NMR studies on interaction of lauryl maltoside with cytochrome c oxidase: a model for surfactant interaction with the membrane protein

Journal of Inorganic Biochemistry 91 (2002) 116–124 www.elsevier.com / locate / jinorgbio

NMR studies on interaction of lauryl maltoside with cytochrome c oxidase: a model for surfactant interaction with the membrane protein 1

2

3

Krishnananda Chattopadhyay , Tapan Kanti Das , Ananya Majumdar , Shyamalava Mazumdar* Department of Chemical Sciences, Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba, Mumbai 400 005, India Received 20 December 2001; received in revised form 22 March 2002; accepted 25 March 2002

Abstract Interaction of lauryl maltoside (LM) surfactant with bovine heart cytochrome c oxidase (CcO) has been studied by NMR techniques. Detailed 2-D 1 H and 13 C NMR techniques were used to assign the NMR signals of the surfactant nuclei. Paramagnetic dipolar shift of the surfactant 13 C NMR signals were used to identify the atoms close to the enzyme. The diamagnetic carbon monoxide complex of CcO did not cause any shift in the surfactant NMR spectra suggesting that the paramagnetic centres of the native CcO cause the shifts by dipolar interactions. The results showed that the polar head groups of the surfactant comprised of two maltoside rings are more affected, while the hydrophobic tail groups did not show any significant change on binding of the surfactant to the enzyme. This indicated that surfactant head groups possibly bind to the enzyme surface and the hydrophobic tail of the surfactant forms micelles and remains away from the enzyme. Based on the results, we propose that the membrane bound enzyme is possibly stabilised in aqueous solution by association with the micelles of the neutral surfactant so that the polar heads of the micelles bind to the polar surface of the enzyme. These micelles might form a ‘belt like’ structure around the enzyme helping it to remain monodispersed in the active form.  2002 Elsevier Science Inc. All rights reserved. Keywords: Cytochrome c oxidase; Lauryl maltoside; Critical micellar concentration; NMR assignments; 2-D NMR

1. Introduction Bovine heart cytochrome c oxidase (CcO) is a large multisubunit membrane protein complex with molecular size of |200 kDa comprising 13 different polypeptides [1–3]. It is the terminal enzyme in the mitochondrial electron transport chain and reduces O 2 to water at a binuclear centre formed by heme a3 and a nearby copper atom, CuB [4–6]. CcO harnesses the chemical energy

*Corresponding author. Fax: 191-22-215-2110. E-mail addresses: [email protected] (S. Mazumdar), tapan.das@ lsp.lsbc.com, [email protected] (T.K. Das), majumdar@ sbnmr1.ski.mskcc.org (A. Majumdar), [email protected] (K. Chattopadhyay). 1 Present address: Department of Biochemistry and Molecular Biophysics, Washington University School of Medicine, 660 South Euclid, St. Louis, MO 63110, USA. 2 Present address: Large Scale Proteomics Corporation, 20451 Goldenrod Lane, Germantown, MD 20876, USA. 3 Present address: Cellular Biochemistry and Biophysics, Box 557, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021, USA.

released by the oxygen reduction to translocate four protons across the inner mitochondrial membrane and activates mitochondrial ATP synthesis [7,8]. The monomeric enzyme consists of three large subunits (I, II and III) coded by mitochondrial DNA [9]. Subunit I contains twelve membrane-spanning helices. Subunit II protrudes to the cytosolic side above the membrane surface [10]. Subunits I and II have been shown to form the catalytic core of the enzyme [6]. In addition to the binuclear centre (heme a 3 and CuB) located at subunit I, CcO contains another heme a at subunit I, one copper homodinuclear centre (Cu A) in subunit II, one magnesium, one zinc and some phospholipids as the intrinsic constituents [10]. Three active metal centres (heme a, heme a 3 and CuB) in the native (resting) form of CcO remain in their higher oxidation states viz. in Fe 13 and Cu 12 states) [6]. The two coppers in CuA form a mixed-valence complex [Cu(1.5) . . . Cu(1.5)] [10–12]. The enzyme exists in the inner mitochondrial membrane with the major parts of almost all the subunits embedded inside the membrane matrix except the soluble CuA domain which forms the electron entry site of CcO. Cytochrome c oxidase, being an integral membrane-

0162-0134 / 02 / $ – see front matter  2002 Elsevier Science Inc. All rights reserved. PII: S0162-0134( 02 )00427-0

