Gene, 96 (1990) 51-57 Elsevier
51
GENE 03780
Correct insertion of a simple eakaryotic plasma-membrane protein into the cytoplasmic membrane of E s c h e r i c h i a coli
(Recombinant DNA;/~-lactamase; protein fusions; ampiciilin resistance; Escherichia coil direct expression vector; membrane protein topology;/~-subunit of sheep-kidney Na, K-ATPase)
Yianbiao Zhang and Jenny K. Broome-Smith Microbial Genetics Group, School of Biological Sciences, Universityof Sussex, Falmer, Brighton, BN1 9QG (U.K.) Received by K.F. Chater: 11 May 1990 Revised: 25 July 1990 Accepted: 26 July 1990
SUMMARY
A genetic system for directly synthesizing eukaryotic membrane proteins in ~.~,.,,~,,,.,,,,,~-..z,~.~,~..,.,.,,""~;~ d assessk-~g"~--.:-u,,.,,ability to insert into the bacterial cytoplasmic membrane is d~.'scribed. The components of this system are the direct expression vector, pYZ4, and the mature/Mactamase (BIaM) cassette plasmid, pYZ5, that can be used to generate translational fusions of BIaM to any synthesized membrane protein. The/~-subunit of sheep-kidney Na, K-ATPase (flNKA), a class-II plasma membrane protein, was synthesized in E. coli using pYZ4, and BIaM was fused to a normally extracellular portion of it. The fusion protein conferred ampicillin resistance on individual host cells, indicating that the BIaM portion had been translocated to the bacterial periplasm, and that, by inference, the eukaryotic plasma-membrane protein can insert into the bacterial cytoplasmic membrane. A series of 31 pNKA: • BIaM fusion proteins was isolated and characterised to map the topology of the eukaryotic plasma membrane protein with respect to the bacterial cytoplasmic membrat~e. This analysis revealed that the organisation of the/~NKA in the E. coli cytoplasmic membrane was indistinguishable from that in its native plasma membrane.
INTRODUCTION
Clear parallels exist between the processes ol protein translocation across the eukaryotic endoplasmic reticulum Correspondence to: Dr. J.K. Broome-Smith, M~erebial Genetics Group, Schod of Biological Sciences, University of Sussex, Brighton .qN 1 9QG (U.K.) Tel. (0273)606755; Fax (0273)678433. Abbreviations: aa, amino acid(s), Ap, ampicillin; pGal,/~-galactosidase; Bla, fi-lactamase; BIaM, mature portion of Bla; blaM, 5'-trun~'ated bla gene coding for BIaM;/~IKA,/~-subunit of sheep-kidney Na,K-ATPase; bp, base pair(s); A, deletion; lg, immunoglobulin; IPTG, isopropyl/~-D-thiogalactopyranoside; kb, kilobase(s) or 1000 bp; Kin, kanarnycin; MCS, multiple cloning site; MIC, minimum inhibitory concentra'!ion; nt, nucleotide(s); ori, origin of DNA replication; PAGE, polyacrylarrdde-gel electrophoresis; s, resistance/resistant; s, sensitive/sensitivity; SDS, sodium dodecyl sulfate; ss, single strand(ed); Tc, tetracycline; TEM, specifies the source of ~-lactarnase; XGal, 5-bromo-4-chloro-3.indolyl/~-D-galactopyranoside; [ ], denotes plasmid-carrier state; : :, novel joint (fusion). d27~-1! 19/90/$03.50 © 1990Elsevier Science Publishers B.V. (BiomedicalDivision)
and bacterial cytoplasmic membranes. For example, several eukaryotic presecretory proteins have been shown to be translocated across the bacterial cytoplasmic membrane and correctly processed when expressed in E. coli (Talmadge et al., 1980; Gray et al., 1985). Several observations suggest that the processes of protein assembly into these two membranes may also share similarities. Both eukaryotic and prokaryotic cytoplasmic membrane protei,s can be subdivided into common topological classes (e.g., see yon Heijne, 1988), and, from the limited e,,idence available, their topogenic signals appear to be recognised by similar criteria (e.g., see Davis and Hsu, 1986). The above findings suggest that many eukaryotic plasmamembrane proteins may be capable of correct insertion into tl:c bacterial cytoplasmic membrane. Two particularly clear examples of this have been described. Expression of the human/~2-adrenergic receptor, fused to a large portion of /~Gal, results in the formation of ~/2-adrenergic receptor
52 ligand-binding sites in E. coli (Marullo et al., 1988), and expression of the human erythrocyte glucose transporter restores glucose transport to a mutant ofE. coli defective in all known glucose-transport pathways (Sarkar et al., 1988), Since the ligand-binding and transport functions of these two proteins depend on the correct juxtaposition of multiple transmembrane segments (and not just on the correct folding of a soluble domain) functional expression in E. coli provides striking evidence for their correct assembly into the bacterial cytoplasmic membrane. Previously we described a gene fusion system for analysing the topology of bacterial cytoplasmic membrane proteins expressed in E. coll. In this system the mature form of TEM Bla is fused to a progressively truncated target membrane protein and cells producing BlaM fusion proteins are identified by their ability to grow when inoculated