Interaction of Human Chloride Intracellular Channel Protein 1 (CLIC1

Jun 14, 2016 - School of Physics, University of New South Wales, Sydney, New South ... direct structural evidence that CLIC1 associates with membranes...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/biochemistry

Interaction of Human Chloride Intracellular Channel Protein 1 (CLIC1) with Lipid Bilayers: A Fluorescence Study Joanna E. Hare,† Sophia C. Goodchild,‡ Samuel N. Breit,§ Paul M. G. Curmi,∥ and Louise J. Brown*,† †

Department of Chemistry and Biomolecular Sciences, Macquarie University, Sydney, New South Wales 2109, Australia Astbury Centre for Structural Molecular Biology and School of Molecular and Cellular Biology, University of Leeds, Leeds LS29JT, United Kingdom § St Vincent’s Centre for Applied Medical Research, St Vincent’s Hospital, Sydney, New South Wales 2010, Australia ∥ School of Physics, University of New South Wales, Sydney, New South Wales 2052, Australia ‡

ABSTRACT: Chloride intracellular channel protein 1 (CLIC1) is very unusual as it adopts a soluble glutathione S-transferase-like canonical fold but can also autoinsert into lipid bilayers to form an ion channel. The conversion between these forms involves a large, but reversible, structural rearrangement of the CLIC1 module. The only identified environmental triggers controlling the metamorphic transition of CLIC1 are pH and oxidation. Until now, there have been no highresolution structural data available for the CLIC1 integral membrane state, and consequently, a limited understanding of how CLIC1 unfolds and refolds across the bilayer to form a membrane protein with ion channel activity exists. Here we show that fluorescence spectroscopy can be used to establish the interaction and position of CLIC1 in a lipid bilayer. Our method employs a fluorescence energy transfer (FRET) approach between CLIC1 and a dansyl-labeled lipid analogue to probe the CLIC1−lipid interface. Under oxidizing conditions, a strong FRET signal between the single tryptophan residue of CLIC1 (Trp35) and the dansyl-lipid analogue was detected. When considering the proportion of CLIC1 interacting with the lipid bilayer, as estimated by fluorescence quenching experiments, the FRET distance between Trp35 and the dansyl moiety on the membrane surface was determined to be ∼15 Å. This FRET-detected interaction provides direct structural evidence that CLIC1 associates with membranes. The results presented support the current model of an oxidation-driven interaction of CLIC1 with lipid bilayers and also propose a membrane anchoring role for Trp35. a dramatic structural transition from a β-sheet to an all-α-helical domain, revealing a hydrophobic surface through which the dimer subunits were noncovalently linked.3 This unique form for CLIC1 suggested that the hydrophobic surface on the oxidized dimer could form a docking interface required for membrane association and subsequent ion channel function. However, the CLICs are fundamentally different from members of classical ion channel protein families as they notably lack both a membrane-targeting leader sequence and hydrophobic regions typically seen in ion channel proteins.13 Hence, to convert between the soluble monomer and the membraneinserted structure, the CLIC proteins must likely undergo a large-scale conformational change to traverse the membrane and form the pore required for ion conductance. However, a high-resolution integral membrane CLIC structure has yet to be described. In lieu of a high-resolution integral membrane structure, numerous structural and functional biophysical techniques have provided some insight into the membrane-inserted structure and metamorphic transitions of the human CLIC1 protein. Support for an integral membrane CLIC form with a single-

A

lthough found mainly in the cytosol, members of the chloride intracellular ion channel (CLIC) protein family (CLIC1−6 in vertebrates) are able to interconvert between a soluble, globular state and a membrane-inserted form capable of transporting anions across the lipid bilayer.1−6 This striking ability to alternate between drastically different conformations has led to CLICs being described as belonging to a small class of proteins termed “metamorphic”.7 The structural dynamism of metamorphic proteins can make them particularly difficult to study using conventional protein structure determination methods.8 Structural determination of the CLIC protein family poses further obstacles, as resolving the CLIC ion channel structure must also consider protein−lipid interactions. The soluble forms of the CLIC family proteins are overall well characterized with the first X-ray crystallography structure described in 2001 for CLIC1 (∼24 kDa).9 The crystal structure obtained under reducing conditions shows CLIC1 to be a small globular protein with a GST-like fold possessing a thioredoxin N-domain and a larger all-α-helical C-domain.9 Similarly, crystal structures of several other mammalian and lower-order CLIC family members share the same monomeric GST-like canonical CLIC fold.5,10−12 Interestingly, under oxidizing conditions, a noncovalently linked dimer form of the CLIC1 member was also created and isolated.3 The crystal structure of this CLIC1 oxidized dimer state showed that the N-domain had undergone © 2016 American Chemical Society

