Induction of raft-like domains by a myristoylated NAP-22 peptide and its Tyr mutant Raquel F. Epand 1 , Brian G. Sayer 2 and Richard M. Epand 1,2 1 Department of Biochemistry and Biomedical Sciences, McMaster University, Hamilton, Canada 2 Department of Chemistry, McMaster University, Hamilton, Canada NAP-22 is a 22-kDa protein found in neurons that is important for neuronal sprouting and plasticity [1]. In addition to the intact 22-kDa protein, significant amounts of N-terminal myristoylated fragments of this protein are also found in many tissues [2]. A protein with a high sequence homology to NAP-22 and prob- ably with very similar properties, cortical cytoskeleton- associated protein (CAP)-23, was first identified by Widmer and Caroni [3]. Myristoylated proteins are commonly found in cholesterol-rich domains in mem- branes [4,5]. Full length NAP-22 partitions into the low density, detergent-insoluble fraction of neuronal membranes [6], suggesting its presence in neuronal rafts. Support for this comes from fluorescence micros- copy studies using both intact biological membranes [7,8] as well as model membranes [9]. The protein binds to liposomes of phosphatidylcholine only when the bilayer contains high mol fractions of cholesterol [10,11]. Several proteins with cationic clusters, including CAP-23 as well as the MARCKS protein and GAP-43, accumulate in rafts, colocalizing with PtdIns(4,5)P2 [8]. Keywords cholesterol; domains; differential scanning calorimetry; MAS ⁄ NMR; phosphatidylinositol (4,5) diphosphate Correspondence R. M. Epand, Department of Biochemistry and Biomedical Sciences, McMaster University, Hamilton, ON Canada L8N 3Z5 Fax: +1 905 521 1397 Tel: 1 905 525 9140, extn 22073 E-mail: epand@mcmaster.ca (Received 10 December 2004, revised 2 February 2005, accepted 14 February 2005) doi:10.1111/j.1742-4658.2005.04612.x The N-terminally myristoylated, 19-amino acid peptide, corresponding to the amino terminus of the neuronal protein NAP-22 (NAP-22 peptide) is a naturally occurring peptide that had been shown by fluorescence to cause the sequestering of a Bodipy-labeled PtdIns(4,5)P2 in a cholesterol-depend- ent manner. The present work, using differential scanning calorimetry (DSC), extends the observation that formation of a PtdIns(4,5)P2-rich domain is cholesterol dependent and shows that it also leads to the forma- tion of a cholesterol-depleted domain. The PtdIns(4,5)P2 used in the present work is extracted from natural sources and does not contain any label and has the native acyl chain composition. Peptide-induced formation of a cho- lesterol-depleted domain is abolished when the sole aromatic amino acid, Tyr11 is replaced with a Leu. Despite this, the modified peptide can still sequester PtdIns(4,5)P2 into domains, probably because of the presence of a cluster of cationic residues in the peptide. Cholesterol and PtdIns(4,5)P2 also modulate the insertion of the peptide into the bilayer as revealed by 1 H NOESY MAS ⁄ NMR. The intensity of cross peaks between the aromatic protons of the Tyr residue and the protons of the lipid indicate that in the presence of cholesterol there is a change in the nature of the interaction of the peptide with the membrane. These results have important implications for the mechanism by which NAP-22 affects actin reorganization in neurons. Abbreviations DH cal , calorimetric enthalpy; Bodipy-TMR-PI(4,5)P2, BODIPY TMR-X C 6 -phosphatidylinositol 4,5-diphosphate; CAP-23, cortical cytoskeleton- associated protein (a protein expressed in chicken having a high degree of homology to NAP-22); DP, direct polarization; DSC, differential scanning calorimetry; LUV, large unilamellar vesicle; MAS, magic angle spinning; NAP-22 peptide, the myristoylated amino terminal 19 amino acids of NAP-22 (myristoyl-GGKLSKKKKGYNVNDEKAK-amide); NAP-22, neuronal axonal membrane protein, also referred to as brain acid soluble protein 1 (BASP1 protein), a 22 kDa myristoylated protein; PC, phosphatidylcholine; PO, 1-palmitoyl-2-oleoyl; PtdIns(4,5)P2, L-a-phosphatidylinositol-4,5-bisphosphate from porcine brain; SO, 1-stearoyl-2-oleoyl; T m , transition temperature. 