- Email: [email protected]
Photoaffinity cross-linking of acyl carrier protein to Escherichiacoli membranes
191
Eiochimica et Biophysics Acta, 1123 (1992) 191-197 0 1992 Elsevier Science Publishers B.V. All rights reserved OOOS-2760/92/$05.00
BBALIP 53800
Photoaffinity cross-linking of acyl carrier protein to Escherichia coli membranes Nicolas Bayan and H&ne
Thhrisod *
Lahoratoire des Biomembranes, UA I1 16, Unirvrsiti Paris Sud, Orsay (France)
(Received 21 May 1991) (Revised manuscript received 19 August 1991)
Key words: Acyl carrier protein; membrane protein; Photoaffinity cross-linking; (E. coti Protein (ACP) is a small acidic protein which interacts with the various enzymes implicated in the biosynthesis of fatty acids in E. coli. It also interacts with the inner membrane proteins implicated in the biosynthesis of phospholipids. Samples
Acyl Carrier
of radioactive ACP were prepared with high specific activities and bearing photoactivable aryl azide derivatives. Two photoactivable reagents were used: para azido phenacyi bromide (PAPA) which reacts with the SH of the ACP prosthetic group and the N-hydrozysuccinimide ester of 4-azido salicilic acid (NH!+ASA) which reacts with the amino groups of the protein. Various methods were used to demonst~te that ACP could be cross-linked specitlcally to an inner membrane protein of E. coti, most probably to the glycerol3-phosphate acyl transferase (GPAT). This covalent link should provide a powerful tool for further analysis of the structure of GPAT and its role in phospholipid biosynthesis. These photoactivable aryl azide derivatives of ACP couid also be very useful for studying the interaction of ACP with the soluble enzymes implicated in fatty acid biosynthesis.
Introduction
Acyl Carrier Protein (ACP), a small acidic protein, occupies a central position in lipid metabolism where it functions as an essential substrate carrier in fatty acid and complex lipid biosynthesis [l]. Broader metabolic functions for this protein have also been reported. In particular, Euglena ACP was found to be a pH-dependent, non-competitive inhibitor of an NADH-depe~dent acetoacetyl-CoA reductase [2] and we have previously shown that it is involved in the biosynthesis of the polyglucose chains of membrane-derived oligosaccharides in E. cd [3,4]. Having many and diverse functions, ACP presumably interacts with many different proteins, particularly with the numerous enzymes of the fatty acid pathway. In E. coli this pathway is cytoplasmic [5]. ACP must also interact with inner membrane proteins, in particular with the enzymes implicated in the biosynthesis and in the reacylation of phospholipids. We have recently shown that the inner membrane of E. coli contains
* Present address: Laboratoire de Chimie Organique Multifonctionnelle, Universitt
Paris Sud, Bit 420. 91405 Orsay Cedex, France.
Correspondence: H. Therisod, Laboratoire de Chimie Organique Multifonctionnelle, Universite Paris Sud, BPt 420. 91405 Orsay Cedex, France.
receptors for ACP and that these receptors are proteins [6]. Two inner membrane proteins have recently been shown to interact with ACP. Cooper et al. have demonstrated that in vitro the 2-acylgiycerolphosphoethanolamine acyltransferase (2-acyl-GPE acyltransferase) implicated in the reacylation of phospholipids has a tight binding site for ACP 173and we have shown that the glycerol-3-phosphate acyltransferase (GPAT) implicated in the bios~thesis of phosphoIipids [S] is an ACP binding protein [9]. All characterised ACPs share common features with calcium binding proteins. Both classes of proteins are rich in acidic residues and are structurally related to one another [lo]. Calcium binding proteins, like ACP are known to interact with a variety of proteins. These interactions are not fully understood. Investigation of the binding between calmodulin and several peptides suggested the involvement of amphiphilic heiices possibly with some electrostatic attractions. The secondary structure of ACP has been studied by two dimensionai ‘H-NMR spectroscopy and a high hehcal content has been deduced Ill]. Ernst-Fonberg and his collaborators have compared the hydrophobicity of helices from four different ACPs and found that the hydrophobic moment of these ACPs helices and those of many EF hand calcium binding proteins (helix-loop-helix) were similar [l&13]. The authors suggest that these amphiphilic helices permit efficient reversible interaction
192 with numerous different proteins and/or membranes [12]. It has been suggested that the configuration of E. co/i ACP is EF-hand based on secondary structure-prediction algorithms [ 131, but the helices predicted do not match those deduced from two dimensional NMR analysis of the secondary structure. We are particularly interested in the study of membrane proteins which interact with ACP. Affinity labelling by means of bifunctional cross-linking reagents, which has proved to be an extremely useful procedure in the biochemical and pharmacological analysis of many receptors, was used in this work. This technique could be a powerful tool in the investigation of the mode of interaction of ACP with its different protein receptors. We have therefore developed methods for preparing radioactive ACP of high specific activities and bearing photoactivable aryl azide derivatives. These photoactivable ACP probes have proved to be highly specific. This paper presents evidence of a covalent association of ACP with an inner membrane protein of E. cofi and our data suggest strongly that this protein is the GPAT. The different techniques used to identify this physical association are discussed. Materials
and Methods
Materials
[ ‘HIP-Alanine was obtained from New England Nuclear. Na”‘I (37 MBq) was from Amersham. Iodogen, lodobeads and NHS-ASA were from Pierce Chemicals. Enzymobeads were from BioRad Laboratories and pAPA from Serva. Bacterial
strains
Strain FB8 (F- prototroph), a derivative of E. coli K12, was grown at 37 “C on medium M63, supplemented with 0.2% glucose and 0.1% casaminoacids. Strain ORV30/pWW20 was a generous gift from R.M. Bell. This strain overexpresses GPAT [14] and was grown in the presence of 50 pg/ml of sodium ampicillin. Strain SJ16 (a p-alanine auxotroph) was grown on minimal medium E [15] containing thiamine (0.001%). methionine (O.Ol%), glucose (0.4%) and /3alanine (8.9 10 -(‘%‘o). Preparation of ACP Purification of ACP.
E. coli ACP was purified as described by Rock and Cronan [16]. This procedure resulted in a sample giving a single band on SDS PAGE. Preparation
of labelled ACP: preparation
[3H]ACP was prepared as described nan [15] except that [“HIP-alanine specific activity of 4 TBq/mmole. Preparation
The
iodination
of labelled ACP:
buffer
used
of [‘H]ACP.
by Rock and Crowas used with a
radioiodination
in all the
of ACP.
methods
de-
scribed was molar ratio iodination.
10 mM PO: , 10 mM Kt of iodine to protein was
pH 7.0. The 1 : 2 in each
lodogen method 1171. Iodogen (1,3.4,5-tetrachloro3,6-diphenyl-glycouril) was dissolved in trichloromethane at 10 kg/ml. Aliquots of 100 ~1 were added to polypropylene tubes and the solvent was evaporated with a gentle stream of dry nitrogen. The coated tubes were stored desiccated in the dark at - 70 v C. ACP ( 10 pg) and iodine (37 MBq) were added to a coated tube and incubated for 10 min at room temperature. The reaction was stopped by removing ACP and iodine from the tube. ACP and iodine were separated on ;I Sephadex G25 column (PD 10 Pharmacia) equilibrated in the iodination buffer. Fractions were collected and the elution profile was estimated by counting aliquots in a scintillation counter. Iodobead method 1181. ACP (10 pg) was mixed with 37 MBq of iodine in a polypropylene tube. One or more iodobeads, previously washed in the iodination buffer were added and the reaction was incubated for 10 min at room temperature. The reaction was stopped by removing the solution from the tube. ACP and iodine were separated as described above. Enqmobead method /19/. The enzymobead reagent consists of an immobilized preparation of lactoperoxidase and glucose oxidasc. The reagent was rehydratcd with 50 ~1 of distilled water an hour before use. The reaction was carried out, as described above, with 10 pug ACP and 37 MBq of sodium iodide in the iodination buffer. The reaction was started by the addition of 250 pug of @o-glucose and was allowed to proceed for 30 min before removing the enzymobeads by crntrifugation. The resulting supernatant was applied to a gel filtration column as previously described. Preparation
of ACP-NHS-ASA
and A CP-pAPA
dcriru-
ti1Y3 Preparation of .4CP-pAPA derirwtille. Since pAPA is activated by UV illumination, all steps were done in minimal light conditions. ACP was incubated overnight at room temperature with 5 mM DTT in order to reduce the protein’s free sulphydryl group. DTT was then eliminated by dialysis. A IO-fold molar excess of pAPA. freshly dissolved in methanol, was slowly added to the reduced ACP dissolved in a Tris buffer at pH 7.0. After 2 h of incubation at room temperature, unreacted pAPA was removed by gel filtration chromatography on Sephadex G25. Preparation of the ACP-NHS-ASA derirwtilx,. A IOfold molar excess of NHS-ASA (dissolved in DMSO) was added to ACP in a phosphate buffer at pH 7.0 and at room temperature. The reaction was conducted at a high concentration of ACP (generally 10 mg/ml) bccause the half life of NHS-ASA is very short (less than 10 min).