K. Chattopadhyay et al. / Journal of Inorganic Biochemistry 91 (2002) 116 – 124

bound protein, is insoluble in water. Several in vitro studies on the enzyme are carried out in nonionic surfactant solutions [7]. These nonionic surfactants are known to maintain the structural integrity of the integral membrane proteins [9]. The extraction of CcO in active form is also carried out in neutral surfactants such as lauryl maltoside (LM), Triton X-100 (TX-100), Tween-80 etc., where the activity of the enzyme is fully retained. LM is able to both disperse and activate purified CcO better than any other surfactant [13]. Some investigators have found it to be the surfactant of choice for studying many other membrane proteins, such as rhodopsin [14] and the photosynthetic reaction centre [15], suggesting that this surfactant may have properties suitable for studies on intrinsic membrane proteins. LM was found to be the most successful activator of beef and neurospora CcO among the many alkyl glycosides tested, resulting in many fold higher activities than in other nonionic surfactants such as Tween-20 or TX-100 [16]. However, the molecular mechanism of the interaction of LM with CcO is not yet known. Rigorous analysis of beef heart CcO by sedimentation equilibrium [17] showed that it exists as a monomer (two heme a, two Cu atoms) in LM with the protein moiety of |194 kDa and associated surfactant equal to |108 kDa (| 200 molecules of LM). The number of bound surfactant molecules has also been reported to be as high as |320 per CcO molecule [18,19]. A micelle formed by aggregation of LM molecules in solution contains approximately 150 surfactant monomers (molecular weight of micelle |76 kDa [20]) and hence more than one micelle is possibly required for wrapping up one oxidase molecule. Studies on sedimentation velocity and equilibrium have recently shown that the extent of dimerisation of CcO in various nondenaturing neutral surfactant solutions depends on the concentration of the surfactant [21]. These studies suggested that the enzyme exists predominantly as monomers at LM concentrations above 0.1% (|2 mM) [21]. LM has also been found to enhance thermal stability of the enzyme [22]. However, no structural information on the association of the surfactant with the enzyme is reported so far. Detergent binding to the integral membrane proteins was shown to involve both detergent–detergent interaction as well as protein–detergent interaction [23,24]. Detergent solubilisation of membrane proteins was suggested to occur by insertion of the delipidated part of the protein into detergent micelles [23,24]. However, for large membrane proteins, the micellar size would not be large enough to accommodate the lipid embedded region of the enzyme, and the surfactant molecules might form a torus or ring around the hydrophobic sector of the protein or be present as a monolayer [25–27]. There is no report to structurally identify the interaction between the protein and surfactant molecule [27], which could provide information on the conformational properties of the surfactant molecules bound to the protein. Studies of the interaction of the surfactant with the

117

membrane enzyme would help to understand the structural and mechanistic aspects of stabilisation of the enzyme by the surfactant in solution. In the present report we have investigated the interaction of CcO with lauryl maltoside using NMR. The NMR spectrum of the surfactant is complicated due to the presence of the long hydrophobic alkyl chain. Detailed assignments of the NMR signals of the surfactants have been achieved by several 2-D NMR techniques. Paramagnetic dipolar shift of the NMR signals of the carbon centres of the surfactant molecule have been used to determine molecular interaction between the enzyme and nonionic surfactant LM and the specific interaction zone between LM and CcO was derived.

2. Materials and methods

2.1. Preparation of samples Lauryl maltoside (LM) was purchased from Fluka. Other reagents used were of the purest grade available commercially. CcO was extracted from beef heart muscle [28] and a part of the final pellets of CcO was suspended in 100 mM Tris buffer, 0.5% (|10 mM) lauryl maltoside, pH 7.4 to carry out the experiments reported here.

2.2. NMR spectroscopic methods All 1-D and 2-D 1 H and 13 C NMR spectra were recorded in D 2 O (99.9%) except those for 1-D 13 C spectra of CcO–LM samples, which were recorded in D 2 O–H 2 O (60:40) mixtures. Experiments were carried out at 303 K on a Bruker AMX 500 instrument equipped with a 5-mm inverse probe operating at 500.13 MHz for 1 H and 125.76 MHz for 13 C [29,30]. The following experiments were carried out to achieve assignment of NMR signals of LM: TOCSY (TOtal Correlation SpectroscopY), ROESY (Rotating-frame Overhauser Enhancement SpectrocopY), DQF-COSY (Double Quantum Filtered COrrelation SpectroscopY), and HSQC (Heteronuclear Single Quantum Coherence). TOCSY spectra were recorded using the TOWNY sequence [31], which is a selfcompensating windowless sequence for suppressing ROESY effects in TOCSY spectra. 1 H– 13 C correlations were obtained using the proton detected HSQC experiment. To improve resolution in the 13 C axis, the spectra were appropriately folded along the 13 C dimension (v2). These NMR experiments were carried out using 20 mM LM in aqueous buffer (10 mM potassium phosphate, pH 7.4) at 303 K to ensure complete formation of micelles. The critical micellar concentration of the surfactant was determined from the variation of the chemical shift of surfactant NMR signals over the concentration range of 0.01–20 mM of the surfactant in aqueous buffer solution in presence and in absence of 25 mM CcO.

