Dec 3, 1993 - Jean-Michel Betton and Maurice Hofnung. Unite de ...... Arkowitz,R.A., Joly,J.C. and Wickner,W. (1993) EMBO J., 12, 243-253. Bassford,P.J. ...
The EMBO Journal vol. 1 3 no. 5 pp. 1 226 - 1234, 1994
In vivo assembly of active maltose binding protein from independently exported protein fragments
Jean-Michel Betton and Maurice Hofnung Unite de Programmation Moldculaire et de Toxicologie Gdndtique (CNRS-UA1444), Ddpartement des Biotechnologies, Institut Pasteur, 25 rue du Docteur Roux, 75015 Paris, France Communicated by M.Hofnung
The maltose binding protein (MBP or MalE) of Escherichia coli is the periplasmic component of the transport system for malto-oligosaccharides. It is synthesized in the cytoplasm with an N-terminal signal peptide that is cleaved upon export. We examined whether active MBP could assemble into an active protein in bacteria, from N- and COOH-terminal complementary protein fragments encoded by distinct, engineered segments of its structural gene. We found export and functional periplasmic assembly of MBP fragments, despite the complex polypeptide chain topology of this protein, if two conditions were satisfied. First, each of the two fragments must carry a signal peptide. Second, the boundaries between the two fragments must correspond to a permissive site within the protein. Functional assembly of active MBP occurred in five cases where these conditions were met: sites after residues 133, 161, 206, 285 and 303; but not in three other cases where the break junction corresponded to a non-permissive site: after residues 31, 120 and 339. Thus, permissive sites which were initially characterized because they could accept extensive genetic insertion/deletion modifications without loss of most biological properties provide a means of defining complementing protein fragments. This observation opens a way to study genetically the relationships between protein export and folding into the periplasm. Key words: complementation/export/maltose binding protein/ protein fragment
Introduction Protein translocation across membranes is a basic cellular all organisms from bacteria to humans. In Escherichia coli, the export of proteins to the periplasm follows the general secretion pathway (GSP; Pugsley, 1993). Considerable progress has been made in the last 10 years in identifying the components of this pathway (Schatz and Beckwith, 1990). However, a number of molecular mechanisms governing the translocation process are still obscure. An important and timely problem in bacterial export is the analysis of the exact relationships between folding and export of proteins (Wickner et al., 1991). In the present paper, we extend for the first time an approach which has been fruitful in studying protein folding and assembly in vitro (Wetlaufer, 1981; Taniuchi et al., 1986) to the case of a protein exported to the periplasm. process common to
1226
Maltose binding protein (MBP or MalE), the malE gene product, serves as the periplasmic receptor for high-affinity membrane transport of maltose and maltodextrins in E. coli (reviewed in Ames, 1986). Because of its role in the maltose
transport system, export of MBP into the periplasm is
essential for cells to utilize maltose as a carbon source. This feature facilitated the development of powerful genetic selections which led to the discovery of several genes involved in secretion: the sec genes (Schatz and Beckwith, 1990). MBP has also been extensively used as a model of protein translocation across the cytoplasmic membrane [see Randall et al. (1987) and Bassford (1990) for reviews]. MBP is synthesized as a precursor, preMBP, with an N-terminal signal sequence that is cleaved off during or shortly after translocation across the cytoplasmic membrane. Translocation requires that the nascent preMBP polypeptide chain reaches a critical molecular weight corresponding to 80% of the final length (Randall, 1983), and that the precursor exists in an export-competent conformation representing a partially unfolded state (Randall and Hardy, 1986). Both the signal peptide (Park et al., 1988) and the binding of SecB, the molecular chaperone involved in protein secretion (Kumamoto, 1991), participate in the maintenance of this initial conformation. Considerable information has accumulated about the translocation machinery of E. coli (Wickner et al., 1991; Pugsley, 1993), but little is known about the conformational state of the precursor polypeptide chain prior to translocation, during its movement through the membrane, or about its processing by leader peptidase on the periplasmic surface of the cytoplasmic membrane. Although Randall and Hardy (1986) showed that MBP folding and export are kinetically competing processes, we know neither the degree of secondary and tertiary structures of precursor proteins, nor the mechanistic constraints that the export apparatus imposes upon translocating polypeptide chains. Another important unsolved problem concerns the final steps in translocation: the release and folding steps. Do specific proteins catalyze or facilitate this step? It has been proposed that SecF and SecD could act at this step, but no mechanism has been hypothesized (Schatz and Beckwith,
1990).
