32
BBALIP
Biochimica et Biophysics
Acta, 1046 (1990) 32-39 Elsevier
53468
Membrane topology of epoxide hydrolase John A. Craft ‘, Susan Baird 2, Michelle Lamont
’
and Brian Burchell 2
’ Department of Biological Sciences, Glasgow College, Glasgow and ’ Department of Biochemical Medicine, Ninewells Hospital Medical School, The Universitv, Dundee (U.K.)
(Revised
Key words:
(Received 7 December 1989) manuscript received 17 April 1990)
Membrane
topology:
Epoxide
hydrolase
The amino acid sequences of epoxide hydrolase from rat, rabbit and human have been subjected to hydropathy analysis and a novel model for the membrane topology of this enzyme is presented. The enzyme would appear to be retained in microsomal membranes by a single transmembrane segment located at the N-terminus and the majority (%%) of the protein is exposed at the cytosolic membrane surface. This model is significantly different from a scheme suggested by analysis of the rat enzyme alone which proposed six transmembrane domains (Porter et al. (1988) Arch. Biochem. Biophys. 248, 121-129). Experiments with rat microsomal membranes were conducted to distinguish between the two models and used proteolytic enzymes and non-permeant chemical probes. Epoxide hydrolase of intact and permeabilised membranes was resistant to digestion by a number of proteinases. However, this is likely to be related to a compact fold of the protein rather than membrane association since purified, delipidated enzyme preparations were also resistant to proteolysis. While the use of proteinases did not provide useful membrane topological information, experiments with the fluorescent probe, 3-azido-2,7+aphthalenedisulphonate strongly support the view that the majority of the protein is indeed exposed at the cytosolic surface of the membranes. The analysis illustrates the caution which must be employed in the formulation of topological models based on hydropathy plots alone and the value of considering homologous proteins.
Introduction The topology of many membrane proteins has been investigated with chemical, immunological and proteolytic probes using intact and permeabilized preparations (discussed in Ref. 1). The approach demands that the probe does not penetrate the intact membrane bilayer and that the target protein has available and susceptible groups on the exposed membrane surface. This type of strategem has been exmployed in studies of the topology of the major xenobiotic metabolising epoxide hydrolase of microsomal membranes. The enhancement of enzyme activity by detergent [2] and the resistance of the enzyme of intact membranes to inhibition by diazobenzenesulphonate and proteinases [3] led to sugges-
Abbreviations: ANDS, 3-azido-2,7-naphthalenedisulphonate; DOC, deoxycholate; SDS-PAGE, sodium dodecyl sulphate polyacrylamide gel electrophoresis. Correspondence: lege, Cowcaddens 0005-2760/90/$03.50
J.A. Craft, Dept. Biological Sciences, Road, Glasgow G4 OBA, U.K. 0 1990 Elsevier Science Publishers
Glasgow
Col-
B.V. (Biomedical
tions that the catalytic site is buried in the membrane. It was inferred that other domains of the protein must extend from the cytoplasmic leaflet of the bilayer since antibodies directed against the enzyme react with intact membranes [4,5]. The results of labelling with “‘1 in the presence of lactoperoxidase suggest that 20-25% of the protein may be exposed at the cytoplasmic membrane surface but permeabilization of the membrane did not expose further groups which could be iodinated [6]. The results of such studies can only give a crude view of the topology of the enzyme and this may be compromised by artifacts of the membrane probes [l]. The availability of amino acid sequences deduced from nucleotide sequences of cDNAs has led to the development of algorithms for the prediction of transmembrane helices and the topology of integral membrane proteins [7,8]. Recently amino acid sequences of the major microsomal epoxide hydrolase have been deduced for the rat [9] and human [lo] enzyme and the rabbit sequence has been determined directly [ll]. A possible topological structure for the enzyme has been suggested based upon the rat sequence and this proposes six transmembrane helices, large extra-membranous domains on both Division)
33 sides of the bilayer and that both amino- and carboxyltermini are located on the cytosolic surface (Ref. 9 and Fig. 2A). This model was based solely on hydropathy analysis, gave no reason for the suggested orientation and no additional biochemical evidence was presented to validate the model. We have analysed the sequence data of the enzyme from three species and suggest a different membrane topology for epoxide hydrolase and support this with separate experimental data. Materials and Methods Materials
3-Azido-2,7_naphthalenedisulphonate (ANDS) was synthesised from 3-amino-2,7_naphthalenedisulphonate (Aldrich Chemical, Poole, U.K.) without recrystallization but otherwise using the method described [12]. l-[l-‘4C]Naphthol (58 mCi/mmol) and [7(n)-3H]styrene oxide (161 mCi/mmol) were purchased from Amersham International. Glucose 6-phosphate, UDP-glucuronic acid, Lubrol PX, carboxypeptidase Y, trypsin, trypsin inhibitor, chymotrypsin, subtilisin and proteinase K were purchased from Sigma (Poole, U.K.). All other reagents were of the Analar grade or the highest grade available. Epoxide hydrolase, fraction CMB2, was purified from rat and used to prepare a monospecific antibody in rabbit [13]. A monospecific antibody to bilirubin UDP-glucuronosyltransferase was produced by immunisation of rabbits with enzyme purified from clofibrate-treated rats as described [14]. Phosphatidylinositol-specific phospholipase C was generously supplied by Dr. M. Low. Membrane preparation
Microsomal membranes were prepared from the livers of male Wistar rats (200 g) in either STKME (0.25 M sucrose containing Tris-HCl (0.05 M), KC1 (25 mM), MgCl, (5 mM) and EDTA (1 mM), pH 7.5 at 4°C) or TKE (Tris-HCl(O.05 M), KC1 (11.5% w/v) and EDTA (1 mM), pH 7.5 at 4°C) [15]. The membranes were resuspended in STKME and were used immediately. Enzymic modification of epoxide hydrolase
Membranes (5 mg/ml) in STKME (pH 7.5) at 30 o C were incubated with trypsin (50 pg/mg membrane protein) for 45 min at 30°C in the presence or absence of various concentrations of deoxycholate [16]. Proteinase activity was inhibited by the addition of trypsin inhibitor (2 pg/pg trypsin) and the membranes were cooled on ice. Control membranes were similarly incubated but in the absence of trypsin. Membranes (5 mg/ml) in STKME (pH 7.4) at 37” C, were also treated with phosphatidylinositol-specific phospholipase C (PI-PLC) (3.25 U/ml: 1 U = 1 pmol/min using [3H]phosphatidylinositol as substrate) for 60 min at 37 o C. The membranes were then harvested by centrifugation at 120000
x g for 60 min and resuspended in STKME to the original volume. The supematants were passed through a membrane filter (0.2 pm) at low pressure to remove small membrane vesicles which do not pellet under the conditions of centrifugation used. Chemical modification of epoxide hydrolase
3-Azido-2,7_naphthalenedisulphonate (ANDS) was used as a membrane-impermeable, photo-activatable, fluorescent membrane probe [12]. Incubation of the membranes with ANDS and subsequent manipulations were carried out in a dark room with safe-light illumination only. Membranes (5 mg/ml) or purified enzyme (360 pg/ml) in sodium phosphate buffer (20 mM, pH 7.4) were irradiated with ultraviolet light (366 nm) at a distance of 5 cm from the light source in the presence of ANDS (1 mM) for 10 min [12] in the absence or presence of DOC (0.05 or 0.5%, w/v). After this time unreacted ANDS was inactivated by irradiation at 366 nm in the presence of glycine (final concentration 10 mM, pH 7.4), for a further 10 min. Samples were used for the determination of styrene-oxide hydrolase activity as below. Protein components of the labelled membranes were resolved by SDS-PAGE as below and the gels were photographed under ultraviolet light [12]. Aliquots of the ANDS-treated membranes were also treated with an equal volume of buffer A (sodium phosphate (0.05 M), EDTA (1 mM), Lubrol PX (0.18, w/v), glycerol (10%) and sodium azide (O.OOS%,w/v), pH 8.0) containing sodium cholate (l%, w/v) prior to quantitative recovery of epoxide hydrolase by immunoprecipitation with an epoxide hydrolase-specific antibody as previously described [15]. The immunoprecipitates were washed with buffer A containing sodium cholate (O.OSW,w/v) and finally dissolved in Tris-HCl (10 mM, pH 6.8 at 20 o C) containing SDS (10% w/v), glycerol (lo%, v/v) and 2-mercaptoethanol (5% v/v). In some experiments, immediately after labelling with ANDS, aliquots were treated with trypsin, as above, in the absence or presence of 0.5% DOC. Samples from these incubations were then immunoprecipitated with the specific anti-epoxide hydrolase antibody. Analytical procedures
Epoxide hydrolase content of control and treated samples was assessed by Western blotting. Membrane proteins were resolved by SDS-PAGE [17] prior to electrophoretic transfer to nitrocellulose filters (Schleicher and Schull) as described by Towbin et al. [18]. Epoxide hydrolase was visualised by immunostaining using a monospecific antibody to the enzyme [13] as described by Dornin et al. [19] using 4-chloro-l-naphthol as substrate. Epoxide hydrolase activity was determined using styrene-oxide as substrate [15]. Glucose6-phosphatase [16] and 1-naphthol-UDP-glucuronos-
34 yltransferase activities methods cited.