193 Irradiations experiments
Irradiation experiments were conducted using a 250 W mercury UV lamp (Applied Photophysics). Samples on ice were directly irradiated for 10 min at a distance of 10 cm from the light.
ACP-SH
Br-CY-g+Y
PAPA
Production of antibodies
Polyclonal antibodies were produced according to the procedure described by Ernst-Fonberg et al. [20]. Antibodies specifically directed against ACP were purified on an ACP-Sepharose column as described by Jackowski and Rock [21].
+ membrane
vesicles
+ hv
N-MEMBRANE
Biochemical methods
RECEPTOR
B
Inverted membrane vesicles were prepared as previously described [6]. Membrane GPAT activities were determined as described [22] and glucosyltransferase activities were determined as described previously
Fig. 1. General
[WI.
mide portion of the molecule reacts specifically with particularly reactive cysteine residues. ACP has only one SH group which is at the extremity of the prosthetic group and expected to be reactive enough to link this reagent. A solution of pure ACP (cold or radioactive) was treated with pAPA as described in Materials and Methods. The stoichiometry of the alkylation was determined by spectrophotometry as described [30]. The difference of absorbance at 300 nm of ACP before and after treatment with pAPA indicates that 80% of the ACP was alkylated by pAPA. This result suggests that the stoichiometry was probably of one molecule of pAPA for one molecule of ACP. Moreover, ACP modified by pAPA completely failed to incorporate [3H]NEM. This result confirms that pAPA bound to the free sulphydryl of the ACP prosthetic group. With MSASA. NHS-ASA (bazidosalicylic acid Nhydroxysuccinimide ester) reacts with the amino groups of proteins. Because this reagent is unstable in water, the reaction must be carried out in organic solvent (formamide and DMSO) [31]. Unfortunately, ACP is partially denatured after incubation in these organic solvents. To avoid these problems, alkylation of ACP by NHS-ASA was performed in a small volume of phosphate buffer and with a high concentration of reagent as described in Materials and Methods. In these conditions, ACP was not inactivated and the degradation of NHS-ASA was partially avoided. To determine the degree of incorporation of this reagent, radioactive NHS-ASA was incubated with ACP in the conditions described previously. ACP was then precipitated by reducing the pH to 3.9 by addition of acetic acid. After several washes, radioactivity associated with ACP was counted (data not shown). The results indicate that approximately one molecule of reagent is linked per molecule of ACP (ACP contains five potential sites of fixation for NHS-ASA: the NH,
Protein concentrations were determined by the method of Lowry et al. [24] or by the microbiuret method described by Munkres and Richards [251. These two methods gave similar results. SDS polyacrylamide gel electrophoresis, was carried out as described by Laemmli [26] with a 4% stacking gel and a 10% separation gel. Samples were denatured in the presence of 2% SDS and 2.5% P-mercaptoethanol in 50 mM Tris-HCl pH 6.8. After electrophoresis, the Coomassie blue gel was dried and autoradiographed. Electroblotting of proteins onto nitrocellulose sheets was carried out at 50 V for 1 h in a Tris (25 mM)-glycine (193 mM) buffer (pH 8.35) containing 20% methanol. ACP was electroblotted onto a 0.22 pm nitrocellulose sheet at 25 V for 30 min. Immunodetection was as described by Pierce and Capron [27] using goat antirabbit IgG conjugated with horseradish peroxidase. Immunoblots were developed with 4-chloro-1-naphtol or with luminol (ECL Western Blotting, Amersham). Radioactive banding patterns after SDS PAGE were analysed as previously described [28]. After electrophoresis was completed, the cylindrical gels were cut into 2 mm slices which were then placed in scintillation vials. 250 ~1 of hydrogen peroxide was added and the slices were incubated overnight. 4 ml scintillation reagent containing 0.7% of acetic acid were added to each vial, which were then counted. Results Modification of ACP with bi’nctional With PAPA. Para azido phenacyl
reagents
bromide (PAPA) (Fig. 1) is a bifunctional reagent which can be covalently linked to specific loci on a macromolecule. It contains a photochemically labile azido group with a reactive cu-bromo ketone moiety [29]. The alkyl bro-
scheme for the synthesis and cross-linking
toactivable
ACP on E. coli membrane
of pho-
vesicles.
194 terminus and the NH, groups of the four lysine side chains). We chose to compare the activities of ACPs modified by pAPA or by NHS-ASA with the activity of unmodified ACP using the in vitro test for the /3 l-2 glucan chain biosynthesis [3,231, since we have previously shown that the prosthetic group of ACP is not necessary for this activity [41. The activity of these modified ACPs was not different from that of unmodified ACP (data not shown). Interaction of ACP with its membrane receptors: Characterisation by anti ACP antibodies Chemical cross-linking followed by Western blot analysis can be used for assessing the formation of specific complexes between proteins. However, ACPs are small acidic proteins and are therefore difficult to trap and to detect on Western blots. The sensitivity of Western blotting in the detection of ACP was therefore assessed. In our hands, less than one picomole of ACP on the nitrocellulose sheet could be detected. However, it is important to note that multiple bands. suggesting ACP oligomers, were detected. These results agree with those obtained by Worsham et al. [32]. ACP interacts with various proteins, so the tendency of ACP to aggregate is not surprising, but the persistence of the oligomers in the presence of SDS was unexpected. Not all proteins undergo complete dissociation in detergent. Nevertheless, for investigating mechanisms of interaction between ACP and target membrane proteins, it is necessary to be aware of distinctive self-aggregation patterns of ACP. For this reason, we developed other procedures for the detection of the interaction of ACP with its membrane receptors. Iodination of E. coli ACP Radioiodination of soluble protein with [ “‘11 has become very popular because of the simplicity and the versatility of the method that often results in proteins of high specific activity. The main disadvantage of this method is that proteins must be treated with very destructive oxidizing agents. The susceptibility of proteins to oxidizing agents and to iodination damage may vary. In some cases. one method of iodination will damage one protein while another method may not. Iodination of Spinach ACP was described by Ohlrogge et al. [33]. Because this ACP contains no tyrosine residues, these authors used the iodinated Bolton Hunter reagent [34]. This reagent can react specifically with NH2 and therefore high specific activities can be achieved. Nevertheless, this method could lead to significant modification of the protein. Direct iodination of E. coli ACP has not been reported but is potentially possible because it contains one tyrosine residue.
TABLE
I
lodirrcrtior1 of .4 C‘P IO pg
of ACP
descrihcd iodine
on a Sephadex
deduced ACP.
were
in Materials
iodinated
Biological
different
methods
ACP
was separated
G?S column
and
the
from the determination activity of ACP
glucosyl transferase
using three
and Methods.
specific
of the radioactivity after
iodination
as
from free
activity
was
associated with
was tchtcd in the
assay.