1792 FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS The importance of electrostatic interactions in the sequestering of PtdIns(4,5)P2 by proteins with a cationic domain has been demonstrated [12]. We have also dem- onstrated the loss of ability of the NAP-22 peptide to sequester Bodipy-labeled PtdIns(4,5)P2 in the presence of high salt concentration [13]. In that work we also demonstrate specificity of the NAP-22-peptide for Bodipy-labeled PtdIns(4,5)P2 compared with Bodipy- labeled PtdIns(3,5)P2 [13]. In addition, using total inter- nal reflectance fluorescence microscopy, we have shown that the sequestering of Bodipy-labeled PtdIns(4,5)P2 into domains can be a cholesterol-dependent pheno- menon [13]. This was demonstrated using a myristoylated N-terminal peptide of NAP-22, Myristoyl-GGKLSK KKKGYNVNDEKAK-amide. It is known that in vivo, in addition to the intact NAP-22 protein, a significant amount of myristoylated N-terminal fragments of this protein are also present [2], indicating that the myristo- ylated N-terminal peptide of NAP-22, such as that used in this work, is also found physiologically. In the present work we demonstrate that not only does cholesterol affect the ability of the NAP-22-peptide to induce the formation of PtdIns(4,5)P2 domains, but it also causes the rearrangement of cholesterol leading to the forma- tion of cholesterol-depleted domains. We also test the role of the aromatic amino acid residue of the peptide in these phenomena. In addition we show that cholesterol also affects the arrangement of the peptide in the bilayer. The present study uses PtdIns(4,5)P2 from porcine brain, a natural form that has long acyl chains enriched in arachidonic acid, and it also does not con- tain any fluorescent probes. Although PtdIns(4,5)P2 from natural sources is highly enriched in arachidonoyl groups that should not interact well with liquid ordered domains of rafts, this lipid nevertheless is found in raft domains of biological membranes [14]. Results Differential scanning calorimetry (DSC) We determined the phase transitions of SOPC and mixtures of this lipid with one or more of the fol- lowing components: cholesterol, PtdIns(4,5)P2 and NAP-22-peptide, using differential scanning calori- metry (DSC). For each sample, six consecutive DSC scans were run, three heating scans and three cooling scans at a scan rate of 2 °CÆmin )1 . Sequential heating and cooling scans were reproducible. In the absence of cholesterol a prominent transition is observed in the region 0–10 °C, corresponding to the chain melting transition of SOPC. This transition is better resolved in cooling than in heating scans, since in some cases the heating scans, initiated at 0 °C, had not reached a steady-state baseline in the temperature range of the transition. The transition of POPC would have been even more difficult to measure, although POPC was used for the NMR experiments (see below) because the NMR results could be more directly compared with our earlier observations on other systems and to avoid any artefacts that may result from storing peptide-lipid mixtures that could attain the gel phase. Nevertheless, we would expect that these two lipids, SOPC and POPC, that differ only by two CH 2 groups on one of the acyl chains, would interact almost identically with peptides. One of the three cooling scans is presented for samples of different compositions (Fig. 1A). In the presence of 40 mol% cholesterol, the chain melting transition of SOPC is broadened