 mapping as described under “Experimental Procedures.” A, the determination of transcription initiation site. Lane 1, the A + G sequencing reaction of the 5’-32P-labeled DdeI-Hind111 fragment used as the DNA probe (see Fig. 1); lane 2, 90 pg of RNA from E. coli SB221/F’lacIq containing pCC9-25 was hybridized to the 32P-labeled DNA probe at 40 “C for 3 h and treated with 150 units of S1 nuclease. The C residue corresponding to the position of the first nucleotide of transcript I as numbered as +1 (see Fig. 2). E, the effect of hybridi- zation temperatures on the ratio of the two transcripts. The following RNAs were hybridized to the same DNA probe and treated with 150 units of S1 nuclease: lane 1,95 pg of RNA from P. rnirabilis, lane 2, 95 pg of RNA from E. coli SB221/F’lacP containing pGC9-25 hy- bridized at 35 ‘C for 3 h; lane 3,95 pg of RNA from P. rnirabilis; lane 4, 95 pg of RNA from E. coli SB221/F’lacIq containing pGC9-25 hybridized at 45 “C for 3 h; lane 5,150 pg of RNA from E. coli JA221 lpp’/F’lacIq (no plasmid) hybridized a t 35 “C for 3 h; lane 6, the DNA probe, alone. I and I I refer to the two transcripta produced from the P. rnirabilis lpp gene (see Fig. 2).

The presence of two P. mirabilis lpp mRNAs of different sizes was further confirmed by Northern blot hybridization. Two RNA species were clearly detected by using the P. mirabilis lpp gene as the probe for both RNA preparations from E. coli harboring pGC9-25 and P. mirabilis (lanes 1 and 2, respectively, in Fig. 5). These two RNA species were not observed in total cellular RNA from E. coli JA221 lpp+fF’laclq and yeast tRNA when hybridized to the same probe (data not shown). Based on their sizes, we concluded that the larger species was transcript I1 and the smaller species was transcript I. In P. mirabilis, transcript I1 was the major species of lpp mRNA as it was in the S1 mapping assay (Fig. 4). In E. coli carrying pGC9-25, transcript I was present in slightly larger amounts than transcript 11, which contradicts the result from

1 2

FIG. 5. Analysis of the P. mirubilis lpp RNA by Northern blot hybridization. Total cellular RNAs from E. coli SB221/F’lacP harboring pGC9-25 and from P. rnirabilis were prepared, subjected to gel electrophoresis under denaturing conditions, and analyzed by Northern blot hybridization as described under “Experimental Pro- cedures.” Lane I , RNA from E. coli SB221/F‘lacIq harboring pGC9- 25; lane 2, RNA from P. rnirabill. I and I1 refer to the transcripts

bp RasI-XbaI DNA fragment, 520-bp HindIII-DdeI DNA fragment, shown in Fig. 4A. Letters a. b. and c indicate the p i t i o n s of the 187-

and 581-bp HindIII-XbaI DNA fragment, respectively, that were used as size markers.

the S1 mapping assay (Fig. 4B). The reason for the discrep- ancy in the ratio of transcripts I1 to I in E. coli between the S1 mapping assay and Northern blot hybridization is not known. In addition to the two lpp mRNAs, there were two bands of larger size seen in the Northern blot of E. coli harboring pGC9-25. These bands were not observed in P. mirabilis and were thought to be the plasmid pGC9-25 DNA that was present in the total cellular RNA preparation, since they disappeared after deoxyribonuclease treatment (data not shown). In Vitro Transcription-To determine if transcripts I and

I1 arise from two tandem P. mirabilis lpp promoters that are active in E. coli, a single-round transcription in vitro was carried out as described under “Experimental Procedures.” As shown in Fig. 6, a 372-bp EcoRI-RsaI fragment (-295 to +77; Figs. 1 and 2) from pGC9-25 carrying both putative lpp promoters gave rise to two transcripts with lengths of 77 and 150 bases (lane 2 ) in good agreement with the initiation sites determined in Fig. 4A; transcript I is initiated at position +1 and transcript I1 at position -69 (or -71) (see Fig. 2).