A classical approach to the understanding of protein folding is to use fragments of protein to study the nature of intermediates in the folding process. For example,
evidence exists that certain isolated fragments behave as
independent folding units (Wetlaufer, 1981). Other protein fragments have been combined to delineate how the different regions of a polypeptide interact to stabilize the secondary and tertiary structures of the native conformation. In vitro complementation of various combinations of overlapping fragments of staphylococcal nuclease and cytochrome c are
the most documented models of this approach (Taniuchi et al., 1986). Many such experiments on fragment complementation have been performed either with fragments produced by limited proteolysis or chemical cleavage, or with © Oxford University Press
Ka. . ~
Complementation and permissive sites of MBP
incomplete polypeptide chains expressed independently by genetic manipulations (Skerra and Pliickthun, 1988; Minard et al., 1989; Matsuyama et al., 1990; Herold et al., 1991). The in vivo assembly of functional proteins from complementing fragments has been demonstrated for several soluble cytoplasmic proteins: isoleucyl tRNA synthetase (Shiba and Schimmel, 1992), alanine racemase (Toyama et al., 1991) and aspartate transcarbamoylase (Yang and Schachman, 1993); and similarly for membrane proteins: lactose permease (Bibi and Kaback, 1990; Wrubel et al., 1990), FhuB protein of the iron hydroxamate transport system (K6ster and Braun, 1990) and muscarinic receptors (Maggio et al., 1993). Most of these studies elucidated the assembly of subdomains or domains within proteins. Here, we used this approach to investigate the folding and assembly properties of fragments of a protein following their complete translocation through a membrane. We examined whether in vivo fragments of MBP could be exported, find one another and assemble in the periplasm into an active conformation. We previously described a set of MBP mutant proteins generated by random insertion of an oligonucleotide linker into malE (Duplay et al., 1987). This study identified several regions of the protein that are involved in export, stability, maltose binding and maltose transport. Yet, a most interesting finding was the identification of 'permissive sites' within MBP: mutations that correspond to insertion or deletion/insertion of several amino acids without loss of export and maltose binding properties (Hofnung et al., 1988). Thus, permissive sites appeared to define segments EcoRI
of polypeptide chain that have little or no role in the formation of the native structure of MBP. In the present work, we genetically engineered some of these sites at which the linker insertion occurred to produce the corresponding pairs of complementary protein fragments, without overlapping sequences. The resulting N-terminal and Cterminal fragments associated together non-covalently in vivo to form active complexes in the periplasm of E. coli, but only when the break point corresponded to a permissive site.
Results
Eight sites, including five permissive sites, were genetically engineered to produce in vivo the corresponding pairs of complementary protein fragments.
Construction of plasmids The strategy of plasmid construction was to use the unique BamHI sites previously introduced into the malE gene (Duplay et al., 1987) to generate several pairs of protein fragments. Initially, we addressed the possibility that the two chains may fold and assemble into an export-competent conformation with a single signal peptide. For this purpose, an oligonucleotide carrying a TAA stop codon, a ribosome binding site and an ATG start codon, flanked by BamHI cohesive ends, was inserted into the selected sites of gene malE (Figure lA). To conserve the malE translational reading frame, we synthesized two double-stranded oligonucleotides corresponding to the two phases of BamHI. The translation of the 5' regions encoding N-fragments,
promlotor
pPD plasmids
GG ATC C '-CC TAG G
N/a
pPD
.z4kb
:
:
Bam HI
pPD65 (MN1alE3 )I)
pPD183 lMaIE339) pPD I ( MalE 12(h BgI II site
-
1>3- TAA
Oligo Comp lphasc 2 +X
pPD346I M1aIE2()6 ) pPD222 Ma1E285) pPD364 (MaIE3 1)
pPDI178 (MaIE133
G GAT CC -t.C CTA GG.
--SA GAT CT-
A
pPD341 eMaIEI61)
_AA GGA
-.
ATG
:.
.7
.-
.x_
>
plmSSidC
plasmids
IIindllIIl
ft?pl I
prnor ra
B
IIf?Ial
tac
2501 hp PCI fragmnclt
L
PI-Ornll(tklJ
RBS. start codon, and sigTnal secquencc of malE
-1
_I
Hind ilI
._I_=__________________________________ .... ... . ..
...... ...