and the enthalpy lowered (Fig. 1B). We also studied the role of the sole aromatic amino acid, Tyr, of the NAP-22-peptide by replacing it with Leu. The temperatures and enthalpies for the phospholipid chain melting transition are shown (Table 1). The temperature of the transition is shifted slightly among the different samples and is low- ered by the presence of peptide. This is probably a result of the peptide partitioning more favorably into the liquid-crystalline phase than into the gel phase. In addition, the enthalpy of this transition in the presence of cholesterol, PtdIns(4,5)P2 and the NAP-22-peptide is increased almost threefold. This indicates that cho- lesterol has been depleted from a domain of the mem- brane that can now undergo a more cooperative and endothermic transition, more like that of the pure phospholipid. Estimates of the transition enthalpy of mixtures containing cholesterol have a higher error because of the low temperature and broadness of the transition. In addition to the phospholipid transition, some samples also exhibit a transition corresponding to the polymorphic transition of anhydrous cholesterol crystals, which appears in the cooling scans at 21 °C. The enthalpy and temperature of this transition was estimated from both cooling and heating scans where this transition occurs at 38 °C. The temperature differ- ence between the heating and cooling curves is charac- teristic of this transition and is caused by the slow rate of interconversion of two forms of anhydrous choles- terol crystals [15]. The polymorphic transition of anhy- drous cholesterol crystals is most clearly seen by DSC in heating scans. We present examples of heating scans of either SOPC ⁄ cholesterol (60 : 40) or SOPC ⁄ choles- terol (50 : 50) containing either 10 or 20 mol% of the NAP-22 peptide or the Y11L NAP-22 (Fig. 2). The transition enthalpies of these peaks, obtained from the areas of the peaks, provide an estimate of the amount of crystalline cholesterol (Table 2). Pure anhydrous R. F. Epand et al. Cholesterol-dependent lipid rearrangement FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS 1793 Fig. 1. DSC cooling scans. (A) SOPC alone (curve 1) and SOPC with 0.2 mol% PtdIns(4,5)P2 added (curve 2); 0.2 mol% PtdIns(4,5)P2 and 10 mol% NAP-22-peptide added (curve 3); 10 mol% NAP-22-peptide added (curve 4). (B) SOPC ⁄ cholesterol 60 : 40 with 0.2 mol% PtdIns(4,5)P2 and 10 mol% NAP-22-peptide added (curve 1); SOPC ⁄ cholesterol 60 : 40 with 10 mol% NAP-22- peptide added (curve 2); SOPC ⁄ cholesterol 60 : 40 with 0.2 mol% PtdIns(4,5)P2 added (curve 3); SOPC ⁄ cholesterol 60 : 40 (curve 4); SOPC ⁄ cholesterol 60 : 40 with 10 mol% mutant Y11L-NAP-22-pep- tide added (curve 5) and SOPC ⁄ cholesterol 60 : 40 with 0.2 mol% PtdIns(4,5)P2 and 10 mol% mutant Y11L-NAP-22-peptide added (curve 6); Scan rate 2°Æmin )1 . Table 1. DSC Transition of SOPC. Transitions observed in cooling scans at 2°Æmin )1 of SOPC with additional components listed in the first three columns. When cholesterol is present it is at a 6 : 4 molar ratio of SOPC:cholesterol PtdIns(4,5)P2 is at 0.2% of total lipids, while NAP-22-peptide is 10 mol% of total lipids when present. Additional components DH (kcalÆmol )1 ) Cholesterol PtdIns(4,5)P2 Peptide T m (°C) - – None 6 4.0 - + None 4.8 4.6 - + NAP-22 peptide 1.7 4 - – NAP-22 peptide 1.7 3.2 + – None 6 0.35 + + None Broad Transition + – NAP-22 peptide 0.7 0.36 + + NAP-22 peptide 1.6 0.85 + – Y11L mutant No transition observed + + Y11L mutant 0.8 0.37 Fig. 2. DSC heating scans. (A) NAP-22 peptide. (B) Y11L-NAP-22- peptide. Curve 1, SOPC ⁄ cholesterol 60 : 40 with 10 mol% peptide; Curve 2, SOPC:cholesterol 60 : 40 with 20 mol% peptide; Curve 3, SOPC ⁄ cholesterol 50 : 50 with 10 mol% peptide; Curve 4, SOPC ⁄ cholesterol 50 : 50 with 20 mol% peptide. Scan