at high density (i.e., patched) onto agar containing Ap. Fusion proteins in which the BIaM portion is translocated across the cytoplasmic membrane can then be distinguished from those in which it is retained in the cytoplasm since only the former class can protect individual cells from lysis by Ap (Broome-Smith and Spratt, 1986). This system has been successfully used to map the cytoplasmic and periplasmic domains of several bacterial cytoplasmic membrane proteins (Broome-Smith and Spratt, 1986; Edelman et al., 1987; Bowler and Spratt, 1989). Since only translocated forms of Bla confer Ap a upon a single host cell, it should be possible to assess whether any foreign membrane protein can insert into the E. coli cytoplasmic membrane simply by fusing BIaM to a portion of it that is normally extracytoplasmic, and determining whether the resultant fusion protein confers Ap R on individual bacterial cells. Where insertion can be demonstrated, a comprehensive series of BlaM fusion proteins can be generated and characterised so that the transmembrane organisation of the protein can be determined. The aim of the present study was to develop a system for directly synthesisipg eukaryotic membrane proteins in E. coli and to assess their ability to insert into the bacterial cytoplasmic membrane using a BIaM fusion approach.
RESULTS AND DISCUSSION (a) Rationale for using BlaM as an indicator of membrane insertion Only translocated forms of Bla confer Ap a on individual cells. Therefore, if the BlaM-coding region (blaM) is fused in-frame to a gene that is directly expressed in E. coli and the resultant fusion protein confers Ap a on individual cells, this indicates that the truncated portion of the target protein directs the translocation of BlaM across the bacterial cytoplasmic membrane and, by inference, that the protein itself
inserts into the bacterial membrane. This test can therefore be used to assess whether any protein made in E. coli is capable of insertion into the bacterial cytoplasmic membrane. Moreover, it can, if desired, be coupled to a direct selection for Ap a transformants. Only cells producing translocated forms of Bla survive and form colonies on agar containing Ap (all transformants producing cytoplasmic forms of Bla, as well as all non-Bla producers, are lysed and thus eliminated in this selection): hence this approach can provide a rapid indication of E. coil cytoplasmic membrane insertion. Plasmid pYZ4 (Fig. la) was therefore constructed as a multipurpose plasmid vector that would facilitate the direct production of eukaryotic membrane proteins in E. coli and the generation of blaM fusions to the expressed membrane protein genes, pYZ4 is a Km a derivative of pBR322 that contains the phage fl ori for ss DNA replication and encodes a modified form of the lacZ~-peptide under the control of the lacUV5 promoter. The modified 0t-peptidecoding region has an Ncol site flanking its start codon and an array of unique cloning sites situated just downstream from the translation start. Since the consensus sequence for translation initiation in higher eukaryotes, GCCACCA TGG, includes the NcoI recognition sequence, the start codons of many eukaryotic coding regions are flanked by naturally occurring NcoI sites (reviewed by Kozak, 1987). Therefore, their cDNAs can be readily subcloned between the NcoI site and any of the 'other restriction sites in the MCS of pYZ4 to yield a derivative which directs bacterial synthesis of the unfused eukaryotic gene product, under the control of the lacUV5 promoter. The use of this inducible promoter, which exhibits only a very low basal level of transcription, should permit the successful subcloning of eukaryotic membrane protein genes even if their overexpression is lethal to E. coli cells. Since subcloning results in the insertional inactivation of the 0c-peptide coding region, by using an appropriate 0¢-complementation host strain ofE. coli(such as TGI; see legend to Fig. 1) transformants containing recombinant plasmids can be distinguished from those containing pYZ4 by their white rather than blue color on agar containing IPTG and XGal. For BlaM fusion analysis pYZ4 is used in conjunction with the BlaM cassette plasmid pYZ5 (Y.Z. and J.K.B.-S., unpublished; Fig. lb). When target genes have been sub~Ioned into pYZ4, so that they are bacterially expressed under lacUV5 promoter control, blaM fusions can be made by ligating the 5' (blunt) PvuII end of the BlaM cassette to specific sites within the target gene, and the 3' end, generated by cutting within the MCS of pYZ5, to the cognate site 3' to the eukaryotic coding region in the pYZ4 derivative. The arrangement of sites in the MCS ofpYZ4 is ideally suited for the unidirectional removal of C-terminal portions of inserted coding regions via exonuclease III/S 1 nuclease
53 E ,
.,_ Bc I~l(ef) i:r~lBT/ -I I II f
Pv
Nc E [B I IhBc
E(ef) I
E(ef) I
Ikll ( e f t / BamH( / /
.
