In vivo and in vitro characterization of ... - Wiley Online Library

Eur. J. Biochem. 207, 687-695 (1992) $> FEBS 1992

In vivo and in vitro characterization of overproduced colicin E9 immunity protein Russell WALLIS ', Ann REILLY I , Arthur ROWE3, Geoffrey R. MOORE', Richard JAMES' and Colin KLEANTHOUS'

'

Molecular and Microbiology Sector, School of Biological Sciences and School of Chemical Sciences, University of East Anglia, Norwich, UK National Centre for Macromolecular Hydrodynamics, Department of Biochemistry, Adrian Building, University of Leicester, UK

(Received March 2, 1992)

-

EJB 92 0282

We report the overproduction of the immunity protein for the DNase colicin E9 and its characterization both in vivo and in vitro. The genes for colicin immunity proteins are normally co-expressed from Coi plasmids with their corresponding colicins. In the context of the enzymatic colicins, the two proteins form a complex, thereby protecting the host bacterium from the antibiotic activity of the colicin. This complex is then released into the medium, whereupon the colicin alone translocates (through the appropriate receptor) into sensitive bacterial strains, resulting in bacterial cell death. The immunity protein for colicin E9 (Im9) has been overproduced in a bacterial host in the absence of its colicin, to enable sufficient material to be isolated for structural studies. As a prelude to such studies, the in-vivo and in-vitro properties of overproduced Im9 were analysed. Electrospray mass spectrometry verified the molecular mass of the purified protein and analytical ultracentrifugation indicated that the native protein approximates a symmetric monomer. Fluorescense-enhancement and gel-filtration experiments show that purified Im9 binds to colicin E9 in a 1: 1 molar ratio and that this binding neutralizes the DNase activity of the colicin. These results lay the foundations for a full biophysical and structural characterization of the colicin E9 DNase inhibitor protein, Im9.

Colicins are a group of plasmid-encoded bacterial proteins which have antibiotic activity and are secreted as part of the SOS stress response of the producing organism [l -41. Three specific functions are generally exhibited by colicin proteins, each associated with a specific domain [5 - 71:Translocation into a foreign cell, associated with the N-terminus of the protein; recognition of a specific extracellular receptor of the target cell prior to import (associated with the central portion of the primary sequence), the most widely studied being the E-group colicins which enter cells via the vitamin B12 receptor (encoded by the btuB gene) [8]; bacteriocidal activity, which is encoded by the extreme C-terminus. Four cytotoxic classes of colicin have thus far been identified; the pore-forming colicins, such as ColEl (and ColA) (col, colicin) which kill cells by causing membrane depolarization (reviewed in [91 I]); RNase colicins, such as ColE3, which specifically cleaves 16s ribosomal RNA [12, 131; DNase colicins such as ColE2, ColE7, ColE8 and ColE9, which are non-specific endonucledses [14 - 161; inhibitors of cell-wall synthesis such as colicin M [17]. Colicins within a class, such as the E group ( M , of 61 000), show substantial sequence similarity across the first two thirds of their primary sequence (which encode the common functions of translocation and receptor recognition), but do not show homology in their C-terminii, unless they encode the same type of bacteriocidal activity [18 -211. Of immediate concern to a colicin producing strain of Escherichia coli is how to avoid committing suicide. This is Correspondence to C. Kleanthous, Molecular and Microbiology Sector, School of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, UK Ahhreviations. Im9, immunity protein for colicin E9; Col, colicin; ColEY complex, heterodimeric complex containing both colicin EY and immunity 9 proteins.

achieved by the coordinate synthesis of an immunity protein ( M , = 9500) which, in the case of the enzymatic colicins, binds to the extreme C-terminus of the protein and, by a littleunderstood process, neutralizes the bacteriocidal activity within the host [22-261. Moreover, within this class of bacteriocin, the immunity protein is secreted in complex with the colicin (a process which involves another protein known as the lysis protein [27]), but itself does not enter bacterial cells. Colicin-immunity specificity is sufficiently discerning that, for example, the immunity protein for ColE8 does not protect a cell from the action of ColE9 [28], even though the sequences for the two immunity proteins are almost 60% identical [29]. Recently, Curtis and James 1301have shown, by mutagenesis, that some of the major specificity differences between ColE8 and ColE9 colicins reside in six residues at the C-terminus. Little is known of the protein-protein interactions between a colicin and its immunity protein, even though the bacteriocins have been the subject of numerous genetic and biochemical studies spanning more than two decades. To date, structural information is only available for the C-terminal domains of the pore-forming colicins, ColA (solved by crystallography) [31] and ColEl (for which the secondary structure elements have been determined by NMR) [32]. As for the enzymatic colicins (ColE2 E9), no structures are available for the isolated proteins or the heterodimer complexes, although crystals of the immunity protein for ColE3 [33] and the ColE3: immunity 3 complex [34] have been reported. We are interested in the protein-protein interaction between the DNase ColE9 and its immunity protein (Im9) and the structures of the two proteins in the unbound form. Due to its small size ( M , of 9583), the Im9 protein is an obvious starting point for solution NMR experiments aimed at de~