In addition to the 77-base and 150-base transcripta, tran- scripts with length of approximately 60 bases were also ob- served (lane 2, Fig. 6). The same transcripts were also synthe- sized from a 245-bp EcoRI-AluI fragment (-295 to -50; Figs. 1 and 2) which lost the 127-bp 3’-fragment carrying the promoter for transcript I (designated P1) ( l a n e 1, Fig. 6). Change in the length of the 5’ sequence (upstream of the lpp promoters) of the DNA template caused change in the length of these transcripts (data not shown). These results indicate that there is another promoter (designated P3) upstream of the P. mimbilis lpp promoters which directs transcription in the direction opposite to that of the P. mirabilis lpp gene (see Fig. 2). When a 127-bp AluI-RsaI fragment (-50 to +77; Figs. 1 and 2) carrying only the P1 promoter was used ae the

4604

1"? 3

t

-60

Proteus mirabilis

FIG. 6. In vitro transcription of the P. mirabilia lpp tem- plates. Three DNA fragments from pGC9-25 were used as the truncated templates in the single-round transcription reaction as described under "Experimental Procedures." The samples were run on an 8% polyacrylamide, 8 M urea gel. The following templates were used: lane 1 , the 245-bp EcoRI-AluI fragment (-295 to -50 region of the P. rnirabilis lpp gene, see Figs. 1 and 2); lone 2, the 372-bp EcoRI- RsaI fragment (-295 to +77); lane 3, the 127-bpp AluI-RsaI fragment (-50 to +77). Numbers 60, 77, and 150 indicate the length of the transcripts. I and 11 refer to the initiation sites of transcripts I and 11, respectively (see Fig. 4).

A B R n

FIG. 7. Analysis of the M. morganii lpp RNA and the E. coli Ipp RNA by Northern blot hybridization. Total cellular RNAs from M. rnorganii and E. coli JA221 Ipp+/F'lacP were prepared, subjected to gel electrophoresis under denaturing conditions, and analyzed by Northern blot hybridization as described under "Exper- imental Procedures." A, RNA from M. morganii; B, RNA from E. coli JA221 lpp+/F'lacIq. Arrows indicate the lpp mRNA.

template, only the 77-base trancript was synthesized (lam 3, Fig. 6). These results clearly demonstrate that there are tandem promoters present in the P. mirabilis lpp gene and that these promoters are active in the E. coli transcription system as well.

Presence of Tandem Promoters in Other lpp Genes-To determine if tandem promoters also exist in other enterobac- terial lpp genes, total cellular RNA preparations from M. morganii and E. coli JA221 Ipp+/F'hIq were analyzed by Northern blot hybridization as described under "Experimen- tal Procedures." As shown in Fig. 7, M. nwrganii contains two lpp sequence-specific transcripts differing by approximately 100 nucleotides in length, whereas E. coli JA221 lpp+/F'lacIq has only one. Since the M. morganii lpp gene seems to have only one transcription termination site (6), it is likely that the M. nwrganii lpp gene also contains two tandem promoters, which give rise to the two transcripts.

FIG. 8. Alignment of P. mirabilia lpp P1 promoter with other lpp promoters. The DNA sequence of P. mirabiliq lpp P1 promoter is compared to the lpp promoter sequences of M. morganii (6), Erwinin ornylouom (5), S. mrcescem (4) and E. coli (3). The translation initiation codon, ATG, and Shine-Dalgarno sequence, GAGG, are underlined. The transcription initiation site is indicated by an arrow. The -10 and -35 regions of the P1 promoter are indicated by open circles, and the same regions of other enterobacterial lpp genes are indicated by closed circles. The sequences that do not match each other are illustrated outside the aligned sequences by open triangles. Nucleotides that are different from those of E. coli and M. morganii are indicated by * and +, respectively. The upper P. rnirabilis lpp sequence is compared to M. rnorganii lpp sequence and the lower P. rnirabilis lpp sequence is compared to the other three lpp sequences. Recent evidence suggests that the -10 region of the E. coli lpp gene is AATACT instead of the previously reported TGTAAT (37). This prompts us to think that the -10 region of the M. morganii lpp gene could be AATATC instead of TGTAAT, but the exact -10 sequence is unknown at present.