~c~/c "¢ V /,Y.
.BH Xb K Sp ~ Hae(ef) "'
'Z
It-
y yN~,BH Xb K Sp
(e)
BHXbKSp ~EcvSsSm
Pv
(b) Fig. 1. Plasmid vectors for the direct expression and BiaM fusion analysis of eukaryotic membrane proteins in E. coll. (a) Construction ofpYZ4. Only the relevant restriction sites are shown. The E. coli strain TGI (Alac-pro, thi, supE, hsdS [F' , proA +B +, lacI° Z AM l 5, traD36] was used for transformations and was grown in L broth or on L agar (Miller, 1972). The plasmid pJBS633, designed for the BlaM fusion analysis of bacterial membrane proteins in E. coii (Broome-Smith and Spratt, 1986), was
digestion (Henikoff, 1984), and thus facilitates the generation of a series of blaM fusions to random positions throughout the gene (see section c). Once fusions are obtained, ss DNA of each fusion derivative can be made and dideoxy sequencing can be primed from within the start of the BlaM-coding region and across the fusion junction (Figs. la and 2). (b) Direct expression of the pNKA in Escherichia coil, and a BlaM fusion test for insertion into the bacterial cytoplasmic membrane The Na,K-ATPase establishes and maintains the Na ÷ and K + gradients across the plasma membrane of animal
digested with AhalIl + EcoRl, and the Tc R and blaM regions were replaced with the EcoRI-HindIII fragment of pAZe3ss, which contains a Shine-Dalgarno sequence (AGGAGG) and appropriately spaced start codon flanked by an Ncol site (C/CATGG) (ZabaUos et al., 1987). An end-filled 104-bp EcoRI fragment carrying the lacUV5 promoter was ligated to pYZ2 DNA, that had been digested with EcoRI and end-filled, to yield the plasmid pYZ3 in which the iacUV5 promoter is flanked by XmnI sites. Finally the BamHI-Haell fragment of M13tgl31, that encodes all except the first few aa of a modified LacZ0t-peptide (Kieny et al., 1983), was ligated to pYZ3 DNA digested with BclI (and end-fiUed) and BamHI, to yield the plasmid pYZ4. The phage fl o;i for ss DNA replication in pYZ4 is oriented such that template DNA can be primed for dideoxy sequencing in a counter-clockwise direction. (b) Restriction map of pYZ5. pYZ5 is a Tc R derivative of pBR322 that contains the unexpressed BlaM coding region from pJBS633, followed by a MCS derived from M 13tg 131. The PvulI-EcoRI blaM cassette is shown boxed. The sequence across the PvulI site (CAG/CTG) and the start of BIaM is: ... CAG CTG CGT CAC CCA GAA ... LeuArgHis ProGlu... +! +2 +3 Abbreviations: A, Ahalll, B, BamHI; Be, Bell; E, EcoRl; Ev, EcoRV; H, HindllI; Hae, Haell; K, KpnI; Nc, Ncol; P, Pstl; Pv, PvulI; S, Sail; Sm, Sinai; Sp, SphI; Ss, Sstl; X, Xmnl; Xb, XbaI; (el'), end-filled; p, lacUV5 promoter; rbs, ribosome-binding site; z, LacZu-peptide coding region. 'z, incomplete LacZu-peptide-coding region; Ap', unexpressed BlaM-coding region.