688 terinining its three-dimensional structure in the absence of its colicin. Such an investigation requires large quantities of protein. Traditionally, immunity proteins have been isolated following induction of colicin-producing bacterial strains containing the appropriate Col plasmid, by the DNA-damaging agent mitomycin C [22, 23, 271. However, these procedures require large batch cultures (100-200 1) and the yield of the immunity protein is poor. In this paper, we describe how we have overcome these difficulties by selectively overexpressing the gene for the Im9 protein in the absence of its colicin, in a bacterial expression vector and purified the protein to homogeneity in just three steps. This is the first report of an immunity protein that has been expressed and purified in this way. The overproduced Im9 protein completely protects a normally sensitive bacterial cell against the action of exogenously added ColE9 and Im9 heterodimeric complex (ColE9 complex). Im9 was purified and its molecular mass analysed, both by mass spectrometry and analytical ultracentrifugation, and it was shown to be a spherical monomer corresponding in molecular mass to the published sequence. The ability of the purified Im9 protein to interact with unbound colicin E9 in vitro was investigated by fluorescence spectroscopy and gel-filtration chromatography and by its ability to inhibit the DNase activity of ColE9. The results of these experiments show that the overproduced Im9 protein binds to its colicin stoichiometrically. These studies further demonstrate that the selectively overproduced Im9 protein confers immunity towards the action of ColE9 both in vivo and in vitro and is eminently suitable for further structural studies. MATERIALS AND METHODS Enzymes and reagents

Restriction enzymes and T4 DNA ligase were obtained from Gibco-BRL, Pharmacia or Northumbria BioLabs. T4 DNA polymerase was purchased from Boehringer Mannheim. Guanidine hydrochloride was obtained from Gibco-BRL, anion and cation exchangers from Whatman, Sephacryl-S100 resin and Superdex-75 column from Pharmacia and miniconcentrators (Centricon and Centriprep) from Amicon. All other reagents were supplied by Sigma Chemical Company, BDH or Aldrich Chemical Company Ltd.

for mutagenesis was an M13mp18 derivative containing a 300-bp insert, carrying the E9 immunity gene (imm9) cloned into the polylinker region. The mutagenic primer was designed to introduce a unique NcoI site containing the ATG start codon of the imm9 gene. NcoI site 5' GAGGTAAGTAGCATGGAACTG 3'. A A A A start site The mutagenic bases are underlined. Following mutagenesis, the 290-bp NcoI - HindIII fragment was cloned into the expression vector pKK233-2 and digested with NcoI and HindIII, to create pRW60. Plasmid restriction,ligation, transformation and electrophoresis

Digestion of DNA with restriction endonucleases, electrophoresis of restriction fragments, ligation of DNA fragments and transformation of competent cells were carried out as described by Maniatis et al. [37]. Protein determinations

Protein concentrations were determined by amino acid analysis and Bradford protein assays which had been previously calibrated by amino acid analysis. Analytical ultracentrifugation

Sedimentation analysis Molecular-mass analysis was performed by means of short-column sedimentation equilibrium analysis [38] using an MSE Mk I1 Analytical Ultracentrifuge run at 24890 rpm and a temperature of 19.5 "C. High-precision optical schlieren traces were recorded [39] to ensure optimal conditions for analysis at the high solute concentrations used (10 mgimi). The enlarged traces were analysed by means of a digitising tablet attached to a microcomputer, using local software. Partial spec@ volume The partial specific volume was computed from the amino acid composition [16], assuming the validity of Traube's additivity rule [38]. A value of 0.723 ml/g was obtained.

Bacterial strains and media E. coli JM83 hsdR, a restriction-deficient derivative of E. coli JM83 (Ara', Lac', Pro, Thi, rpsL, @ODlncZMlS) was used as the host strain for the plasmids. Cultures containing recombinant plasmids were grown in Luria-Bertani broth or on plates of Lurk-Bertani agar, supplemented where necessary with ampicillin (100 pg/ml). Plasmids

M13mp18 has been described previously [35] as has the plasmid pKC67, which contains the HincII fragment (originally from the ColE9-J plasmid) encoding the ColE9 operon [36]. The expression vector pKK233-2 was purchased from Pharmacia. Site directed mutagenesis and construction of pRW60

Site-directed mutagenesis was performed using a method adapted in this laboratory [30]. The single-stranded template

D i f f ~ i o ncoeffiient from intensit4if~uctuationspectroscopy Translational diffusion coefficients were estimated using an OROS M801 detector. Samples were injected into the cell and at least 12 successive estimated values logged after the gross photon count had been checked for stability. The values obtained were corrected for solvent viscosity to water at 20°C. Biological assay of colicin E9 activity

Colicin activity, determined both during the purification of the ColE9 complex and in testing strains for the degree of immunity conferred by Im9 expression, was examined quantitatively by the spot test described by Reeves [40]. This test involves pipetting drops of ColE9 complex onto a LuriaBertani agar plate overlayed with a lawn of JMX3 (approximately lo7 cells) in molten agar (0.7%), followed by growth for 15 h at 37°C. Colicin activity is indentified by the appearance of clear zones (indicating cell death), and the level of this

689

1

2 3 4

5 6 7 8

kDa

+66 -45

-36

-29 -24

-20.1

-14.2 -Dye Front

Fig. 1. Expression of the imm9 gene and purification of the In19 protein. SDS/pol yacrylamide gel electrophoresis (16%) showing expression and purification of Im9 from JM83. Lanes 1 and 8 contain marker proteins. Lanes 2 - 7 each contain 2 pg protein (from Bradford-protein-assay determinations). Lane 2, purified ColE9 complex indicating the migration position of Im9; lane 3, extract of JM83 containing the plasmid pKK233-2; lane 4, extract of JM83 containing the plasmid pKK233-2 and the 1mm9 gene encoding the Im9 protein (pRW60); lane 5 , 50-95% ammonium sulphate fraction; lane 6, post-anionexchange chromatography; lane 7, post-gel filtration chromatography.

activity (the titre) is taken from the dilution factor required to abolish the appearance of these zones.