118

K. Chattopadhyay et al. / Journal of Inorganic Biochemistry 91 (2002) 116 – 124

Titration of 13 C peaks of LM was carried out by gradual addition of stock CcO (45 mM in 10 mM potassium phosphate buffer, pH 7.4) to a 400-ml aliquot of 20 mM LM, to have a variation of CcO concentration in the range of 0–25 mM. CO complex of reduced CcO was prepared by passing freshly prepared carbon monoxide (CO) into a solution of CcO reduced by excess sodium dithionite. The CO complex of reduced CcO (CO–CcO) was passed through a Sephadex G25 column under anaerobic conditions to remove any excess sodium dithionite before the NMR experiments.

surfactant shows a larger number of peaks compared to those in the 1 H NMR spectrum. The hydrophobic tail of the surfactant shows broad resonances in the 35–15 ppm range of the 1-D 13 C NMR spectrum of LM, which is similar to that of SDS reported earlier [30]. Larger chemical shift dispersion in the 13 C NMR spectrum of the surfactant could help to identify any effects on the NMR resonances due to addition of CcO to the surfactant solution. However, assignment of the NMR signals of the surfactant molecule is necessary before investigation of its interaction with CcO.

3.1. Assignment of 1 H and lauryl maltoside

13

C NMR resonances of

3. Results and discussion The NMR spectrum of LM is complicated because of resence of eleven CH 2 groups in the hydrophobic tail of the surfactant and the maltose rings (Fig. 1). The NMR spectra of the surfactant (Fig. 2) are required to be assigned in order to identify resonances from individual nuclei. Comparison of the NMR spectra of LM with those of sodium dodecyl sulphate (SDS) [30] could give qualitative identification of the resonances of the surfactant. The proton NMR spectra (lower panel in Fig. 2) show upfield shifted peaks (1–2 ppm) corresponding to the hydrophobic tail part of the surfactant (protons attached to C 14 to C 24 , Fig. 1). The maltose ring protons reside in the hydrophilic region of the molecule and show resonances at 3–5 ppm region. The 1-D proton NMR of the surfactant is very broad and ill-resolved. The 1-D 13 C NMR spectrum of the

The resonance assignments were obtained based on a combination of 1 H DQF-COSY, TOCSY, ROESY and 1 H- 13 C HSQC spectra. Complete assignments were achieved for the maltoside ring and the resonances from the lauryl side chain could only be partially assigned because of extensive overlap of resonances in the centre of the hydrocarbon chain (see Figs. 1 and 2 and Table 1).

3.2. Spin system identification 3.2.1. Ring protons A combination of DQF-COSY and low-mixing time (30 ms) TOCSY was employed for identifying spin-systems belonging to each ring. A short-mixing time (30 ms) TOCSY and a DQF-COSY were used for sequential connectivities, while a long-mixing time (100 ms) TOCSY was used for relayed connectivities. The most downfieldshifted protons are attached to the aldehyde group i.e. 6H and 10H in Fig. 1 and both of them show intensities expected for single protons. Starting from the most downfield-shifted protons (6H and 10H in the two rings, Fig. 1), other ring protons were identified by sequential connectivities as shown for ring 1 in Figs. 3 and 4. 3.2.2. Side chain protons Complete resonance assignment of the side chain protons were difficult because the resonances of the CH 2 protons are highly overlapped with each other appearing near 1.4 ppm. The protons (13H, 14H etc.) in the side chain, which are shifted downfield due to proximity to ring 2, were assigned by chemical shifts. The most upfieldshifted protons could be assigned to terminal methyl protons.

Fig. 1. Schematic structure and numbering scheme of lauryl maltoside.