Received: February 1, 2016 Revised: June 10, 2016 Published: June 14, 2016 3825

DOI: 10.1021/acs.biochem.6b00080 Biochemistry 2016, 55, 3825−3833

Article

Biochemistry

Figure 1. (A) Crystal structure of the soluble reduced CLIC1 monomer form (PDB entry 1K0M9). The putative TM region (Cys24−Val46) is shaded black. The single native tryptophan residue (Trp35) is shown (stick model). (B) Cartoon illustrating the FRET strategy used to position the depth of Trp35 of CLIC1 within the bilayer of fluorescently labeled liposomes. Trp35 is the intrinsic fluorophore used as the donor probe for the measurements of the FRET distance to the dansyl acceptor probe located on the lipid headgroups. The system under investigation can be described as having a restricted two-dimensional geometry where FRET is measured as the “distance of closest approach” between Trp donors in an infinite plane and randomly and uniformly distributed acceptors in an infinite parallel plane (the bilayer surface).

gating mechanism is difficult as measuring the amount of CLIC1 bound to the membrane is particularly challenging.2,16,25 As a result, we continue to have a limited understanding of the connection between environmental factors that promote interaction with the bilayer and subsequent ion channel activity.26 One current model is that in which oxidation of CLIC1 in the presence of membranes promotes membrane interaction,16 possibly through formation of the intermediate “half-dimer” structural state. Oligomerization then likely follows, forming the active ion channel.2,27 Thus, directly, the redox environment can change the structure of CLIC1 but is also observed to modify ion channel activity.3 A better understanding of the factors that trigger CLIC1− bilayer association and insertion, as distinct from ion channel gating, is required to build a model of a CLIC1 membrane state. A combination of fluorescence measurements can offer a reliable and noninvasive approach to studying the interaction and arrangement of proteins within lipid environments (Figure 1). Förster resonance energy transfer (FRET), which measures the proximity of two fluorescent molecules in a donor− acceptor pair, is one such biophysical approach. We have previously performed FRET measurements to obtain distances both within, and between, CLIC1 subunits.25,27 These measurements revealed intrasubunit conformational changes and allowed the modeling of a CLIC1 oligomeric state, likely present in the membrane environment.27 In the study presented here, we use the FRET approach for the first time to establish direct structural evidence of the interaction between CLIC1 and the membrane. FRET measurements were obtained between the single native Trp35 residue of CLIC1 and a dansyl label located on the headgroup of the lipid bilayer. Modified Stern−Volmer acrylamide quenching of Trp35 fluorescence was also performed to assess the CLIC1− membrane interaction and allow for an estimate of the proportion of CLIC1 interacting with the bilayer under our experimental conditions. We describe how these approaches can together be used to begin to map the interaction of CLIC1 with the bilayer. This approach is an important step toward obtaining a better understanding of the elusive CLIC membrane state.

pass transmembrane (TM) domain near the N-terminus (Cys24−Val46) first came from proteolytic digestion studies14 and was later supported by electrophysiological studies in which FLAG epitopes, fused to either the N- or C-terminus of CLIC1, were detected on opposite sides of the membrane.15 Fluorescence quenching studies of CLIC1 then showed that Trp35 in the TM region was positioned in a buried environment upon the addition of membranes;16 a result later supported by peptide studies representing the TM domain region of CLIC1.17−19 While these structural studies of a membrane state have started to provide support for an integral membrane form, the electrophysiological characterization of the individual CLIC channels is also far from complete. To date, the best-studied CLIC channel is the human CLIC1 that has been characterized independently by three research groups.1−4,9,15,20,21 Together, these studies and others have demonstrated that the channel properties of bacterially expressed CLIC1 proteins in artificial lipid bilayers are similar to those of channels detected in CLIC1-transfected whole cells. Likewise, a similar mechanism of membrane integration and pH-dependent conductance was also found for both in vitro and cellular CLIC1 channels, with evidence emerging of an oligomeric state likely to be responsible for the fully functional ion channel (reviewed in ref 22). In addition to the oxidation-triggered monomer−dimer structural transition identified by crystallography, a pHtriggered intermediate state of CLIC1 possessing a solventexposed hydrophobic surface has also been identified through acid-induced destabilization measurements.23,24 Although the experiments were performed in the absence of a membrane mimetic, the authors proposed that the acidic environment encountered by CLIC1 at the surface of membranes would also trigger a similar structural transition, priming the TM region of CLIC1 for insertion into the bilayer. Supporting this view are the many functional studies of CLIC members, which also display the pH dependence of the ion channel activity. These studies have shown channel activity to be minimal around neutral pH but increasing rapidly as the pH is decreased or, in some studies, also increased.1,2,5,10,12 However, deciphering whether the pH dependency of the channel activity is related to the proportion of membrane-inserted CLIC1 or a subsequent 3826