cells. It consists of a catalytic at-subunit (that probably spans the plasma membrane six times or more) and a p-subunit that is essential for activity of the holoenzyme. The pNKA is a 302-aa glycoprotein. Inspection of the aa sequence reveals only one hydrophobic segment that is long enough to form an 0t-helical membrane-spanning segment (aa 34-60), and this is preceded by a short, highly charged, net basic, domain and followed by a large C-terminal domain containing three potential glycosylation sites (Shull et al., 1986). Biochemical and genetic analyses of the structure and assembly of two homologous mammalian p-subunit polypeptides have since confirmed that the/~-subunit is a class-II membrane protein (van Heijne, 1988), consist-
54 Nc
'LB HxbKSp EEvSsSm BHXb~mSP
Nc
Pstl( e ~ P
[I-//~"@ pYZ4 '] Ncol Hindll, (el) ' ~ '(~ , It ;-. / .t-Ncol\1~ "i'- *" 07/ fragment Ahalll l ~ l i
pYZ,I "l"qlil I~:Ebvl~Slm _A } ~ _
Xbil
(,f)
EcoFtl
blaPI cassel:l:e N ~ ' ~ / P .,., , /"
I1,
pYZiilblaM / I
'""'
il
Fig. 2. The expression of the flNKA in E. coil, and construction offlNKA : : BlaM fusions. See sections b and e, for details. Only the relevant restriction sites are shown, blaM was fused in-frame after codon 147 of the flNKA gene by ligating the PvulI-geoRl BIaM cassette of pYZ5 to pYZI 1 DNA that had been digested to completion with PstI, end-trimmed with S 1 nuclease to.xemove the resultant Y-overhanging ends, and then digested with £coRI. The BIaM portion offlNKA : : BIaM fusion proteins is shown by thick arrows. The dashed line represents positions at which blaM could be fused within the C-terminally resected flNKA-coding region. Plasmid ss DNA was obtained from transformants that appeared to produce in-frame fusion proteins by :.nfection with variant R408 of phage fl (Russel et al., 1986), and the nt sequence across the fusion junction was determined by dideoxy sequencing using a primer (5'-dCTCGTGCACCCAACTGA) that is complementary to codons 14-18 of blaM. Abbreviations as Fig. 1, and (et), end-trimmed.
ing of a short N-terminal cytoplasmic domain, a single hydrophobic transmembrane segment that functions both to translocate and anchor the bulk of the protein to the membrane, and a large C-terminal extracellular domain (Ovchinnik6v et al., 1987; Kawakami and Nagano, 1988). A plasmid-cloned eDNA for the flNKA was obtained from Drs. G.E. Shull and J.B. Lingrel (of the University of Cincinnati, U.S.A.) and the/~NKA-coding region was subcloned into the vector pYZ4 as an NcoI-AhalII fragment (Fig. 2). The resultant plasmid, pYZ 11, directs the synthesis of the flNKA under iacUV5 promoter control. Production of this protein has no o,bvious deleterious effects on the E. coil lacl Q strain TG1. Moreover, pYZ 11 can be introduced into, and maintained in, a derivative of JM 101 cured of its F' (JMI01 F - ; see legend to Fig. 3.) that does not produce any lac represser and hence constitutively synthesises the fi-subunit polypeptide. To determine whether the flNKA inserts into the E. coli cytoplasmic membrane, BIaM was fused to a position (aa 147) in the protein that is normally exposed on the extracellular side of the plasma membrane (yielding the plasmid pYZ 11-147/blaM; Fig. 2). Well-separated cells of
TGl[pYZII-147/blaM] were inoculated onto agar containing a range of Ap concentrations. The MICs of Ap were 1600 pg/ml and 25 #g/ml in the presence or absence of 1 mM IPTG, respectively. Since induced expression of the fusion protein confers Ap R on individual cells (plasmidfree cells, or cells producing cytoplasmic forms of Bla, are killed by 5 #g Ap/ml), we conclude that the truncated form of this eukaryotic membrane protein directs the translocation of BIaM to the periplasm and in turn we infer that the flNKA inserts into the E. coli cytoplasmic membrane. The above results also demonstrate that the lacUV5 promoter of pYZ4 is tightly controlled, since it yielded only a low basal level, and a substantial (and approx. 60-fold greater) induced level, of Ap R. (c) BlaM fusion analysis of the organisation of the pNKA in the Escherichia cell cytoplasmic membrane Fusions of blaM to random positions throughout the flNKA-coding region were constructed by digesting pYZ 11 with XbaI + SstI and then resecting unidirectionally into the gene with exonuclease III followed by S1 nuclease (Fig. 2). The resected ends were made blunt by end-filling
55
51
GENE 03780
Correct insertion of a simple eakaryotic plasma-membrane protein into the cytoplasmic membrane of E s c h e r i c h i a coli
(Recombinant DNA;/~-lactamase; protein fusions; ampiciilin resistance; Escherichia coil direct expression vector; membrane protein topology;/~-subunit of sheep-kidney Na, K-ATPase)