cation, ColE9-containing fractions were identified by the biological assay described above and by SDS/PAGE. Complex was prepared from a shaking 1-1 culture of JM83 cells transformed with the plasmid pKC67. At an A 5 5 0of 0.2, the culture was induced with mitomycin C (0.2 pg/ml) and incubated with shaking for a further 5 h. Cells were removed by centrifugation and the secreted colicin precipitated with ammonium sulphate (65% saturation). The precipitate was resuspended in, then dialysed against, potassium phosphate (0.05 M, pH 7.0,4"C) containing EDTA (1 mM) and 2-mercaptoethanol (1 mM). The extract was then applied to a DE-52 anion exchange column (50 ml), equilibrated in the same buffer. Colicin E9 complex does not bind to this column (the complex elutes in the unbound fraction) but many contaminating proteins are retained, resulting in purification of the ColE9 complex. The extract was then dialysed against potassium phosphate/sodium citrate (10 mM, pH 5.5,4"C), containing EDTA (1 mM) and 2-mercaptoethanol (1 mM) and loaded onto a CM-52 cation exchange column (20 ml), equilibrated in the same buffer. Following washing with the loading buffer, the ColE9 complex was eluted with a gradient of potassium chloride (0 400 mM). Pooled fractions were concentrated by lyophilization, following dialysis against distilled water, redissolved in buffer A containing potassium chloride (0.05 M) and loaded onto a Sephacryl-S100 gel-filtration column (250 ml), equilibrated in the same buffer. ColE9-complex-containing fractions were pooled, dialysed against water, lyophilized and finally resuspended in a small volume of potassium phosphate (0.05 M, pH 7.0) and stored frozen. ~

Purification of overproduced Im9 protein

Overproduced Im9 protein was purified from 5 1 cultures of E. coli JM83 cells containing pRW60, which had been grown for 16 h with vigorous aeration in Luria-Bertani broth containing ampicillin. Cells were harvested by centrifugation and resuspended in a small volume of Tris/HCl (SOmM, pH 7.5,4"C), EDTA (1 mM) and 2-mercaptoethanol(l mM) (buffer A) containing the protease inhibitor phenylmethylsulphonyl fluoride (1 mM). The cells were broken using a French press (SLM Aminco) at 900-1000 psi. Following incubation with 0.3 mg each of DNase and RNase (at 37°C for 30 min), the cell debris was removed by centrifugation. Im9-containing fractions in this and subsequent steps were identified by SDSjPAGE (16% polyacrylamide) using either a Hoefer Mighty small gel apparatus (as in Fig. 1) or a Pharmacia Phast system (20% polyacrylamide gels). The Im9 protein was precipitated by ammonium sulphate fractionation (50-95%), then dialysed against buffer A containing 50 mM potassium chloride. The extract was then applied to a DE52 anion-exchange column (50 ml), equilibrated in the same buffer at 4°C. The column was washed extensively in loading buffer prior to being eluted with a gradient of potassium chloride (50-400 mM) in buffer A. Im9 fractions were pooled, dialysed against buffer A, concentrated using an Amicon centriprep-I0 column, then applied to a SephacrylSlOO gel-filtration column (250 ml) equilibrated in buffer A containing potassium chloride (50 mM) at 4°C. Im9 fractions were pooled, concentrated using an Amicon centriprep-10 column and dialysed against SO mM potassium phosphate (50 mM, pH 7.0) and stored frozen. Purification of ColE9 complex

A modified procedure to that used by Chak et al. [16] was used to purify the ColE9 complex. Throughout the purifi-

Preparation of free colicin

Colicin E9 protein and Im9 were separated by incubating purified complex for 2 h at 25°C in potassium phosphate (0.05 M, pH 7.0) containing dithiothreitol(1 mM) and guanidine hydrochloride (2.5 M). These conditions, which were established by fluorescence spectroscopy (see below), denature both proteins and allow separation by gel-filtration chromatography (Superdex-75 ;see Results). The two proteins (colicin is the first to elute off the column, followed by the immunity protein) were collected and refolded by dialysis against potassium phosphate (0.05 M, pH 7.0), containing dithiothreitol (1 mM). The identity of the two proteins was confirmed by SDSjPAGE and by the fact that all the cytotoxic activity was associated with the first protein peak (data not shown). Fluorescence spectroscopy; protein denaturation, renaturation and subunit association

Fluorescence spectroscopy was used to monitor the denaturation of the ColE9 complex, the renaturation of the separated subunits and the binding of the overproduced Im9 protein to the refolded colicin. All experiments were conducted with a Shimadzu RF5000 spectrofluorimeter, thermostated at 25°C and in filtered potassium phosphate (0.05 M, pH 7.0) containing dithiothreitol (1 mM) using an excitation wavelength of 295 nm to ensure only tryptophan excitation, and an emission wavelength of 334 nm. The excitation bandwidth was 5 nm, the emission bandwidth was 10 nm and 3 ml quartz cuvettes were used throughout. Denaturation of ColE9 complex by guanidine hydrochloride was investigated by incubating protein (19 pg/ml) with increasing concentrations of guanidine hydrochloride (and equilibrated for a minimum of 2 h at 25"C), followed by fluorescence measurements. Colicin E9 protein was separated from its immunity

690 protein, as described above, under denaturing conditions (2.5 M guanidine hydrochloride), then refolded by dialysis against potassium phosphate at room temperature (0.05 M, pH 7.0) containing dithiothreitol (1 mM). Aliquots of this refolded material were analysed by fluorescence (at 25°C) to ensure refolding had occured (see Results). The association of the overproduced lm9 protein to this unbound colicin was also monitored by fluorescence. Typically, increasing increments (up to a total of 3.5 -4.0 pg; 0.37 -42 nmol), of Im9 were added to unbound colicin (21 - 25 pg; 0.34- 0.40 nmol) in a total volume of 1.5 ml in potassium phosphate (0.05 M, pH 7.0) containing dithiothreitol (1 mM) at 25°C and the change in emission intensity observed at 334 nm.

sensitive strain of E. coli JM83. Fig.2 shows the result of such an experiment. JM83 containing the expression vector pKK233-2 (Fig. 2A) is very sensitive to the bacteriocidal activity of purified ColE9 complex, with ‘zones’ evident at nanogram levels of protein. This level of sensitivity is identical for JM83 without pKK233-2 (data not shown). Constitutive expression of Im9 in JM83 (Fig. 2B) results in complete resistance towards any concentration of ColE9 complex (the highest used in this experiment being 8 mgiml). Thus, expression of the gene for the Im9 protein confers full biological immunity towards the action of its target colicin, ColE9.

