Relevant Problems in Lipid Protein Interaction and Fluorescence

Partition to the Membrane

The extent of the partition of a peptide between the lipid and water phases is very variable and is usually given by the partition coefficient, Kp. The determination of Kp is usually the first step in the study of the interaction of a peptide with a given membrane system. It is defined by (for an alternative definition and interconversion see Santos et al. 2003)

K = nSiL/VL P ns,w/Vw where Vi are the volumes of the phases, and ws,i are the moles of solute present in each phase (i = W, aqueous phase; i = L, lipid phase).

Most of the applications of fluorescence spectroscopy to the quantification of the degree of interaction of the peptide with a model membrane rely on the difference in a fluorescence parameter of the partitioning molecule, such as the quantum yield, fluorescence anisotropy or fluorescence lifetime, in the two media. The additivity of the fluorescence emission intensity, I, and steady-state anisotropy, r, lead to the following equations, which are both useful in the calculation of Kp:

rw ((M)'1 -1) + rLKpeL®L/(ew®w) (VUL])"1 -1 + KpzL9L/(zw9w)

where [I] is the lipid concentration, VLis the lipid molar volume, e; is the molar absorption coefficient at the excitation wavelength and is the fluorescence quantum yield of the peptide in phase i. Equation 1.21 is also valid if lifetime-weighted quantum yields, <t> (see Sect. 1.2.1), are used instead of steady-state intensities (in these cases, 1, 7w and h should be replaced by <t>, <t>w and <t>l, respectively - the latter values are the lifetime-weighted quantum yields in pure water and membrane phases, respectively), because both <t> and I are proportional to the fluorescence quantum yield. In fact, the use of data coming from time-resolved experiments is advantageous because they are less prone to artefacts such as light scattering or inner filter effects, and are preferred to the use of steady-state intensities. Anisotropy measurements are also strongly affected by light scattering, which can be critical for the highest lipid concentrations (e.g. Castanho and Prieto 1992).

Measurements are usually carried out by titration, i.e. addition of successive amounts oflipid to the solution keeping the solute concentration constant, apart from the dilution effect. However, this can lead to significant photobleaching, especially if the intrinsic Trp or Tyr fluorescence is being measured. In that case, preparation of aliquots with constant solute concentrations and varying lipid amounts should be considered (see Santos et al. 1998 for an example).

In the case that Kv is not too high, the molar fraction of solute in water, %w> can be significant. If the peptide fluorophore fluoresces in water (e.g. Trp), the experimental spectrum, J(A)L+W, is the sum of the fractions in water, I(A)w, and in the membrane, J(A)L. The latter can be obtained from:

The recovery of the anisotropy decay parameters for the peptide in the lipid phase is more complex. However, in any case, the residual anisotropy of the bound species, (see Sect. 1.2.2), is readily obtained (see Contreras et al. 2001, for an example):

where r^ is the experimentally determined value in the presence of lipid at time ^ œ.

FRET can also be used as evidence of peptide interaction with membranes (e.g. Rosomer et al. 1996). Typically, the increased efficiency of FRET between a Trp residue in the peptide and a membrane-bound acceptor probe reveals the partition of the peptide towards the bilayer. In this way, the extent of the interaction of the HIV fusion inhibitor T1249 with POPC/cholesterol vesicles was recently established, using dehydroergosterol as acceptor, despite the fact that the

fluorescence parameters of the peptide are almost unchanged in the presence of the vesicles (Veiga et al. 2004).

Protein/Peptide Aggregation

There are two main techniques to monitor the peptide/protein aggregation or local enrichment within the membrane: FRET between identical or non-identical protein units and self-quenching of the peptide/protein fluorescence.

The basic formalisms of FRET were described in Sect. 1.2.1, where the strong dependence of this phenomenon on the distance between fluorophores was emphasized. Energy homotransfer between Trp residues is limited to peptide aggregation studies, because of the small value for R0 for the Trp/Trp pair (6-10 A; Table 1.1 and de Almeida et al. 2004). In any case, it has proved useful, as in a recent study on the organization of the peptide yM4 from the muscle AChR (de Almeida et al. 2004), for which significant aggregation in POPC/cholesterol (3:2) bilayers was excluded, as the decrease in the anisotropy could be explained assuming a uniform distribution of peptide. Energy homotransfer between extrinsic labels has also been used to monitor peptide aggregation within membranes (e.g. Runnels and Scarlata 1995).