DOI: 10.1021/acs.biochem.6b00080 Biochemistry 2016, 55, 3825−3833

Article

Biochemistry

■

EXPERIMENTAL PROCEDURES Expression and Purification of CLIC1-WT and CLIC1W35F. Recombinant wild-type (WT) CLIC1 (GenBank accession number NP_001279) was expressed in the pET28a vector (Novagen) as a fusion protein with an N-terminal sixHis tag followed by a thrombin cleavage site as previously described.27 Briefly, Escherichia coli BL21(DE3) cells were grown in 2× YT medium at 37 °C and induced with 1 mM isopropyl thio-β-galactoside (IPTG) at midlog growth phase. The cell culture was grown for a further 16 h, at the reduced temperature of 20 °C, before cell lysis. Soluble CLIC1 was purified from the total cellular lysate by binding to Ni-NTA His-Bind resin (Novagen). CLIC1 was eluted via cleavage of the His tag following a 16 h incubation with 100 NIH units of bovine plasma thrombin per liter of cell culture (Sigma). Following incubation with 5 mM dithiothreitol (DTT) for 0.5 h, the CLIC1 reduced WT monomer was further isolated by size exclusion chromatography on a Superdex-75 prep grade high-performance column (GE Healthcare). The expressed CLIC1 protein consisted of the CLIC1 sequence with an extra three residues at the N-terminus (Gly-Ser-His) as a result of the thrombin cleavage site in the fusion construct. The QuikChange site-directed mutagenesis kit (Stratagene) was used to create a “Trpless” CLIC1-W35F construct for FRET experiments (“acceptor only” sample) by mutating the single native tryptophan to phenylalanine (W35F). The Trpless CLIC1 construct was purified as described above for WT. Liposome Vesicles. Liposome vesicles (100 nm) were prepared as previously described.25 For fluorescence quenching experiments, soybean phosphatidylcholine (Sigma catalog no. P5638) and cholesterol (Sigma catalog no. C8667) were dissolved in chloroform and mixed in a molar ratio of 85:15. For the protein to lipid FRET experiments, dansyl-PE [1,2dioleoyl-sn-glycero-3-phosphoethanolamine-N-(5-dimethylamino-1-naphthalenesulfonyl), Avanti Polar Lipids catalog no. 810330] was incorporated into liposomes as the FRET acceptor. A series of acceptor-containing liposomes were formed by adding the dansyl-PE to the chloroform mix at a 0.5, 1.0, 1.5, or 2.0% molar percentage of the total lipid. The surface density of dansyl-PE molecules in acceptors per square angstrom (σ) is calculated by assuming that each cholesterol and phospholipid molecule in the cholesterol−phospholipid bilayer occupies 37 and 48 Å2 of surface area, respectively.28 The liposome vesicles were formed by evaporating the chloroform under a stream of nitrogen gas with further drying under vacuum for several hours. Lipids were then rehydrated in assay buffer [150 mM NaCl, 1 mM EDTA, and 50 mM Na3PO4 (pH 6.0)] before extrusion through a 100 nm polycarbonate filter using a LiposoFast Basic apparatus (Avestin). Unilamellar vesicles of 100 nm were confirmed by a transmission electron microscope (Philips CM10) and dynamic light scattering with a 633 nm laser source (Malvern Instruments Zetasizer Nano-ZS). No notable changes in the lipid integrity or size of liposomes were observed by transmission electron microscopy (TEM) following the addition of CLIC1 protein or oxidation (data not shown). Liposome vesicles were used within 24 h of rehydration to ensure lipid integrity was maintained. Quenching of Trp35 Fluorescence with Acrylamide. Tryptophan quenching by acrylamide was performed on both the soluble form of CLIC1 and upon the addition of phospholipid liposome vesicles. Acrylamide is an effective quenching agent of solvent-exposed Trp fluorescence. Fortu-

