Proc. Natl. Acad. Sci. USA Vol. 74, No. 12, pp. 5411-5415, December 1977
Biochemistry
Use of gene fusions to study outer membrane protein localization in Escherichia coli (A receptor/,-galactosidase/hybrid proteins/immunofluorescence)
THOMAS J. SILHAVY*, HOWARD A. SHUMAN*, JON BECKWITH*, AND MAXIME SCHWARTZt *Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, Massachusetts 02115; and tUnite de G6n6tique Molculaire,
Department de Biologie Mo 6culaire, Institut Pasteur, Paris, France
Communicated by Eugene P. Kennedy, September 22,1977
ABSTRACT Escherichia coli strains have been isolated that produce hybrid proteins comprised of an NH2-terminal sequence from the a gene product (an outer membrane protein) and a major portion of the COOH-terminal sequence of -galactosidase (8-D-galactoside galactohydrolase, EC 3.2.1.23; a cytoplasmic protein). These proteins exhibit fl-galactosidase activity. One such strain, 3105, produces a hybrid protein containing very little of the JamB gene protein; the protein is found in the cytoplasm. The protein found in a second strain, pop 3186, contains much more of the lamB gene protein; a substantial fraction of the -galactosidase activity is found in the outer membrane, probably facing outward. These results indicate that information necessary to direct the lamB gene product to its outer membrane location is located within the amB gene itself. The properties of such fusion strains open up the prospect of a precise genetic analysis of the genetic components involved in protein transport. The outer membrane of Escherichia coli contains several distinct proteins not found elsewhere in the bacterial cell (7). We are interested in determining the mechanisms responsible for directing these proteins specifically to this membrane. As a prototype for an outer membrane protein, we have chosen the product of gene lamB. This protein acts as a receptor for bacteriophage A adsorption (8) and also plays a role in the transport of maltose and maltodextrins (9, 10) and in the chemotaxis towards these sugars (11). The lamB gene is located in one of the two divergent operons constituting the malB region of the chromosome (12). As such, it is controlled by the positive regulatory gene of the maltose system, malT (13). We wish to determine which particular components of the lamB gene or its controlling elements are responsible for insuring that this protein is directed to its proper outer membrane location. Recently we suggested that gene fusion could be used for studying problems of protein localization (3). This technique enables one to fuse two structural genes to produce a new gene which codes for a hybrid protein. We demonstrated that a hybrid protein composed of a functional COOH-terminal sequence from fl-galactosidase (f3-D-galactoside galactohydrolase, EC 3.2.1.23; a cytoplasmic enzyme) and an NH2-terminal sequence from another of the proteins involved in maltose transport (the maiF product) was tightly bound to the cytoplasmic membrane. In this paper we report the construction of strains producing two different types of proteins that are hybrids of fl-galactosidase and the lamB gene product. The properties of these strains yield information on the components of the lamB gene determining the localization of its product. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact. 5411
MATERIALS AND METHODS Media and Chemicals are described elsewhere (14-16). Amylose was from Koch Light Laboratories. Phage and Bacteria. The bacterial strains and phage used are listed in Table 1. Bacterial and phage techniques have been described (15). Gene fusions were constructed as described (3, 17). Assays. fl-Galactosidase was assayed according to Miller (15), and NADH oxidase (NADH:02 oxidoreductase; EC 1.6.99.3) according to Osborn et al. (19). Antisera. Rabbit anti-f3-galactosidase was the gift of N. Guiso, Institut Pasteur, Paris. Rabbit anti-RNA polymerase was from S. Austin, Fort Detrick Cancer Research Center, Frederick, MD. Fluorescein conjugated IgG from sheep anti-rabbit IgG (fluorescein