Experimental
Specific
Glucosyltransferase
conditions
activity
activity
Iodogrn
1-N GBq/mmole
t
Iodoheads I bead
370 GBq/mmole
t
2 heads
I I IO GBq/mmole
t
5 heads
‘7Yhll GBq/mmole
Enzymobeads IS min
?YO GBq/mmolr
30 min
7JO GBq/mmole
+ _
The different methods used to iodinate E. coli ACP are described in Materials and Methods. The different specific activities obtained with the different procedures are given in Table I. The biological activity of ACP after iodination was verified. The p l-2 linked polyglucose chain elongation activity of ACP was tested using the glucosyl transferase assay (Table I) [3,23]. The iodobeads method was chosen for subsequent experiments. This method gave ACP with a specific activity of 370 GBq/mmole and no significant loss of functionality. This precaution is necessary to be sure that ACP can still interact with its receptors. The stability of iodoproteins differ markedly from one to another. Iodinated ACP dramatically lost activity when stored at 4’ C. To avoid this problem [ “5]ACP was freshly prepared before each experiment. Photoaffinity label& of E. coli membrane llesicles Photolabelling with /“slJACP-pAPA and [“‘IIACPNHS-ASA. The synthesis and cross-linking of photoactivable ACP on E. coli membrane vesicles are summarized in Fig. 1. Right side out membrane vesicles of i?. coli strain ORV3O/pWW20, which overexpresses GPAT, were incubated for 20 min with 0.5 PM [““IIACP-pAPA. This concentration is close to the K,, determined for the interaction of ACP with the membrane [6]. Vesicles were washed with buffer, then irradiated for 10 or 20 min under UV, washed again with buffer in the presence of 10 PM unlabelled ACP to remove unbound [“‘IIACP. The samples were then analysed by SDS PAGE and the Coomassie blue stained gel was dried and autoradiographed. The activity in the glucosyl transferase assay of ACP and those of membrane vesicles [3,23] are not altered by illumination. Fig. 2 shows the results obtained. Lanes 3 and 4 represent, respectively, the analysis of membrane pro-
195 teins after 10 min and 20 min of irradiation. Each lane contains three radioactive bands. The top band corresponds to a material which did not enter the gel. This material of high molecular weight is not a specific complex between ACP and a membrane protein because this band was present in control experiments in the presence of an excess of unlabelled ACP. This material was probably aggregates resulting from the large amounts of protein (200 pg) loaded in each lane. A second band, of about 20 kDa protein was presumably iodinate ACPpAPA (lane 6). The third band, which has an apparent molecular weight of 100 kDa, was not present in controls: when the cross-linking was performed in the presence of an excess of unlabelled ACP (lane 21, or without photoactivating light (data not shown). This band probably corresponds to GPAT (83 kDa) which has been cross-linked to ACP. In similar experiments with vesicles from wild type E. cofi, which contains much less (eight-fold) GPAT, this 100 kDa band was not detected. The same experiments were conducted with [‘251]ACP modified with NHS-ASA. The results are presented in Fig. 3. Radioactive cross-linked complex with a molecular weight of about 100 kDa was again detected. ACP modified with NHS-ASA resulted in significant non-specific binding. This non-specific binding was not observed when the pAPA reagent was used. Photoaffinity label&g with [3HIACP-pAPA. [ 3H]ACP with a specific activity of 3-4 TBq/mmole was prepared. Unlike iodination, this procedure allowed the
123456
46~
.:
Fig. 2. Autoradiograph of iodinated ACP-pAPA cross-linked on ORV30/pWW20 membrane vesicles after SDS PAGE. Lane 1: molecular weight markers (kDa); Lane 2: ORV30/pWW20 membrane vesicles after 10 min irradiation in the presence of 0.5 PM of iodinated ACP-pAPA and 50 PM of unlabelled ACP-pAPA; Lane 3: ORV30/pWW20 membrane vesicles after 10 min irradiation in the presence of 0.5 PM of iodinated ACP-pAPA; Lane 4: ORV30/pWW20 membrane vesicles after 20 min irradiation in the presence of 0.5 PM of iodinated ACP-pAPA; Lane 5: this lane represents the migration of GPAT as visualized after coomassie staining. Lane 6: iodinated ACP-pAPA.
123 !/
4 j.
‘,. ;
:,.
.:
”
,,
4TOP
200 t;r 46304 FRONT
Fig. 3. Autoradiograph of iodinated ACP-NHSASA cross-lurked on ORV30/pWW20 membrane vesicles after SDS PAGE. Lane 1: This lane represents the migration of GPAT as visualized after coomassie staining; Lane 2: ORV30/pWW20 membrane vesicles after 10 min irradiation in the presence of 0.5 I.LM of iodinated ACP-NHSASA; Lane 3: ORV30/pWW20 membrane vesicles after IO min irradiation in the presence of 0.5 PM of iodinated ACP-NHSASA and 50 PM of unlabelled ACP-NHSASA; Lane 4: iodinated ACP-NHSASA.
synthesis of an L3H]ACP chemically identical to unlabelled ACP. Therefore the affinity of 13HlACP for its membrane receptors was not modified by the label. [ 3H]ACP was modified with pAPA and cross-linking was done as described above. Because it is difficult to obtain large amounts of [3H]ACP, it was used at a concentration of 0.1 PM instead of 0.5 PM. Autoradiography is less sensitive to tritium than to [‘25I] and therefore semi preparative gel electrophoresis was used to analyse the results of cross-linking: 1 mg of membrane protein was loaded onto a cylindrical polyacrylamide gel (9 mm diameter and 15 cm length). After electrophoresis, gels were sliced into 2 mm slices which were placed in scintillation vials and counted. Fig. 4A shows the radioactive profile after electrophoresis of FB8 (wild type) membrane vesicles after irradiation in the presence of L3HlACP modified with pAPA, and Fig. 4B shows that of 0RV30 (rich in GPAT) membrane vesicles. In each case, two major peaks can be distinguished. The first one, which migrates like a protein of 20 kDa is free [3H]ACP. The second peak has an apparent molecular weight of 100 kDa. This peak presumably represents a specific covalent attachment between ACP and a membrane receptor because it was absent when the experiment was done in the presence of an excess of unlabelled ACP or in the absence of irradiation. These results are similar to those obtained by autoradiography with iodo ACP
0
10
20
30
40
50
60
A
FRACTIONS 250
200
150
100
50
0 0
10
20
30
FRACTIONS
40
50
60
B
Fig. 4. Cross-linking of [‘HIACP-pAPA on E. coli membrane vesicles. 0.1 PM of [3H]ACP-pAPA was added to membrane vesicles, and the sample irradiated for 10 min to obtain cross-linking. Membrane proteins were separated on a semi preparative cylindrical l&1.5% polyacrylamide gradient gel. Radioactivity was detected after cutting the gel into 2 mm slices as described in Materials and Methods. A control experiment in the presence of an excess (SO PM) of unlabelled ACP was performed. (A) Represents the radioactive profile obtained with wild type (FB8) vesicles after correction from the control experiment. (B) Represents the radioactive profile obtained with ORV30/pWW20 vesicles after correction from the control experiment. The migration of molecular weight markers is represented on the top of each graph.