Heterotransfer experiments, involving two distinct labelled derivatives of proteins or peptides, were described more than twenty years ago. In a now historical paper, using a simple formalism which neglected intermolecular FRET but took into account different oligomerization schemes (dimmers/trimers/tetramers), Veatch and Stryer (1977) verified that the hypothesis of formation of gramicidin dimers (donor: dansyl-gramicidin C; acceptor: 4-(diethylamino)-phenyla-zobenzene-4-sulfonyl chloride derivative of gramicidin C) led to better model fits than both trimer and tetramer formation scenarios. Aggregation of proteins of the chromaffin granule membrane, labelled with maleimide iodoaminonaph-thyl sulfonate (donor) and fluorescein mercury acetate or fluorescein-5-maleim-ide (acceptor), was qualitatively verified by Morris et al. (1982) upon addition of calcium ion. Following these studies, FRET has become an important tool in the characterization of protein/peptide aggregation (e.g. Fagan and Dewey 1986; Adair and Engelman 1994; Yano et al. 2002; Sal-Man et al. 2004). However, self-association is not the only possible cause for increased FRET efficiency between proteins/peptides in membranes, as verified in a study of the aggregation state of the M13 major coat protein using an IAEDANS-labelled protein as donor and a BODIPY-labelled protein as acceptor, carried out in vesicles of different lipid mixtures, for which it was concluded that lipid domain formation leads to protein compartmentalization and more efficient FRET, without formation of aggregates (Fernandes et al. 2003; see also below in this section).

In addition to spectroscopic studies in membrane model systems, studies of FRET in cell membranes, carried out under the microscope (including single-pair studies) are becoming common. Kubitscheck et al. (1991) have in this manner ruled out self-association of FcE receptors of type 1 on the surface of mast cells. Homoassociation of erbB2 (a member of the epidermal growth factor receptor-

type tyrosine kinase receptor family) was verified in several types of breast tumour cells (Nagy et al. 1998). Gramicidin dimerization has also been observed in single-pair FRET, and correlated with single-channel current recordings (Borisenko et al. 2003; Harms et al. 2003).

The putative existence of lipid rafts (domains enriched in (glyco)sphingolipids, cholesterol, specific membrane proteins and glycosylphosphatidylinositol (GPI)-anchored proteins) has raised, since its proposal (Simons and Ikonen 1997), considerable interest in the membrane biophysics community, because of their possible implication in a variety of cell processes (see, e.g. Simons and Toomre 2000; Anderson and Jacobson 2002, and Sect. 1.1 of this chapter). The search for lipid rafts in live cells has been carried out using FRET microscopy techniques, where the fluorophores include GPI-anchored fluorescent proteins. In some literature studies, no evidence of protein clustering using hetero-FRET was observed (Ken-worthy and Edidin 1998; Kenworthy et al. 2000; Glebov and Nichols 2004). However, homo-FRET microscopy analysis yielded density-independent values of the fluorescent anisotropy of GPI-anchored folate-receptor isoform (at variance with that from transmembrane-anchored folate-receptor isoform), which can only be explained by the clustering of the proteins in domains of < 70-nm size (Varma and Mayor 1998), which are disrupted by the removal of cholesterol. Further experiments by the same group, and attempts to model the resulting complex homo- and hetero-FRET data suggest that cell surface GPI-anchored proteins are present mainly as monomers and in a minor fraction (20-40%) as very small (< 5 nm) clusters containing no more than four GPI-anchored protein molecules (which can be of the same or different species) (Sharma et al. 2004). The authors argue that this cluster distribution would not be detected in the other less-sensitive hetero-FRET studies mentioned above.