anti-IgG) (lot S343) was purchased from Miles-Yeda Ltd., Rehovot, Israel. All three reagents were absorbed with an equal volume of packed MC 4100 cells. Antibody reactivity was confirmed by Ouchterlony double diffusion. Fluorescence Microscopy. Exponential cultures grown in the presence of maltose for 3 hr were harvested and washed twice with minimal medium (M63) at 0-40 and resuspended to 109 cells per ml. Aliquots of washed cells were incubated with 50 ,g of anti-f3-galactosidase, anti-RNA polymerase, or M63 for 1 hr at 0-4o. Bacteria were then washed twice with M63 and samples were placed on microscope slides. The bacteria were examined with a Leitz-ortholux phase contrast microscope using a X54 oil objective and a X10 eyepiece. Fluorescence was observed by illumination with an ultraviolet light source. Photographs were obtained by exposing high-speed Ektachrome type B film (Kodak) for l/4 sec for phase contrast and 1 min for fluorescence pictures. Special Features of Selection for lamB-lacZ Fusions. In general, the techniques used for the steps outlined in Fig. 1 are identical with those described before (3, 17). What follows are procedures specific for the lamB system. (i) Strains carrying a mu prophage inserted in gene lamB were selected as resistant to phage Avir and Mal+ (20). Abbreviations: Genetic nomenclature is from ref. 1. The symbol cf designates fusions (2, 3). Fusions described in this paper are caused by the fusion of two structural genes and result in the production of a hybrid protein. These are referred to as protein fusions and are designated 'I (lamB-lacZ)hyb (3). Fusion components are written in the order of transcription. Bacteriophage A nomenclature is from ref. 4. The symbol "::" is the site of a Mu genome insertion. A "+" or -" before "Mu" indicates the orientation of the Mu genome in a lysogen. In the "+" orientation the Mu genes A through S are in a clockwise orientation (5, 6) on the standard E. coli map. Mu DNA on phage is oriented relative to bacterial genes carried by the phage. Primes, as in Mu' or 'trp, indicate that the DNA for the genetic region referred to is deleted on the side the prime is written.
5412
Proc. Natl. Acad. Sci. USA 74 (1977)
Biochemistry: Silhavy et al. Table 1. Bacterial strains and phage Genotype or bacterial genes carried
Strain MC4100
pop 3024 pop 3032 pop 3105 pop 3186 E7150 pop 468 (KLF12) HfrG6A(malB)
Phage Xpl(209, 118) (Xap malB13)
F
E
This study This study This study This study This study Hofnung Hofnung
::(+Mu')trp 'BA' - W209 lac 'O Z U118Y A' malF malE malK lamB'
17 Casadaban Marchal & Hofnung 16
P,,,,1
K
P.,,,,,1
K
F
b
/amB ImB'
Mu
*
'lamB
Xpl (209,118) c
K
am
i
,'*"'/amB
I
I
P
/Im8' Mu
/OC
to
N{
A',,
K
/amB'Mu
*'B
IamB ''
Y
A'
O'
Z
Y
AIgm8 '
f
A'
,IOMB
d FIG. 1. Scheme for isolating lamB-lacZ protein fusions. (a) The two divergent operons that comprise the malB locus. Arrows indicate direction of transcription. The promoter here is designed Pmai. (b) A Mu (c ts) phage is inserted in lamB (see text for details). The phage can be inserted in either orientation; however, only those inserted with the immunity end (designated by *) promoter distal ("-" orientation) will yield fusions with the Xpl(209, 118) phage used. (c) The Mu lysogen is then lysogenized with Xpl(209, 118). This phage was derived from a trp-lac fusion (W209) and, as such, it carries some trp DNA. It does not contain an intact lac promoter and, in addition, it carries the early lacZ nonsense mutation U118 (17). The parental strain, MC4100, contains a lac deletion; Xpl(209, 118) phage does not contain Xatt. Therefore lysogenization occurs primarily by recombination between the homologous Mu DNA. Since the Mu (c ts) prophage is temperature-inducible, selection for Lac+ survivors at 420 in nearly all cases will yield fusions of the lacZ gene to Pmai. The purpose of the early lacZ nonsense is to select for deletions that extend into the lacZ gene, producing a gene that codes for a hybrid protein (d). Deletions that do not extend past the nonsense mutation will remain Lac-.