DISCUSSION

In contrast to all the enterobacterial prolipoproteins studied so far, the M. mirabilis prolipoprotein consists of a 19-amino acid signal peptide and a 59-amino acid mature lipoprotein sequence (3-6,141. In addition to 26-amino acid substitutions, the P. mirabilis prolipoprotein has a deletion of 1 amino acid at position -17 of the signal peptide, an insertion of 2-amino acid residues, serine and asparagine, immediately downstream of the signal peptide cleavage site, and a deletion of a lysine residue between positions 40 and 41 of the lipoprotein se- quence as compared to the E. coli prolipoprotein (14). In spite of these differences in the amino acid sequence, the P. mira- bilk prolipoprotein expressed in E. coli is normally modified and processed, and the resulting mature lipoprotein is assem- bled in the E. coli outer membrane as is the E. coli lipoprotein. The highly conserved regions within the P. mirebilk are likely to play important roles in the assembly of the lipoprotein in the E. coli outer membrane. These highly conserved regons include the signal peptide, the sequence around the signal peptide cleavage site, the sequence between positions 43 and 52, the tyrosine residue at position 57, and the C-terminal lysine residue (14). Furthermore, the entire lipoprotein struc- ture is highly conserved because nearly half of the amino acid substitutions are replaced by functionally similar amino acid residues.

The existence of two tandem promoters in the P. mimbilia Ipp gene, designated P1 and P2, which give rise to the two RNA transcripts, I and 11, respectively, was clearly demon- strated. Both promoters are active in E. coli From the S1

Proteus mirabilis Lipoprotein Gene 4605

- V I - I1

/ u \ IV C G U U A

- A' A\A

U - A A - A _ A u - A A - t r c: *c A - U U - A : $1 A - U, U - A u -&

U' U G - C c \ A C A - U A U

'A - U' A' U U A - U C C 11 U - A G / A G A - U 'C - G A / U - A A - U U - A U - A - V U - A 111 G - C G - C

\

A A

- A \ A - U I

U'A - u, / c \ A G A G -

C G G U A G U I A C U I

U C

\ 1

U A A U U C C U U G U G A - - G c / A' 'G

C - G U A A - U U A G - C U U U - A A - U C A G - C

v ' \

G A Y j U - A A - Us C - G' U U G - C

" / U - A U - A C - G A - U

A - U c: A A A A - U

A - U A - U A c G - C c A A - U U - A A - U U A A - U, U A C - G G - C 'U - A

G - C A - U G C - G A - U A - U G A - U - v' G - C A - U U - A k - G e - C C - G A - U A - U G - C G - C U - A G - C C - G A - U

C C - G A '\ C U C - G G - C. G U - A

G' I U - A /) A c G I

5'Av!AAAG-A - U-AGU .......G G-U - A-GG ........v U-C - G-CU . . . . . e A-A - U - P A " - U - A- &.........AU-A - U-UU-OH -70 +1 +82 +116 +156 +216 +223 +256 +320 +345 438 1

FIG. 9. Possible secondary structures of the P. mirabilis lpp mRNAs. Bases are numbered as shown in Fig. 2. The first nucleotide of transcript I is numbered as +I and that of transcript I1 as -71. The termination codon, UAA, is boxed. Transcript I shares stem-loop Structures I-V with transcript 11. The AG values for the six stem-loop structures shown (I-VI) are calculated according to Tinoco et al. (38) to be -19.4, -9.8, -14.0, -8.4, -14.2, and -19.6 kcal, respectively.