[18] were
determined
by
the
Results and Discussion Amino-acid sequences of the rat [9], human [lo] and rabbit [ll] microsomal epoxide hydrolases were analysed by the Goldman-Engelman-Stietz algorithm [8] for the prediction of transmembrane a-helices and the resulting free energy plots are shown in Fig. 1. In such plots, peaks with a free energy change of greater than +20 kcal/mol and consisting of about 20 amino acids are considered to represent putative transmembrane seg-
73
\E 5
ments [8]. Examination of Fig. 1 reveals that each plot contains only two peaks which meet these criteria. One such peak is present in each plot and is located at the N-terminus. Another peak is common to the plots of the rat and rabbit enzymes (Fig. 1A and B) and starts at residue 144 and 141, respectively. The other peak in the human plot which meets the criteria for a transmembrane segment starts at residue 245 (Fig. 1C). In the plots shown in Fig. 1, the computations were carried out with a scanning window of 20 amino acids. While this number is considered to be the minimum size necessary for a transmembrane a-helix it is possible that shorter helices may span the membrane [8]. How-
’ -10
2 -20 E-l c 2 -30 ” 2 b 6
-40 -50
: I=
-60 I
-70
0
I
I
100 First
I
200 residue
I
I 300
I
I
400
in window
30
4or
0 E \
I
r
C
10 o
0”
-10
5 8l
-20
c5 ”
-30
,” ki z
-40 -50
L I=
-60 I
0 First
residue
in window
I 100 First
I
I
I
200 residue
I 300
I
I 400
I
in window
Fig. 1. Free energy plots of sequences of epoxide hydrolase from rat (A), rabbit (B) and human (C). Free energy changes were calculated from the epoxide hydrolase sequences by the Goldman-Engelman-Steitz algorithm using a scanning window of 20 amino acids.
35 (A)
cl 265
310
368
250
326
354
t L, Lumen
(W cytoso1 #PI
i t
m
,
l
Il 20
1
1 L"rnM
+ NH3
Fig. 2. Schematic diagrams of the predicted membrane topology of epoxide hydrolase. Models for the possible membrane topology of epoxide hydrolase are presented as described by Porter et al. [9] (A) (adapted from 12) and as deduced from Fig. 1 and text (B). The diagrams show the putative transmembrane segments (black bars) and the numbers refer to the residues of the rat enzyme. Potential sites for trypsin action are shown (* ).
ever, recalculation of the free energy values using successive increases in the length of the scanning window from 10 to 19 amino acids did not lead to the appearance in the plots of additional, significant peaks. The amino acid sequences of epoxide hydrolase from the three species are highly homologous [9,10] and it seems unlikely that the membrane topology of these proteins will be different. Comparison of the hydropathy plots of the three species reveals only one common feature in the three proteins and suggests that epoxide hydrolase is anchored to the membrane bilayer by a single transmembrane helix located at the N-terminus. A schematic representation of the protein assuming this explanation is shown in Fig. 2 and is consistent with this protein being synthesised with an unprocessed Nterminal ‘signal’ [30]. The other peaks in the profiles which meet the criteria for putative transmembrane domains do not occur in the enzyme from each of the species and thus are unlikely to represent membrane spanning segments. The model of the topology of
epoxide hydrolase developed here differs significantly from that suggested by Porter et al. [9] (Fig. 2A) but such differences can be tested by biochemical experimentation. The six-helix model exposes significant proportions of the polypeptide on both luminal and cytosolid membrane surfaces, while if the enzyme is attached by a single transmembrane helix located at the N-terminus, no residues are found on the luminal surface and the majority of the protein is located on the cytosolic surface. These different structures would generate distinctive patterns of fragments following proteolysis and different reaction products would be formed with chemical probes. The topology experiments utilised intact microsomal membranes and membranes treated with deoxycholate (DOC) (0.05%1.0% w/v). DOC at 0.05% disrupts the membrane bilayer permeability barrier but without causing the dissociation of membrane components [16,21]. Higher concentrations of the detergent cause membrane solubilisation. The integrity of the permeability barrier of the intact membranes was verified by measurements of the activities of two enzymes known to be disposed on the luminal membrane surface. Glucose6-phosphatase activity of the membranes was shown to be resistant to tryptic-inactivation in the absence, but not the presence of DOC (O.OS%,w/v) (data not shown) [16]. The activity of UDP-glucuronosyltransferase of the membranes, using 1-naphthol as acceptor substrate was increased by a factor of 16.5 when measured in the presence of Lubrol PX or when membranes were pretreated with DOC (O.OSW),consistent with this enzyme being at least 94% latent in intact membranes [22]. 1-Naphthol-UDP-glucuronosyltransferase activity was also resistant to tryptic digest in the absence of DOC but was greatly reduced, down to 20% of control activity, when DOC (0.05% w/v) was present (data not shown). The effect of trypsin on epoxide hydrolase of membrane and purified enzyme preparations was investigated in the absence and presence of DOC. The trypsin-treated samples were analysed by Western blot using an epoxide hydrolase-specific antibody. Trypsin did not affect the migration or intensity of the immunostained epoxide hydrolase of native membranes (Fig. 3A) or of the purified enzyme (Fig. 3B) and this result was not altered by the presence of a low concentration of DOC (0.05%). In the presence of higher concentrations of DOC, both the membrane and purified preparation were hydrolysed by trypsin to give two strongly stained fragments with an M, estimated to be close to 32000 (Figs. 3A and B). At least two other fragments were observed with M, 29000 and 21000 but these appeared faintly on the immunoblots. Similar results were found when other proteolytic enzymes (chymotrypsin, subtilisin and proteinase K) replaced trypsin (data not shown). Epoxide hydrolase activity of
36 TABLE
I
Effect of trypsin and ANDS microsomal epoxide hydrolase
on siyrene-oxide
hydrolase activity of
Microsomal membranes were treated with trypsin or ANDS as described in Materials and Methods at the concentration of DOC indicated and styrene oxide hydrolase activity subsequently determined. The data shows mean f SD. for triplicate determination. The data is taken from one experiment but similar results were obtained with two other membrane preparations. Addition
None Trypsin ANDS
Styrene-oxide hydrolase DOC (% w/v)
activity (nmol/min
per mg);
0
0.05
0.5
4.99 f 0.29 4.74 It 0.28 4.30*0.15
4.62 + 0.41 4.57 f 0.21 3.12 f 0.60
8.44&0.30 2.78 f 0.38 2.28 f 0.26