rate 2°Æmin )1 . Cholesterol-dependent lipid rearrangement R. F. Epand et al. 1794 FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS cholesterol crystals have an enthalpy of 910 calÆ mol )1 [16]. In some cases the height of the transition peak is not proportional to the area because the peaks differ in their breadth (cooperativity). At SOPC ⁄ cholesterol (60 : 40) it is clear that the NAP-22 peptide promotes the formation of a larger amount of cholesterol crys- tals than does the Y11L mutant peptide. However, at SOPC ⁄ cholesterol (50 : 50) the difference between the two peptides in this regard largely disappears. In the absence of peptide (pure SOPC ⁄ cholesterol 60 : 40 or 50 : 50), no peak is observed corresponding to the formation of cholesterol crystallites (not shown), but SOPC ⁄ cholesterol (50 : 50) is close to the solubility limit of cholesterol [17]. Fluorescence quenching Addition of the Y11L-NAP-22-peptide to large uni- lamellar vesicles (LUVs) containing 0.1 mol% Bodipy- TMR-PI(4,5)P2 results in quenching of the Bodipy fluorescence (Fig. 3). Self-quenching of the fluorescence of Bodipy-TMR-PI(4,5)P2 by the MARCKS peptide has been shown to be caused by sequestering of the labeled lipid into domains [12]. We show that the quenching of the Bodipy fluorescence by the Y11L-NAP-22-pep- tide is not significantly affected by cholesterol (Fig. 3), unlike the case of the unmodified NAP-22-peptide [13] that is shown in this figure for comparison. In addition, in the presence of cholesterol the native sequence with Tyr is more potent than the Y11L- NAP-22-peptide in causing quenching of the Bodipy- TMR-PI(4,5)P2. NMR The 1 H NMR spectra of various lipid mixtures in the presence of 10 mol% of the NAP-22-peptide show predominantly the resonances of the protons of POPC (Fig. 4). Because they are well resolved from other peaks, very small peaks arising from the aroma- tic protons of Tyr can also be seen in the region of 7 p.p.m. The chemical shifts of these, as well as the major resonances from the phospholipid are summar- ized in Table 3. Each of the aromatic peaks was split into a doublet with a vicinal coupling constant of 7.5 Hz. Table 2. DSC transition of anhydrous cholesterol crystallites. Tran- sitions observed in heating scans at 2°Æmin)1 of SOPC with 40 or 50 mol% cholesterol, as well as with added peptide. % Cholesterol Peptide DH (calÆmol cholesterol )1 ) 40 10% NAP-22 peptide 20 40 20% NAP-22 peptide 33 40 10% Y11L mutant peptide 0 40 20% Y11L mutant peptide 12 50 10% NAP-22 peptide 74 50 20% NAP-22 peptide 130 50 10% Y11L mutant peptide 97 50 20% Y11L mutant peptide 110 Fig. 3. Quenching of the fluorescence emission by the NAP-22-pep- tide (dashed lines) or by the Y11L-NAP-22-peptide (solid lines) of Bodipy-TMR-PI(4,5)P2. LUVs composed of POPC with added NAP- peptide (h) or Y11L-NAP-22-peptide (d). POPC with 40 mol% cho- lesterol with added NAP-peptide (n) or Y11L-NAP-22-peptide (.). Maximum emission intensity at 571 nm is plotted against the pep- tide to lipid molar ratio (P ⁄ L). LUVs were present in the cuvette at a concentration of 50 l M and the Bodipy-labelled lipids were pre- sent as 0.1 mol percentage of the total lipid. Fig. 4. 1-D 1 H MAS ⁄ NMR spectra of several lipid mixture (as indi- cated on the right of each spectrum) and also containing 10 mol% NAP-22-peptide. PC, POPC. See Table 3 for assignments. R. F. Epand et al. Cholesterol-dependent lipid rearrangement FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS 1795 Static 31 P NMR powder patterns demonstrated that all of the samples used for magic angle spinning (MAS) studies were in bilayer arrangement (not shown). Two-dimensional 1 H MAS NOESY spectra were recorded at 25 °C for four lipid samples, each with 10 mol% NAP-22-peptide. The lipid component was either POPC; POPC