12 18; Schwartz, unpublished
lacZ-lamB fusions were spread on lactose minimal agar and incubated (for 1 day at 42° and then at 370), lac + colonies -continued to appear for as long as the agar plates were incubated. Colonies were picked from these plates up to 35 days after spreading. Pop 3105 and pop 3186 were picked after 11 and 32 days, respectively. Mapping of Mu Insertions and Fusion Joints in Gene lamB. The lamB protein plays an essential role in the growth of E. coli on maltodextrins (10). Maltodextrins can therefore be used to select lamB+ recombinants on a background of lamB bacteria. The maltodextrins were prepared as follows: A 1% suspension of amylose was hydrolyzed with 0.5 M H2SO4 for 20 min at 100°; the suspension was cooled down rapidly, neutralized with Ba(OH)2, and centrifuged. The supernatant was concentrated 10-fold over a UM2 Amicon membrane, which retains molecules of 103 molecular weight or heigher. The concentrate was clarified by centrifugation, sterilized, and used as a substrate for selection of lamB+ recombinants. The amount to be used was determined empirically so as to give good growth of a lamB+ strain and very poor growth of a lamB nonsense mutant. All the Mu insertions and gene fusions used in this work were isolated in the same F- strain, MC-4100, which is araD. They were crossed with a series ofpreviously described his Hfr strains (18) carrying deletions extending from malK into lamB. A deletion was considered to recombine with a Mu insertion or a fusion when the number of lamB+ his+ recombinants, selected on EM maltodextrin agar (21) was equal or superior to 10-6 of the number of ara+his+ recombinants, selected on M63 arabinose agar. RESULTS Construction of lamB-lacZ Fusions. The technique we have used for construction of protein fusions has been described in detail elsewhere (3, 17). As a first step it involves generating strains in which the lacZ gene is transposed to a position in or close to the gene of interest, in this case lamB. In order to do this, insertions of the genome of bacteriophage Mu into lamB were isolated (Fig. lb). When infected by a X transducing phage that carries the lac region and a portion of the Mu genome [Xpl(209, 118)], the strains with the insertions can become lysogenic for the X phage via homologous recombination within the Mu DNA sequences (Fig. ic). Such a lysogen is Lac- for two reasons: it lacks a promoter region for the lac operon, and the lacZ gene contains a chain-terminating mutation (lacZU118) close to the beginning of the gene. Simultaneous selection for the loss of Mu phage DNA at 420 (the Mu used is thermoinducible) and for ability to grow on lactose reveals strains in which the lacZ gene I
P,,4/
17
araD139A(lac)U169 strA relA thi araD139A(lac)U169 strA relA thi (lamB::-Mu(c ts)61) araD139A(1ac)U169strA relA thi (lamB::-Mu(c ts)42) araD139A(lac) U169 strA relA thi C(lamB-lacZ)hyb 61-4 araDl39A(lac)U169 strA relA thi cI(IamB-lacZ)hyb 42-1 F'1acZ+lacYNG328 pro+/A(lac-pro)X111 VaIR spc F'argH+ malB+/argH thyA strA recA his, strains carrying A(maIB) 1, 112, 227, 214, 17, 105, 10, 5, 3, 18, 107
(ii) Lysogenization by Xpl(209, 118) of strains carrying a Mu insertion in lamB is complicated by the X-resistance of those strains. This problem was overcome by preparing a mixed lysate of Xpl(209, 118) and XCIh8OAint9 which has the tail, and therefore the host range, of phage 080. Such a lysate contains a certain proportion of particles with the Xpl(209, 118) genome incorporated in the coat of k80 (21). These particles can infect and lysogenize X-resistant bacteria. Cultures of the strains carrying Mu DNA inserted in lamB were infected with the mixed lysate [total multiplicity of infection was 3; about onethird of the particles had the Xpl(209, 118) genome]. After 45 min at 30° they were streaked on Luria broth agar plates coated with about 109 XCIh8OAint9 particles. The Xpl(209, 118) lysogens were recognized among the survivors as immune to XCIh80Aint9 and sensitive to 480 vir. (iii) When cultures of the strains to be used for selecting a
Ref.
Origin
Biochemistry: Silhavy et al. H
M
Proc. Natl. Acad. Sci. USA 74 (1977)
Li L2
5413
A
400F
-
300
z UJ
200
V) 0
FIG. 3. Surface immunofluorescence of lamB-lacZ fusion strains. Bacteria were labeled with antisera and fluorescein anti-IgG. (A) Phase contrast photomicrograph of strain pop 3186 grown in M63 containing glycerol and maltose and stained with anti-fl-galactosidase and fluorescein anti-IgG. (B) Fluorescence photomicrograph of the same field as A9 (C) Phase contrast photomicrograph of strain pop 3105 grown in M63 containing glycerol and maltose and treated as in A. (D) Fluorescence photomicrograph of the same field as C.