the trypsin-treated membranes was also determined and the results are shown in Table I. Styrene oxide hydrolase activity of native membranes or membranes incubated in the presence of a low concentration of DOC (O.OS%, w/v) was not affected by trypsin. Addition of a high concentration of DOC (O.SW, w/v) increased styrene oxide hydrolase activity as previously observed [2] but enzyme activity was reduced by approx. 70% when these membranes were treated with trypsin. The observations on the effects of trypsin with membranes and purified enzyme suggest that susceptible bonds are not accessible to trypsin in either preparation. This could be a consequence of protein association with lipids of the membrane bilayer or in the case of the purified enzyme with micelles of lipid and detergent. This latter explanation is considered unlikely since the purification procedure yields a preparation with very low amounts of phospholipid and detergent [13]. The resistance of the protein to proteolysis is more likely to be the result of the protein being folded compactly and not a consequence of membrane association although this latter possibility cannot be formally discounted. It is only when high concentrations of DOC are added that a limited proteolysis occurs. This is in contrast to the effect of trypsin on purified epoxide hydrolase which has been heat- and SDS-denatured when many tryptic fragments are produced [23,24]. When DOC is added at concentrations above O.l%, the 52000 protein may be hydrolysed to two fragments with M, values of about 32000 and 21000. These may then be further hydrolysed; the 32000 fragment to yield a band on the immunoblot of similar staining intensity of only slightly lower M, and a faint band at 29000. The 21000 fragment may be extensively hydrolysed to products which do not appear on the blots. A possible explanation of the limited proteolysis found when high concentrations of detergent are added is that the detergent removes lipids which protect the enzyme in the intact membrane. This explanation is considered unlikely since the hy-
dropathy analysis provides no evidence for a transmembrane domain at the position indicated by the sizes of the tryptic fragments. It is considered more reasonable that the detergent induces in the protein a conformational change which exposes previously buried tryptictarget sites. The location of the carboxy-terminus of the enzyme has been investigated by incubation of membranes with carboxypeptidase Y in the absence and presence of DOC (0.05% and 0.5%, w/v) followed by immunoblot analysis. Treatment of the various membrane preparations with carboxypeptidase did not alter the electrophoretic mobility or immuno-staining of the enzyme or the subsequent susceptibility of the enzyme to tryptic digestion (data not shown). This provides further evidence that the enzyme is tightly folded and suggests that the carboxyl-terminus is not accessible to the enzymatic probe in any of the preparations tested. The carboxyl-terminus of the enzyme may be inaccessible if it were involved in attachment of the enzyme to the membrane via an inositol phospholipid [25,26]. While this possibility may be unlikely because of the hydrophilic nature of the carboxyl-terminus, purified enzyme preparations that have been extensivly delipidated, still contain residual phospholipid at least a part of which is phosphatidylinositol [24,27]. To assess the possibility that epoxide hydrolase is attached to membranes via an inositol phospholipid, membranes were treated with a phosphatidylinositol-specific phospholipase C. The membranes were then recovered by centrifugation and the resulting pellet and supernatant fractions analysed for epoxide hydrolase by Western blot and styrene oxide hydrolase activity. Samples of the purified enzyme preparation were also used as potential substrates for the PI-PLC. The immunoblot analysis (Fig. 4) showed no alteration of membrane epoxide hydrolase mobility during SDS-PAGE or decrease in intensity of the immunostain and no immunoreactive protein was recovered in the supematant fractions. The electrophoretic mobility of the purified enzyme was also unaltered by PI-PLC treatment. Styrene oxide hydrolase activity was undiminished in the treated membranes or purified enzyme and no enzyme activity was recovered in the membrane supematants (data not shown). Similar results were obtained whether the membranes had been prepared in 0.25 M STKME or in Tris/KCl or if a DOC-solubilised membrane sample was used rather than the purified enzyme. These observations suggest it is unlikely that inositol phospholipids are involved in the membrane binding of epoxide hydrolase. The topological model of epoxide hydrolase presented here places 96% of amino acid residues exposed on the cytosolic membrane face. This contrasts with the model suggested by Porter et al. (cf. Figs. 2A and B) in which only 33% of the residues are located on the cytosolic side of the membrane while a similar propor-
tion, 35% are located on the luminal membrane face. These differences should be distinguished by the use of membrane-impermeable probes and we have used the photo-activatable reagent 3-azido-2,7-naphthalenedisulphonate. Incubation of membranes with this reagent under ultraviolet illumination results in the labelling of many proteins whose fluorescence can be viewed after resolution by SDS-PAGE (not shown). To determine the permeability of the membranes to ANDS, the labelling of UDP-glucuronosyltransferase was investigated. Hydropathy analysis and biochemical experimentation of the glucuronosyltransferases suggests that these proteins are anchored in the membrane by a single transmembrane domain located near the carboxy-terminus [14]. The majority of the amino acids of these proteins are located on the luminal membrane surface and only a short segment including the carboxyl-terminus is exposed at the cytosolic membrane face. Intact membranes and membranes treated with 0.05% and 0.5% DOC were labelled with ANDS, the membranes solubilised and UDP-glucuronyltransferases immunoprecipitated with an antibody raised against bilirubin UDPglucuronosyltransferase. When intact membranes were used for fluorescence labelling, no fluorescence signal was observed in the precipitated antigen (data not shown). Only when the labelling reaction was carried out in the presence of DOC (O.OSW, w/v) did the resulting immunoprecipitate show detectable fluorescence and the extent of the signal was not increased by
Trypsin
0
+
+
m
0
0
0.05
PI-PLC
-
Fraction
P
S
+
+
P
S
+ EH
w
Fig. 4. Effects of PI-PLC on the electrophoretic mobility of epoxide hydrolase of membrane preparations and purified protein. Microsomal membranes, prepared in STKME, or purified epoxide hydrolase were incubated with ( + ) or without (- ) PI-PLC as described in Materials and Methods. The membranes were harvested and the pellets (P), supematant fractions (S) or purified enzyme (EH) electrophoresed in the presence of SDS and immunoblotted with an antibody specific for microsomal epoxide hydrolase.
the inclusion of DOC at a higher concentration (0.5%) (data not shown). The extent of ANDS labelling of epoxide hydrolase was investigated by SDS-PAGE resolution of im-
+
+
+
+
+
+
0.1
0.2
0.3
0.4
0.5
1.0
Trypsin
0
+
+
+
m
0
0
0.05
0.5
Fig. 3. Effects of trypsin on the electrophoretic mobility of epoxide hydrolase of membrane preparations (A) and purified protein (B). Microsomal membranes or purified epoxide hydrolase were treated with trypsin as described in Materials and Methods in the presence or absence of DOC at the concentrations (w/v) indicated. Incubation mixtures were electrophoresed in the presence of SDS and immunoblotted with an antibody specific for microsomal epoxide hydrolase.