Self-quenching of peptide intrinsic or extrinsic fluorescence has often been invoked as evidence for aggregation (e.g. Pecheur et al. 1999; Haro et al. 2003). However, to be certain of this, one should carry out both steady-state and time-resolved experiments to be able to distinguish between dynamic self-quenching (which may reflect local enrichment of peptide, but not necessarily formation of aggregates) and static self-quenching (which comes from aggregation; see Sect. 1.2.3). For example, de Almeida et al. (2004) observed an increased dynamic quenching (but no static quenching) of the above-mentioned yM4 peptide fluorescence in POPC/cholesterol mixtures, which (also taking into account the reduced FRET efficiency between the yM4 Trp residue and dehydroergosterol) they interpreted as reflecting the formation of peptide-rich patches, in which the peptide is not necessarily dimeric. Likewise, whereas enhanced dynamic self-quenching is verified for the BODIPY-labelled M13 major coat protein in mixtures of DMoPC/DOPC and DEuPC/DOPC (see Table 1.2 for abbreviations), from the application ofEq. 1.19, a static quenching component characterized by a reasonable active sphere radius (14 A in both cases, the same as in pure DOPC) is recovered. This provides evidence that there is no specific protein aggregation in these systems, and confirms that the enhanced FRET efficiency for this system is probably due to protein segregation to DOPC (the hydrophobically matching phospholipid in both mixtures)-enriched heterogeneities (Fernandes et al. 2003). On the other hand, an unreasonable active sphere radius (> 20 A) is obtained in both DEuPC

and DMoPC pure vesicles, showing that the protein aggregates when there is no hydrophobically matching phospholipids in the bilayer composition. These examples show that separation of the static and dynamic quenching components (which requires that both steady-state and time-resolved experiments be carried out) allows the identification and characterization of peptide oligomerization or segregation in the bilayer. In the case that only one of these phenomena is present, this is readily apparent from the model analysis of the data.

Besides self-quenching, a quenching-based methodology was developed (Mall et al. 2001) and applied to the study of oligomerization of peptides type Ac-LysLysGlyLeu(m)XLeu(n)LysLysAla-amide in PC bilayers, where X is Trp or the quencher 3,5-dibromotyrosine. Formation of heterodimers leads to quenching of Trp fluorescence, from which the characterization of the aggregation state is achieved, for different numbers of Leu residues in the peptides, acyl chain lengths and lipid phases.

Lipid Selectivity: the Annular Region

Proteins incorporated in model membranes containing lipids with different electrostatic properties or hydrophobic lengths may show selectivity to one lipid component at the protein-lipid interface or preferential phase partitioning, depending on the lipid miscibility (Dumas et al. 1997, Lehtonen and Kinnunen 1997; Fahsel et al. 2002). Characterization of the degree of selectivity of a protein for different lipids has been traditionally addressed using electron spin resonance spectroscopy (see Marsh and Horvath 1998, for a review). However, essentially identical information can be obtained using fluorescence spectroscopy tools.

Doxyl spin labels were used by Barrantes et al. (2000) to compare selectivity of the whole acetylcholine receptor and some of its transmembrane fragments (in all cases, Pyrene-labelled derivatives) for fatty acid and sterol spin-labelled probes. The determination of relative binding constants of different phospholipids for a given protein has been carried out by Lee and co-workers, using a methodology based on the quenching of Trp by brominated phospholipids, for more than two decades (East and Lee 1982). This method is still being actively used for selectivity studies in many lipid-protein/peptide systems. For example, (di(Ci4:i)PC) is the monounsaturated PC phospholipid with the strongest binding for the outer membrane porin OmpF from Escherichia coli, with smaller binding constants being measured for both longer or shorter fatty acyl chain lengths (O'Keeffe et al. 2000). The authors propose that in the chain-length range from C14 to C20, hydrophobic matching is achieved mostly by distortion of the lipid bilayer around the OmpF, whereas for chains longer than C20, distortion of both the lipid bilayer and of the protein is required. Whereas PC and PE bind with equal affinity to OmpF, the affinity for PG is about half that for PC.

ESR results report the fraction of motionally restricted lipids, whereas fluorescence collisional quenching depends on molecular contact. On the other hand, FRET only depends on donor-acceptor distances and is an alternative technique to quantify lipid selectivity. Several studies where this is exploited in a qualitative manner have been described (Wang et al. 1988; Narayanaswami and McNamee 1993; Albert et al. 1996). Gutierrez-Merino derived approximate analytical expressions for the average rate of FRET, <kr>, in membranes undergoing phase separation or protein aggregation (Gutierrez-Merino i98ia,b) and extended this formalism to the study of protein-lipid selectivity (Gutierrez-Merino et al. 1987). His approximate model (which considers only transfer to neighbouring acceptor molecules, and calculates the average rate of FRET, <kr>, whose relation to the FRET efficiency is not straightforward) has proved to be useful to the study of the lipid annulus around the oligomeric acetylcholine receptor (Poveda et al. 2002; Antollini et al. 1996).