itously, CLIC1 possesses a single native tryptophan (Trp35) residue within the TM region. Trp35 of CLIC1 was excited at 290 nm and the observed fluorescence emission (Fobs) recorded at 345 nm. Small aliquots from a 5 M stock solution of an ultrapure acrylamide stock solution (Sigma) were added to CLIC1 and CLIC1/liposome suspensions (molar ratio of 1:200) for final acrylamide quencher concentrations ([Q]) ranging from 10 to 200 mM. Acrylamide quenching profiles were recorded for the reduced CLIC1 monomer (5 mM DTT) in the presence and absence of liposomes, and oxidized CLIC1 (addition of H2O2 to a final concentration of 2 mM), also in the presence and absence of liposomes. The absorbance of the liposome vesicles and acrylamide was less than 0.1 cm−1 at the emission wavelength of interest (345 nm). Thus, light scattering and inner filter effects were minimal on the fluorescence emission spectra, although still not ignored. All Fobs spectra were therefore corrected for dilution, inner filter effects produced by the light scattering from lipid vesicles, and the increasing concentration of acrylamide. The final background-subtracted fluorescence emission (F) was calculated according to30 ⎛ OD + ODem ⎞ ⎟ F = Fobs exp⎜ ex ⎝ ⎠ 2

(1)

where ODex and ODem are the absorbances of the sample at 290 and 345 nm, respectively. Quenching data were initially analyzed according to the Stern−Volmer relationship F0 = 1 + KSV[Q] F

(2)

where F0 is the corrected fluorescence intensity in the absence of any quenching agent and F is the background-subtracted fluorescence emission as recorded at each increasing concentration of acrylamide quencher [Q]. KSV is the Stern−Volmer quenching constant, which reflects the accessibility of Trp35 to acrylamide. However, as acrylamide does not readily partition into a membrane bilayer, the KSV value is considered to best reflect the bimolecular rate constant for dynamic quenching of a Trp residue in the aqueous phase. A modified Stern−Volmer plot was therefore employed to resolve the “accessible” (a) from the “inaccessible”, or buried (b), population of Trp35 residues that arise as a result of interaction with liposome vesicles.29 In this case, the total fluorescence in the absence of quencher (F0) is the sum of the fluorescence intensities arising from both the accessible (F0a) and buried (F0b) fractions, i.e. F0 = F0a + F0b

(3)

The modified form of the Stern−Volmer equation F0 1 1 = + F0 − F fa K a[Q] fa

30

(4)

can then be used to determine the fraction of molecules that are accessible (fa) to the quencher agent and the Stern−Volmer quenching constant (Ka) of the accessible population. In this study, the fraction of accessible residues (fa) will be the proportion of CLIC1 in solution or the “soluble fraction”. Fluorescence Emission Spectra: Trp35 to Dansyl-PE. Steady-state fluorescence intensity spectra were recorded at 22 °C using a PerkinElmer LS50B fluorimeter operating in ratio mode with spectral bandwidths of 2.8−3.2 nm for maximal fluorescence of the Trp donor. CLIC1-Trp35 to dansyl-PE 3827