(Figs. 2 and 3). However, this method also detected the 100 kDa band in a wild type strain of E. coli, expressing low levels of GPAT. The intensity of the 100 kDa peak is seven times higher in the ORV30/pWW20 samples than in the wild type samples. This factor of seven is in good agreement with the increase of activity and with the increase of ACP binding observed in this strain [9,14]. This is consistent with ACP being crosslinked to GPAT. Discussion ACP interacts with many proteins, both soluble and membrane bound [.5]. The nature of these interactions are not understood and a general study of this problem
could be very interesting to elucidate how ACP interacts with such a wide variety of proteins. The method of choice for studying these interactions is the use of cross-linking reagents to link the receptors with their ligand. This method is now very attractive because of the development of many photoreactive bifunctional reagents which facilitate specific cross-linking. In this study we used two photosensitive reagents: pAPA and NHS-ASA. These two reagents bind stoichiometrically to ACP and the biological activity of the protein was not altered. Thus. these two modified ACPs arc very useful for studying the interaction of ACP with proteins. Several methods were used to detect the complexes between ACP and its receptors. First. the complexes between ACP and its receptors were detected with anti ACP antibodies after Western blotting. Unfortunately, in this case, the use of anti ACP antibodies was not satisfactory. The sensitivity of the detection was not sufficient to detect the few cross-linked complexes ohtained after irradiation of membrane vesicles. Moreover, we confirm the results obtained by Worsham et al. [32] indicating that ACP forms oligomers even at low concentration and after SDS denaturation. These structures which were observed on Western blots could hide minor specific cross-linking. The second approach used was to prepare radioactive ACP of high specific activity. In this case the covalent complexes were detected by analysing the radioactive profile after SDS PAGE. Radioiodination of proteins gives high specific activity but, unfortunately, ACP is very sensitive to the oxidizing agents used in the iodination reaction. Consequently [ ““IIACP of relatively high specific activity lost its functionality. This inactivation could be the result of a dimerisation of ACP or more probably, oxidation of the single methionine to methionine sulfone. To avoid this alteration, [“HIACP with a specific activity of 3 or 4 TBq/mmole was prepared. In this third approach, the biological activity of the protein was not altered and the high specific activity allowed reasonable detection of ACP and its covalent complexes. We demonstrated photochemical cross-linking bctween ACP and one E. cofi membrane protein. Our data suggest strongly that this protein is the GPAT. These results agree with those we recently obtained showing that GPAT was an ACP receptor 191. This cross-linkage is specific because it is ( 1) inhibited by unlabelled ACP, (2) dependent on activating light and (3) significantly higher on GPAT-rich membrane vesicles than wild type vesicles. The molecular weight of the cross-linked GPATACP complex was estimated to be around 100 kDa. The molecular weight of GPAT is around 83 kDa [14] and ACP 9 kDa [5]. Nevertheless, it is difficult to predict the number of molecules of ACP which are
197 bound to GPAT because in SDS electrophoresis ACP migrates as a 23 kDa protein [3,35]. E. coli membranes contain other membrane proteins which can bind ACP. 2-acyl GPE binds ACP in vitro [7] and the glucosyltransferase, implicated in the MD0 biosynthesis, is also believe to bind ACP [3,231. These proteins are probably present in lower concentrations than GPAT, which is probably the major ACP binding protein in E. co/i membranes (60-80% of the total ACP binding) as indicated by binding experiments of ACP to E. co/i membrane [6,9] and by the enzymatic activity of membranes from different strains of E. cofi [141. These results underscore the difficulty of covalently labelling the ACP membrane protein receptors because of their low abundance. Appropriate spacing and chemical accessibility of the reactive groups on both receptor and ligand are required for successful crosslinking. These conditions are difficult to obtain in cases of protein-protein interactions and generally a very low yield of cross-links (around 1%) is obtained. Nevertheless, this covalent linking should provide a powerful tool for the future study of the structure and function of GPAT in relation to phospholipid biosynthesis and in particular for the analysis of the ACP binding site. This could be done after proteolysis of the ACP-GPAT complex. This technique could also be very useful for studying the interaction of ACP with the soluble enzymes implicated in the biosynthesis of fatty acids in E. coli. A comparative study of all these ACP binding proteins and more particularly, the regions implicated in ACP could help elucidate the interaction mechanism. It would be particularly interesting to know if this interaction between ACP and the ACP-binding proteins involves amphiphilic helices as proposed by Ernst-Fonberg et al. [12], thus providing an explanation for the mechanisms of recognition for its different receptors. Acknowledgements
We thank Dr. J.P. Tenu for helpful assistance in photolabelling techniques and for his continued interest in this work.
4 Therisod,
6 7 8 9 IO 11 I2 13 I4 15 16 17 18 19 20 21 22 23 24 25 26 27
28
29 30 31 32 33
References 1 Vagelos, P.R. (1973) in The Enzymes, Vol. 8 (Bayer, P.D., ed.), pp. 155-199, Academic Press, New York. 2 Ernst-Fonberg, M.L. (1986) Plant Physiol 82, 978-984. 3 Thtrisod, H., Weissborn, A.C. and Kennedy, E.P. (1986) Proc. Natl. Acad. Sci. USA 83, 7236-7240.
H. and
Kennedy,
USA 84, 8235-8238. 5 Cronan, J.E. and Rock,
34 35
E.P. (1987) C.O.
(1987)
Proc.
Natl.
in Escherichia
Acad.
Sci.
coli and
Salmonella typhimurium Cellular and Molecular Biology I: 31-55, (Neidhardt, F.C., ed.) pp. 474-497, American Society for Microbiology, Washington. Bayan, N. and Therisod, H. (1989) FEBS Lett. 253, 155-159. Cooper, C.L., Hsu. L., Jackowski, S. and Rock, C.O. (1989) J. Biol. Chem. 264, 7384-7389. Bell, R.M., Mavis, R.D.. Osborn, M.J. and Vagelos. P.R. (1971) Biochem. Biophys. Acta 249, 628-635. Bayan N. and ThCrisod, H. (1989) FEBS Lett. 255. 330-334. Krestsinger, R.H. (1980) CRC Cr. Rev. Biochem. 8, 119-174. Holak, T.A., Nilges. M., Prestegard, J.H.. Gronenborn, A.M. and Clore. G.M. (1988) Eur. J. Biochem. 175, 9-15. Ernst-Fonberg, M.L.. Tucker, M.M. and Fonberg, I.B. (1987) FEBS Lett. 215, 261-265. Argos, P. (1977) Biochemistry. 16, 665-672. Wilkison, W.O., Walsh, J.P., Corless, J.M. and Bell, R.M. (1986) J. Biol. Chem. 261, 9951-9958. Rock, C.O. and Cronan, J.E. (1981) Methods Enzymol. 71, 341351. Rock, C.O. and Cronan, J.E. (1980) Anal. Biochem. 102,362-364. Fraker, P.J. and Speck Jr., J.C. (1978) Biochem. Biophys. Res. Commun. 80, 849-857. Markwell, M.A.K. (1983) Anal. Biochem. 125, 427-432. Thorell, J.I. and Johansson, B.G. (1971) Biochim. Biophys. Acta 251, 363-368. Ernst-Fonberg, M.L., Schongalla, A.W. and Walker, T. (1977) Arch. Biochem. Biophys. 178, 166-173. Jackowski, S. and Rock, CO. (1983) J. Biol. Chem. 258, 1518615191. Green, P.R., Merrill, A.H. and Bell, R.M. (1981) J. Biol. Chem. 256, 11151-11159. Weissborn, A.C. and Kennedy, E.P. (1984) J. Biol. Chem. 259, 12644-12651. Lowry, O.H., Rosebrough, N.J., Farr. A.L. and Randall, R.J. (1951) J. Biol. Chem. 193, 265-275. Munkres, K.D. and Richards, F.M. (1965) Arch. Biochim. Biophys. 109, 466-479. Laemmli, U.K. (1970) Nature 227, 680-685. Pierce, R.J. and Capron, A. (1987) in Techniques Biochimiques d’Analyse des Antigenes de Membrane (Charron, D., ed) pp. 43-54, INSERM, France. Chirio, M.C., Brithes, D., Napias, C., Grandier-Vazeille, X., Rakotomanana, F. and Chevallier, J. (1990) Eur. J. Biochem. 194, 293-299. Hixson, H.H. and Hixson, S.S. (1975) Biochemistry 14,4251-4254. Moreland, R.B. and Dockter, M.E. (1981) Biochim. Biophys. Res. Commun. 99, 339-346. Ji, T.H. and Ji, I. (1982) Anal. Biochem. 121, 286-289. Worsham, L.M.S., Tucker, M. and Ernst-Fonberg, M.L. (1990) Biochem. Biophys. Acta 1043, 198-202. Ohlrogge, J.B., Kuhn, D.N. and Stumpf, P.K. (1979) Proc. Natl. Acad. Sci. USA 76, 1194-1198. Bolton, A.E. and Hunter, W.N. (1973) Biochem. J. 133, 529-539. Rock. C.O. and Cronan, J.E. (1979) J. Biol. Chem. 254, 97789785.