Subunit association observed by gel filtration

Purification of Im9 was accomplished from 5 1 cultures of E. coli JM83 containing pRW60 in rich medium, as described in Materials and Methods. The yield of pure protein from this volume of cells is, routinely, 80- 100 mg. The purification of the protein was followed by SDSjPAGE (16% polyacrylamide) using the migration of Im9 from isolated ColE9 complex as a control. Fig. 1 (lanes 4-7) shows a typical purification profile for this protein, with samples being run at each of the three steps in the purification (ammonium sulphate fractionation, anion-exchange chromatography and gel-filtration chromatography). The purity of this material was determined to be > 99% by laser densitometry (data not shown). The migration of Im9 on SDSjPAGE corresponds to a protein of M , 11000-12000, which is almost 2000 units greater than the anticipated M , deduced from the amino acid sequence ( M , 9583). To determine whether this high value was a result of post-translational modification or simply the migration properties of Im9 in an SDS gel, the monomeric molecular mass of overproduced Im9 was determined by electrospray mass spectrometry. This technique is now widely used to determine the M , of proteins [42]. The observed M , of isolated Im9 is 9582.47 T 0.39, from two independent experiments (expected M , is 9583.45). It should be emphasized that, while these results show unambiguously that the overproduced protein is not post-translationally modified in any way and corresponds almost exactly to the M , calculated from the cloned sequence [21], they do not provide any information on the oligomeric structure of the isolated protein (since the folding of the protein after injection into the instrument is unclear) [42] or its ability to interact with its native colicin.

Samples for gel filtration, in filtered potassium phosphate (0.05 M, pH 7.0, room temperature) containing dithiothreitol (1 mM), were applied to a Superdex-75 gel filtration column ( M , fractionation range of 5000-80000) in 100 p1 aliquots. All protein samples were incubated at 25°C for 30 min prior to injection. DNase activity Protein samples (15 pl) were incubated at 25°C for 30 min in Tris/HCl (0.05 M, pH 8.0) containing sodium chloride (80 mM) and magnesium sulphate (10 mM). Substrate DNA ( 5 p1 containing 150 ng linear pUC18) in the same buffer was then added to each reaction mixture and incubated a t 37°C for 1 h. The reactions were stopped with loading buffer and electrophoresed in an agarose gel (1 YO). RESULTS Overproductionof the ColE9 immunity protein (Im9) In order that the iflzrn9 gene could be cloned into a bacterial expression vector (pKK233-2; Pharmacia), a site-directed mutant was first constructed which introduced an NcoI site at the 5’ end of the gene (see Material and Methods). The gene was then cloned into pKK233-2 following digestion with NcoI and HindIII. The recombinant plasmid (pRW60) was then transformed into the E. coli strain, JM83. Constitutive overexpression of the Im9 protein was accomplished by overnight growth in rich medium. SDS/polyacrylamide gels (16% polyacrylamide) were run of crude extracts from cells containing either recombinant (pRW60) and non-recombinant plasmid (pKK233-2) to ascertain whether the Im9 protein could easily be identified and the approximate level of expression. Fig. 1 (lanes 3 and 4) shows the direct comparison of two such extracts. The Im9 protein is clearly visible as the lowest-molecular-mass species on the gel (Fig. 1, lane 4) and its migration position corresponds to Im9 from purified ColE9 complex (lane 2; see Materials and Methods for further details). Laser densitometry of such gels showed that the lm9 protein, under the growth conditions used, was approximately 10% of the total cell protein (data not shown). Expression of the ColEl immunity protein (a membrane bound immunity protein) has also been accomplished by similar procedures, but in yields which only allow identification of the protein following [35S]methioninelabelling [41]. In-vivo biological activity of individually expressed Im9

The biological efficacy of overproduced Im9 was tested by establishing the level of immunity conferred to a (normally)

Purification of overproduced Im9

Physical characterization of the Im9 protein

In order to determine the native molecular mass of purified Im9, analytical ultracentrifugation experiments were conducted. The apparent molecular mass at 10 mg/ml was estimated as 7780 Da and 7508 Da from two independent preparations of Im9. The ‘ideal’ molecular mass (i.e. at infinite dilution) for one of these preparations was estimated using the Z-T difference equation [38] to be 9400 500 Da, which does not differ significantly from the formula mass. This latter method of data analysis assumes that a single (second) virial coefficient only need be considered in accounting for the non ideality. The protein is thus clearly monomeric, even at this relatively high concentration. The non-ideal term is however unusually large; for a typical globular protein an apparent molecular mass closer to about 8500 Da would have been expected on the basis of the excluded volume term. This suggests that either the protein is less compact in conformation than a normal globular protein (i.e. it is strongly asymmetric, swollen or both) or there is a very significant charge effect to