mapping experiments (Fig. 4A), the -35 regions of the P1 and P2 promoters are identified as -33CTGTAT-28 and "04TTGCTT-99, respectively (see Fig. 2), which show some homology to the consensus sequence for the E. coli -35 region TTGACA (31,32). The -10 regions are -11CATACT-6 for the P1 promoter and -s2CATATT-77 or -saTATTGT-75 for the P2 promoter, which also show homology to the E. coli -10 region consensus sequence, TATAAT (31,32). It appears that the P. mirabilis Ipp P1 promoter and its flanking regions show some degree of sequence homology to the promoters of the other four lpp genes (Fig. 8), while the P. mirabilis lpp P2 promoter does not appear to share any significant homology to the P. mirabilis P1 promoter or to any sequence upstream of the promoter regions of the other four lpp genes.

The expression of the P. mirabilis lpp gene is thus controlled by two tandem promoters. The prominent production of tran- script I1 over transcript I seen in P. mirabilis grown in a rich medium (Figs. 4B and 5) suggests that the P2 promoter is more active than the P1 promoter under these growth condi- tions. However, the P1 promoter appears to be more active in E. coli carrying pGC9-25 under the same growth conditions (Fig. 5 ) as well as in an E. coli cell-free system (Fig. 6). The expression of the lpp gene by tandem promoters was also found in M. morganii but not in E. coli (Fig. 7). Existence of tandem promoters has been shown for genes influenced by catabolite repression such as the gal operon (33), genes that

are nitrogen-regulated such as the glnA gene (34), and genes required to produce a large amount of the product such as the genes for ribosomal RNA (35). The tandem lpp promoters may have similar roles or another as yet undetermined role.

Possible stem-loop structures of transcripts I and I1 are shown in Fig. 9. It is noteworthy that the first 70 nucleotides of transcript 11 can form a stem-loop structure with a AG of approximately -20 kcal (Structure VI, Fig. 9). This Structure VI may have an important role in the stability of transcript 11, or have a regulating role for the lpp gene expression. The two lpp mRNAs share the other five secondary structures. Structures I and I1 are common to all the other enterobacterial lpp mRNAs. Structure I1 has the termination codon UAA (boxed in Fig. 9) at or near the loop (AG = -9.8 kcal for P. mirabilis, -10.8 kcal for M. morganii, -8.8 kcal for E. amy- louora, -15.5 kcal for S. marcescens, and -11.3 kcal for E. coli). Structure I probably acts as a rho-independent transcrip- tion termination signal ( AG = -19.4 kcal for P. mirabilis, -21 kcal for E. amylouora, S. marcescens, and E. coli, and -9.4 kcal for M. morganii) . The total AG is approximately -85 and -65 kcal for transcripts I1 and I, respectively.

Acknowledgments-We thank Dr. Pamela J. Green and Dorothy Comeau for their critical reading of the manuscript.

REFERENCES 1. Nakamura, K., Pirtle, R. M., and Inouye, M. (1979) J. Bacterid

137,595-604

4606 Proteus mirabilis Lipoprotein Gene 2. Inouye, M. (1979) in Bwmembranes (Manson, L. A,, ed) Vol. 10, 21. Maniatis, T. T., Fritsch, E. F., and Sambrook, J. (1982) in

pp. 141-208, Plenum' Publishing Corp., New York 3. Nakamura, K., and Inouye, M. (1979) Cell 18, 1109-1117

Moleculnr Cloning: A Laboratory Manual, pp. 431-433, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York