DOI: 10.1021/acs.biochem.6b00080 Biochemistry 2016, 55, 3825−3833

Article

Biochemistry donor−acceptor pair samples (DA) were excited at a wavelength of 290 nm (λex), and emission (λem) was monitored from 300 to 460 nm. The dansyl label on the phospholipids does not contribute to the fluorescence signal over this range. CLIC1 and liposome vesicles were mixed in a 1:200 molar ratio in assay buffer to a final lipid mixture concentration of 1.6 mM. Each fluorescence spectrum was the accumulation of four measurements with a scan rate of 300 nm/min. Four biochemically equivalent samples were prepared in parallel: (i) Blank (B) sample, 8 μM Trpless W35F CLIC1 in the presence of 1.6 mM unlabeled lipid; (ii) Donor only (D) sample, 8 μM WT CLIC1 (Trp35 donor) in the presence of 1.6 mM unlabeled lipid; (iii) Acceptor only (A) sample, 8 μM Trpless CLIC1-W35F in the presence of labeled lipids (varying percentage incorporation of dansyl-PE acceptor, 0.5, 1.0, 1.5, or 2.0%); and (iv) Donor & Acceptor (D+A) sample, 8 μM WT CLIC1 (Trp35 donor) in the presence of labeled lipids (similarly, varying percentage incorporation of dansyl-PE acceptor, 0.5, 1.0, 1.5, or 2.0%). Following the addition of CLIC1, samples were incubated for a minimum of 0.5 h under reducing conditions or 4.0 h under oxidizing conditions, before spectral measurements were taken.27 All fluorescence emission spectra were corrected for background fluorescence and scatter by subtraction of the corresponding Blank (B) sample. FRET: “Distance of Closest Approach”. In general, FRET is a distance-dependent interaction between the electronic excited states of two chromophores in which excitation is transferred from an excited donor (D) molecule (e.g., Trp) to an acceptor (A) molecule (e.g., dansyl) without emission of a photon. The characteristic distance at which the donor fluorescence and FRET are equally probable is defined as Ro, i.e., distance between the (D+A) pair that gives a FRET efficiency of 50%. A relatively simple application of FRET in membrane systems containing proteins bearing fluorescent residues, such as Trp, is the determination of the transverse location of the latter relative to the bilayer. In our case, Trp35 is the energy transfer donor (D) to dansyl acceptor (A), which is located on the lipid headgroup (dansyl-PE). However, the situation is complex in membranes as each CLIC1-Trp35 donor molecule is likely to be surrounded by an ensemble of dansyl acceptors at noncorrelated distances (Figure 1B). Thus, analysis of energy transfer as a single (D+A) distance in this situation is not feasible as the decay of the donor emission is dependent on the topology of the system under study, as well as the concentration of acceptors. In this situation, the use of simplified analytical treatments for the energy transfer efficiency can be applied as a function of Ro and the surface density of acceptors (σ). In such a situation, the FRET efficiency on a surface is determined by the ratio of the donor emission quantum yield in the absence (QD) and presence (QDA) of the acceptor according to the equation E = 1 − QDA/QD.31 This ratio is calculated from the fluorescence emission spectra according to QD Q DA

=

FD FDA − FA

QD Q DA

⎛ πσR 2 ⎞⎛ R ⎞4 0 ⎜ 0⎟ ⎟ =1+⎜ ⎝ 2 ⎠⎝ L ⎠

(6)

The slope of the plot of QD/QDA as a function of σRo2 allows for calculation of L. For the Trp to dansyl-PE FRET probe pair, Ro is 24 Å.34 As CLIC1 is added to preformed liposome vesicles, insertion is unidirectional, and because of the effective range of this FRET probe pair, the extent of energy transfer to acceptors at the inner surface of a bilayer with a thickness of ∼50 Å is negligible and, thus, not considered in the calculations. FRET, and consequently L, was also corrected for the fractional labeling of the acceptor, typically defined in the literature as “fa” but here termed “facc” (note that fa is also commonly used in the modified Stern−Volmer equation to represent the fraction of accessible residues). This term facc is the population of D molecules that are sufficiently close to A molecules for energy transfer (E) to occur according to ⎛ F ⎞ 1 E = ⎜1 − DA ⎟ FD ⎠ facc ⎝

(7)

In our case, facc represents the fraction of CLIC1 protein that is in the proximity of the dansyl acceptors in the bilayer for FRET to occur, that is, the membrane-associated fraction. This fraction can be estimated using facc = 1 − fa

(8)

where fa was obtained from the Stern−Volmer quenching data (eq 4). To verify our method for establishing interaction of CLIC1 with the membrane, the nonionic detergent Triton X-100 (Sigma) was added to FRET samples to a final concentration of 0.2% (v/v) after the final measurement (24 h) was recorded for each data set. This was to ensure disruption of the (D+A) FRET interaction occurred and resulted in the summative recovery of each Donor and Acceptor equivalent spectrum (data not shown).