3.2.3. Identification of ring 1 and ring 2 Unambiguous assignments of resonances of ring 1 and ring 2 were made on the basis of observed ROE between 10H and 13H in a 150 ms ROESY spectrum (Fig. 5). However, positive cross peaks between some of the resonances were observed which appear to be indicative of

K. Chattopadhyay et al. / Journal of Inorganic Biochemistry 91 (2002) 116 – 124

Fig. 2.

13

119

C (top) and 1 H (bottom) NMR spectra of lauryl maltoside (2.5%) in aqueous medium.

some sort of chemical exchange or secondary TOCSY effects.

3.2.4. 13 C Assignments 13 C NMR assignments were obtained from 1 H– 13 C HSQC (Fig. 6) spectra. Well-resolved spectrum from the ring region allowed unambiguous correlation of 1 H resonances with the corresponding 13 C resonances from the carbon attached by a single bond to hydrogen. Very high resolution in HSQC spectra was obtained by appropriately folding the spectrum along the 13 C dimension, thereby reducing the spectral width along the 13 C axis. Side chain 13 C assignments could also be obtained directly from the HSQC spectrum. 3.3. 1 H and 13 C NMR of LM in presence of native CcO and the carbon monoxide derivative of reduced CcO The 1 H as well as 13 C NMR signals of LM in aqueous solution showed distinct dependence of the chemical shift as well as the line width on the concentration in the concentration range of 0.01–0.2 mM of the surfactant (data

not shown). Such concentration dependence of the NMR signals is characteristic of the formation of micelles by self-aggregation of the monomeric surfactant molecules in solution. Analogous concentration dependence of the chemical shift and the line width of surfactant NMR signals are also observed in other surfactants and are used to determine the critical micellar concentration (CMC) of the surfactant micelles in solution. The value of the CMC determined from the concentration dependence of the chemical shift was found to be |0.15 mM for LM micelles, which agrees with earlier reports [13]. The chemical shift as well as line width of the surfactant NMR resonances was found to remain independent of the surfactant concentration above |0.5 mM indicating complete formation of micelles. Solubilisation of CcO in aqueous solution required at least 0.02% (|0.4 mM) LM solution while the enzyme has been shown to remain monomeric [21] at LM concentrations above 0.1% (2 mM). The stabilisation of the enzyme in aqueous solution thus depends on its interaction with the surfactant molecules. The native form of CcO has the CuA, the heme a and the CuB centres carrying one

K. Chattopadhyay et al. / Journal of Inorganic Biochemistry 91 (2002) 116 – 124

120

Table 1 1 H and 13 C NMR resonance assignments for lauryl maltoside in aqueous solution at 303 K Carbon number

Type of carbon

d ( 13 C) (ppm)a

Hydrogen

d ( 1 H) (ppm)a

1C 2C 3C 4C 5C 6C 7C 8C 9C 10C 11C 12C 13C

–CH 2 OH –CH –CHOH –CHOH –CHOH –OCHO– –CHO– –CHOH –CHOH –OCHO– –CHO– –CH 2 OH –CH 2 O–

75.65 63.31 71.95 80.68 74.64 103.20 77.53 79.07 75.70 105.70 75.89 63.57 73.10

14C 15–23C

–CH 2 –CH 2

32.45 |32.5 b

24C

–CH 3

16.64

1H 2H 3H 4H 5H 6H 7H 8H 9H 10H 11H 12H 13Ha 13Hb 14H 15H 16–23H 24H

3.80 3.92 3.56 3.79 3.70 5.44 3.53 3.85 3.45 4.47 3.78 3.95 3.67 3.99 1.75 1.43 |1.4 b 0.96

a

Using trimethylsilyl propionate (TSP) as internal standard. Assignments of individual resonances could not be achieved because of extreme overlap. b

unpaired electron each and the heme a3 carrying five unpaired electrons on the metal centre, making the enzyme highly paramagnetic. All these paramagnetic centres would affect the NMR signals of the surfactant bound to the

Fig. 4. Selective regions of 1 H TOCSY spectra of LM. The composite pulse sequence TOWNY was used for TOCSY, with an r f field strength of 9 kHz and mixing time 30 ms; 2048 and 200 complex points were acquired along t 2 and t 1 , respectively with t max 5 682.6 ms and t max 5 2 1 66.6 ms. Sixteen transients were collected per t 1 increment. After zerofilling, final 2-D matrix dimensions were 2048 (v2)3512 (v1). Quadrature detection long v1 was achieved using the TPPI method.

enzyme. A shift in the position of the NMR signal may be observed in molecules containing paramagnetic ions and these shifts are generally a consequence of the large hyperfine fields that nuclei experience due to the magnetic

Fig. 3. Selective regions of 1 H DQF-COSY spectra of LM. 2048 and 200 complex points were acquired along t 2 and t 1 , respectively, with t max 5 2 682.6 ms and t max 5 66.6 ms. 16 transients were collected per t 1 1 increment. Data were multiplied by sine bell window functions shifted by 22.58 along t 2 and 458 along t 1 , prior to 2-D Fourier transformation. After zero-filling, final 2-D matrix dimensions were 2048 (v2)3512 (v1). Quadrature detection along v1 was achieved using the TPPI method.