0 -j
cD
100
.P
0~
FRACTION NUMBER
FIG. 2. Isopycnic sucrose density gradient centrifugation of a total membrane fraction from strain pop 3186. A fresh overnight culture of the strain grown in minimal medium containing glycerol was inoculated into 50 ml of fresh medium containing maltose and grown for 2.5 hr to O.D. 600 = 1.3. Cells were harvested and the membrane fraction was prepared and subjected to sucrose density gradient centrifugation (19). Hi, Ml, and L1 and L2 indicate the location in the gradient of outer membrane, M-band, and inner membrane, respectively. #l-Galactosidase units are expressed as nmol of o-nitrophenylgalactoside hydrolyzed per min; NADH oxidase units as Amol of NADH oxidized per min.
has been fused to a nearby gene, such as lamB. Since the portion of the lacZ gene that includes lacZU118 is not essential for galactosidase activity (2), deletions of the appropriate extent can generate hybrid proteins with f3-galactosidase activity. A Lac+ thermoresistant clone was classified as a lanB-lacZ fusion strain when its fl-galactosidase was maltose-inducible. Starting from 20 independent Mu insertions, lamB-lacZ fusions were obtained from 6 clones. Two of the lamB-lacZ fusion strains producing hybrid proteins were chosen for further study: pop 3186 and pop 3105. Both strains exhibit maltose control of ,B-galactosidase synthesis. At 300, strain pop 3186 produces about 75 units of f3-galactosidase in M63 glycerol and 974 units in M63 glycerol maltose. In the same media and conditions pop 3105 produces 150 and 904 units, respectively. (At 370, strain pop 3186 produces much less fl-galactosidase only in the uninduced condition for reasons not understood.) Localization of fl-Galactosidase Activity in Fusion Strains. In order to determine if the localization of the fl-galactosidase activity was altered by gene fusion, we prepared whole cell membrane fractions of these two fusion strains as described (3). Results of this experiment showed that the membrane fraction from strain pop 3105 contained only 10% of the #-galactosidase activity. This is identical to what is seen with other fusion strains when the NH2 terminus of the fl-galactosidase hybrid is comprised of a normally cytoplasmic protein (3). With strain pop 3186, however, different results were obtained; 40-70% of the fl-galactosidase activity was in the crude membrane fraction. The cold osmotic shock procedure of Neu and Heppel (22) released essentially no ,B-galactosidase activity from either strain, suggesting that the soluble activity is not periplasmic, i.e., located between the inner and outer membranes. These results 13-
indicate that the hybrid protein of strain pop 3105 is cytoplasmic. When the inner and outer membranes of strain pop 3186 were separated by the method of Osborn et al. (19), approximately 50% of the membrane-bound f3-galactosidase activity sedimented with the outer membrane fraction in the sucrosegradient (Fig. 2). These results are in contrast to those obtained with the strain described in a previous article, where the fgalactosidase activity was converted to an inner membranebound state by gene fusion. In that case, no activity was seen in the outer membrane fraction (3). A strain possessing some fl-galactosidase activity in the outer membrane may be able to grow on lactose even in the absence of lactose permease. A derivative of pop 3186 was constructed that carried a mutation in the lacY gene by recombination with an episome from strain E 7150. The lacY derivative grows on minimal lactose agar although not as well as does the original lacY+ strain. This is seen most clearly at 300. In contrast, a lacY derivative of pop 3105, the fusion strain that produces a cytoplasmic fl-galactosidase, does not grow on lactose at all. In order to test whether the fl-galactosidase portion of the hybrid protein is facing outward from the cell, we performed immunofluorescence experiments using anti-f3-galactosidase and fluorescein-labeled anti-IgG ("sandwich" technique). Although the results (Fig. 3) are much less dramatic than most similar experiments with typical surface antigens, they are nevertheless strongly suggestive. The maltose-induced cells from strain pop 3186 can clearly be seen to be covered by dim fluorescence. No such fluorescence can be seen with induced pop 3105, the strain that produces cytoplasmic hybrid protein. Control experiments with anti-RNA polymerase were likewise negative. Genetic Mapping of Fusion Strains. Like lamB nonsense mutants (10), strains carrying either Mu insertions or fusions in lamB grow very poorly on maltodextrins. This observation provided a basis for the selection of lamB+ strains, and thus for deletion mapping of the insertions and the fusion joints. (A more detailed lamB map will be published elsewhere.) It is clear (Fig. 4) that both the fusion joint in pop 3105 and the Mu insertion from which the fusion is derived lie very early in lamB. On the other hand, the fusion joint in pop 3186 maps rather late in
Biochemistry: Silhavy et al.