■

RESULTS Oxidation Promotes Association of CLIC1 with the Membrane: Stern−Volmer Quenching of Trp35. Acrylamide, a quencher of tryptophan fluorescence, was used to probe the solvent accessibility of the single native Trp35 residue of CLIC1 in the presence and absence of lipid vesicles. Stern−Volmer plots of the quenching of Trp35 were performed under both reducing (5 mM DTT) and oxidizing (2 mM H2O2) conditions, as shown in Figure 2. The Trp fluorescence decreased in a concentration-dependent manner upon titration with acrylamide for all samples. Samples in which the Trp residue is least accessible to quencher, or within a buried environment, result in the smallest KSV values (eq 2). The Stern−Volmer plots in Figure 2 show that the Trp35 residue is most accessible to the acrylamide quencher in solution, and under reducing conditions (KSV = 15.9 M−1; r2 = 0.985). There was no significant change in this observed KSV value upon the addition of liposome vesicles (15.5 M−1; r2 = 0.995). There was therefore little to no change in the local environment surrounding Trp35 upon addition of lipids under reducing conditions. In contrast, the Stern−Volmer quenching constant for CLIC1 under oxidizing conditions in solution was lower than that observed for reducing conditions (KSV = 13.1

(5)

where FD, FA, and FDA are the maximal fluorescence emission intensities taken at 343 nm. The “distance of closest approach (L)” between donors in an infinite plane and randomly and uniformly distributed acceptors in an infinite parallel plane is then obtained from32,33 3828

DOI: 10.1021/acs.biochem.6b00080 Biochemistry 2016, 55, 3825−3833

Article

Biochemistry

Figure 2. Stern−Volmer plots for CLIC1-Trp35 fluorescence quenching by acrylamide in the absence (empty symbols) and presence (filled symbols) of liposomes. Quenching of CLIC1-Trp35 was performed under reducing conditions (5 mM DTT, blue circles) and oxidizing conditions (2 mM H2O2, red squares). Trp35 was excited at 290 nm, and the fluorescence emission was monitored at 345 nm. Linear fits to the quenching data are shown.

Figure 3. Modified Stern−Volmer plot of acrylamide quenching of CLIC1-Trp35 in the presence of liposomes. Quenching was performed under reducing conditions (5 mM DTT, blue circles) and oxidizing conditions (2 mM H2O2, red squares). Trp35 was excited at 290 nm, and the fluorescence emission was monitored at 345 nm. The modified Stern−Volmer plot under oxidizing conditions shows evidence of two different types of fluorophores with different accessibility to acrylamide. The intercept of the slope with the y-axis (values shown) provides a measure of the accessible fraction ( fa). The fraction of nonaccessible sites was ∼5% for reducing conditions and ∼30% for oxidizing conditions.

M−1; r2 = 0.981). This small decrease in the KSV value upon oxidation of CLIC1 in the absence of lipids may be attributed to the oxidation-triggered conformational change where a more buried environment is observed for Trp35 in the dimer crystal structure.3 A considerably lower KSV value of 8.9 M−1 was obtained for CLIC1 under oxidizing conditions in the presence of liposome vesicles. This small KSV value suggests that the Trp35 residue in this lipid environment is notably more buried. However, these data also displayed a poor linear correlation (r2 = 0.948) with a notable downward curvature in the Stern− Volmer quenching profile. Hence, data for CLIC1 under oxidizing conditions in the presence of liposomes were analyzed by the Modified Stern−Volmer method. Trp35 of CLIC1 Is in Different Environments: Modified Stern−Volmer Quenching Profile. The deviation from linearity for the Stern−Volmer plots for CLIC1 under oxidizing conditions in the presence of liposomes suggests the Trp35 residue may not be in a single environment but instead has differing accessibilities according to eq 3