Fig. 5. Part of the 1 H ROESY spectrum. A cw r f field of 1.3 kHz strength was applied for 150 ms. Other conditions were similar to those employed in TOCSY experiments (see Fig. 4).

K. Chattopadhyay et al. / Journal of Inorganic Biochemistry 91 (2002) 116 – 124

Fig. 6. Part of the 1 H– 13 C HSQC spectrum. Acquisition and processing parameters used were (t 2 : t 1 ): carrier (80:4.8) ppm, data points (1024:164) complex, acquisition time (0.34:0.025) ms, 2-D matrix dimension (2048:512). 64 transients per t 1 were acquired. Quadrature detection along v1 was achieved using the states-TPPI method. The 13 C dimension was folded, thus increasing the digital resolution along v1 by almost a factor of 2.

moments of the unpaired electrons of the paramagnetic centre [32]. However, dipolar interactions between the electron spin of the paramagnetic centre and nuclear spin of the proton or carbon may also cause broadening of the signals. Hyperfine-shifted resonances in NMR spectra have been very useful in providing insights into the properties of many metal-containing systems [29,30,32]. Paramagnetic shift of NMR signals can be suitably used to study ligand binding [33,34] or association of substrate analogues [35] to the active site (metal centre) of various metalloproteins. The 1 H NMR spectrum of LM (20 mM) in presence of native CcO (data not shown) was found to become slightly broadened compared to that in absence of the enzyme (Fig. 2). The dipolar broadening of the 1 H signals of LM by the paramagnetic centres of the enzyme was more than the dipolar shift. The CMC of the surfactant (|0.15 mM) was however found to be almost unchanged in presence of the enzyme (25 mM). In order to determine whether the broadening of the surfactant NMR signals in presence of CcO were due to the paramagnetic centres of the enzyme or not, studies were carried out with the carbon monoxide complex of the fully reduced enzyme. The carbon monoxide complex of reduced CcO (CO–CcO) consists of reduced CuA, reduced heme a and reduced CuB centres and the CO complex of reduced heme a3 centre, which makes it diamagnetic. Addition of CO–CcO (|20 mM) to the LM solution (20 mM) did not have any detectable effect on the 1 H NMR spectrum of the surfactant, which

121

confirmed that the paramagnetic centres of the enzyme are responsible for the observed broadening of the surfactant proton resonances in presence of the native enzyme. Analogous to the 1 H NMR, the 13 C NMR spectra of LM in aqueous solution showed distinct concentration dependence characteristic of micellisation. The 13 C NMR signals were independent of surfactant concentration above |0.5 mM as observed in case of 1 H NMR signals indicating complete formation of the micelles. The 13 C NMR spectra of LM micelles (20 mM) however showed small but distinct shift of resonances on addition of a small amount of native CcO. The 13 C NMR spectrum of the surfactant was however, not affected on addition of CO–CcO, indicating that the shift in the 13 C NMR signals on addition of native CcO to the surfactant indeed arises due to interaction of the surfactant with the paramagnetic enzyme. 13 C-NMR spectra of LM were recorded in the presence of increasing concentrations of native CcO. The change in chemical shift [Dd ( 13 C)] of each carbon resonance of LM was followed as a function of added CcO concentration and was plotted as shown in Fig. 7. The observed change in chemical shift, Dd, of the micelle in presence of the paramagnetic enzyme in the fast exchange limit can be given as [32] Dd 5 (dobs 2 df ) 5 fb (db 2 df )