5414
Proc. Natl. Acad. Sci. USA 74 (1977) A
malK
B
C
D
E
F
lamB 1
Mu insertions
fusion joint
21
22
41 61 72
51 82 31 102 52
61-4
91 101
92
62 42
32 71
-
-
42 -I
1'
6112
AlI a 17
A214 A227
65
WO
A 107 ASAS~~~~~~~~~~~~~~~~~~~~~~~~~~~ A18_I _AO_
FIG. 4. Location of the fusion joints in gene lamB. The Mu insertions were all isolated during the course of this work. When they start with the same digit (i.e., 11 and 12 for instance), they were isolated from the same Mu-treated culture and could be identical. Within each interval defined by deletion endings, the mutations have been ordered according to their frequency of recombination with the deletion that ends closest to the left. Only 10-fold differences in recombination have been considered. Deletion A112 is shown to end within malK because it recombines with some known malK point mutations (18; Schwartz, unpublished data). Strain pop 3105 contains fusion number 61-4; strain pop 3186 contains fusion number 42-1.
FIG. 5. Sodium dodecyl sulfate/polyacrylamide gel electrophoresis of whole cell extracts of the IamB-lacZ fusions. Electrophoresis was performed as described (24). Whole cell extracts were prepared by resuspending a washed exponential culture in -sodium dodecyl sulfate sample buffer (24) and heating to 1000 for 2 min. A volume of extract containing approximately 150 ,tg of protein was loaded in each gel slot. (A) 13-Galactosidase, 2.5,Mg; (B) pop 3105, uninduced; (C) pop 3105, induced; (D) pop 3186, induced; (E)op 3186, uninduced; (F) 03-galactosidase, 2.5,Mg. The hybrid protein in each case is indicated
lamB. It is important to note, however, that this fusion joint is at a site earlier in the lamB gene than the Mu insertion from
is added to growing cultures of strain pop 3186, after a short time we observe that the cells do not divide properly and often elongate to four or five times the normal cell length (Fig. 3). They appear to form septa, but do not separate. Growth in the presence of maltose for extended periods of time (more than 3 hr at 370) leads to some cell lysis. When a wild-type copy of the maiB region is introduced into pop 3186, either with an
which the fusion is derived. Therefore the deletion that generated the fusion removed part of the lamB gene and all of the Mu prophage (see Fig. ld). Past experience has taught us that deletions that generate protein fusions with the Xpl(209, 118) phage all end in the same region of the lacZ gene, probably resulting in the removal of approximately 40 amino acids from the NH2 terminus of galactosidase. The same holds true for the two fusion strains described here. Details and results of the mapping procedure will be published elsewhere. Since the genetic mapping results indicate that there is considerably more of the lamB gene present in the fusion of pop 3186 than in that of pop 3105 but approximately the same amount of lacZ gene, we might expect the hybrid protein from the two strains to be very different in molecular weight. Due to its exceptionally high molecular weight (116,000) (23) wild-type fl-galactosidase can be easily detected on sodium dodecyl sulfate/polyacrylamide gel electrophoresis of whole cell extracts. When extracts of pop 3105 and pop 3186 were run on such gels, maltose-inducible proteins of high molecular weight appeared in each case, and we presume these to be the hybrid proteins (Fig. 5). According to the gel pattern, the protein produced by pop 3105 has about the same molecular weight as wild-type /3-galactosidase, while that produced by pop 3186 is larger by 10,0000-15,000 daltons. For comparison, the molecular weight of purified X receptor is approximately 55,000 (Schwartz, unpublished data). Unusual Properties of a lamB-lacZ Fusion Strain. Strain pop 3186, which has hybrid protein in the outer membrane, exhibits certain abnormal growth properties. lamB strains are normally Mal+ because the lamB product is not necessary for the transport of maltose at the concentrations normally used in growth media (10). Yet, strain pop 3186 grows extremely slowly on maltose as carbon source. However, this strain appears to be maltose sensitive rather than maltose negative. Indeed, although it grows normally on various minimal or complete media (e.g., M63 glycerol or Luria broth), its growth is strongly inhibited by the addition of maltose. Further, when this sugar 1-
by an arrow.
episome (KLF12) or a transducing phage (XpmalB13), the resulting merodiploid is still sensitive to the presence of maltose.