Yianbiao Zhang and Jenny K. Broome-Smith Microbial Genetics Group, School of Biological Sciences, Universityof Sussex, Falmer, Brighton, BN1 9QG (U.K.) Received by K.F. Chater: 11 May 1990 Revised: 25 July 1990 Accepted: 26 July 1990
SUMMARY
A genetic system for directly synthesizing eukaryotic membrane proteins in ~.~,.,,~,,,.,,,,,~-..z,~.~,~..,.,.,,""~;~ d assessk-~g"~--.:-u,,.,,ability to insert into the bacterial cytoplasmic membrane is d~.'scribed. The components of this system are the direct expression vector, pYZ4, and the mature/Mactamase (BIaM) cassette plasmid, pYZ5, that can be used to generate translational fusions of BIaM to any synthesized membrane protein. The/~-subunit of sheep-kidney Na, K-ATPase (flNKA), a class-II plasma membrane protein, was synthesized in E. coli using pYZ4, and BIaM was fused to a normally extracellular portion of it. The fusion protein conferred ampicillin resistance on individual host cells, indicating that the BIaM portion had been translocated to the bacterial periplasm, and that, by inference, the eukaryotic plasma-membrane protein can insert into the bacterial cytoplasmic membrane. A series of 31 pNKA: • BIaM fusion proteins was isolated and characterised to map the topology of the eukaryotic plasma membrane protein with respect to the bacterial cytoplasmic membrat~e. This analysis revealed that the organisation of the/~NKA in the E. coli cytoplasmic membrane was indistinguishable from that in its native plasma membrane.
INTRODUCTION
Clear parallels exist between the processes ol protein translocation across the eukaryotic endoplasmic reticulum Correspondence to: Dr. J.K. Broome-Smith, M~erebial Genetics Group, Schod of Biological Sciences, University of Sussex, Brighton .qN 1 9QG (U.K.) Tel. (0273)606755; Fax (0273)678433. Abbreviations: aa, amino acid(s), Ap, ampicillin; pGal,/~-galactosidase; Bla, fi-lactamase; BIaM, mature portion of Bla; blaM, 5'-trun~'ated bla gene coding for BIaM;/~IKA,/~-subunit of sheep-kidney Na,K-ATPase; bp, base pair(s); A, deletion; lg, immunoglobulin; IPTG, isopropyl/~-D-thiogalactopyranoside; kb, kilobase(s) or 1000 bp; Kin, kanarnycin; MCS, multiple cloning site; MIC, minimum inhibitory concentra'!ion; nt, nucleotide(s); ori, origin of DNA replication; PAGE, polyacrylarrdde-gel electrophoresis; s, resistance/resistant; s, sensitive/sensitivity; SDS, sodium dodecyl sulfate; ss, single strand(ed); Tc, tetracycline; TEM, specifies the source of ~-lactarnase; XGal, 5-bromo-4-chloro-3.indolyl/~-D-galactopyranoside; [ ], denotes plasmid-carrier state; : :, novel joint (fusion). d27~-1! 19/90/$03.50 © 1990Elsevier Science Publishers B.V. (BiomedicalDivision)
and bacterial cytoplasmic membranes. For example, several eukaryotic presecretory proteins have been shown to be translocated across the bacterial cytoplasmic membrane and correctly processed when expressed in E. coli (Talmadge et al., 1980; Gray et al., 1985). Several observations suggest that the processes of protein assembly into these two membranes may also share similarities. Both eukaryotic and prokaryotic cytoplasmic membrane protei,s can be subdivided into common topological classes (e.g., see yon Heijne, 1988), and, from the limited e,,idence available, their topogenic signals appear to be recognised by similar criteria (e.g., see Davis and Hsu, 1986). The above findings suggest that many eukaryotic plasmamembrane proteins may be capable of correct insertion into tl:c bacterial cytoplasmic membrane. Two particularly clear examples of this have been described. Expression of the human/~2-adrenergic receptor, fused to a large portion of /~Gal, results in the formation of ~/2-adrenergic receptor
52 ligand-binding sites in E. coli (Marullo et al., 1988), and expression of the human erythrocyte glucose transporter restores glucose transport to a mutant ofE. coli defective in all known glucose-transport pathways (Sarkar et al., 1988), Since the ligand-binding and transport functions of these two proteins depend on the correct juxtaposition of multiple transmembrane segments (and not just on the correct folding of a soluble domain) functional expression in E. coli provides striking evidence for their correct assembly into the bacterial cytoplasmic membrane. Previously we described a gene fusion system for analysing the topology of bacterial cytoplasmic membrane proteins expressed in E. coll. In this system the mature form of TEM Bla is fused to a progressively truncated target membrane protein and cells producing BlaM fusion proteins are identified by their ability to grow when inoculated at high density (i.e., patched) onto agar containing Ap. Fusion proteins in which the BIaM portion is translocated across the cytoplasmic membrane can then be distinguished from those in