(1)

where the chemical shift of the free micelles in absence of the paramagnetic enzyme is df and that of the micelles bound to the enzyme is db . The observed chemical shift (dobs ) would be a linear combination of those of the free and bound forms weighted by the fractions of bound ( fb ) and free micelles. Addition of the paramagnetic enzyme to the micellar solution of LM would increase fb leading to a systematic change in the observed chemical shift with increase in the enzyme concentration. The resonances due to two rings (rings 1 and 2) and the side chain (Table 1) were plotted separately (Fig. 7) to show the shifts of the polar head groups and the nonpolar side chain. In the case of the ring 1, all the 13 C resonances were found to undergo significant changes on addition of the enzyme, some in the positive side and some in the negative side. Except the 7C and 13C (see Table 1 and Fig. 1 for nomenclature) resonances in ring 2, all other carbons also show significant changes in chemical shift due to the addition of CcO. However, the side chain resonances were found to undergo only very small changes in their positions. The results shown in Fig. 7 thus demonstrate that the atoms of the surfactant, which are at the polar head group, are affected more significantly by the paramagnetic dipolar interactions on surfactant binding to the enzyme than the atoms at the hydrophobic tail of the surfactant. Thus the results indicate that the association of LM with CcO possibly occurs primarily through the interaction of its polar head groups with the surface of CcO. The fact that a sharp difference is observed between the

122

K. Chattopadhyay et al. / Journal of Inorganic Biochemistry 91 (2002) 116 – 124

Fig. 7. Titration of 1-D 13 C resonances of LM by CcO. The relative change in the chemical shift [Dd ( 13 C)] of the 13 C NMR signals of LM are plotted against CcO concentration. Ring 1: 13 C resonances assigned to the first ring of the maltose group of the surfactant. Ring 2: 13 C resonances assigned to the second ring of the maltose group of the surfactant. Side chain: 13 C resonances assigned to the hydrophobic tail group of the surfactant. See Fig. 1 for numbering of the carbon centres in the surfactant.

CcO-induced shift of resonances in the rings (head groups of the surfactant) and the side chain (hydrophobic tail of the surfactant), indicates that the conglomeration of maltoside rings in the micellar structure possibly keeps the side chain away from the enzyme surface. The interaction of the surfactant with the enzyme indeed involves the membrane intensive part of the enzyme. This would help the amino acids in the hydrophobic membrane intensive surface of the enzyme to remain away from water in the aqueous micellar solution. The neutral surfactants form micelles with polar surface and hydrophobic core. The NMR results suggest that the polar surface of these micelles probably binds to polar patches on the enzyme in the membrane intensive domain. This observation provides further support to an earlier proposal [13] that the maltoside head groups are immobilized on the enzyme surface. Considering the large membrane embedded segments of CcO, one might consider that the micelles formed by aggregation of LM might surround the enzyme in such as way that they maintain the enzyme–micelles complex dispersed in aqueous solution. Fig. 8 shows a schematic model of the interaction of CcO with LM, with several micelles surrounding the enzyme making it ‘soluble’ in water. A ‘belt like’ structure around the CcO by cylindrical

Fig. 8. Model for the association of lauryl maltoside surfactant micelles with CcO.

K. Chattopadhyay et al. / Journal of Inorganic Biochemistry 91 (2002) 116 – 124

micelles of lauryl maltoside may be a likely structure of the micellar layer around the enzyme in solution. The micellar phase is in dynamic equilibrium with the monomers and hence the NMR spectra shows only one set of signals, which is the average of the micellar and monomeric forms (fast exchange regime). Thus the surfactant molecules involved in solubilizing the enzyme also undergo fast exchange with the free surfactants. The fast exchange of the bound and free surfactant in the micelles around the CcO would possibly stabilise the ‘belt like’ structure of surfactants around the enzyme with the surfactant head groups interacting with the enzyme as well as with water. Several factors might be responsible to make LM a good surfactant for solubilizing CcO so that the enzyme shows high rates of electron transfer and high enzyme stability in the surfactant solution. Alkyl glycosides (such as LM) form unusually small micelles compared to most nonionic surfactants [36], which are efficient in dispersing the enzyme in solution. In the case of LM-solubilised rhodopsin, it was earlier suggested that the large size and high packing density of nonionic head groups of the surfactant were responsible for preventing protein denaturation [14]. The weak dipolar interaction between the polar head groups of the surfactant and the polar amino acids residues would possibly provide enough stabilisation to the enzyme to solubilise it in the aqueous surfactant solution but would not disrupt the three-dimensional structure of the enzyme which is stabilised by critical balance of hydrophobic, ionic and other interactions within the protein. These are also advantageous for gel-filtration as well as for kinetic and physical studies on the enzyme in solution. The smaller micellar size also helps in quicker surfactant removal by dialysis. The lower enzymatic activities of CcO observed in some other surfactants might be due to less effective dispersion of the enzyme in the surfactant solution with a resulting limited accessibility to the reducing substrate, rather than inhibition of the intrinsic electron-transfer ability of the enzyme [16]. The ability of LM to stabilise an active monomeric form of CcO under relatively mild condition illustrates the unusual effectiveness of this surfactant as a dispersing agent [16,17,20]. Structural information obtained from the present NMR study indicated that the polar head groups of the surfactants play a very important role in their binding to the enzyme. A previous study [22] on the influence of surfactant structure on CcO activity had also highlighted the importance of the polar head groups. Studies have also revealed that the observed electron transfer rate in various surfactant environments was not dependent on cytochrome c concentration [22] suggesting that cytochrome c binding and subsequent electron transfer is not affected by binding of the surfactant to CcO. The alternative model [25–27] that the hydrophobic tail of the surfactant interacts with the hydrophobic patch of the membrane intensive domain of the enzyme may have