Eiochimica et Biophysics Acta, 1123 (1992) 191-197 0 1992 Elsevier Science Publishers B.V. All rights reserved OOOS-2760/92/$05.00
BBALIP 53800
Photoaffinity cross-linking of acyl carrier protein to Escherichia coli membranes Nicolas Bayan and H&ne
Thhrisod *
Lahoratoire des Biomembranes, UA I1 16, Unirvrsiti Paris Sud, Orsay (France)
(Received 21 May 1991) (Revised manuscript received 19 August 1991)
Key words: Acyl carrier protein; membrane protein; Photoaffinity cross-linking; (E. coti Protein (ACP) is a small acidic protein which interacts with the various enzymes implicated in the biosynthesis of fatty acids in E. coli. It also interacts with the inner membrane proteins implicated in the biosynthesis of phospholipids. Samples
Acyl Carrier
of radioactive ACP were prepared with high specific activities and bearing photoactivable aryl azide derivatives. Two photoactivable reagents were used: para azido phenacyi bromide (PAPA) which reacts with the SH of the ACP prosthetic group and the N-hydrozysuccinimide ester of 4-azido salicilic acid (NH!+ASA) which reacts with the amino groups of the protein. Various methods were used to demonst~te that ACP could be cross-linked specitlcally to an inner membrane protein of E. coti, most probably to the glycerol3-phosphate acyl transferase (GPAT). This covalent link should provide a powerful tool for further analysis of the structure of GPAT and its role in phospholipid biosynthesis. These photoactivable aryl azide derivatives of ACP couid also be very useful for studying the interaction of ACP with the soluble enzymes implicated in fatty acid biosynthesis.
Introduction
Acyl Carrier Protein (ACP), a small acidic protein, occupies a central position in lipid metabolism where it functions as an essential substrate carrier in fatty acid and complex lipid biosynthesis [l]. Broader metabolic functions for this protein have also been reported. In particular, Euglena ACP was found to be a pH-dependent, non-competitive inhibitor of an NADH-depe~dent acetoacetyl-CoA reductase [2] and we have previously shown that it is involved in the biosynthesis of the polyglucose chains of membrane-derived oligosaccharides in E. cd [3,4]. Having many and diverse functions, ACP presumably interacts with many different proteins, particularly with the numerous enzymes of the fatty acid pathway. In E. coli this pathway is cytoplasmic [5]. ACP must also interact with inner membrane proteins, in particular with the enzymes implicated in the biosynthesis and in the reacylation of phospholipids. We have recently shown that the inner membrane of E. coli contains
* Present address: Laboratoire de Chimie Organique Multifonctionnelle, Universitt
Paris Sud, Bit 420. 91405 Orsay Cedex, France.
Correspondence: H. Therisod, Laboratoire de Chimie Organique Multifonctionnelle, Universite Paris Sud, BPt 420. 91405 Orsay Cedex, France.
receptors for ACP and that these receptors are proteins [6]. Two inner membrane proteins have recently been shown to interact with ACP. Cooper et al. have demonstrated that in vitro the 2-acylgiycerolphosphoethanolamine acyltransferase (2-acyl-GPE acyltransferase) implicated in the reacylation of phospholipids has a tight binding site for ACP 173and we have shown that the glycerol-3-phosphate acyltransferase (GPAT) implicated in the bios~thesis of phosphoIipids [S] is an ACP binding protein [9]. All characterised ACPs share common features with calcium binding proteins. Both classes of proteins are rich in acidic residues and are structurally related to one another [lo]. Calcium binding proteins, like ACP are known to interact with a variety of proteins. These interactions are not fully understood. Investigation of the binding between calmodulin and several peptides suggested the involvement of amphiphilic heiices possibly with some electrostatic attractions. The secondary structure of ACP has been studied by two dimensionai ‘H-NMR spectroscopy and a high hehcal content has been deduced Ill]. Ernst-Fonberg and his collaborators have compared the hydrophobicity of helices from four different ACPs and found that the hydrophobic moment of these ACPs helices and those of many EF hand calcium binding proteins (helix-loop-helix) were similar [l&13]. The authors suggest that these amphiphilic helices permit efficient reversible interaction
192 with numerous different proteins and/or membranes [12]. It has been suggested that the configuration of E. co/i ACP is EF-hand based on secondary structure-prediction algorithms [ 131, but the helices predicted do not match those deduced from two dimensional NMR analysis of the secondary structure. We are particularly interested in the study of membrane proteins which interact with ACP. Affinity labelling by means of bifunctional cross-linking reagents, which has proved to be an extremely useful procedure in the biochemical and pharmacological analysis of many receptors, was used in this work. This technique could be a powerful tool in the investigation of the mode of interaction of ACP with its different protein receptors. We have therefore developed methods for preparing radioactive ACP of high specific activities and bearing photoactivable aryl azide derivatives. These photoactivable ACP probes have proved to be highly specific. This paper presents evidence of a covalent association of ACP with an inner membrane protein of E. cofi and our data suggest strongly that this protein is the GPAT. The different techniques used to identify this physical association are discussed. Materials
and Methods
Materials
[ ‘HIP-Alanine was obtained from New England Nuclear. Na”‘I (37 MBq) was from Amersham. Iodogen, lodobeads and NHS-ASA were from Pierce Chemicals. Enzymobeads were from BioRad Laboratories and pAPA from Serva. Bacterial
strains
Strain FB8 (F- prototroph), a derivative of E. coli K12, was grown at 37 “C on medium M63, supplemented with 0.2% glucose and 0.1% casaminoacids. Strain ORV30/pWW20 was a generous gift from R.M. Bell. This strain overexpresses GPAT [14] and was grown in the presence of 50 pg/ml of sodium ampicillin. Strain SJ16 (a p-alanine auxotroph) was grown on minimal medium E [15] containing thiamine (0.001%). methionine (O.Ol%), glucose (0.4%) and /3alanine (8.9 10 -(‘%‘o). Preparation of ACP Purification of ACP.
E. coli ACP was purified as described by Rock and Cronan [16]. This procedure resulted in a sample giving a single band on SDS PAGE. Preparation
of labelled ACP: preparation
[3H]ACP was prepared as described nan [15] except that [“HIP-alanine specific activity of 4 TBq/mmole. Preparation
The
iodination
of labelled ACP:
buffer
used
of [‘H]ACP.
by Rock and Crowas used with a
radioiodination
in all the
of ACP.
methods
de-
scribed was molar ratio iodination.
10 mM PO: , 10 mM Kt of iodine to protein was
pH 7.0. The 1 : 2 in each
lodogen method 1171. Iodogen (1,3.4,5-tetrachloro3,6-diphenyl-glycouril) was dissolved in trichloromethane at 10 kg/ml. Aliquots of 100 ~1 were added to polypropylene tubes and the solvent was evaporated with a gentle stream of dry nitrogen. The coated tubes were stored desiccated in the dark at - 70 v C. ACP ( 10 pg) and iodine (37 MBq) were added to a coated tube and incubated for 10 min at room temperature. The reaction was stopped by removing ACP and iodine from the tube. ACP and iodine were separated on ;I Sephadex G25 column (PD 10 Pharmacia) equilibrated in the iodination buffer. Fractions were collected and the elution profile was estimated by counting aliquots in a scintillation counter. Iodobead method 1181. ACP (10 pg) was mixed with 37 MBq of iodine in a polypropylene tube. One or more iodobeads, previously washed in the iodination buffer were added and the reaction was incubated for 10 min at room temperature. The reaction was stopped by removing the solution from the tube. ACP and iodine were separated as described above. Enqmobead method /19/. The enzymobead reagent consists of an immobilized preparation of lactoperoxidase and glucose oxidasc. The reagent was rehydratcd with 50 ~1 of distilled water an hour before use. The reaction was carried out, as described above, with 10 pug ACP and 37 MBq of sodium iodide in the iodination buffer. The reaction was started by the addition of 250 pug of @o-glucose and was allowed to proceed for 30 min before removing the enzymobeads by crntrifugation. The resulting supernatant was applied to a gel filtration column as previously described. Preparation
of ACP-NHS-ASA
and A CP-pAPA
dcriru-
ti1Y3 Preparation of .4CP-pAPA derirwtille. Since pAPA is activated by UV illumination, all steps were done in minimal light conditions. ACP was incubated overnight at room temperature with 5 mM DTT in order to reduce the protein’s free sulphydryl group. DTT was then eliminated by dialysis. A IO-fold molar excess of pAPA. freshly dissolved in methanol, was slowly added to the reduced ACP dissolved in a Tris buffer at pH 7.0. After 2 h of incubation at room temperature, unreacted pAPA was removed by gel filtration chromatography on Sephadex G25. Preparation of the ACP-NHS-ASA derirwtilx,. A IOfold molar excess of NHS-ASA (dissolved in DMSO) was added to ACP in a phosphate buffer at pH 7.0 and at room temperature. The reaction was conducted at a high concentration of ACP (generally 10 mg/ml) bccause the half life of NHS-ASA is very short (less than 10 min).