These results are consistent with the proposal that induction of the synthesis of the hybrid protein by maltose interferes with cellular growth. Further strong support for this proposal comes from a study of mutants that are no longer sensitive to maltose. Among these, we found strains carrying amber mutations either in the lamB or in the lacZ portion of the hybrid gene. Suppression of the amber mutations restores the maltose sensitivity. Finally, strain pop 3105, which produces a cytoplasmic hybrid protein, exhibits no abnormal growth properties. Although we have no further information on the mechanism of growth inhibition in strain pop 3186, the results are consistent with the hypothesis that transport across or incorporation into the membranes of this unusually large protein may be responsible for the abnormal growth properties.
DISCUSSION We have isolated two types of strains in which the structural gene for the cytoplasmic enzyme (3-galactosidase (lacZ) is fused with a portion of the structural gene for an outer membrane protein, the product of gene lamB. Genetic mapping shows that the hybrid gene of strain pop 3186 contains much more of the lamB gene than that of strain pop 3105 (Fig. 4). Accordingly, the hybrid protein produced by strain pop 3186 is considerably larger than that of strain pop 3105 (Fig. 5). Using three different methods we have obtained evidence that a fraction of the l-galactosidase of strain pop 3186 is extracytoplasmic, and probably in the outer membrane. (i) By genetic manipulation we could show that this strain could grow on lactose even in the absence of a lactose transport
Biochemistry: Silhavy et al. system. This result indicates the presence of fl-galactosidase activity outside of the cytoplasmic membrane barrier. (ii) Biochemically, using sucrose density gradients, we could demonstrate the presence of a significant amount of fl-galactosidase activity in an outer membrane fraction (Fig. 2). (iii) Finally, using immunofluorescence, we obtained evidence for the presence of some 03-galactosidase crossreacting material at the cell surface. When the same techniques were used with strain pop 3105, the results indicated that the hybrid protein synthesized by the
strain was cytoplasmic, as is
wild-type fl-galactosidase. From
the genetic and biochemical data, we presume that the hybrid gene in strain pop 3105 contains a nucleotide sequence, albeit small, from the beginning of the lamB gene. The observation made with the two fusion strains therefore strongly suggests that information that directs the lamB gene product to its outer membrane location lies within the lamB gene. The interpretation of these results is somewhat clouded by the finding that not all of the hybrid protein produced by pop 3186 is found in the outer membrane location. We do not understand the heterogeneous location of this protein. However, it is possible that either the unusually large size of the protein or the interference with cell division and growth that results from the transport of this protein may account for this finding. The mechanism of transport of proteins to extracytoplasmic locations in bacteria is unknown. There does exist, however, a specific model to explain protein excretion into the endoplasmic reticular lumen in eukaryotic cells. According to this model (25, 26) proteins destined to be excreted are synthesized as precursors with a sequence of hydrophobic amino acids at their NH2-terminal end. This sequence, the signal sequence, is involved in the transport through the membrane and is cleaved either during or after transport, revealing the mature protein. Recently, evidence has been presented that, in E. coli, amino acid sequences analogous to the signal sequence may be involved in the excretion of alkaline phosphatase into the periplasm (27) and the incorporation of a lipoprotein into the outer membrane (28). The evidence, however, does not allow the conclusion that a signal sequence is sufficient or even necessary for transport of these proteins. Further, models to explain protein transport in E. coli must take into account the several noncytoplasmic protein locations: the cytoplasmic membrane, the periplasmic space, and the outer membrane. Entirely different mechanisms may explain the localization of proteins in these different locations. For the lamB gene product, the evidence presented here