--

69 1

Fig. 2. Overproduction of Im9 confers resistance towards the action of ColE9 complex. A lawn of E. coli JM83 containing the overexpression vector pKK233-2 (without insert) is shown in A and the same strain and plasmid containing the imm9 gene is shown in B. Onto both plates were dropped serial dilutions (lo-’ - lo-’) of purified ColE9 complex (8 mgiml). Zones, indicating cell death, were clearly apparent in the left-hand plate up to dilutions of

be added to the excluded volume term. Isoelectric focussing gels indicate that the Im9 protein has a p i of 4.5 (Kleanthous, unpublished results) and so is significantly anionic, consistent with the latter suggestion. To confirm this idea, the translational diffusion coefficient was determined at a lower protein concentration, since charge effects ‘extrapolate out’ on dilution, whereas the effect of asymmetry and swelling remain. At 3.0 mg/ml the translational diffusion coefficient was estimated as D z 0 = 13.0 0.5 x lo-’ cm2 . s- From this value, the frictional ratio may be estimated (using the formula mass and computed partial specific volume), giving the dimensionless ratioflf, of 1.15. Since the concentration dependence of D is generally small, we can take this as close to an ‘infinite-dilution’ estimate, and note that it is within the range conventionally found for a wide range of globular proteins 1431. Hence, we can conclude that this protein is itself close to globular in conformation and that its unusually high virial coefficient is to be accounted for in terms of charge (Donnan) effects. Due to the small size of Im9, an accurate determination of its sedimentation coefficient is difficult to obtain. Nevertheless, a value of s:o,w = 1.14s was determined which, in combination with the measured diffusion coefficient and calculated partial specific volume, yields an apparent molecular mass of 9300. Although an approximate value, this is in satisfactory agreement with the above data and again emphasizes the point that Im9 is monomeric.

’.

In-vitro characterization of the isolated Im9 protein

The biological function of Im9 is to protect a ColE9 producing cell from the DNase activity exhibited by this colicin. While our results show that the overproduced Im9 protein functions in vivo, it does not follow that the independently isolated protein will function in vitro, following purification from crude extracts. Since our ultimate aim is to determine the structure of Im9, it was essential that its binding to unbound ColE9 be analysed to ensure that it did indeed exhibit

the in-vitro properties associated with an immunity protein, namely stoichiometric binding to uncomplexed ColE9, resuIting in complete inhibition of its DNase activity. The first stage of such a strategy involved the purification of unbound ColE9 protein. isolation of unbound coficin E9 It is generally recognised that the interaction between a colicin and its immunity protein is sufficiently strong that separation can only be achieved under denaturing conditions [23]. Typical conditions which have been used include gel filtration in 6 M urea or guanidine hydrochloride or electroelution from SDS/polyacrylamide gels. The interaction between ColE9 and Im9 has never been the subject of any biophysical study and so conditions were first sought under which the two subunits, from purified complex, could be separated. The purification of ColE9 complex is described in Materials and Methods. Denaturation by guanidine hydrochloride was chosen as the method to disrupt the interaction between the two subunits and the appropriate concentration required to denature (and so separate) the subunits was determined by fluorescence spectroscopy. Tryptophan fluorescence is a very sensitive probe of protein tertiary structure and so is often used to monitor protein denaturation [44]. Colicin E9 contains six tryptophan residues, whereas Im9 has only one and so the fluorescence signal from the complex is likely to be dominated by the colicin tryptophans. The fluorescence spectrum for the ColE9 complex (shown in the inset to Fig. 3) has an emission maximum at 337 nm, indicative of one or more buried tryptophan residues, [44]. This signal provided a convenient measure of the denaturation process (shown in Fig. 3) since the fluorescence decreased 3 -4-fold as the concentration of guanidine hydrochloride was raised. The resulting spectrum was consistent with that of free tryptophan (inset to Fig. 3), which has an emission maximum at 354 nm, suggesting that the proteins had denatured. Several points are worthy of note from these

692

-: 0 . 8 -: 1.0

a

0

.

0 . 6 -$ Y

0.4

1

1 I

0.2

o . L = 0

0

0

000,

*

I

1

2

1 3

I

I

I 450

400

350

300

W a v e l e n g t h (nm)

[Gn.HCI] (M) Fig. 3. Guanidine hydrochloride denaturation curve for purified ColE9 complex. Denaturation of the ColE9 complex by guanidine hydrochloride(Gn. HCI) was monitored by fluorescence spectroscopy (AEx = 295 nm; iE, = 334 nm) in potassium phosphate (0.05 M, pH 7.0, 25 C) containing 1 mM dithiothreitol plotted as a function of the total change in emission intensity. The two spectra shown in the top left-hand corner of the figure were obtained in 0 M guanidine hydrochloride (native) and 2.5 M guanidine hydrochloride (unfolded), respectively. The proteins could be separated easily under denaturing conditions (as described in Materials and Methods) by gel-filtration chromatography(Superdex-75;Pharmacia), an example of which is shown in the bottom right-hand corner of the figure.