which it is retained in the cytoplasm since only the former class can protect individual cells from lysis by Ap (Broome-Smith and Spratt, 1986). This system has been successfully used to map the cytoplasmic and periplasmic domains of several bacterial cytoplasmic membrane proteins (Broome-Smith and Spratt, 1986; Edelman et al., 1987; Bowler and Spratt, 1989). Since only translocated forms of Bla confer Ap a upon a single host cell, it should be possible to assess whether any foreign membrane protein can insert into the E. coli cytoplasmic membrane simply by fusing BIaM to a portion of it that is normally extracytoplasmic, and determining whether the resultant fusion protein confers Ap R on individual bacterial cells. Where insertion can be demonstrated, a comprehensive series of BlaM fusion proteins can be generated and characterised so that the transmembrane organisation of the protein can be determined. The aim of the present study was to develop a system for directly synthesisipg eukaryotic membrane proteins in E. coli and to assess their ability to insert into the bacterial cytoplasmic membrane using a BIaM fusion approach.
RESULTS AND DISCUSSION (a) Rationale for using BlaM as an indicator of membrane insertion Only translocated forms of Bla confer Ap a on individual cells. Therefore, if the BlaM-coding region (blaM) is fused in-frame to a gene that is directly expressed in E. coli and the resultant fusion protein confers Ap a on individual cells, this indicates that the truncated portion of the target protein directs the translocation of BlaM across the bacterial cytoplasmic membrane and, by inference, that the protein itself
inserts into the bacterial membrane. This test can therefore be used to assess whether any protein made in E. coli is capable of insertion into the bacterial cytoplasmic membrane. Moreover, it can, if desired, be coupled to a direct selection for Ap a transformants. Only cells producing translocated forms of Bla survive and form colonies on agar containing Ap (all transformants producing cytoplasmic forms of Bla, as well as all non-Bla producers, are lysed and thus eliminated in this selection): hence this approach can provide a rapid indication of E. coil cytoplasmic membrane insertion. Plasmid pYZ4 (Fig. la) was therefore constructed as a multipurpose plasmid vector that would facilitate the direct production of eukaryotic membrane proteins in E. coli and the generation of blaM fusions to the expressed membrane protein genes, pYZ4 is a Km a derivative of pBR322 that contains the phage fl ori for ss DNA replication and encodes a modified form of the lacZ~-peptide under the control of the lacUV5 promoter. The modified 0t-peptidecoding region has an Ncol site flanking its start codon and an array of unique cloning sites situated just downstream from the translation start. Since the consensus sequence for translation initiation in higher eukaryotes, GCCACCA TGG, includes the NcoI recognition sequence, the start codons of many eukaryotic coding regions are flanked by naturally occurring NcoI sites (reviewed by Kozak, 1987). Therefore, their cDNAs can be readily subcloned between the NcoI site and any of the 'other restriction sites in the MCS of pYZ4 to yield a derivative which directs bacterial synthesis of the unfused eukaryotic gene product, under the control of the lacUV5 promoter. The use of this inducible promoter, which exhibits only a very low basal level of transcription, should permit the successful subcloning of eukaryotic membrane protein genes even if their overexpression is lethal to E. coli cells. Since subcloning results in the insertional inactivation of the 0c-peptide coding region, by using an appropriate 0¢-complementation host strain ofE. coli(such as TGI; see legend to Fig. 1) transformants containing recombinant plasmids can be distinguished from those containing pYZ4 by their white rather than blue color on agar containing IPTG and XGal. For BlaM fusion analysis pYZ4 is used in conjunction with the BlaM cassette plasmid pYZ5 (Y.Z. and J.K.B.-S., unpublished; Fig. lb). When target genes have been sub~Ioned into pYZ4, so that they are bacterially expressed under lacUV5 promoter control, blaM fusions can be made by ligating the 5' (blunt) PvuII end of the BlaM cassette to specific sites within the target gene, and the 3' end, generated by cutting within the MCS of pYZ5, to the cognate site 3' to the eukaryotic coding region in the pYZ4 derivative. The arrangement of sites in the MCS ofpYZ4 is ideally suited for the unidirectional removal of C-terminal portions of inserted coding regions via exonuclease III/S 1 nuclease
53 E ,
.,_ Bc I~l(ef) i:r~lBT/ -I I II f
Pv
Nc E [B I IhBc
E(ef) I
E(ef) I
Ikll ( e f t / BamH( / /
.