is consistent with a signal mechanism for transport. However, further characterization of the hybrid proteins and of the lamB gene product itself will be necessary before the applicability of this model can be tested. Whatever the mechanism, it is clear that the sequence in gene lamB responsible for the localization of its product is capable of directing even a large soluble molecule such as f3-galactosidase to an outer membrane location. Strain pop 3186, producing a hybrid protein, a portion of which is found in the outer membrane, is sensitive to the presence of maltose. The inhibition of bacterial division is apparently due to the synthesis of hybrid protein. As a result, it is possible to select for mutants resistant to the effect of maltose that are due to mutations altering the production of the hybrid protein. Among the mutants, we might expect to find some that are defective in the export process for this protein. Such a direct selection for "export-defective" mutants in combination with
Proc. Natl. Acad. Sci. USA 74 (1977)
5415
additional lamB-lacZ fusions should allow us to better understand the molecular mechanism of the export process. We thank E. Brickman, R. MacGillivray, and A. McIntosh for assistance and M. Hofnung for providing bacterial strains. We also thank D. Ault and E. Unanue for their help with the fluorescence microscopy and A. Ullmann, who suggested the use of UM2 Amicon membrane for purification of maltodextrins. This work was supported by a National Science Foundation (PCM76-21955) and an American Cancer Society Grant (VC-13F) to J.B. and by a grant of the Delegation Generale a la Recherche Scientifique et Technique (75 7 0039) to M.S. H.A.S. was supported by a National Research Service Award (5T32GM07306-02). T.J.S. is a fellow of the Jane Coffin Childs Memorial Fund for Medical Research. This investigation has been aided by a grant from the Jane Coffin Childs Memorial Fund for Medical Research. The "Unite de Genetique Moleculaire" is part of the Laboratoire Associe No. 270 (Centre Nationale Recherche Scientifique). 1. Bachmann, B. J., Low, K. B. & Taylor, A. L. (1976) Bacteriol. Rev. 40, 116-167. 2. Muller-Hill, B. & Kania, J. (1974) Nature 249,561-563. 3. Silhavy, T. J., Casadaban, M. J., Shuman, H. A. Beckwith, J. R. (1976) Proc. Natl. Acad. Sci. USA 73,3423-3427. 4. Hershey, A. D., ed. (1971) The Bacteriophage Lambda (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY). 5. Zeldis, J., Bukhari, A. & Zipser, D. (1973) Virology 55, 289294. 6. Abelson, J., Boran, W., Bukhari, A. I., Fallen, M., Howe, M., Metlay, M., Taylor, A. L., Toussaint, A., Van De Putte, P., Westmaas, G. C. & Wijffelman, A. C. (1973) Virology 54,9092. 7. Schnaitman, C. A. (1970) J. Bacteriol. 104,890-891. 8. Randall-Hazelbauer, L. & Schwartz, M. (1973) J. Bacteriol. 116, 1436-1446. 9. Szmelcman, S. & Hofnung, M. (1975) J. Bacteriol. 124, 112118. 10. Szmelcman, S., Schwartz, M., Silhavy, T. J. & Boos, W. (1976) Eur. J. Biochem. 65,13-19. 11. Hazelbauer, G. L. (1975) J. Bacteriol. 124, 119-126. 12. Hofnung, M. (1974) Genetics 76, 169-184. 13. Hofnung, M. & Schwartz, M. (1971) Mol. Gen. Genet. 112, 117-132. 14. Gottesman, S. & Beckwith, J. R. (1969) J. Mol. Biol. 44, 117127. 15. Miller, J. H. (1972) Experiments in Molecular Genetics (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY). 16. Hatfield, D., Hofnung, M. & Schwartz, M. (1969) J. Bacteriol. 98,559-567. 17. Casadaban, M. J. (1976) J. Mol. Biol. 104, 541-555. 18. Hofnung, M., Hatfield, D. & Schwartz, M. (1974) J. Bacteriol. 117,40-47. 19. Osborn, M. J., Gander, J. E., Parisi, E. & Carson, J. (1972)J. Biol. Chem. 247,3962-3972. 20. Hofnung, M., Jezierska, A. & Braun-Breton, C. (1976) Mol. Gen. Genet. 145, 207-213. 21. Appleyard, R., MacGregor, J. & Baird, K. (1956) Virology 2, 565-574. 22. Neu, H. C. & Heppel, L. A. (1965) J. Biol. Chem. 240, 36853692. 23. Fowler, A. V. & Zabin, I. (1977) Proc. Natl. Acad. Sci. USA 74, 1507-1510. 24. Laemmli, U. K. (1970) Nature 227,680-685. 25. Blobel, G. & Sabatini, D. D. (1971) in Biomembranes, ed. Manson, L. A. (Plenum Press, New York), Vol. 2, pp. 193-195. 26. Blobel, G. & Dobberstein, D. (1975) 1. Cell Biol. 67,835-851. 27. Inouye, H. & Beckwith, J. (1977) Proc. Natl. Acad. Sci. USA 74, 1440-1444. 28. Inouye, S., Wang, S., Sekizawa, J., Halegoua, S. & Inouye, M. (1977) Proc. Natl. Acad. Sci. USA 74, 1004-1008.