B ai

.-

c

data. Denaturation (reproducibly) proceeds by a two-step = mechanism, the first step at [guanidine 0.95 M and the second at [guanidine hydro~hloride],,~= 1.5 M. The first transition is not due to the dissociation of Im9 since gel-filtration experiments show that concentrations greater than 1.5 M guanidine hydrochloride are required for dissociation. Gel-filtration experiments, (see below) show that, at 2.5 M guanidine hydrochloride, the two subunits can be separated cleanly. The denatured complex could be refolded simply by dilution to yield the same spectrum as the native complex and retain complete biological activity when tested against a sensitive strain of E. coli (data not shown). Uncomplexed colicin E9 protein was purified by gel filtration in 2.5 M guanidine hydrochloride (an example of which is shown in the inset to Fig. Z). At this concentration of guanidine hydrochloride, the protein is denatured since its fluorescence spectrum is identical to the unfolded spectrum shown in Fig. 3. The protein was renatured by dialysis (or dilution) against potassium phosphate containing dithiothreitol. RenatUrdtlOn was monitored by fluorescence spectroscopy which showed that a signal very similar t o that of complex was generated with a n emission maximum at 334 nm (an example of such a spectrum is shown in Fig. 4A, spectrum b). The interpretation that renaturation had proceeded to a native structure is further strengthened by the observation that this uncomplexed material retains full biological activity (when compared with native ColE9 complex) against a sensitive strain of E. coli, which signifies that all the functional regions of the protein must have refolded to their native states, regions which include domains for receptor recognition, membrane translocation and DNase activity. The implications of this observation in terms of colicin/immunity function are further addressed in the Discussion. Further in-

: :I ;, L

A

, * ,

0.0 0.0

0.2

,

,

,

,

,

,

0.6 0.8 1.0 Stoichiometry (mol Im9/rnol ColE9) 0.4

,I 1.2

Fig. 4. Binding of overproducedIm9 protein to free colicin E9 monitored by fluorescence enhancement. (A) Fluorescence emission spectra showing the changes induced by Im9 binding to colicin E9 protein (IEy = 295 nm) in potassium phosphate (0.05 M, pH 7.0, 25°C) containing 1 m M dithiothreitol. (a) Im9 protein (0.125 nmol); (b) refolded colicin E9 protein (0.390 nmol). Spectra c-f show increasing additions of Im9 (0.125 nmol aliquots) to colicin E9 (0.390 nmol) in the molar ratio of (c) , 0.32: 1 ; (d) 0.64: 1 ; (e) 0.96: 1 ; (0 1.28:1. (B) Plot of the fractional change in fluorescence as a function of the Im9: colicin E9 ratio. Smaller increments of Im9 than shown in (A) above were added to the same fixed concentration of colicin E9. Each point represents the average of duplicate observations.

vitro evidence that the protein had refolded to its native structure comes from assays (see below) which indicate that the refolded colicin E9 protein retains its DNase activity. Association of overproduced and purified Im9 protein with unbound colicin E9

The binding of purified Im9 to unbound colicin E9 was investigated in three separate experiments; binding-induced fluorescence changes, gel filtration and inhibition o f DNase activity. Each set of experiments show that the overproduced Im9 protein binds to its colicin with a 1 : 1 stoichiometry. Fluorescence changes induced by Im9 binding to colicin E9 The fluorescence spectrum of refolded colicin E9 is shown in Fig. 4A (curve b). The spectrum is very similar to that of

693 native ColE9 complex but i , , is ,shifted to 334 nm and the intensity of the signal is approximately 85% that of the complex. It was found that adding increasing concentrations of Im9 protein to unbound colicin E9 resulted in an increase in fluorescence (Fig. 4A) which reached a value, similar to that of the native complex at this concentration, when the two proteins were in a 1 : 1 molar ratio (Fig. 4B). Furthermore, the emission maximum for this reconstituted 1 : 1 complex shifts to 337 nm, identical to that of native ColE9 complex. It is likely that the spectroscopic change emanates from the colicin, since the fluorescence of the Im9 protein (Fig. 4A, curve a) is very much smaller than that of the native colicin. Similar reports of spectroscopic changes have been reported by Suzuki et al. [45] in their study of the circular dichroic changes induced by colicin E3 immunity protein binding to its colicin. These workers found no evidence of gross conformational changes but concluded that local changes in the environment of aromatic residues had occured as a result of forming the complex.

I

c I

I i / h \

'

I

I

I

GeLfiltration experiments Size-exclusion chromatography was also used to follow the association of unbound colicin E9 and purified Im9 (Fig. 5). Purified ColE9 complex (Fig. 5A), containing both colicin E9 and its endogenous immunity protein, migrates as a single peak on a Superdex-75 FPLC column (fractionation range 5000-80000 Da) with an apparent molecular mass of 72000- 73000 Da. The two proteins separate in 2.5 M guanine hydrochloride (Figs 3 and 5B). The colicin E9 protein was purified and refolded (as described above and in Materials and Methods). One equivalent of the overproduced Im9 protein (run in isolation in Fig. 5C) was then incubated with this material, as described in Materials and Methods, and injected onto the column (Fig. SD). A single species was observed (there is no evidence of unbound Im9) which migrated in the same position as native complex. This peak was collected, run on SDSjPAGE (16% polyacrylamide) and shown to contain both the colicin and the added immunity protein (data not shown). To ensure Im9 binding had been saturated another equivalent was added (Fig. 5E). All of this extra protein appeared as unbound Im9, showing that the colicin was indeed fully bound with added immunity protein. Inhibition of ColE9 DNase activity

I ~ I , U ' I I O N V O L U M E lml)

Fig. 5. Association of overproducedIm9 protein with colicin E9 observed by gel-filtration chromatography. The physical association of the two proteins was monitored by gel-filtration chromatography (Superdex75) in potassium phosphate (0.05 M, pH 7.0, room temperature) containing 1 mM dithiothreitol, with the exception of panel (B) which also included 2.5 M guanidine hydrochloride. All experiments were conducted at a full-scale deflection of 0.1, using a constant injection volume of 100 p1 and at a final protein concentration (of complex) of 0.30-0.35 mg/ml. (A) Native ColE9 complex; (B) complex unfolded in 2.5 M guanidine hydrochloride; (C) purified Im9 protein alone (one equivalent); (D) refolded colicin E9 to which one equivalent of purified Im9 had been added; (E) as in (D), except two equivalents of Im9 were added. The dashed lines indicate the elution positions of folded ColE9 complex and unbound Im9 protein, respectively.