~c~/c "¢ V /,Y.
.BH Xb K Sp ~ Hae(ef) "'
'Z
It-
y yN~,BH Xb K Sp
(e)
BHXbKSp ~EcvSsSm
Pv
(b) Fig. 1. Plasmid vectors for the direct expression and BiaM fusion analysis of eukaryotic membrane proteins in E. coll. (a) Construction ofpYZ4. Only the relevant restriction sites are shown. The E. coli strain TGI (Alac-pro, thi, supE, hsdS [F' , proA +B +, lacI° Z AM l 5, traD36] was used for transformations and was grown in L broth or on L agar (Miller, 1972). The plasmid pJBS633, designed for the BlaM fusion analysis of bacterial membrane proteins in E. coii (Broome-Smith and Spratt, 1986), was
digestion (Henikoff, 1984), and thus facilitates the generation of a series of blaM fusions to random positions throughout the gene (see section c). Once fusions are obtained, ss DNA of each fusion derivative can be made and dideoxy sequencing can be primed from within the start of the BlaM-coding region and across the fusion junction (Figs. la and 2). (b) Direct expression of the pNKA in Escherichia coil, and a BlaM fusion test for insertion into the bacterial cytoplasmic membrane The Na,K-ATPase establishes and maintains the Na ÷ and K + gradients across the plasma membrane of animal
digested with AhalIl + EcoRl, and the Tc R and blaM regions were replaced with the EcoRI-HindIII fragment of pAZe3ss, which contains a Shine-Dalgarno sequence (AGGAGG) and appropriately spaced start codon flanked by an Ncol site (C/CATGG) (ZabaUos et al., 1987). An end-filled 104-bp EcoRI fragment carrying the lacUV5 promoter was ligated to pYZ2 DNA, that had been digested with EcoRI and end-filled, to yield the plasmid pYZ3 in which the iacUV5 promoter is flanked by XmnI sites. Finally the BamHI-Haell fragment of M13tgl31, that encodes all except the first few aa of a modified LacZ0t-peptide (Kieny et al., 1983), was ligated to pYZ3 DNA digested with BclI (and end-fiUed) and BamHI, to yield the plasmid pYZ4. The phage fl o;i for ss DNA replication in pYZ4 is oriented such that template DNA can be primed for dideoxy sequencing in a counter-clockwise direction. (b) Restriction map of pYZ5. pYZ5 is a Tc R derivative of pBR322 that contains the unexpressed BlaM coding region from pJBS633, followed by a MCS derived from M 13tg 131. The PvulI-EcoRI blaM cassette is shown boxed. The sequence across the PvulI site (CAG/CTG) and the start of BIaM is: ... CAG CTG CGT CAC CCA GAA ... LeuArgHis ProGlu... +! +2 +3 Abbreviations: A, Ahalll, B, BamHI; Be, Bell; E, EcoRl; Ev, EcoRV; H, HindllI; Hae, Haell; K, KpnI; Nc, Ncol; P, Pstl; Pv, PvulI; S, Sail; Sm, Sinai; Sp, SphI; Ss, Sstl; X, Xmnl; Xb, XbaI; (el'), end-filled; p, lacUV5 promoter; rbs, ribosome-binding site; z, LacZu-peptide coding region. 'z, incomplete LacZu-peptide-coding region; Ap', unexpressed BlaM-coding region.