38
DOC (?A) during ANDS labelling
DOC (%) during ANDS labelling
0
0
Posl-ANDS
-
+
trypsin
0.05
0.05
+ Membranes
0.5
0.5
+
0
0
+ Purified enzyme
Fig. 5. Fluorescence of ANDS-labelled, SDS-PAGE-resolved, immune-precipitated rat microsomal epoxide hydrolase. Rat hepatic microsomal membranes were labelled with 3-Azido-2,7-naphthalenedisulphonate (ANDS) by photoactivation with ultraviolet light in the absence or presence of deoxycholate (DOC) (0.05% or 0.58, w/v). The membranes were solubilised with cholate prior to immunoprecipitation with anti-rat-epoxide hydrolase and SDS-PAGE (A) or were treated with trypsin in the presence of DOC (0.5%) prior to immunoprecipitation and electrophoresis (B). Purified mEH1 was also labelled and electrophoresed without immunoprecipitation.
munoprecipitates. Epoxide hydrolase can be quantitatively recovered from the membrane preparations by immunoprecipitation and a single fluorescent band was observed (Figs. 5A and B). This band migrates slightly faster than purified enzyme which had been labelled with ANDS but not subjected to immunoprecipitation (Fig. 5B). The slightly higher mobility of the immunoprecipitated protein is probably due to the presence of high amounts of IgG heavy chain which has a similar molecular weight to epoxide hydrolase. Attempts to quantify the fluorescence of the immunoprecipitates by fluorospectrometry were not successful due to the persistent presence of low molecular weight fluorescent material. The fluorescence signal of the protein immunoprecipitated from intact membranes (Fig. 5A, channel 1) was not increased by addition of DOC at 0.05% (channel 2) or 0.5% (channel 3) to membranes during reaction with ANDS (Fig. 5A). When the ANDS labelled membranes were treated with trypsin in the presence of 0.5% DOC, only weakly fluorescent bands with M, about 31500, 27500 and 23000 were observed (Fig. 5B). The membranes were also assayed for styrene oxide hydrolase activity after treatment with ANDS and the results are shown in Table I. The pattern of results was similar to that found for trypsin treatment and the enzyme was relatively resistant to inhibition by ANDS in intact preparations or preparations treated with low DOC (0.05%). Enzyme activity was however, inhibited
by 70% when both ANDS and a high concentration DOC (0.5%) were added. These observations with ANDS suggest that the majority of the epoxide hydrolase molecule is accessible in intact membranes to this membrane-impermeant probe and thus is disposed on the cytosolic membrane surface (Fig. 2B). Membrane permeabilisation or solubilization does not significantly increase the extent of fluorescent labelling. This data is totally inconsistent with a model of epoxide hydrolase with six transmembrane domains and with similar, significant proportions of the enzyme amino acid residues exposed on the luminal and cytosolic membrane faces. Although the majority of the enzyme appears to be exposed at the cytosolic surface it must, none the less, be compactly folded since peptide bonds which are potentially susceptible to proteolysis, were found to be resistant to trypsin and other proteolytic enzymes in both membranes and purified protein. The carboxy-terminus of the protein also appears to be inaccessible and native epoxide hydrolase of membranes or purified preparations is not a substrate for carboxypeptidase Y. The protein can be a substrate for tryptic digest but only in the presence of a high concentration of DOC. This suggests that the detergent causes a conformational change which causes an increase of enzyme activity, renders the enzyme sensitive to inhibition by ANDS (Table I) and exposes a ‘hinge’ site to a limited proteolysis. These suggestions
39
are also consistent with the observation that the fluorescent label incorporated into the protein in native membranes is lost when the membranes are subsequently treated with trypsin in high DOC (Fig. 5B). The analysis of the epoxide hydrolase sequences illustrates the caution which must be employed in the interpretation of hydropathy plots and the value of considering homologous proteins. It is frequently forgotten that peaks in hydropathy profiles do not necessarily represent transmembrane domains and models for membrane topology should be validated by separate biochemical experimental data. It is interesting to note that a cytochrome P-450, originally thought to be anchored by multiple transmembrane segments based upon hydropathy analysis, is more likely to be attached by a single segment at the N-terminus [28,29]. Acknowledgements
We thank Drs. Macnab and Goldman for the computer software used for the predictive calculations and Dr. M. Low for gifts of phosphatidylinositol-specific phospholipase C. This work was supported in part by the Cancer Research Campaign and the Wellcome Trust. References Hargrave, P.A. (1986) in Techniques for the analysis of membrane proteins (Ragan, C.I. and Cherry, R.J., eds.), pp. 129-151, Chapman and Hall, London. Burchell, B., Bentley, P. and Oesch, F. (1976) B&him. Biophys. Acta 444, 531-538. Siedegard, J., Moron, M.S., Et&son, L.G. and De Pierre, J. (1978) Biochim. Biophys. Acta 543, 29-40. Oesch, F. and Bentley, P. (1976) Nature 259, 53-55. Levin, W., Thomas, P.E., Koreniowski, D., Jerina, D.M. and Lu, A.Y.H. (1978) Mol. Pharmacol. 14, 1107-1120. Siedegard, S., DePierre, J., Guenther, T.M. and Oesch, F. (1982) Acta Chem. Stand. B36, 555-557.
7 Kyte, J. and Doolitle, R.F. (1982) J. Mol. Biol. 157, 105-132. 8 Engleman, D.M., Stietz, T.A. and Goldman, A. (1986) Annu. Rev. Biophys. Biophys. Chem. 15, 321-353. 9 Porter, T.D., Beck, T.W. and Kasper, C.B. (1986) Arch. B&hem. Biophys. 248, 121-129. 10 Jackson, M.R., Craft, J.A. and Burchell, B. (1987) Nucleic Acid Res. 15, 7188. 11 Heinemarm, F.S. and Ozols, J. (1984) J. Biol. Chem. 259, 797-804. 12 Dockter, M.E. and Koseki, T. (1983) Biochemistry 22, 3954-3961. 13 Bulleid, N.J., Graham, A.B. and Craft, J.A. (1986) B&hem. J. 233, 607-611. 14 Shepherd, S.R.P., Baird, S.J., Hallinan, T. and Burchell, B. (1989) B&hem. J. 259, 617-620. 15 Bulleid, N.J. and Craft, J.A. (1984) Biochem. Pharmacol. 33, 1451-1457. 16 Cooper, M.B., Craft, J.A., Estall, M. and Rabin, B.R. (1980) Biochem. J. 190, 737-746. 17 Laemmli. U.K. (1970) Nature (London) 227, 680-685. 18 Towbin, H., Staehlin, T. and Gordon, J. (1979) Proc. Natl. Acad. Sci. USA 76, 4350-4354. 19 Domin. B.A., Serabjit-Singh, C.J. and Philpot, R.M. (1984) Anal. B&hem. 136, 390-396. 20 Otani, G., Abou-el-Makarem, M.M. and Bock, K.W. (1976) Biothem. Pharmacol. 25, 1293-1297. 21 Kreibich, G., Debey, P. and Sabatini, D.D. (1973) J. Cell. Biol. 56, 436-462. 22 Dutton, G.J. (1980) Glucuronidation of Drugs and Other Compounds, CRC Press, Boca Raton. 23 DuBois, G.C., Appella, E., Ryan, D.E., Jerina, D.M. and Levin, W. (1982) J. Biol. Chem. 257, 270X-2712. 24 Bulleid, N.J. (1985) PhD Thesis, CNAA, Glasgow College. 25 Low, M.G. (1987) B&hem. J. 244, l-13. 26 Low. M.G. and Saltiel, A.R. (1988) Science 239, 268-275. 27 Griffin, M.J. and Palakoderty, R.B. (1985) in Tight association between rat liver microsomal epoxide hydrolase and phosphatidylinositol in inositol and phosphoinositides: Metabolism and regulation (Bleasdale, J.E., Eichberg, J. and Hauser, G., eds.), p. _._ __ 669, Humana, Clifton. 28 De Lemos-Chiarandini, D., Frey, A.B., Sabatini, D.D. and Kreibich, G. (1987) J. Cell. Biol. 104, 209-219. 29 Vergeres, G., Winterhaler, K.H. and Richter, C. (1989) Biochemistry 328, 3650-3655. 30 Gonzalez, F.J. and Kaspar, C.B. (1980) B&hem. Biophys. Res. Commun. 93, 125441258.