of the modified Stern−Volmer plot revealed the presence of a significant proportion of inaccessible Trp35 residues. The intercept of 1.45 (fa of 0.69) suggests approximately 30% of the CLIC1 protein is in a buried environment, presumably, because of interactions with the lipid bilayer. The Close Interaction between Trp35 of CLIC1 and Dansyl-Labeled Lipids: FRET Distance of Closest Approach. Our quenching studies support the model in which oxidation in the presence of lipid bilayers promotes interaction of CLIC1 with bilayers.16 Under oxidative conditions, up to 30% of the CLIC1 protein moves to a buried environment, likely because of association and insertion into the lipid bilayer. Other approaches have also shown that the interaction of CLIC1 with artificial bilayers is also enhanced when the protein is oxidized.3,10,12 However, what cannot be revealed by all these studies is how CLIC1 positions itself with respect to the membrane. Other approaches must therefore be sought. Fö rster resonance energy transfer (FRET) is a wellestablished procedure for measuring distances in biological systems and detecting conformational changes within a distance range of 20−100 Å.35,36 Here we have used the FRET technique to seek direct evidence of the interaction between CLIC1 and the membrane. Specifically, we used a simple quantitative FRET-based approach to study the spatial relationship between the single intrinsic tryptophan residue of CLIC1 (Trp35) to a lipid-labeled phospholipid (dansyl-PE). The efficiency of energy transfer from the CLIC1-Trp donor to the dansyl acceptor group naturally depends on many factors, including the extent of overlap of the Trp fluorescence emission with the absorption of dansyl, the relative orientation of the D and A transition dipoles, and the distance between the D and A labels. CLIC1 was titrated into liposome vesicles containing varying acceptor concentrations, and thus varying surface density (σ), of dansyl-labeled phospholipids (Figure 4). FRET was detected

F0 = F0a + F0b

The inaccessible environment detected under these conditions can be assumed to be as a result of the association and/ or insertion of CLIC1 into the lipid bilayer, while the accessible environment arises from CLIC1 that remains in solution. In such a case, quenching of two populations of fluorophores can then be analyzed by a modified form of the Stern−Volmer equation (eq 4) to quantify the fraction of fluorophores accessible to the quencher.29 The plots of the acrylamide quenching data with the modified Stern−Volmer equation for CLIC1 in the presence of lipids under both reducing and oxidizing environments are shown in Figure 3. Both fits are linear in nature with the yintercept providing a measure of the fraction of “accessible” residues (fa) or the soluble CLIC1 fraction. For CLIC1 under reducing conditions in the presence of lipids, the y-intercept value of 1.05 (fa = 0.95) suggests that only a very small population (∼5%), if any, of the Trp35 residue of the CLIC1 protein would be found in an inaccessible or “buried” environment. However, under oxidizing conditions, analysis 3829

DOI: 10.1021/acs.biochem.6b00080 Biochemistry 2016, 55, 3825−3833

Article

Biochemistry

From our quenching results and other studies, it was reasonable to assume that any FRET we observed was due to insertion of CLIC1 into the bilayer. Therefore, to obtain the CLIC1-dansyl distance (L) from the energy transfer observed in Figure 4, we assumed that the CLIC1 donor molecules participating in FRET are associated with lipids and are not in their alternative soluble states. Figure 5 shows the plot of the

Figure 4. FRET between CLIC1-Trp35 and dansyl-PE-labeled liposomes with an increasing concentration of acceptor label incorporation (from 0 to 2.0%). FRET Trp emission profiles under reducing conditions (5 mM DTT, solid lines) and oxidative conditions (2 mM H2O2, dotted lines). Data for the Donor only sample are shown as a black line.

Figure 5. Energy transfer between CLIC1-Trp35 and the membranepartitioned dansyl-PE acceptor label. QD/QDA values (±standard deviation) for CLIC1-Trp35 are plotted as an increasing percentage of dansyl-PE lipid for a reduced (blue circles) or oxidized (red squares) CLIC1 environment. Data plotted were corrected for the proportion of CLIC1 interacting with the lipid environment. The proportion of CLIC1 interacting with liposomes in an oxidizing environment was ∼30%. For comparison, a series of gray dashed lines show the relationship between QD/QDA and the surface density of acceptors (σ) for transverse distances (L) ranging from 15 to 35 Å (labeled on the right ordinate).