193 Irradiations experiments
Irradiation experiments were conducted using a 250 W mercury UV lamp (Applied Photophysics). Samples on ice were directly irradiated for 10 min at a distance of 10 cm from the light.
ACP-SH
Br-CY-g+Y
PAPA
Production of antibodies
Polyclonal antibodies were produced according to the procedure described by Ernst-Fonberg et al. [20]. Antibodies specifically directed against ACP were purified on an ACP-Sepharose column as described by Jackowski and Rock [21].
+ membrane
vesicles
+ hv
N-MEMBRANE
Biochemical methods
RECEPTOR
B
Inverted membrane vesicles were prepared as previously described [6]. Membrane GPAT activities were determined as described [22] and glucosyltransferase activities were determined as described previously
Fig. 1. General
[WI.
mide portion of the molecule reacts specifically with particularly reactive cysteine residues. ACP has only one SH group which is at the extremity of the prosthetic group and expected to be reactive enough to link this reagent. A solution of pure ACP (cold or radioactive) was treated with pAPA as described in Materials and Methods. The stoichiometry of the alkylation was determined by spectrophotometry as described [30]. The difference of absorbance at 300 nm of ACP before and after treatment with pAPA indicates that 80% of the ACP was alkylated by pAPA. This result suggests that the stoichiometry was probably of one molecule of pAPA for one molecule of ACP. Moreover, ACP modified by pAPA completely failed to incorporate [3H]NEM. This result confirms that pAPA bound to the free sulphydryl of the ACP prosthetic group. With MSASA. NHS-ASA (bazidosalicylic acid Nhydroxysuccinimide ester) reacts with the amino groups of proteins. Because this reagent is unstable in water, the reaction must be carried out in organic solvent (formamide and DMSO) [31]. Unfortunately, ACP is partially denatured after incubation in these organic solvents. To avoid these problems, alkylation of ACP by NHS-ASA was performed in a small volume of phosphate buffer and with a high concentration of reagent as described in Materials and Methods. In these conditions, ACP was not inactivated and the degradation of NHS-ASA was partially avoided. To determine the degree of incorporation of this reagent, radioactive NHS-ASA was incubated with ACP in the conditions described previously. ACP was then precipitated by reducing the pH to 3.9 by addition of acetic acid. After several washes, radioactivity associated with ACP was counted (data not shown). The results indicate that approximately one molecule of reagent is linked per molecule of ACP (ACP contains five potential sites of fixation for NHS-ASA: the NH,
Protein concentrations were determined by the method of Lowry et al. [24] or by the microbiuret method described by Munkres and Richards [251. These two methods gave similar results. SDS polyacrylamide gel electrophoresis, was carried out as described by Laemmli [26] with a 4% stacking gel and a 10% separation gel. Samples were denatured in the presence of 2% SDS and 2.5% P-mercaptoethanol in 50 mM Tris-HCl pH 6.8. After electrophoresis, the Coomassie blue gel was dried and autoradiographed. Electroblotting of proteins onto nitrocellulose sheets was carried out at 50 V for 1 h in a Tris (25 mM)-glycine (193 mM) buffer (pH 8.35) containing 20% methanol. ACP was electroblotted onto a 0.22 pm nitrocellulose sheet at 25 V for 30 min. Immunodetection was as described by Pierce and Capron [27] using goat antirabbit IgG conjugated with horseradish peroxidase. Immunoblots were developed with 4-chloro-1-naphtol or with luminol (ECL Western Blotting, Amersham). Radioactive banding patterns after SDS PAGE were analysed as previously described [28]. After electrophoresis was completed, the cylindrical gels were cut into 2 mm slices which were then placed in scintillation vials. 250 ~1 of hydrogen peroxide was added and the slices were incubated overnight. 4 ml scintillation reagent containing 0.7% of acetic acid were added to each vial, which were then counted. Results Modification of ACP with bi’nctional With PAPA. Para azido phenacyl
reagents
bromide (PAPA) (Fig. 1) is a bifunctional reagent which can be covalently linked to specific loci on a macromolecule. It contains a photochemically labile azido group with a reactive cu-bromo ketone moiety [29]. The alkyl bro-
scheme for the synthesis and cross-linking
toactivable
ACP on E. coli membrane
of pho-
vesicles.
194 terminus and the NH, groups of the four lysine side chains). We chose to compare the activities of ACPs modified by pAPA or by NHS-ASA with the activity of unmodified ACP using the in vitro test for the /3 l-2 glucan chain biosynthesis [3,231, since we have previously shown that the prosthetic group of ACP is not necessary for this activity [41. The activity of these modified ACPs was not different from that of unmodified ACP (data not shown). Interaction of ACP with its membrane receptors: Characterisation by anti ACP antibodies Chemical cross-linking followed by Western blot analysis can be used for assessing the formation of specific complexes between proteins. However, ACPs are small acidic proteins and are therefore difficult to trap and to detect on Western blots. The sensitivity of Western blotting in the detection of ACP was therefore assessed. In our hands, less than one picomole of ACP on the nitrocellulose sheet could be detected. However, it is important to note that multiple bands. suggesting ACP oligomers, were detected. These results agree with those obtained by Worsham et al. [32]. ACP interacts with various proteins, so the tendency of ACP to aggregate is not surprising, but the persistence of the oligomers in the presence of SDS was unexpected. Not all proteins undergo complete dissociation in detergent. Nevertheless, for investigating mechanisms of interaction between ACP and target membrane proteins, it is necessary to be aware of distinctive self-aggregation patterns of ACP. For this reason, we developed other procedures for the detection of the interaction of ACP with its membrane receptors. Iodination of E. coli ACP Radioiodination of soluble protein with [ “‘11 has become very popular because of the simplicity and the versatility of the method that often results in proteins of high specific activity. The main disadvantage of this method is that proteins must be treated with very destructive oxidizing agents. The susceptibility of proteins to oxidizing agents and to iodination damage may vary. In some cases. one method of iodination will damage one protein while another method may not. Iodination of Spinach ACP was described by Ohlrogge et al. [33]. Because this ACP contains no tyrosine residues, these authors used the iodinated Bolton Hunter reagent [34]. This reagent can react specifically with NH2 and therefore high specific activities can be achieved. Nevertheless, this method could lead to significant modification of the protein. Direct iodination of E. coli ACP has not been reported but is potentially possible because it contains one tyrosine residue.
TABLE
I
lodirrcrtior1 of .4 C‘P IO pg
of ACP
descrihcd iodine
on a Sephadex
deduced ACP.
were
in Materials
iodinated
Biological
different
methods
ACP
was separated
G?S column
and
the
from the determination activity of ACP
glucosyl transferase
using three
and Methods.
specific
of the radioactivity after
iodination
as
from free
activity
was
associated with
was tchtcd in the
assay.