BBALIP
Biochimica et Biophysics
Acta, 1046 (1990) 32-39 Elsevier
53468
Membrane topology of epoxide hydrolase John A. Craft ‘, Susan Baird 2, Michelle Lamont
’
and Brian Burchell 2
’ Department of Biological Sciences, Glasgow College, Glasgow and ’ Department of Biochemical Medicine, Ninewells Hospital Medical School, The Universitv, Dundee (U.K.)
(Revised
Key words:
(Received 7 December 1989) manuscript received 17 April 1990)
Membrane
topology:
Epoxide
hydrolase
The amino acid sequences of epoxide hydrolase from rat, rabbit and human have been subjected to hydropathy analysis and a novel model for the membrane topology of this enzyme is presented. The enzyme would appear to be retained in microsomal membranes by a single transmembrane segment located at the N-terminus and the majority (%%) of the protein is exposed at the cytosolic membrane surface. This model is significantly different from a scheme suggested by analysis of the rat enzyme alone which proposed six transmembrane domains (Porter et al. (1988) Arch. Biochem. Biophys. 248, 121-129). Experiments with rat microsomal membranes were conducted to distinguish between the two models and used proteolytic enzymes and non-permeant chemical probes. Epoxide hydrolase of intact and permeabilised membranes was resistant to digestion by a number of proteinases. However, this is likely to be related to a compact fold of the protein rather than membrane association since purified, delipidated enzyme preparations were also resistant to proteolysis. While the use of proteinases did not provide useful membrane topological information, experiments with the fluorescent probe, 3-azido-2,7+aphthalenedisulphonate strongly support the view that the majority of the protein is indeed exposed at the cytosolic surface of the membranes. The analysis illustrates the caution which must be employed in the formulation of topological models based on hydropathy plots alone and the value of considering homologous proteins.
Introduction The topology of many membrane proteins has been investigated with chemical, immunological and proteolytic probes using intact and permeabilized preparations (discussed in Ref. 1). The approach demands that the probe does not penetrate the intact membrane bilayer and that the target protein has available and susceptible groups on the exposed membrane surface. This type of strategem has been exmployed in studies of the topology of the major xenobiotic metabolising epoxide hydrolase of microsomal membranes. The enhancement of enzyme activity by detergent [2] and the resistance of the enzyme of intact membranes to inhibition by diazobenzenesulphonate and proteinases [3] led to sugges-
Abbreviations: ANDS, 3-azido-2,7-naphthalenedisulphonate; DOC, deoxycholate; SDS-PAGE, sodium dodecyl sulphate polyacrylamide gel electrophoresis. Correspondence: lege, Cowcaddens 0005-2760/90/$03.50
J.A. Craft, Dept. Biological Sciences, Road, Glasgow G4 OBA, U.K. 0 1990 Elsevier Science Publishers
Glasgow
Col-
B.V. (Biomedical
tions that the catalytic site is buried in the membrane. It was inferred that other domains of the protein must extend from the cytoplasmic leaflet of the bilayer since antibodies directed against the enzyme react with intact membranes [4,5]. The results of labelling with “‘1 in the presence of lactoperoxidase suggest that 20-25% of the protein may be exposed at the cytoplasmic membrane surface but permeabilization of the membrane did not expose further groups which could be iodinated [6]. The results of such studies can only give a crude view of the topology of the enzyme and this may be compromised by artifacts of the membrane probes [l]. The availability of amino acid sequences deduced from nucleotide sequences of cDNAs has led to the development of algorithms for the prediction of transmembrane helices and the topology of integral membrane proteins [7,8]. Recently amino acid sequences of the major microsomal epoxide hydrolase have been deduced for the rat [9] and human [lo] enzyme and the rabbit sequence has been determined directly [ll]. A possible topological structure for the enzyme has been suggested based upon the rat sequence and this proposes six transmembrane helices, large extra-membranous domains on both Division)
33 sides of the bilayer and that both amino- and carboxyltermini are located on the cytosolic surface (Ref. 9 and Fig. 2A). This model was based solely on hydropathy analysis, gave no reason for the suggested orientation and no additional biochemical evidence was presented to validate the model. We have analysed the sequence data of the enzyme from three species and suggest a different membrane topology for epoxide hydrolase and support this with separate experimental data. Materials and Methods Materials
3-Azido-2,7_naphthalenedisulphonate (ANDS) was synthesised from 3-amino-2,7_naphthalenedisulphonate (Aldrich Chemical, Poole, U.K.) without recrystallization but otherwise using the method described [12]. l-[l-‘4C]Naphthol (58 mCi/mmol) and [7(n)-3H]styrene oxide (161 mCi/mmol) were purchased from Amersham International. Glucose 6-phosphate, UDP-glucuronic acid, Lubrol PX, carboxypeptidase Y, trypsin, trypsin inhibitor, chymotrypsin, subtilisin and proteinase K were purchased from Sigma (Poole, U.K.). All other reagents were of the Analar grade or the highest grade available. Epoxide hydrolase, fraction CMB2, was purified from rat and used to prepare a monospecific antibody in rabbit [13]. A monospecific antibody to bilirubin UDP-glucuronosyltransferase was produced by immunisation of rabbits with enzyme purified from clofibrate-treated rats as described [14]. Phosphatidylinositol-specific phospholipase C was generously supplied by Dr. M. Low. Membrane preparation
Microsomal membranes were prepared from the livers of male Wistar rats (200 g) in either STKME (0.25 M sucrose containing Tris-HCl (0.05 M), KC1 (25 mM), MgCl, (5 mM) and EDTA (1 mM), pH 7.5 at 4°C) or TKE (Tris-HCl(O.05 M), KC1 (11.5% w/v) and EDTA (1 mM), pH 7.5 at 4°C) [15]. The membranes were resuspended in STKME and were used immediately. Enzymic modification of epoxide hydrolase
Membranes (5 mg/ml) in STKME (pH 7.5) at 30 o C were incubated with trypsin (50 pg/mg membrane protein) for 45 min at 30°C in the presence or absence of various concentrations of deoxycholate [16]. Proteinase activity was inhibited by the addition of trypsin inhibitor (2 pg/pg trypsin) and the membranes were cooled on ice. Control membranes were similarly incubated but in the absence of trypsin. Membranes (5 mg/ml) in STKME (pH 7.4) at 37” C, were also treated with phosphatidylinositol-specific phospholipase C (PI-PLC) (3.25 U/ml: 1 U = 1 pmol/min using [3H]phosphatidylinositol as substrate) for 60 min at 37 o C. The membranes were then harvested by centrifugation at 120000
x g for 60 min and resuspended in STKME to the original volume. The supematants were passed through a membrane filter (0.2 pm) at low pressure to remove small membrane vesicles which do not pellet under the conditions of centrifugation used. Chemical modification of epoxide hydrolase
3-Azido-2,7_naphthalenedisulphonate (ANDS) was used as a membrane-impermeable, photo-activatable, fluorescent membrane probe [12]. Incubation of the membranes with ANDS and subsequent manipulations were carried out in a dark room with safe-light illumination only. Membranes (5 mg/ml) or purified enzyme (360 pg/ml) in sodium phosphate buffer (20 mM, pH 7.4) were irradiated with ultraviolet light (366 nm) at a distance of 5 cm from the light source in the presence of ANDS (1 mM) for 10 min [12] in the absence or presence of DOC (0.05 or 0.5%, w/v). After this time unreacted ANDS was inactivated by irradiation at 366 nm in the presence of glycine (final concentration 10 mM, pH 7.4), for a further 10 min. Samples were used for the determination of styrene-oxide hydrolase activity as below. Protein components of the labelled membranes were resolved by SDS-PAGE as below and the gels were photographed under ultraviolet light [12]. Aliquots of the ANDS-treated membranes were also treated with an equal volume of buffer A (sodium phosphate (0.05 M), EDTA (1 mM), Lubrol PX (0.18, w/v), glycerol (10%) and sodium azide (O.OOS%,w/v), pH 8.0) containing sodium cholate (l%, w/v) prior to quantitative recovery of epoxide hydrolase by immunoprecipitation with an epoxide hydrolase-specific antibody as previously described [15]. The immunoprecipitates were washed with buffer A containing sodium cholate (O.OSW,w/v) and finally dissolved in Tris-HCl (10 mM, pH 6.8 at 20 o C) containing SDS (10% w/v), glycerol (lo%, v/v) and 2-mercaptoethanol (5% v/v). In some experiments, immediately after labelling with ANDS, aliquots were treated with trypsin, as above, in the absence or presence of 0.5% DOC. Samples from these incubations were then immunoprecipitated with the specific anti-epoxide hydrolase antibody. Analytical procedures
Epoxide hydrolase content of control and treated samples was assessed by Western blotting. Membrane proteins were resolved by SDS-PAGE [17] prior to electrophoretic transfer to nitrocellulose filters (Schleicher and Schull) as described by Towbin et al. [18]. Epoxide hydrolase was visualised by immunostaining using a monospecific antibody to the enzyme [13] as described by Dornin et al. [19] using 4-chloro-l-naphthol as substrate. Epoxide hydrolase activity was determined using styrene-oxide as substrate [15]. Glucose6-phosphatase [16] and 1-naphthol-UDP-glucuronos-
34 yltransferase activities methods cited.