123

several shortcomings. These hydrophobic groups of the surfactant could penetrate inside the hydrophobic core of the enzyme thereby destabilizing the internal hydrophobic balance between different segments of the enzyme. Moreover, if the hydrophobic interaction between the surfactant and the enzyme were the mechanism for stabilisation of the enzyme in surfactant medium then SDS with the same hydrophobic tail as that in LM, would also have stabilised the enzyme, but SDS is known to denature the enzyme. The results obtained from our studies indicate that stabilisation of the membrane intensive enzyme in aqueous solution takes place by dipolar interaction between the polar surface of the micelles of neutral surfactant and probably the surface amide groups of the enzyme. Ionic surfactants, such as SDS, would not participate in such dipolar interactions with the surface amide bonds of the enzyme and might rather lead to penetration of the hydrophobic tails of the SDS molecules inside the enzyme causing disintegration of the multisubunit enzyme. The hydrophobic tail of the neutral surfactant, on the other hand, remains away from the enzyme forming the micellar core outside the enzyme. The binding of the polar groups of LM can thus stabilise the enzyme in a fully active conformation while SDS denatures the enzyme. It is important to note that the present results do not rule out possibility of one or two monomeric surfactant molecules penetrating inside the hydrophobic core of the enzyme. Such a strongly bound site will have a very weak and broad NMR signal compared to the fast exchanging site. However, since the enzyme is solubilised only at above the CMC of the surfactant [21] hence interaction of the micelles is important for the activation of the enzyme, and any irreversible interaction of the monomeric surfactant is unlikely to have an effect on the mechanism of activation and stabilisation of the enzyme in the surfactant solution.

4. Conclusions CcO retains its full enzymatic activity in lauryl maltoside medium. We have determined the interaction zone of LM with CcO by identifying the 1 H and 13 C resonances due to the rings and side chain of LM. It was observed that micelles of LM interact with CcO through its polar head groups. The results thus suggested that the polar head groups of LM might interact with the polar amino acids of the enzyme, and the hydrophobic tail groups of the surfactant remain away from the enzyme, possibly forming the micellar core outside the enzyme. Hydrogen bonding between the maltoside protons of surfactant and amide groups of polar, membrane intensive amino acids might play an important role in such interaction. This model of stabilisation of the membrane enzyme by neutral surfactant would allow the enzyme to remain in solution as a monodispersed form. The other probable model where the

124

K. Chattopadhyay et al. / Journal of Inorganic Biochemistry 91 (2002) 116 – 124

hydrophobic tails of the surfactant interact with the hydrophobic segments of the enzyme would affect the ‘hydrophobic balance’ of the enzyme, and might affect the structure and stability of the multi subunit enzyme.

Acknowledgements This work was supported by the Tata Institute of Fundamental Research. The NMR experiments were carried out at the National Facility of High Field NMR, TIFR, which is gratefully acknowledged.