between the CLIC1-Trp35 donor and the dansyl acceptor label as observed by a decrease in the intensity of the Trp donor emission peak (∼340 nm) by dansyl, which has an absorption range of 310−380 nm. As shown in Figure 4, FRET was observed under both reducing and oxidizing conditions. As expected for CLIC1 associating or being incorporated into a bilayer environment, the higher the concentration of dansyl acceptors in liposome vesicles, the greater the amount of energy transfer observed. The extent of energy transfer was greatest upon the addition of H2O2, with up to ∼25% energy transfer measured in the presence of 2% dansyl (Figure 4). The occurrence of protein to lipid energy transfer in our experiments is simultaneously simple to detect but complex to analyze. Complexity arises as the detected donor to acceptor energy transfer may be due to the presence of multiple distances and distance distributions resulting from the nonrandom distribution of the acceptors around each donor label (Figure 1). Several different theoretical FRET models that can consider energy transfer efficiencies in situations where randomly distributed molecules exist have been developed.31−33,37,38 Following previous studies of FRET in two dimensions,31,39 we also sought to obtain an estimate of the distance between our D and A pair by making the assumption that all membrane-associated CLIC1 adopts the same conformation. This somewhat simplifies the FRET analysis as all Trp donors are then thought to be located at the same height above, or below, the membrane, thereby forming a donor plane. The second plane is then formed from the acceptors, which are localized at the aqueous membrane interface via their attachment to the headgroup of the phospholipid molecule. Previously reported quenching studies using brominated lipids indicated that Trp35 is located within the hydrophobic core of the bilayer close to the membrane surface.16 We therefore used the observed protein−lipid energy transfer in our experiments to determine a location or depth of Trp35 relative the lipid headgroups of the bilayer. Furthermore, by considering the proportion of CLIC1 that interacts with the bilayer, as suggested by our quenching data (Figure 3), and therefore can contribute to the FRET signal, a lower distance limit on this interaction could then be estimated.

ratio of the Trp35 fluorescence in the absence of acceptor to that in the presence of acceptor (QD/QDA) as a function of increasing dansyl acceptor-labeled lipid surface density (σR02). Values for the QD/QDA ratio were obtained using eq 5 following correction for the proportion of CLIC1 interacting with the lipid bilayer (eq 7), as determined by the modified Stern− Volmer analysis (Figure 3). Under reducing conditions, the amount of CLIC1 interacting with lipids ( facc) was ∼5% and, upon oxidation, increased to 30%. Standard errors shown in Figure 5 were therefore obtained by considering an error of 5% when determining the amount of CLIC1 interacting with the bilayer from the quenching experiments. Likewise, for oxidizing conditions, the FRET was corrected for the 30% proportion of CLIC1 interacting with lipid. Errors shown for oxidizing conditions were calculated by also considering an error in the interacting population of CLIC1 with the bilayer of 5% (range of 25−35%). The amount of energy transfer is observed to be near zero for a very low acceptor surface density and increases monotonically with an increase in the level of incorporation of acceptor into the bilayer. As QD/QDA was found to be proportional to σR02 over a range of σ values, and as indicated by the linear fits in Figure 5, the approximation of Dewey and Hammes33 was then used to calculate L. For comparison, theoretical plots of QD/ QDA versus σR02 were also calculated and are displayed on Figure 5 (dashed lines) for various values of transverse distance L. For FRET under reducing (circles) and oxidizing (squares) conditions, the data in Figure 5 can be accounted for by a straight line in both environments (r2 = 0.827 and 0.923, 3830

DOI: 10.1021/acs.biochem.6b00080 Biochemistry 2016, 55, 3825−3833

Article

Biochemistry respectively). Analysis of the slopes of the experimental data in relation to the theoretical lines yielded a distance (L) value of 24.8 Å for the CLIC1-Trp35 to dansyl-PE pairs under reducing conditions with upper and lower values of 24.4 and 25.3 Å, respectively. Under oxidizing conditions, the distance obtained was 15.2 Å with lower and upper values of 13.1 and 16.6 Å, respectively. This distance range suggests the Trp35 residue is located in the vicinity to the lipid headgroups. The results also suggest that the Trp35 residue moves to a position closer to the bilayer surface upon oxidation. However, FRET distances must be interpreted with a degree of caution because of assumptions made during their calculation. As with any FRET measurement, there are known limitations with assumptions that are made for κ2, which describes the freedom of rotation of the fluorescent labels. R0 values are calculated by assuming that the transition dipoles of the D and A are dynamically randomized during the excited-state lifetime of D, and an accepted value of κ2 that is then adopted is 2/3. Further, the contribution of the size of the extrinsic labels themselves can also add uncertainty to the distance L being measured. Assuming a value of 2/3 usually yields distances that differ by