[18] were
determined
by
the
Results and Discussion Amino-acid sequences of the rat [9], human [lo] and rabbit [ll] microsomal epoxide hydrolases were analysed by the Goldman-Engelman-Stietz algorithm [8] for the prediction of transmembrane a-helices and the resulting free energy plots are shown in Fig. 1. In such plots, peaks with a free energy change of greater than +20 kcal/mol and consisting of about 20 amino acids are considered to represent putative transmembrane seg-
73
\E 5
ments [8]. Examination of Fig. 1 reveals that each plot contains only two peaks which meet these criteria. One such peak is present in each plot and is located at the N-terminus. Another peak is common to the plots of the rat and rabbit enzymes (Fig. 1A and B) and starts at residue 144 and 141, respectively. The other peak in the human plot which meets the criteria for a transmembrane segment starts at residue 245 (Fig. 1C). In the plots shown in Fig. 1, the computations were carried out with a scanning window of 20 amino acids. While this number is considered to be the minimum size necessary for a transmembrane a-helix it is possible that shorter helices may span the membrane [8]. How-
’ -10
2 -20 E-l c 2 -30 ” 2 b 6
-40 -50
: I=
-60 I
-70
0
I
I
100 First
I
200 residue
I
I 300
I
I
400
in window
30
4or
0 E \
I
r
C
10 o
0”
-10
5 8l
-20
c5 ”
-30
,” ki z
-40 -50
L I=
-60 I
0 First
residue
in window
I 100 First
I
I
I
200 residue
I 300
I
I 400
I
in window
Fig. 1. Free energy plots of sequences of epoxide hydrolase from rat (A), rabbit (B) and human (C). Free energy changes were calculated from the epoxide hydrolase sequences by the Goldman-Engelman-Steitz algorithm using a scanning window of 20 amino acids.
35 (A)
cl 265
310
368
250
326
354
t L, Lumen
(W cytoso1 #PI
i t
m
,
l
Il 20
1
1 L"rnM
+ NH3
Fig. 2. Schematic diagrams of the predicted membrane topology of epoxide hydrolase. Models for the possible membrane topology of epoxide hydrolase are presented as described by Porter et al. [9] (A) (adapted from 12) and as deduced from Fig. 1 and text (B). The diagrams show the putative transmembrane segments (black bars) and the numbers refer to the residues of the rat enzyme. Potential sites for trypsin action are shown (* ).
ever, recalculation of the free energy values using successive increases in the length of the scanning window from 10 to 19 amino acids did not lead to the appearance in the plots of additional, significant peaks. The amino acid sequences of epoxide hydrolase from the three species are highly homologous [9,10] and it seems unlikely that the membrane topology of these proteins will be different. Comparison of the hydropathy plots of the three species reveals only one common feature in the three proteins and suggests that epoxide hydrolase is anchored to the membrane bilayer by a single transmembrane helix located at the N-terminus. A schematic representation of the protein assuming this explanation is shown in Fig. 2 and is consistent with this protein being synthesised with an unprocessed Nterminal ‘signal’ [30]. The other peaks in the profiles which meet the criteria for putative transmembrane domains do not occur in the enzyme from each of the species and thus are unlikely to represent membrane spanning segments. The model of the topology of
epoxide hydrolase developed here differs significantly from that suggested by Porter et al. [9] (Fig. 2A) but such differences can be tested by biochemical experimentation. The six-helix model exposes significant proportions of the polypeptide on both luminal and cytosolid membrane surfaces, while if the enzyme is attached by a single transmembrane helix located at the N-terminus, no residues are found on the luminal surface and the majority of the protein is located on the cytosolic surface. These different structures would generate distinctive patterns of fragments following proteolysis and different reaction products would be formed with chemical probes. The topology experiments utilised intact microsomal membranes and membranes treated with deoxycholate (DOC) (0.05%1.0% w/v). DOC at 0.05% disrupts the membrane bilayer permeability barrier but without causing the dissociation of membrane components [16,21]. Higher concentrations of the detergent cause membrane solubilisation. The integrity of the permeability barrier of the intact membranes was verified by measurements of the activities of two enzymes known to be disposed on the luminal membrane surface. Glucose6-phosphatase activity of the membranes was shown to be resistant to tryptic-inactivation in the absence, but not the presence of DOC (O.OS%,w/v) (data not shown) [16]. The activity of UDP-glucuronosyltransferase of the membranes, using 1-naphthol as acceptor substrate was increased by a factor of 16.5 when measured in the presence of Lubrol PX or when membranes were pretreated with DOC (O.OSW),consistent with this enzyme being at least 94% latent in intact membranes [22]. 1-Naphthol-UDP-glucuronosyltransferase activity was also resistant to tryptic digest in the absence of DOC but was greatly reduced, down to 20% of control activity, when DOC (0.05% w/v) was present (data not shown). The effect of trypsin on epoxide hydrolase of membrane and purified enzyme preparations was investigated in the absence and presence of DOC. The trypsin-treated samples were analysed by Western blot using an epoxide hydrolase-specific antibody. Trypsin did not affect the migration or intensity of the immunostained epoxide hydrolase of native membranes (Fig. 3A) or of the purified enzyme (Fig. 3B) and this result was not altered by the presence of a low concentration of DOC (0.05%). In the presence of higher concentrations of DOC, both the membrane and purified preparation were hydrolysed by trypsin to give two strongly stained fragments with an M, estimated to be close to 32000 (Figs. 3A and B). At least two other fragments were observed with M, 29000 and 21000 but these appeared faintly on the immunoblots. Similar results were found when other proteolytic enzymes (chymotrypsin, subtilisin and proteinase K) replaced trypsin (data not shown). Epoxide hydrolase activity of
36 TABLE
I
Effect of trypsin and ANDS microsomal epoxide hydrolase
on siyrene-oxide
hydrolase activity of
Microsomal membranes were treated with trypsin or ANDS as described in Materials and Methods at the concentration of DOC indicated and styrene oxide hydrolase activity subsequently determined. The data shows mean f SD. for triplicate determination. The data is taken from one experiment but similar results were obtained with two other membrane preparations. Addition
None Trypsin ANDS
Styrene-oxide hydrolase DOC (% w/v)
activity (nmol/min
per mg);
0
0.05
0.5
4.99 f 0.29 4.74 It 0.28 4.30*0.15
4.62 + 0.41 4.57 f 0.21 3.12 f 0.60
8.44&0.30 2.78 f 0.38 2.28 f 0.26