References [1] R.A. Capaldi, Ann. Rev. Biochem. 59 (1990) 569–596. [2] G. Buse, G.C.M. Steffens, R. Biewald, B. Bruch, S. Hensel, in: S. Papa, B. Chance, L. Ernster (Eds.), Cytochrome Systems, Molecular Biology and Bioenergetics, Plenum Press, New York, 1987. ¨ [3] M. Wikstrom, K. Krab, M. Saraste, Cytochrome Oxidase. A Synthesis, Academic Press, London, 1981. [4] M. Saraste, Q. Rev. Biophys. 23 (1990) 331–366. [5] Y. Hatefi, Ann. Rev. Biochem. 54 (1985) 1015–1069. ¨ Nature 356 (1992) 301–309. [6] G.T. Babcock, M. Wikstrom, [7] R.A. Copeland, S.I. Chan, Ann. Rev. Phys. Chem. 40 (1989) 671–698. ¨ Chem. Rev. 90 (1990) 1247–1260. [8] B.G. Malmstrom, [9] M. Brunori, G. Antonini, F. Malatesta, P. Sarti, M.T. Wilson, in: G.L. Eichhorn, L.G. Marzilli (Eds.), Heme Proteins, Vol. 7, Elsevier, New York, 1988, pp. 93–153. [10] T. Tsukihara, H. Aoyama, E. Yamashita, T. Tomizaki, H. Yamaguchi, K. Shinzawa-Itoh, R. Nakashima, R. Yaono, S. Yoshikawa, Science 269 (1995) 1069–1074. [11] R.B. Gennis, S. Ferguson-Miller, Science 269 (1995) 1063–1064. [12] G.C.M. Steffens, T. Soulimane, G. Wolff, G. Buse, Eur. J. Biochem. 213 (1993) 1149–1157. [13] T. VanAken, S. Foxall-VanAken, S. Castleman, S. Ferguson-Miller, Methods Enzymol. 125 (1986) 27–35. [14] P. Knudsen, W.L. Hubbell, Membr. Biochem. 1 (1978) 297–322.

[15] M.W. Kendall-Tobias, M. Seibert, Arch. Biochem. Biophys. 216 (1982) 255–258. [16] P. Rosevear, T. VanAken, J. Baxter, S. Ferguson-Miller, Biochemistry 19 (1980) 4108–4115. [17] M.D. Suarez, A. Revzin, R. Narlock, E.S. Kempner, D.A. Thompson, S. Ferguson-Miller, J. Biol. Chem. 259 (1984) 13791–13799. [18] K.A. Nalecz, R. Bolli, B. Ludwig, A. Azzi, Biochim. Biophys. Acta 808 (1985) 259–272. [19] R. Bolli, K.A. Nalecz, A. Azzi, Arch. Biochem. Biophys. 240 (1985) 102–116. [20] D.A. Thompson, S. Ferguson-Miller, Biochemistry 22 (1983) 3178– 3187. [21] A. Musatov, J. Ortega-Lopez, N.C. Robinson, Biochemistry 39 (2000) 12996–13004. [22] N.C. Robinson, J. Neumann, D. Wiginton, Biochemistry 24 (1985) 6298–6304. [23] N.C. Robinson, C. Tanford, Biochemistry 25 (1975) 369–378. [24] S. Makino, J.L. Woolford Jr., C. Tanford, R.E. Webster, J. Biol. Chem. 250 (1975) 4327–4332. [25] J.V. Moller, M. le Maire, J.P. Anderson, in: A. Watts, J.J.H.M. de Ponts (Eds.), Progress in Protein–lipid Interactions, Vol. 2, Elsevier, Amsterdam, 1986, pp. 147–197. [26] M. le Maire, S. Kwee, J.P. Anderson, J.V. Moller, Eur. J. Biochem. 129 (1983) 525–532. [27] J.V. Moller, M. le Maire, J. Biol. Chem. 268 (1993) 18659–18672. [28] T.K. Das, S. Mazumdar, Biochim. Biophys. Acta 1209 (1994) 227–237. [29] S. Mazumdar, O.K. Medhi, S. Mitra, Inorg. Chem. 27 (1988) 2541–2543. [30] S. Mazumdar, J. Phys. Chem. 94 (1990) 5947–5953. [31] M. Kadkhodaei, T.L. Hwang, J. Tang, A.J. Shaka, J. Magn. Reson. A 105 (1) (1993) 104–107. [32] G.N. LaMar, W.D. HorrocksJr., R.H. Holm (Eds.), Nmr of Paramagnetic Molecules: Principles and Applications, Academic Press, New York, 1973. [33] L. Banci, I. Bertini china, C. Luchinat, Methods Enzymol. 239 (1994) 485–514. [34] I. Morishima, S. Ogawa, T. Inubushi, T. Yonezawa, T. Iizuka, Biochemistry 16 (1977) 5109–5115. [35] S. Modi, D.V. Behere, S. Mitra, J. Biol. Chem. 264 (1989) 19677– 19684. [36] A. Helenius, K. Simons, Biochim. Biophys. Acta 415 (1975) 29–79.