Another FRET methodology, described in Sect. 1.2.1, was recently proposed and applied to lipid selectivity of the M13 major coat protein (Fernandes et al. 2004). In one experiment, the protein selectivity for the acceptor (DOPE deri-vatized with NBD at the head group) was measured in bilayers of either DOPC, DEuPC or DMoPC. In a second set of measurements, several probes were used as acceptors, all studies being made in DOPC vesicles. The probes used as acceptors were phospholipids of identical acyl chains (18:1 and 12:0) and different head-groups, derivatized with NBD at the 12:0 chain. The complete set of experiments is described in Table 1.2, where the recovered association constants (Ks, see Sect. 1.2.1) are summarized. Regarding the hydrophobic matching study, the lower value recovered in DOPC relative to those in DMoPC and DEuPC bilayers confirms the larger protein selectivity towards the hydrophobic matching unlabelled phospholipid (DOPC). On the other hand, from the varying acceptor headgroup study, a larger selectivity for the anionic labelled phospholipids is inferred. The latter results confirm those of Peelen et al. (1992), obtained using ESR and aggregated

Table 1.2. Relative association constant, Ks, of labelled phospholipids towards Ml3 major coat protein. KS(PC) is the relative association constant of (18:1-(12:0-NBD)-PC)

Labelled phospholipid

Bilayer composition

KS

KS/KS(PC)

((18:1)2-PE-NBD)

DOPC ((18:1)2PC)

1.4

-

((18:1)2-PE-NBD)

DEuPC ((22:1)2PC)

2.1

-

((18:1)2-PE-NBD)

DMoPC ((14:1)2PC)

2.9

-

(18:1-(12:0-NBD)-PE)

DOPC

2.0

1.0

(18:1-(12:0-NBD)-PC)

DOPC

2.0

1.0

(18:1-(12:0-NBD)-PG)

DOPC

2.3

1.1

(18:1-(12:0-NBD)-PS)

DOPC

2.7

1.3

(18:1-(12:0-NBD)-PA)

DOPC

3.0

1.5

Reprinted with permission from Fernandes et al. (2004). Copyright 2004 Biophysical Society.

Reprinted with permission from Fernandes et al. (2004). Copyright 2004 Biophysical Society.

protein, and in fact the relative association constants ratios (KS(PX)/KS(PC)) obtained in the two works are almost identical.

Protein/Peptide Dynamics

Time-resolved fluorescence anisotropy studies of protein and peptides in solution containing an intrinsic or extrinsic fluorophore have been carried out on numerous occasions to get an insight into the shape and size of the macromol-ecule. In addition, these experiments are capable of providing direct information on protein structural fluctuations occurring during the lifetime of the fluorophore that are believed to be fundamental for protein biological activity. However, little information is known about the structural fluctuations of protein and peptides inserted in lipid membranes due to the complex interpretation of the experimental results. Most of these works have explored the dynamics of peptides inserted in membranes but not of proteins, mainly because the main motions of proteins occur out of the time-scale of the fluorescence experiment (generally in microseconds or milliseconds). One of the first dynamical studies of peptides inserted in membranes was made by Maliwal et al. (1986), who labelled the peptide melittin with an anthraniloyl probe, analysing its anisotropy decay in solution and upon incorporation in the membrane. They found that melittin monomer is highly flexible in solution, with greater than 90% of its anisotropy being lost by the local motions. These internal motions drastically decrease upon binding to lipids, being very sensitive to the phase state of the lipid complexes. In 1988, Vogel et al. carried out some very interesting work in which the structural fluctuations of five membrane-incorporated 21 amino acid helical synthetic peptides containing a single Trp residue at sequence positions 1,6,11,16 and 21, were analysed. In all cases the experimental anisotropy decay was fitted, according to Eq. 1.14, by a sum of two exponentials with two characteristic rotational correlation times and 92, and a residual anisotropy r^ different from zero, indicative of the fact that the reorien-tational motion of the Trp side-chain, irrespective of its sequence position, is restricted on the time-scale of the fluorescence experiment. Since and differ by more than an order of magnitude, they were assumed to reflect two independent molecular motions, one due to fast movements of the peptide segment containing the Trp residue (r'(i)) and the other related to the global rotational motion of the whole peptide into the lipid bilayer (Ichiye and Karplus 1983):

where Si and S2 are the order parameters characterizing the internal and the whole peptide fluctuations. Since » the above expression can be simplified to the sum of two exponential functions plus a constant term:

r(t)=r'(t)[(l-Si)etl^ + Si] r' (f) = r (0)[(1- S12)e"i/<Pse8mental + Si2]

r (f) = r (0)[(1- Sft exp (-i/cpsegme„tal) + (1- S22)Si2 exp (-t/<pgiobal) + Sfo2] (1.26)

and the experimental rotational correlation times, ^ and can be directly related to tegmental and 9gi0bai> respectively. Using this model, Vogel et al. found that the amplitude of the fast fluctuations was modulated by the physical state of the membrane and were greater near the ends of the peptide rather than at the centre. Similar dynamical studies from different synthetic peptides containing a single Trp residue located at different positions have been carried out during the last fewyears (Clayton et al. 2000; Talbot et al. 2001; de Foresta et al. 2002).

The intrinsic fluorescence of Tyr has been rarely used to determine structural fluctuations of peptides inserted in lipid membranes due to its low extinction coefficient and its short absorption wavelength. However, as mentioned in Sect. 1.3.2, this amino acid shows a high intrinsic anisotropy and a fluorescence lifetime which should be appropriate to characterize nanosecond and subnanosecond motions of transmembrane peptides. Recently, Poveda et al. (2003) have exploited the intrinsic fluorescence of a 20 amino acid synthetic peptide having a single Tyr residue and no Trp in its sequence to characterize the interaction, insertion and dynamical properties of the peptide into anionic membranes. Following the same approach as Vogel et al. (1988) they found that the rotational mobility of Tyr examined from its fluorescence anisotropy decay could be described by two different rotational correlation times and a residual constant value. The short correlation time should correspond to fast rotation reporting on local Tyr mobility while the longest rotational correlation time and the high residual anisotropy indicated that the peptide diffused in a viscous and anisotropic medium compatible with the aliphatic region of a lipid bilayer. These results supported the hypothesis that the peptide was inserted into the membrane as a monomer to configure an intramolecular p-hairpin structure. Assuming that this hairpin structure behaved like a rigid-body they estimated its dimensions and rotational dynamics and a model for the peptide inserted into the bilayer was proposed.

Protein/PeptideTopography

The transverse location of a fluorophore in a membrane can be determined either from energy transfer or quenching methodologies, but the latter methods are by far much more common. For peptides the aim is to determine the position of the fluorophore, and from this information some global geometry for the mem-brane-peptide system would be inferred, e.g. a transmembrane situation versus a peptide located at the membrane surface.

The quenching methodologies are based on using a lipophilic membrane and inserting molecules in which the quencher moiety is located at selected positions. This can be reached either using derivatized fatty acids or lipids with nitroxide labels, or brominated. In the case that the fluorophore under study is at a similar position, the effective quencher concentration is higher, so a higher Stern-Volmer constant (KSy = kq t0 in Eq. 1.15) is obtained. This approach is also called the "differential quenching methodology", and a study with at least a pair of quenchers, one near the surface and the other at an inner location, is required. In the case that qualitative information is required, direct inspection of the different plots is enough, and there is no need to obtain time-resolved data. In all cases the effective quencher concentration should be taken into account and a reference for the nitroxide-labelled fatty acid partition constants (Eq. 1.20) is the work of Blatt et al. (1984). Lipid molar volumes can be found on the book by Marsh (1990).

However, from the above-described quenching data, quantitative information on the position of the fluorophore can be obtained using the so-called "parallax" method of Chattopadhyay and London (1987). This method, which became very popular in membrane biophysics, was later refined by the same authors (Abrams and London 1992). It was originally based on a two-dimensional sphere-of-ac-tion (the three-dimensional formalism was presented in Sect. 1.2.3), thus ignoring dynamic contributions for quenching. These contributions, in fact, exist, but if a high quencher concentration is used, their effect is not critical. Very good examples of its application to proteins/peptide topography are described in the literature, so these will not be discussed here in detail. The present state-of-the-art of this methodology was advanced by Fernandes et al. (2002), who obtained the quenching profile distribution by Brownian dynamics, making possible the recovery of both the average location and width of the fluorophore position distribution.