the trypsin-treated membranes was also determined and the results are shown in Table I. Styrene oxide hydrolase activity of native membranes or membranes incubated in the presence of a low concentration of DOC (O.OS%, w/v) was not affected by trypsin. Addition of a high concentration of DOC (O.SW, w/v) increased styrene oxide hydrolase activity as previously observed [2] but enzyme activity was reduced by approx. 70% when these membranes were treated with trypsin. The observations on the effects of trypsin with membranes and purified enzyme suggest that susceptible bonds are not accessible to trypsin in either preparation. This could be a consequence of protein association with lipids of the membrane bilayer or in the case of the purified enzyme with micelles of lipid and detergent. This latter explanation is considered unlikely since the purification procedure yields a preparation with very low amounts of phospholipid and detergent [13]. The resistance of the protein to proteolysis is more likely to be the result of the protein being folded compactly and not a consequence of membrane association although this latter possibility cannot be formally discounted. It is only when high concentrations of DOC are added that a limited proteolysis occurs. This is in contrast to the effect of trypsin on purified epoxide hydrolase which has been heat- and SDS-denatured when many tryptic fragments are produced [23,24]. When DOC is added at concentrations above O.l%, the 52000 protein may be hydrolysed to two fragments with M, values of about 32000 and 21000. These may then be further hydrolysed; the 32000 fragment to yield a band on the immunoblot of similar staining intensity of only slightly lower M, and a faint band at 29000. The 21000 fragment may be extensively hydrolysed to products which do not appear on the blots. A possible explanation of the limited proteolysis found when high concentrations of detergent are added is that the detergent removes lipids which protect the enzyme in the intact membrane. This explanation is considered unlikely since the hy-
dropathy analysis provides no evidence for a transmembrane domain at the position indicated by the sizes of the tryptic fragments. It is considered more reasonable that the detergent induces in the protein a conformational change which exposes previously buried tryptictarget sites. The location of the carboxy-terminus of the enzyme has been investigated by incubation of membranes with carboxypeptidase Y in the absence and presence of DOC (0.05% and 0.5%, w/v) followed by immunoblot analysis. Treatment of the various membrane preparations with carboxypeptidase did not alter the electrophoretic mobility or immuno-staining of the enzyme or the subsequent susceptibility of the enzyme to tryptic digestion (data not shown). This provides further evidence that the enzyme is tightly folded and suggests that the carboxyl-terminus is not accessible to the enzymatic probe in any of the preparations tested. The carboxyl-terminus of the enzyme may be inaccessible if it were involved in attachment of the enzyme to the membrane via an inositol phospholipid [25,26]. While this possibility may be unlikely because of the hydrophilic nature of the carboxyl-terminus, purified enzyme preparations that have been extensivly delipidated, still contain residual phospholipid at least a part of which is phosphatidylinositol [24,27]. To assess the possibility that epoxide hydrolase is attached to membranes via an inositol phospholipid, membranes were treated with a phosphatidylinositol-specific phospholipase C. The membranes were then recovered by centrifugation and the resulting pellet and supernatant fractions analysed for epoxide hydrolase by Western blot and styrene oxide hydrolase activity. Samples of the purified enzyme preparation were also used as potential substrates for the PI-PLC. The immunoblot analysis (Fig. 4) showed no alteration of membrane epoxide hydrolase mobility during SDS-PAGE or decrease in intensity of the immunostain and no immunoreactive protein was recovered in the supematant fractions. The electrophoretic mobility of the purified enzyme was also unaltered by PI-PLC treatment. Styrene oxide hydrolase activity was undiminished in the treated membranes or purified enzyme and no enzyme activity was recovered in the membrane supematants (data not shown). Similar results were obtained whether the membranes had been prepared in 0.25 M STKME or in Tris/KCl or if a DOC-solubilised membrane sample was used rather than the purified enzyme. These observations suggest it is unlikely that inositol phospholipids are involved in the membrane binding of epoxide hydrolase. The topological model of epoxide hydrolase presented here places 96% of amino acid residues exposed on the cytosolic membrane face. This contrasts with the model suggested by Porter et al. (cf. Figs. 2A and B) in which only 33% of the residues are located on the cytosolic side of the membrane while a similar propor-
tion, 35% are located on the luminal membrane face. These differences should be distinguished by the use of membrane-impermeable probes and we have used the photo-activatable reagent 3-azido-2,7-naphthalenedisulphonate. Incubation of membranes with this reagent under ultraviolet illumination results in the labelling of many proteins whose fluorescence can be viewed after resolution by SDS-PAGE (not shown). To determine the permeability of the membranes to ANDS, the labelling of UDP-glucuronosyltransferase was investigated. Hydropathy analysis and biochemical experimentation of the glucuronosyltransferases suggests that these proteins are anchored in the membrane by a single transmembrane domain located near the carboxy-terminus [14]. The majority of the amino acids of these proteins are located on the luminal membrane surface and only a short segment including the carboxyl-terminus is exposed at the cytosolic membrane face. Intact membranes and membranes treated with 0.05% and 0.5% DOC were labelled with ANDS, the membranes solubilised and UDP-glucuronyltransferases immunoprecipitated with an antibody raised against bilirubin UDPglucuronosyltransferase. When intact membranes were used for fluorescence labelling, no fluorescence signal was observed in the precipitated antigen (data not shown). Only when the labelling reaction was carried out in the presence of DOC (O.OSW, w/v) did the resulting immunoprecipitate show detectable fluorescence and the extent of the signal was not increased by
Trypsin
0
+
+
m
0
0
0.05
PI-PLC
-
Fraction
P
S
+
+
P
S
+ EH
w
Fig. 4. Effects of PI-PLC on the electrophoretic mobility of epoxide hydrolase of membrane preparations and purified protein. Microsomal membranes, prepared in STKME, or purified epoxide hydrolase were incubated with ( + ) or without (- ) PI-PLC as described in Materials and Methods. The membranes were harvested and the pellets (P), supematant fractions (S) or purified enzyme (EH) electrophoresed in the presence of SDS and immunoblotted with an antibody specific for microsomal epoxide hydrolase.
the inclusion of DOC at a higher concentration (0.5%) (data not shown). The extent of ANDS labelling of epoxide hydrolase was investigated by SDS-PAGE resolution of im-
+
+
+
+
+
+
0.1
0.2
0.3
0.4
0.5
1.0
Trypsin
0
+
+
+
m
0
0
0.05
0.5
Fig. 3. Effects of trypsin on the electrophoretic mobility of epoxide hydrolase of membrane preparations (A) and purified protein (B). Microsomal membranes or purified epoxide hydrolase were treated with trypsin as described in Materials and Methods in the presence or absence of DOC at the concentrations (w/v) indicated. Incubation mixtures were electrophoresed in the presence of SDS and immunoblotted with an antibody specific for microsomal epoxide hydrolase.