4. Nakamura, K., and Inouye, MI (1980) Proc. Natl. Acad. Sci. 22. Thomas, P. (1983) Methods Enzymol. 100, 255-266

5. Yamagata, H., Nakamura, K., and Inouye, M. (1981) J. Bid. Mol. Biol. 113, 237-251

6. Huang, Y.-X., Ching, G., and Inouye, M. (1983) J. Biol. Chem. 686

7. Tokunaga, M., Tokunaga, H., and Wu, H. C. (1982) Proc. Natl. Acad. Sci. U. S. A. 79, 6156-6160

8. Inouye, S., Franceschini, T., Sato, M., Itakura, K., and Inouye, 27. Lee, N., Nakamura, K., and Inouye, M. (1981) J. Bacteriol. 146,

9. Inukai, M., Takeuchi, M., Shimizu, K., and Arai, M. (1978) J. 28. Inukai, M., and Inouye, M. (1983) Eur. J. Biochem. 130, 27-32 Antibiot. (Tokyo) 31,1203-1205 29. Inouye, M., Shaw, J., and Shen, C. (1972) J. Biol. Chem. 247,

10. Hussain, M., Ichihara, S., and Mizushima, S. (1980) J. Biol. 8154-8159 Chem. 255,3707-3712 30. Green, M. R., and Roeder, R. G. (1980) Cell 22,231-242

11. Buchanan, R. E., and Gibbons, N. E. (1974) Bergey's Manual of 31. Rosenberg, M., and Court, D. (1979) Annu. Rev. Genet. 13,319- Determinative Bacteriology, 8th Ed., Williams and Wilkins, 353 Baltimore 32. Hawley, D., and McClure, W. (1983) Nucleic Acids Res. 11,2237-

12. Sanderson, K. E. (1976) Annu. Rev. Microbiol. 30, 327-349 2255 13. Katz, E., Loring, D., Inouye, S., and Inouye, M. (1978) J. Bate - 33. Aiba, H., Adhya, S., and de Crombrugghe, B. (1981) J. Biol.

14. Ching, G., and Inouye, M. (1986) J. Mol. BWZ. 185,501-507 34. Reitzer, L. J., and Magasanik, B. (1985) Proc. Natl. Acad. Sci. 15. Halegoua, S., Hirashima, A., and Inouye, M. (1974) J. Bacteriol. U. S. A. 82, 1979-1983

16. Nakamura, K., Masui, Y., and Inouye, M. (1982) J. Mol. Appl. Biochem. 53,75-117 Genet. 1,289-299 36. Nakamura, K., Pirtle, R. M., Pirtle, I. L., Takeichi, K., and

17. Coleman, J., Green, P. J., and Inouye, M. (1984) Cell 37, 429- Inouye, M. (1980) J. Biol. Chem. 255,210-216 436 37. Inouye, S., and Inouye, M. (1985) Nucleic Acids Res. 13, 3101-

18. Filip, C., Fletcher, G., Wulff, J. L., and Earhart, D. F. (1973) J. 3109 Bacteriol. 115, 717-722 38. Tinoco, I., Bores, P., Dengler, B., Levine, M., Uhlenbeck, O.,

19. Braun, V., and Sieglin, J. (1970) Eur. J. Biochem. 13, 336-346 Crothers, D., and Gralla, J. (1973) Nature New Biol. 246, 40- 20. Berk, A. J., and Sharp, P. A. (1977) Cell 12,721-732 41

U. S. A. 77, 1369-1373 23. Rigby, P. W., Deickmann, M., Rhodes, C., and Berg, P. (1977) J.

Chem. 256,2194-2198 24. Kajitani, M., and Ishihama, A. (1983) Nucleic Acids Res. 11,671-

258,8139-8145 25. Friden, P., Newman, T., and Freundlich, M. (1982) Proc. Natl.

Acad. Sci. U. S. A. 79,2255-2259 26. Aiba, H. (1983) Cell 32, 141-149

M. (1983) EMBO J. 2,87-91 861-866

riol. 134, 674-676 Chem. 256,11905-11910

120,1204-1208 35. Nomura, M., Gourse, R., and Baughman, G. (1984) Annu. Rev.