cells. It consists of a catalytic at-subunit (that probably spans the plasma membrane six times or more) and a p-subunit that is essential for activity of the holoenzyme. The pNKA is a 302-aa glycoprotein. Inspection of the aa sequence reveals only one hydrophobic segment that is long enough to form an 0t-helical membrane-spanning segment (aa 34-60), and this is preceded by a short, highly charged, net basic, domain and followed by a large C-terminal domain containing three potential glycosylation sites (Shull et al., 1986). Biochemical and genetic analyses of the structure and assembly of two homologous mammalian p-subunit polypeptides have since confirmed that the/~-subunit is a class-II membrane protein (van Heijne, 1988), consist-
54 Nc
'LB HxbKSp EEvSsSm BHXb~mSP
Nc
Pstl( e ~ P
[I-//~"@ pYZ4 '] Ncol Hindll, (el) ' ~ '(~ , It ;-. / .t-Ncol\1~ "i'- *" 07/ fragment Ahalll l ~ l i
pYZ,I "l"qlil I~:Ebvl~Slm _A } ~ _
Xbil
(,f)
EcoFtl
blaPI cassel:l:e N ~ ' ~ / P .,., , /"
I1,
pYZiilblaM / I
'""'
il
Fig. 2. The expression of the flNKA in E. coil, and construction offlNKA : : BlaM fusions. See sections b and e, for details. Only the relevant restriction sites are shown, blaM was fused in-frame after codon 147 of the flNKA gene by ligating the PvulI-geoRl BIaM cassette of pYZ5 to pYZI 1 DNA that had been digested to completion with PstI, end-trimmed with S 1 nuclease to.xemove the resultant Y-overhanging ends, and then digested with £coRI. The BIaM portion offlNKA : : BIaM fusion proteins is shown by thick arrows. The dashed line represents positions at which blaM could be fused within the C-terminally resected flNKA-coding region. Plasmid ss DNA was obtained from transformants that appeared to produce in-frame fusion proteins by :.nfection with variant R408 of phage fl (Russel et al., 1986), and the nt sequence across the fusion junction was determined by dideoxy sequencing using a primer (5'-dCTCGTGCACCCAACTGA) that is complementary to codons 14-18 of blaM. Abbreviations as Fig. 1, and (et), end-trimmed.
ing of a short N-terminal cytoplasmic domain, a single hydrophobic transmembrane segment that functions both to translocate and anchor the bulk of the protein to the membrane, and a large C-terminal extracellular domain (Ovchinnik6v et al., 1987; Kawakami and Nagano, 1988). A plasmid-cloned eDNA for the flNKA was obtained from Drs. G.E. Shull and J.B. Lingrel (of the University of Cincinnati, U.S.A.) and the/~NKA-coding region was subcloned into the vector pYZ4 as an NcoI-AhalII fragment (Fig. 2). The resultant plasmid, pYZ 11, directs the synthesis of the flNKA under iacUV5 promoter control. Production of this protein has no o,bvious deleterious effects on the E. coil lacl Q strain TG1. Moreover, pYZ 11 can be introduced into, and maintained in, a derivative of JM 101 cured of its F' (JMI01 F - ; see legend to Fig. 3.) that does not produce any lac represser and hence constitutively synthesises the fi-subunit polypeptide. To determine whether the flNKA inserts into the E. coli cytoplasmic membrane, BIaM was fused to a position (aa 147) in the protein that is normally exposed on the extracellular side of the plasma membrane (yielding the plasmid pYZ 11-147/blaM; Fig. 2). Well-separated cells of
TGl[pYZII-147/blaM] were inoculated onto agar containing a range of Ap concentrations. The MICs of Ap were 1600 pg/ml and 25 #g/ml in the presence or absence of 1 mM IPTG, respectively. Since induced expression of the fusion protein confers Ap R on individual cells (plasmidfree cells, or cells producing cytoplasmic forms of Bla, are killed by 5 #g Ap/ml), we conclude that the truncated form of this eukaryotic membrane protein directs the translocation of BIaM to the periplasm and in turn we infer that the flNKA inserts into the E. coli cytoplasmic membrane. The above results also demonstrate that the lacUV5 promoter of pYZ4 is tightly controlled, since it yielded only a low basal level, and a substantial (and approx. 60-fold greater) induced level, of Ap R. (c) BlaM fusion analysis of the organisation of the pNKA in the Escherichia cell cytoplasmic membrane Fusions of blaM to random positions throughout the flNKA-coding region were constructed by digesting pYZ 11 with XbaI + SstI and then resecting unidirectionally into the gene with exonuclease III followed by S1 nuclease (Fig. 2). The resected ends were made blunt by end-filling
55