_
p
j
_ ppSSSSC
__~~~~~~~~~~~~~~~paniidsl
Fig. 1. Construction of plasmids pSSNC and plasmids pSSNSSC. The starting plasmids belong to a collection of plasmids (pPDs) which carry a single BamHI site, previously constructed by random linker insertion, into the malE gene (Duplay et al., 1987). (A) pSSNC plasmids express Nfragments with signal peptide and C-fragments without signal peptide, both under the control of the pmal promoter. Double-stranded oligonucleotides (Comp) in two different coding phases, 1 and 2, containing a stop codon, a ribosome binding site (RBS) and a start codon were inserted in the unique BamHI site of pPD plasmids with the corresponding reading frame. For clarity, only the sequence of oligonucleotide Comp in phase 2 is given. The natural BglII site of pPDI was also used for these constructions. (B) pSSNSSC plasmids express both protein fragments (N and C) with signal peptides. An expression cassette constructed by PCR from pMAL-p vector DNA (Biolabs) was cloned between the HpaI and HindHI sites of pSSNCs. This led to the periplasmic expression of C-fragments under the control of the tac promoter. 1227
N =} >- |(
J.-M.Betton and M.Hofnung
under the control of the maltose promoter, was therefore interrupted, and the C-fragments were expressed without a signal sequence. The oligonucleotides were designed to accept in-frame insertion of a signal sequence encoding DNA fragment in the HindIII site. This site was chosen because: (i) it was absent in pPD1, the parent plasmid, and thus facilitated the subsequent step of insertion in plasmid construction; and (ii) the reading frame, used in both Comp oligonucleotides (phase 1 and 2), encodes a Lys-Ile sequence which is similar to the mature N-terminus of wild-type MBP (Lys-Leu). In the next step, a DNA fragment containing the tac promoter and encoding the complete signal sequence of MBP was inserted between the HpaI and Hindm sites (Figure 1B). The resulting constructs, called pSSNSSC, produce both the N- and C-fragments with a signal sequence, under two different promoters. Insertions in Hindm were in-frame with the previously added start codon (see above). The HindIll coding sequence, which fits well with the minimum substrate sequence recognized by signal peptidase I of E.coli (Dev et al., 1990), conserved the same N-terminal sequence on the C-fragments. Permissive sites in MBP provided an opportunity to determine whether reconstitution is possible when breaks are introduced at these regions. Among the 10 linker insertion/deletion mutants which have little or no effect on the maltose binding activity of MBP (Duplay et al., 1987),
malEL N3T
te,.x
,*i
.
[3B (XIIII DB
~
chose five mutations for their broad distribution over the
303. Three additional sites that did not correspond to
permissive sites were studied: after residues 31, 120 and 339. MalE31 corresponds to a double amino acid substitution which was previously identified as an abnormally exported protein (Duplay and Hofnung, 1988). Site 120 corresponds to the amino acid residue position of the natural Bgll restriction site in the malE gene. MalE339 carries a large deletion of 15 amino acid residues, including Trp340, which belongs to the maltose binding site (Spurlino et al., 1991). Figure 2A summarizes all the constructions. The fragment pairs are designated N31C, N120C, N133C, etc. The numbers corresponding to the N-terminal flanking positions of the originally inserted BamHI linker (Duplay et al., 1987; Betton et al., 1993), indicate the new termini of protein fragments. The letters N and C denote the natural N- and C-termini, respectively. When a gene fragment encoding a signal sequence was present, the corresponding fragment was designated by 'SS' (Figure 2B). In most cases, the original linker insertions were accompanied by small deletions corresponding to the loss of five or six residues on average (Duplay et al., 1987). Except for the two recombinants constructed from pSSNSSCs (pSSN120-SS31C and pSSN339-SS303C), the largest overlapping sequence among these pairs included five residues (N285C) and the largest deletion nine residues (N161C).
120) 133 161285" 303) 206' 339 I [ll l-f4 itl.IVl Ot.lI /tl1 II I\ U 13.1 >VtIT1K3 ii. dx1 't\ I3(X 3 M\it.x.\H> [J1\3
31
A
we
polypeptide chain: sites after residues 133, 161, 206, 285 and
3 'iIlp ,d11i:fii :
3
x
:3
6
.V
1.
u
II[K
u
-a-1= 7
-
(Lj
=@-1 _--_
U-
. . . . . . . . . . . . . ._
N12( ni-l('
N12(tf
N 203 (
: :
:. :: [:::
_______-
N2I1(W(' L.
N!
-
M::.
E(
N.1B)
.--..-.... ----
... ..
::
:.
::,
': .::
:::
:::::.
..
:.-
:... .:::
::.
:::
::.
::
..
No ..ME
.:
.::..
''
:,
.;1).3('. I ....~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
B
ss
N fragment ....
Co IYagrnlnCt +LZ_Z]
p SSNSC' p Ilas m id.k
ss N Iraa-mient
_
s.s C' 1iraimeni j : ST
pSSNSSC plasni1ds Fig. 2. Schematic representation of the fragment pairs of MBP. (A) The MBP and derived protein fragments are representated in a linear fashion without the corresponding signal peptides. The secondary structure elements and the three peptide segments connecting the two domains (black boxes) are shown at the top. Numbers refer to residue positions in the mature MPB after which the discontinuity of the polypeptide chain was introduced by mutagenesis. Asterisks indicate the original permissive site. Vertical bars correspond to break points with C-terminal addition of 2-5 residues on the N-fragments. Stippled rectangles are proportional to the size of the deletions (1-12 residues) introduced by the linker insertion mutagenesis. Residues or groups of residues involved in maltose binding are indicated as black dots. (B) Schematic representation of the protein fragments explaining the plasmid nomenclature.
1228
Complementation and permissive sites of MBP
A solubilized whole cells
solubilized whole cells
A
du u
u U}
(1-
2 c;
z V=tjz C4 C
-
-
-C
z
z
C; CA
Z z
Z Z z U7
V)
&n
U:
K 9