It is not unusual in this type of study that a downward curvature is observed in the Stern-Volmer plots. This can be due to the fraction of peptide that remains in the water under the experimental conditions where the quenching study was carried out, since this fraction is not quenched by the above-described lipophilic probes. This fraction is easily known from the peptide membrane/partition study (see Sect. 1.4.1), and in this case a Lehrer-type formalism can be fitted to the data:

Both Ksv and the non-accessible fraction/B are obtained, and the latter parameter should be close to the above-described fraction in the water, and be identical for all the quenchers used, irrespective of their location in the membrane. However this is not always verified, and, for example, in the study of inactivating peptides of the Shaker B potassium channel (Poveda et al. 2003), the clear downward curvature could not be rationalized on the basis of the very small fraction of peptide in water (5% at the most), as determined from independent methodologies. In these cases another process implying complex accessibility to the lipophilic quencher should be present (e.g. peptide aggregation), but this does not preclude obtaining the peptide location from the linear part of the plot.

Other quenching methodologies rely on the use of water-soluble quenchers such as acrylamide and iodide. Shielding from the quencher happens in the case of peptide internalization in the membrane, and in this way a smaller quenching efficiency as compared to that in water should be obtained. However, from our experience, this can only show that there is an interaction with the membrane, for which other approaches are more informative (see Sect. 1.4.1), and no real topographical information as compared to the above-described "differential quenching" can be obtained.

Regarding FRET, the approach is similar to differential quenching, but the probes deactivate the peptide's emission via a dipolar and not a molecular contact interaction. Both tryptophan and tyrosine act as energy transfer donors, since they absorb at short wavelengths, so suitable acceptors with known positions in the membrane should be used (see Table 1.1). Since the efficiency of transfer depends on the donor-acceptor interplanar separation, quantitative information can be obtained from Eqs. 1.5 and 1.6, or again it is possible to qualitativelylocate the fluorophores (Moreno and Prieto 1993). A series of stearic acids derivatized with the anthroyl chromophore is available, and is characterized in great detail in the literature (see, e.g. Blatt et al. 1984). Since there is almost invariance of the absorption spectra of the different acceptors, the Förster radius is constant for all of them, and for tryptophan as a donor it is close to R0 = 25 Â.

Modulation ofMembrane Properties

The properties of a lipid bilayer membrane can be significantly affected by the interaction with proteins or peptides. These can range from local perturbations, up to drastic alterations in the membrane structure.

Although the dividing line related to the type of induced effects is difficult to establish, the traditional classification of transmembrane and non-transmembrane peptidic systems can be useful for this purpose.

For the transmembrane systems, the so-called "annular region" has been the object of studies for a long time (Sect. 1.4.3). The annular region has been assumed to consist of a single ring of lipids around the protein, but eventually it could be a larger region. In this case the process would be lipid-mediated, and it could be considered that there was a longer-range perturbation induced by the peptide. An interaction of this type was reported for the interaction of the nicotinic acetylcholine receptor, which would generate a cholesterol-rich region in its vicinity (Poveda et al. 2002).

For the case of non-transmembrane peptidic systems bound to the membrane interface, clear phase separation in the case of anionic lipid has been reported. Examples are the interaction ofPep-i, a protein carrier (e.g. Henriques and Castan-ho 2004), the sequestering effect on PIP2 of protein basic domains (Gambhir et al. 2004), or strong alterations of the thermotropic behaviour of binary membrane model systems induced by a peptide hormone (Contreras et al. 2001). Cholera toxin subunit B was found to increase the size of the lo domains (rafts) for certain membrane lipid compositions (de Almeida et al. 2005). However, much stronger alterations than lipid lateral redistribution are also verified. Alterations of the hydrocarbon region of the membrane were reported (Giudici et al. 2003). The imbalance due to peptide interaction with the outer leaflet of a vesicle can induce an increase in the membrane curvature (Pokorny and Almeida 2004). This induced strain is concomitant with peptide aggregation and is followed by translocation of the aggregate. During this process the membrane undergoes very strong local perturbations. The same happens for the so-called "cell penetrating peptides", used to induce protein insertion into cells. For these systems vesicle aggregation, lipid fusion and asymmetric lipid flip-flop are reported (Henriques and Castanho 2004).

Was this article helpful?

0 0
Healthy Chemistry For Optimal Health

Healthy Chemistry For Optimal Health

Thousands Have Used Chemicals To Improve Their Medical Condition. This Book Is one Of The Most Valuable Resources In The World When It Comes To Chemicals. Not All Chemicals Are Harmful For Your Body – Find Out Those That Helps To Maintain Your Health.

Get My Free Ebook


Post a comment