Experimental
Specific
Glucosyltransferase
conditions
activity
activity
Iodogrn
1-N GBq/mmole
t
Iodoheads I bead
370 GBq/mmole
t
2 heads
I I IO GBq/mmole
t
5 heads
‘7Yhll GBq/mmole
Enzymobeads IS min
?YO GBq/mmolr
30 min
7JO GBq/mmole
+ _
The different methods used to iodinate E. coli ACP are described in Materials and Methods. The different specific activities obtained with the different procedures are given in Table I. The biological activity of ACP after iodination was verified. The p l-2 linked polyglucose chain elongation activity of ACP was tested using the glucosyl transferase assay (Table I) [3,23]. The iodobeads method was chosen for subsequent experiments. This method gave ACP with a specific activity of 370 GBq/mmole and no significant loss of functionality. This precaution is necessary to be sure that ACP can still interact with its receptors. The stability of iodoproteins differ markedly from one to another. Iodinated ACP dramatically lost activity when stored at 4’ C. To avoid this problem [ “5]ACP was freshly prepared before each experiment. Photoaffinity label& of E. coli membrane llesicles Photolabelling with /“slJACP-pAPA and [“‘IIACPNHS-ASA. The synthesis and cross-linking of photoactivable ACP on E. coli membrane vesicles are summarized in Fig. 1. Right side out membrane vesicles of i?. coli strain ORV3O/pWW20, which overexpresses GPAT, were incubated for 20 min with 0.5 PM [““IIACP-pAPA. This concentration is close to the K,, determined for the interaction of ACP with the membrane [6]. Vesicles were washed with buffer, then irradiated for 10 or 20 min under UV, washed again with buffer in the presence of 10 PM unlabelled ACP to remove unbound [“‘IIACP. The samples were then analysed by SDS PAGE and the Coomassie blue stained gel was dried and autoradiographed. The activity in the glucosyl transferase assay of ACP and those of membrane vesicles [3,23] are not altered by illumination. Fig. 2 shows the results obtained. Lanes 3 and 4 represent, respectively, the analysis of membrane pro-
195 teins after 10 min and 20 min of irradiation. Each lane contains three radioactive bands. The top band corresponds to a material which did not enter the gel. This material of high molecular weight is not a specific complex between ACP and a membrane protein because this band was present in control experiments in the presence of an excess of unlabelled ACP. This material was probably aggregates resulting from the large amounts of protein (200 pg) loaded in each lane. A second band, of about 20 kDa protein was presumably iodinate ACPpAPA (lane 6). The third band, which has an apparent molecular weight of 100 kDa, was not present in controls: when the cross-linking was performed in the presence of an excess of unlabelled ACP (lane 21, or without photoactivating light (data not shown). This band probably corresponds to GPAT (83 kDa) which has been cross-linked to ACP. In similar experiments with vesicles from wild type E. cofi, which contains much less (eight-fold) GPAT, this 100 kDa band was not detected. The same experiments were conducted with [‘251]ACP modified with NHS-ASA. The results are presented in Fig. 3. Radioactive cross-linked complex with a molecular weight of about 100 kDa was again detected. ACP modified with NHS-ASA resulted in significant non-specific binding. This non-specific binding was not observed when the pAPA reagent was used. Photoaffinity label&g with [3HIACP-pAPA. [ 3H]ACP with a specific activity of 3-4 TBq/mmole was prepared. Unlike iodination, this procedure allowed the
123456
46~
.:
Fig. 2. Autoradiograph of iodinated ACP-pAPA cross-linked on ORV30/pWW20 membrane vesicles after SDS PAGE. Lane 1: molecular weight markers (kDa); Lane 2: ORV30/pWW20 membrane vesicles after 10 min irradiation in the presence of 0.5 PM of iodinated ACP-pAPA and 50 PM of unlabelled ACP-pAPA; Lane 3: ORV30/pWW20 membrane vesicles after 10 min irradiation in the presence of 0.5 PM of iodinated ACP-pAPA; Lane 4: ORV30/pWW20 membrane vesicles after 20 min irradiation in the presence of 0.5 PM of iodinated ACP-pAPA; Lane 5: this lane represents the migration of GPAT as visualized after coomassie staining. Lane 6: iodinated ACP-pAPA.
123 !/
4 j.
‘,. ;
:,.
.:
”
,,
4TOP
200 t;r 46304 FRONT
Fig. 3. Autoradiograph of iodinated ACP-NHSASA cross-lurked on ORV30/pWW20 membrane vesicles after SDS PAGE. Lane 1: This lane represents the migration of GPAT as visualized after coomassie staining; Lane 2: ORV30/pWW20 membrane vesicles after 10 min irradiation in the presence of 0.5 I.LM of iodinated ACP-NHSASA; Lane 3: ORV30/pWW20 membrane vesicles after IO min irradiation in the presence of 0.5 PM of iodinated ACP-NHSASA and 50 PM of unlabelled ACP-NHSASA; Lane 4: iodinated ACP-NHSASA.
synthesis of an L3H]ACP chemically identical to unlabelled ACP. Therefore the affinity of 13HlACP for its membrane receptors was not modified by the label. [ 3H]ACP was modified with pAPA and cross-linking was done as described above. Because it is difficult to obtain large amounts of [3H]ACP, it was used at a concentration of 0.1 PM instead of 0.5 PM. Autoradiography is less sensitive to tritium than to [‘25I] and therefore semi preparative gel electrophoresis was used to analyse the results of cross-linking: 1 mg of membrane protein was loaded onto a cylindrical polyacrylamide gel (9 mm diameter and 15 cm length). After electrophoresis, gels were sliced into 2 mm slices which were placed in scintillation vials and counted. Fig. 4A shows the radioactive profile after electrophoresis of FB8 (wild type) membrane vesicles after irradiation in the presence of L3HlACP modified with pAPA, and Fig. 4B shows that of 0RV30 (rich in GPAT) membrane vesicles. In each case, two major peaks can be distinguished. The first one, which migrates like a protein of 20 kDa is free [3H]ACP. The second peak has an apparent molecular weight of 100 kDa. This peak presumably represents a specific covalent attachment between ACP and a membrane receptor because it was absent when the experiment was done in the presence of an excess of unlabelled ACP or in the absence of irradiation. These results are similar to those obtained by autoradiography with iodo ACP
0
10
20
30
40
50
60
A
FRACTIONS 250
200
150
100
50
0 0
10
20
30
FRACTIONS
40
50
60
B
Fig. 4. Cross-linking of [‘HIACP-pAPA on E. coli membrane vesicles. 0.1 PM of [3H]ACP-pAPA was added to membrane vesicles, and the sample irradiated for 10 min to obtain cross-linking. Membrane proteins were separated on a semi preparative cylindrical l&1.5% polyacrylamide gradient gel. Radioactivity was detected after cutting the gel into 2 mm slices as described in Materials and Methods. A control experiment in the presence of an excess (SO PM) of unlabelled ACP was performed. (A) Represents the radioactive profile obtained with wild type (FB8) vesicles after correction from the control experiment. (B) Represents the radioactive profile obtained with ORV30/pWW20 vesicles after correction from the control experiment. The migration of molecular weight markers is represented on the top of each graph.