38
DOC (?A) during ANDS labelling
DOC (%) during ANDS labelling
0
0
Posl-ANDS
-
+
trypsin
0.05
0.05
+ Membranes
0.5
0.5
+
0
0
+ Purified enzyme
Fig. 5. Fluorescence of ANDS-labelled, SDS-PAGE-resolved, immune-precipitated rat microsomal epoxide hydrolase. Rat hepatic microsomal membranes were labelled with 3-Azido-2,7-naphthalenedisulphonate (ANDS) by photoactivation with ultraviolet light in the absence or presence of deoxycholate (DOC) (0.05% or 0.58, w/v). The membranes were solubilised with cholate prior to immunoprecipitation with anti-rat-epoxide hydrolase and SDS-PAGE (A) or were treated with trypsin in the presence of DOC (0.5%) prior to immunoprecipitation and electrophoresis (B). Purified mEH1 was also labelled and electrophoresed without immunoprecipitation.
munoprecipitates. Epoxide hydrolase can be quantitatively recovered from the membrane preparations by immunoprecipitation and a single fluorescent band was observed (Figs. 5A and B). This band migrates slightly faster than purified enzyme which had been labelled with ANDS but not subjected to immunoprecipitation (Fig. 5B). The slightly higher mobility of the immunoprecipitated protein is probably due to the presence of high amounts of IgG heavy chain which has a similar molecular weight to epoxide hydrolase. Attempts to quantify the fluorescence of the immunoprecipitates by fluorospectrometry were not successful due to the persistent presence of low molecular weight fluorescent material. The fluorescence signal of the protein immunoprecipitated from intact membranes (Fig. 5A, channel 1) was not increased by addition of DOC at 0.05% (channel 2) or 0.5% (channel 3) to membranes during reaction with ANDS (Fig. 5A). When the ANDS labelled membranes were treated with trypsin in the presence of 0.5% DOC, only weakly fluorescent bands with M, about 31500, 27500 and 23000 were observed (Fig. 5B). The membranes were also assayed for styrene oxide hydrolase activity after treatment with ANDS and the results are shown in Table I. The pattern of results was similar to that found for trypsin treatment and the enzyme was relatively resistant to inhibition by ANDS in intact preparations or preparations treated with low DOC (0.05%). Enzyme activity was however, inhibited
by 70% when both ANDS and a high concentration DOC (0.5%) were added. These observations with ANDS suggest that the majority of the epoxide hydrolase molecule is accessible in intact membranes to this membrane-impermeant probe and thus is disposed on the cytosolic membrane surface (Fig. 2B). Membrane permeabilisation or solubilization does not significantly increase the extent of fluorescent labelling. This data is totally inconsistent with a model of epoxide hydrolase with six transmembrane domains and with similar, significant proportions of the enzyme amino acid residues exposed on the luminal and cytosolic membrane faces. Although the majority of the enzyme appears to be exposed at the cytosolic surface it must, none the less, be compactly folded since peptide bonds which are potentially susceptible to proteolysis, were found to be resistant to trypsin and other proteolytic enzymes in both membranes and purified protein. The carboxy-terminus of the protein also appears to be inaccessible and native epoxide hydrolase of membranes or purified preparations is not a substrate for carboxypeptidase Y. The protein can be a substrate for tryptic digest but only in the presence of a high concentration of DOC. This suggests that the detergent causes a conformational change which causes an increase of enzyme activity, renders the enzyme sensitive to inhibition by ANDS (Table I) and exposes a ‘hinge’ site to a limited proteolysis. These suggestions
39
are also consistent with the observation that the fluorescent label incorporated into the protein in native membranes is lost when the membranes are subsequently treated with trypsin in high DOC (Fig. 5B). The analysis of the epoxide hydrolase sequences illustrates the caution which must be employed in the interpretation of hydropathy plots and the value of considering homologous proteins. It is frequently forgotten that peaks in hydropathy profiles do not necessarily represent transmembrane domains and models for membrane topology should be validated by separate biochemical experimental data. It is interesting to note that a cytochrome P-450, originally thought to be anchored by multiple transmembrane segments based upon hydropathy analysis, is more likely to be attached by a single segment at the N-terminus [28,29]. Acknowledgements
We thank Drs. Macnab and Goldman for the computer software used for the predictive calculations and Dr. M. Low for gifts of phosphatidylinositol-specific phospholipase C. This work was supported in part by the Cancer Research Campaign and the Wellcome Trust. References Hargrave, P.A. (1986) in Techniques for the analysis of membrane proteins (Ragan, C.I. and Cherry, R.J., eds.), pp. 129-151, Chapman and Hall, London. Burchell, B., Bentley, P. and Oesch, F. (1976) B&him. Biophys. Acta 444, 531-538. Siedegard, J., Moron, M.S., Et&son, L.G. and De Pierre, J. (1978) Biochim. Biophys. Acta 543, 29-40. Oesch, F. and Bentley, P. (1976) Nature 259, 53-55. Levin, W., Thomas, P.E., Koreniowski, D., Jerina, D.M. and Lu, A.Y.H. (1978) Mol. Pharmacol. 14, 1107-1120. Siedegard, S., DePierre, J., Guenther, T.M. and Oesch, F. (1982) Acta Chem. Stand. B36, 555-557.
7 Kyte, J. and Doolitle, R.F. (1982) J. Mol. Biol. 157, 105-132. 8 Engleman, D.M., Stietz, T.A. and Goldman, A. (1986) Annu. Rev. Biophys. Biophys. Chem. 15, 321-353. 9 Porter, T.D., Beck, T.W. and Kasper, C.B. (1986) Arch. B&hem. Biophys. 248, 121-129. 10 Jackson, M.R., Craft, J.A. and Burchell, B. (1987) Nucleic Acid Res. 15, 7188. 11 Heinemarm, F.S. and Ozols, J. (1984) J. Biol. Chem. 259, 797-804. 12 Dockter, M.E. and Koseki, T. (1983) Biochemistry 22, 3954-3961. 13 Bulleid, N.J., Graham, A.B. and Craft, J.A. (1986) B&hem. J. 233, 607-611. 14 Shepherd, S.R.P., Baird, S.J., Hallinan, T. and Burchell, B. (1989) B&hem. J. 259, 617-620. 15 Bulleid, N.J. and Craft, J.A. (1984) Biochem. Pharmacol. 33, 1451-1457. 16 Cooper, M.B., Craft, J.A., Estall, M. and Rabin, B.R. (1980) Biochem. J. 190, 737-746. 17 Laemmli. U.K. (1970) Nature (London) 227, 680-685. 18 Towbin, H., Staehlin, T. and Gordon, J. (1979) Proc. Natl. Acad. Sci. USA 76, 4350-4354. 19 Domin. B.A., Serabjit-Singh, C.J. and Philpot, R.M. (1984) Anal. B&hem. 136, 390-396. 20 Otani, G., Abou-el-Makarem, M.M. and Bock, K.W. (1976) Biothem. Pharmacol. 25, 1293-1297. 21 Kreibich, G., Debey, P. and Sabatini, D.D. (1973) J. Cell. Biol. 56, 436-462. 22 Dutton, G.J. (1980) Glucuronidation of Drugs and Other Compounds, CRC Press, Boca Raton. 23 DuBois, G.C., Appella, E., Ryan, D.E., Jerina, D.M. and Levin, W. (1982) J. Biol. Chem. 257, 270X-2712. 24 Bulleid, N.J. (1985) PhD Thesis, CNAA, Glasgow College. 25 Low, M.G. (1987) B&hem. J. 244, l-13. 26 Low. M.G. and Saltiel, A.R. (1988) Science 239, 268-275. 27 Griffin, M.J. and Palakoderty, R.B. (1985) in Tight association between rat liver microsomal epoxide hydrolase and phosphatidylinositol in inositol and phosphoinositides: Metabolism and regulation (Bleasdale, J.E., Eichberg, J. and Hauser, G., eds.), p. _._ __ 669, Humana, Clifton. 28 De Lemos-Chiarandini, D., Frey, A.B., Sabatini, D.D. and Kreibich, G. (1987) J. Cell. Biol. 104, 209-219. 29 Vergeres, G., Winterhaler, K.H. and Richter, C. (1989) Biochemistry 328, 3650-3655. 30 Gonzalez, F.J. and Kaspar, C.B. (1980) B&hem. Biophys. Res. Commun. 93, 125441258.