Colicin E9 is known to be a non-specific endonuclease [16]. Refolded colicin E9 was tested for DNase activity against linearized pUC18 (Fig. 6, lane 2); complete digestion was observed after 1 h at pH 8.0 and 37°C. Fig. 6, lane 3 is a control of Im9 added to the substrate DNA to demonstrate that no DNase activity is associated with this protein. Fig. 6, lanes 4-6 show the effect of increasing the molar ratio of added six other known enzymatic E-type colicins (three RNases and Im9/colicin E9 protein; the substrate DNA is only protected three DNases) of which little is known except their nucleotide from digestion when the colicin is completely bound with and amino acid sequences. This paper represents the first immunity (1 : 1 stoichiometry). Fig. 6, lane 7 shows that, under of a series of investigations into the biochemical basis for the same conditions and at the same protein concentration, immunity towards the non-specific DNase ColE9. Our results native ColE9 complex (with endogenous Im9 bound) also does show interesting correlations with earlier studies on the nonspecific endonuclease ColE2 system and add to the growing not possess any detectable DNase activity. body of information on the solution properties of colicin immunity proteins. To obtain substantial quantities of the Im9 protein suitable DISCUSSION for structural studies, the imm9 gene was cloned into a bacBiochemical investigations on the enzymatic colicins and terial expression vector and overexpressed, in the absence of their protein inhibitors have centred, principally, on ColE3 its colicin to which Im9 is normally bound. While this provided [24, 45 -481 and, to a lesser extent, ColE2 [25]. Yet there are a ready source of immunity protein, it neccessitated a series

694

1 2 3 4 5 6 7

Royal Society, Society of General Microbiology and the University of East Anglia Research Promotion Fund.

REFERENCES

Fig. 6. Purified lm9 inhibits the DNase activity of colicin E9. Ethidiumbromide-stained agarose gel (1 % polyacrylamide) of electrophoresed target DNA (linearized pUC18, 150 ng) following a 1-h incubation at 3 7 ' C with; lane 1, digestion buffer (described in Materials and Methods); lane 2, refolded colicin E9 (0.73 pg); lane 3, overproduced Im9 (0.12 pg); lanes 4-6, increasing molar ratio of Xm9/colicin E9 (0.2: 1 , 0.5: 1 and 1 :1, respectively); lane 7, native ColE9 complex (0.85 Pg).

of experiments aimed at demonstrating that the overproduced protein was active both in vivo and in vitro. In the process, solution properties of a hitherto unstudied immunity protein have been determined. The in-vivo activity of the overproduced Im9 protein is clearly observed by the complete protection it affords E. cofi JM83, a strain which is normally sensitive towards the cytotoxic effects of ColE9. The in-vitro activity of purified Im9 was demonstrated by its ability to bind stoichiometrically to unbound ColE9 protein, resulting in the complete inhibition of its DNase activity. Similarly, Yamamoto et al. [25]showed that, following the separation of the two subunits of ColE2 complex, the Im2 protein provided complete DNase protection when added back to the unbound colicin E2 protein. An interesting observation stemming from these studies concerns the bacteriocidal activity of unbound ColE9. Conflicting reports have appeared in the literature regarding the bacteriocidal activity of unbound enzymatic colicins and the role of immunity proteins in receptor recognition [24,49]. Our results demonstrate clearly that, for ColE9, bound immunity protein is not a a pre-requisite for cell attachment or translocation, since the free colicin has exactly the same bacteriocidal activity (when analysed by the zone spot test shown in Fig. 2) as colicin complexed to its immunity protein. In conclusion, we have demonstrated that the Im9 protein can be overproduced in a bacterial host, in isolation of its native colicin, purified to homogeneity and shown to interact specifically with unbound ColE9 in a one to one molar ratio, resulting in complete inhibition of colicin DNase activity. Isolated Im9 is monomeric, even at high protein concentrations, which will greatly benefit future solution NMR studies aimed at determining the three dimensional structure of this DNase inhibitor protein. Thc authors would like to thank Andrea Shneier and Dr Chris Abell (University of Cambridge) for conducting the electrospray mass-spectrometry experiments, Dr David Campbell (University of Dundee) for amino acid analyses and Mrs Purnima Mistry (University of Leicester) for technical assistance with the analytical ultracentrifLigation experiments. This work was funded by the Science and Engineering Research Council of the United Kingdom under the Molecular Recognition Initiative and by research grants (to C. K.) from The