(Figs. 2 and 3). However, this method also detected the 100 kDa band in a wild type strain of E. coli, expressing low levels of GPAT. The intensity of the 100 kDa peak is seven times higher in the ORV30/pWW20 samples than in the wild type samples. This factor of seven is in good agreement with the increase of activity and with the increase of ACP binding observed in this strain [9,14]. This is consistent with ACP being crosslinked to GPAT. Discussion ACP interacts with many proteins, both soluble and membrane bound [.5]. The nature of these interactions are not understood and a general study of this problem
could be very interesting to elucidate how ACP interacts with such a wide variety of proteins. The method of choice for studying these interactions is the use of cross-linking reagents to link the receptors with their ligand. This method is now very attractive because of the development of many photoreactive bifunctional reagents which facilitate specific cross-linking. In this study we used two photosensitive reagents: pAPA and NHS-ASA. These two reagents bind stoichiometrically to ACP and the biological activity of the protein was not altered. Thus. these two modified ACPs arc very useful for studying the interaction of ACP with proteins. Several methods were used to detect the complexes between ACP and its receptors. First. the complexes between ACP and its receptors were detected with anti ACP antibodies after Western blotting. Unfortunately, in this case, the use of anti ACP antibodies was not satisfactory. The sensitivity of the detection was not sufficient to detect the few cross-linked complexes ohtained after irradiation of membrane vesicles. Moreover, we confirm the results obtained by Worsham et al. [32] indicating that ACP forms oligomers even at low concentration and after SDS denaturation. These structures which were observed on Western blots could hide minor specific cross-linking. The second approach used was to prepare radioactive ACP of high specific activity. In this case the covalent complexes were detected by analysing the radioactive profile after SDS PAGE. Radioiodination of proteins gives high specific activity but, unfortunately, ACP is very sensitive to the oxidizing agents used in the iodination reaction. Consequently [ ““IIACP of relatively high specific activity lost its functionality. This inactivation could be the result of a dimerisation of ACP or more probably, oxidation of the single methionine to methionine sulfone. To avoid this alteration, [“HIACP with a specific activity of 3 or 4 TBq/mmole was prepared. In this third approach, the biological activity of the protein was not altered and the high specific activity allowed reasonable detection of ACP and its covalent complexes. We demonstrated photochemical cross-linking bctween ACP and one E. cofi membrane protein. Our data suggest strongly that this protein is the GPAT. These results agree with those we recently obtained showing that GPAT was an ACP receptor 191. This cross-linkage is specific because it is ( 1) inhibited by unlabelled ACP, (2) dependent on activating light and (3) significantly higher on GPAT-rich membrane vesicles than wild type vesicles. The molecular weight of the cross-linked GPATACP complex was estimated to be around 100 kDa. The molecular weight of GPAT is around 83 kDa [14] and ACP 9 kDa [5]. Nevertheless, it is difficult to predict the number of molecules of ACP which are
197 bound to GPAT because in SDS electrophoresis ACP migrates as a 23 kDa protein [3,35]. E. coli membranes contain other membrane proteins which can bind ACP. 2-acyl GPE binds ACP in vitro [7] and the glucosyltransferase, implicated in the MD0 biosynthesis, is also believe to bind ACP [3,231. These proteins are probably present in lower concentrations than GPAT, which is probably the major ACP binding protein in E. co/i membranes (60-80% of the total ACP binding) as indicated by binding experiments of ACP to E. co/i membrane [6,9] and by the enzymatic activity of membranes from different strains of E. cofi [141. These results underscore the difficulty of covalently labelling the ACP membrane protein receptors because of their low abundance. Appropriate spacing and chemical accessibility of the reactive groups on both receptor and ligand are required for successful crosslinking. These conditions are difficult to obtain in cases of protein-protein interactions and generally a very low yield of cross-links (around 1%) is obtained. Nevertheless, this covalent linking should provide a powerful tool for the future study of the structure and function of GPAT in relation to phospholipid biosynthesis and in particular for the analysis of the ACP binding site. This could be done after proteolysis of the ACP-GPAT complex. This technique could also be very useful for studying the interaction of ACP with the soluble enzymes implicated in the biosynthesis of fatty acids in E. coli. A comparative study of all these ACP binding proteins and more particularly, the regions implicated in ACP could help elucidate the interaction mechanism. It would be particularly interesting to know if this interaction between ACP and the ACP-binding proteins involves amphiphilic helices as proposed by Ernst-Fonberg et al. [12], thus providing an explanation for the mechanisms of recognition for its different receptors. Acknowledgements
We thank Dr. J.P. Tenu for helpful assistance in photolabelling techniques and for his continued interest in this work.
4 Therisod,
6 7 8 9 IO 11 I2 13 I4 15 16 17 18 19 20 21 22 23 24 25 26 27
28
29 30 31 32 33
References 1 Vagelos, P.R. (1973) in The Enzymes, Vol. 8 (Bayer, P.D., ed.), pp. 155-199, Academic Press, New York. 2 Ernst-Fonberg, M.L. (1986) Plant Physiol 82, 978-984. 3 Thtrisod, H., Weissborn, A.C. and Kennedy, E.P. (1986) Proc. Natl. Acad. Sci. USA 83, 7236-7240.
H. and
Kennedy,
USA 84, 8235-8238. 5 Cronan, J.E. and Rock,
34 35
E.P. (1987) C.O.
(1987)
Proc.
Natl.
in Escherichia
Acad.
Sci.
coli and
Salmonella typhimurium Cellular and Molecular Biology I: 31-55, (Neidhardt, F.C., ed.) pp. 474-497, American Society for Microbiology, Washington. Bayan, N. and Therisod, H. (1989) FEBS Lett. 253, 155-159. Cooper, C.L., Hsu. L., Jackowski, S. and Rock, C.O. (1989) J. Biol. Chem. 264, 7384-7389. Bell, R.M., Mavis, R.D.. Osborn, M.J. and Vagelos. P.R. (1971) Biochem. Biophys. Acta 249, 628-635. Bayan N. and ThCrisod, H. (1989) FEBS Lett. 255. 330-334. Krestsinger, R.H. (1980) CRC Cr. Rev. Biochem. 8, 119-174. Holak, T.A., Nilges. M., Prestegard, J.H.. Gronenborn, A.M. and Clore. G.M. (1988) Eur. J. Biochem. 175, 9-15. Ernst-Fonberg, M.L.. Tucker, M.M. and Fonberg, I.B. (1987) FEBS Lett. 215, 261-265. Argos, P. (1977) Biochemistry. 16, 665-672. Wilkison, W.O., Walsh, J.P., Corless, J.M. and Bell, R.M. (1986) J. Biol. Chem. 261, 9951-9958. Rock, C.O. and Cronan, J.E. (1981) Methods Enzymol. 71, 341351. Rock, C.O. and Cronan, J.E. (1980) Anal. Biochem. 102,362-364. Fraker, P.J. and Speck Jr., J.C. (1978) Biochem. Biophys. Res. Commun. 80, 849-857. Markwell, M.A.K. (1983) Anal. Biochem. 125, 427-432. Thorell, J.I. and Johansson, B.G. (1971) Biochim. Biophys. Acta 251, 363-368. Ernst-Fonberg, M.L., Schongalla, A.W. and Walker, T. (1977) Arch. Biochem. Biophys. 178, 166-173. Jackowski, S. and Rock, CO. (1983) J. Biol. Chem. 258, 1518615191. Green, P.R., Merrill, A.H. and Bell, R.M. (1981) J. Biol. Chem. 256, 11151-11159. Weissborn, A.C. and Kennedy, E.P. (1984) J. Biol. Chem. 259, 12644-12651. Lowry, O.H., Rosebrough, N.J., Farr. A.L. and Randall, R.J. (1951) J. Biol. Chem. 193, 265-275. Munkres, K.D. and Richards, F.M. (1965) Arch. Biochim. Biophys. 109, 466-479. Laemmli, U.K. (1970) Nature 227, 680-685. Pierce, R.J. and Capron, A. (1987) in Techniques Biochimiques d’Analyse des Antigenes de Membrane (Charron, D., ed) pp. 43-54, INSERM, France. Chirio, M.C., Brithes, D., Napias, C., Grandier-Vazeille, X., Rakotomanana, F. and Chevallier, J. (1990) Eur. J. Biochem. 194, 293-299. Hixson, H.H. and Hixson, S.S. (1975) Biochemistry 14,4251-4254. Moreland, R.B. and Dockter, M.E. (1981) Biochim. Biophys. Res. Commun. 99, 339-346. Ji, T.H. and Ji, I. (1982) Anal. Biochem. 121, 286-289. Worsham, L.M.S., Tucker, M. and Ernst-Fonberg, M.L. (1990) Biochem. Biophys. Acta 1043, 198-202. Ohlrogge, J.B., Kuhn, D.N. and Stumpf, P.K. (1979) Proc. Natl. Acad. Sci. USA 76, 1194-1198. Bolton, A.E. and Hunter, W.N. (1973) Biochem. J. 133, 529-539. Rock. C.O. and Cronan, J.E. (1979) J. Biol. Chem. 254, 97789785.