1. Reeves, P. (1972) The Bacteriocins, Chapman and Hall, London. 2. Jakes, K. S. (1982) in Molecular action oftoxins & viruses (Cohen, P. & Van Heyningen, S., eds) pp. 131- 167, Elsevier, Amsterdam. 3. Konisky, J. (1982) Ann. Rev. Microbiol. 36, 125-144. 4. Luria, S. E. & Suit, J . L. (1987) in Escherichia cofi & Salmonella typhirnurium, cellular & molecular biology (Neidhardt, F. C., ed.) vol. 2, pp. 1615- 1624, Am. SOC.Microbiol., Washington DC. 5. Ohno-Iwashita, Y. & Imahori, K. (1980) Biochemistry 19, 652659 6. Dankert, J. R., Uratani, Y., Grabau, C., Cramer, W. A. & Hermondson, M. (1982) J . Biol. Chem. 257, 3857-3863. 7. BenCdetti, H., Frenette, M., Baty, D., Knibiehler, M.. Pattus, F. & Lazdunski, C. (1991) J . Mol. Bid. 217,429-439. 8. DiMasi, D. R., White, J. C., Schnaitman, C. A. & Bradbeer, C. (1973) J . Bacteriol. 115, 506-513. 9. Cramer, W. A,, Dankert, J. R. & Uratani, Y . (1983) Biochim. Biophys. Acta. 737, 113 - 193. 10. Lazdunski, C., Baty, D., Geli, V., Cavard, D., Morlon, J., Lloubks, R., Howard, P., Knibiehler, M., Chartier, M., Varenne, S., Frenette, M., Dasseux, J . L. & Pattus, F. (1988) Biochim. Biophys. Actu. 947,445 - 464. 11. Pattus, F., Masotte, D., Wilmsen, H. U., Lakey, J., Tsernoglou, D., Tucker, A. & Parker, M. W. (1990) Experientiu 46, 180192. 12. Senior, B. W. & Holland, I. B. (1971) Proc. Nut1 Acad. Sci. USA 68,959-963. 13. Bowman, C. M., Dahlaberg, J. E., Ikemura, T., Konisky, J. & Nomura, M. (1971) Proc. Nut1 Acad. Sci. USA 68,964-968. 14. Schaller, K. & Nomurd, M. (1976) Proc. Nut1 Acad. Sci. U S A 73, 3989- 3993. 15. Toba, M., Masaki, H. & Ohta, T. (1988) J . Bacteriol. 170, 3237 3242. 16. Chak, K-F., Kuo, W-S., Lu, F.-M. &James, R. (1991) J . Gen. Microbiol. 137, 91 - 100. 17. Schaller, K., Holtje, J.-V. & Braun, V. (1982) J . Bacteriof. 152, 994 - 1000. 18. Cole, S. T., Saint-Joanis, B. & Pugsley, A. P. (1985) Mol. Gen. Genet. 198,465-472. 19. Uchimura, T. & Lau, P. (1987) Mol. Gen. Genet. 209, 489-493. 20. Curtis, M. D., James, R. & Coddington, A. (1989) J . Gen. Microhiol. 135, 2783 - 2788. 21. Eaton, T. &James, R. (1989) Nucleic Acids Res. 17, 1761 -1761. 22 Sidikaro, J. & Nomura, M. (1974) J . Biol. Chem. 249,445-453. 23 Jakes, K. S. & Zinder, N. D. (1974) Proc. Nut1 Acud. Sci. USA 71,3380-3384. 24 Ohno, S., Ohno-Iwashita, Y., Suzuki, K. & Imahori, K. (1977) J . Biochem. 82,1045 - 1053. 25 Yamamoto, H., Nishida, K.-J., Beppu, T. & Arima, K. (1978) J . Biochem. 83,827 -834. 26 Jakes, K. S. & Lazdunski, C. (1992) in Bacteriocins, microcins & lantibiotics (James, R., Lazdunski, C. & Pattus, F., eds) in the press, NATO AS1 Series H, Springer, Heidelberg. 27 Pugsley, A. P. & Schwartz, M. (1983) Mol. Gen. Genet. 190,366372. 28 Cooper, P. C. & James, R. (1984) J . Ger?. Microhiol. 130, 209215. 29 James, R., Jarvis, M. & Barker, R. F. (1987) J . Gen. Microbiol. 133, 1553- 1562. 30 Curtis, M. D. & James, R. (1991) Mol. Microbiol. 5,2727-2733. 31 Parker, M. W., Pattus, F., Tucker, A. D. & Tsernoglou, D. (1989) Nature 337, 93-96. 32 Wormald, M. R., Merrill, A. R., Cramer, W. A. & Williams, R. J. P. (1990) Eur. J . Biochem. 191, 155-161. 33 Shoham, M., Levinson, B. L. & Richards. F. M. (1984) J . Mol. Biol. 177, 563 - 565. -

695 34. Frolow, F. & Shoham, M. (1990) J . Bid. Chem. 265, 1019610197. 35. Vieira, J. & Messing, J. (1982) Gene ( A m s t . ) 19, 259-2611. 36. Chak, K.-F. &James, R. (1986) J . Gen. Microbiol. 132,61 -71. 37. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular cloning. A laboratory manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. 38. Rowe, A. J. (1984) Techniques for determining molecular weight in Techniques in Protein and Enzyme Biochemistry (Tipton, K. F., ed.) vol. Bljl supplement, pp. 12-14, Elsevier. 39. Rowe, A. J., Wynne-Jones, S., Thomas, D. & Harding, S. E. (1989) Proc. SPIE 1163, 138-148. 40. Reeves, P. (1965) Ausf. J . Exp. Biol. Med. Sci. 43, 191 -200. 41. Shanafelt, A. B., Goldman, K.M., Kastelein, R. A. & Kayalar, C. (1987) Plasmid 17,261 -264.

42. Shneier, A., Kleanthous, C., Deka, R., Coggins, J. R. & Abell, C. (1991) J . Am. Chem. Soc. 113,9416-9418. 43. Squire, P. G. & Himmel, M. (1979) Arch. Biochem. Biophys. 196, 165 - 177. 44. Schmid, F. X. (1989) in Protein structure; a practical approach (Creighton, T. E., ed.) pp. 251 -285, Oxford University Press. 45. Suzuki, K., Ohno, S. & Imahori, K.(1980) J . Biochem. 87, 761 769. 46. Jakes, K . S., Zinder, N. D. & Boon, T. (1974) J . Biol. Chem. 249, 438 - 444. 47. Suzuki, K. & Imahori, K. (1978) J . Biochem. 84,1021 - 1029. 48. Levinson, B. L., Pickover, C. A. & Richards, F. M. (1983) J . Bid. Chem. 258,10967- 10972. 49. Schaller, K. & Nomura, M. (1976) Proc. Nut1 Acad. Sci. USA 73, 3989 - 3993.