with 0.2 mol% PtdIns(4,5)P2; POPC ⁄ cholesterol (6 : 4); POPC ⁄ cholesterol ⁄ PtdIns(4, 5)P2 (60 : 40 : 0.2). No resonances assignable to cho- lesterol could be detected either with or without the peptide, in agreement with earlier observations [18]. The peptide is in relatively low concentration and many of its resonances would not be well resolved from those of the lipid, except for the Tyr aromatic protons. We have focused on the relative strength of the NOE interactions between the Tyr aromatic protons and other atoms. Stronger NOE interactions between two atoms are a measure of their closer approach. These are observed as peaks in the 2D NOESY spectra. Sli- ces of the NOESY at the resonance position of the aromatic protons are shown for several lipid mixtures containing 10 mol% NAP-22-peptide using a delay time of 50 or 300 ms (Fig. 5). The longer delay times can result in larger NOEs by allowing more complete energy transfer through dipolar interactions. However, longer delay times can also allow NOE effects to be observed between two groups that are not physically close to each other as a result of spin diffusion. It is likely, however, that at least with a 50-ms delay time, spin diffusion does not contribute greatly to the observed dipolar interactions [19]. Qualitatively one can conclude that the aromatic residue of the peptide inserts into the bilayer with all of the lipid mixtures, as indicated by the fact that most of the protons of the phospholipid show cross-peaks with the aromatic pro- tons. In addition, the presence of cholesterol allows a closer proximity of the Tyr side chain of the peptide with the terminal methyl group of the acyl chain of the lipid as shown by the observation that the intensity of the cross-peak with the terminal CH 3 group (at 1 p.p.m.) relative to that of the CH 2 resonances at 1.4 p.p.m. is larger in the presence than in the absence of cholesterol (Fig. 5). In order to specifically assess how PtdIns(4,5)P2 affects the location of the Tyr residue of the NAP- 22-peptide in the membrane, we calculated difference spectra by taking a pair of spectra that were identical except for the presence of PtdIns(4,5)P2. Prior to sub- traction the two spectra were adjusted for small dif- ferences in intensity and resonance position so as to visually give the maximal overlap of the two spectra. Difference spectra were calculated for pairs of spectra with PtdIns(4,5)P2 minus the spectra for the same lipid mixture without PtdIns(4,5)P2 using a delay time of 50 ms (Fig. 6) or 300 ms (Fig. 7). Slices from the 2D NOESY spectrum at the two resonance posi- tions for the aromatic residues for pairs of samples with or without cholesterol are shown. Peaks of higher intensity, such as the aromatic peaks at 6.9 and 7.2 p.p.m., the HDO peak at 4.8 p.p.m. and the quaternary ammonium peak at 3.3 p.p.m. show some residual intensity in the difference spectra, that we do not consider significant because the intensity of the Table 3. Assignment of 1 H NMR resonances. Assignment a Chemical Shift (p.p.m.) Meta phenolic CH 7.2 Ortho phenolic CH 6.9 Glycerol C2 5.4 HDO 4.8 Glycerol C3 4.5 Choline a 4.4 Glycerol C1 4.1 Choline b 3.7 Quaternary CH 3 3.3 CH 2 CO 2.4 CH 2 CCO 1.7 CH 2 1.4 CH 3 1.0 a Groups correspond to POPC, except for HDO that are the residual protons of the water and the phenolic CH of the Tyr aromatic pro- tons from the NAP-22-peptide. Fig. 5. 1D slices from the MAS 1 H NOESY spectrum at the chemical shifts of the aro- matic protons using a mixing time of 50 and 300 ms. The slice at 7.2 p.p.m. corresponds to the meta CH of Tyr and that at 6.9 p.p.m. to the phenolic ortho CH. Cholesterol-dependent lipid rearrangement R. F. Epand et al. 1796 FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS difference spectra peaks represent a small fraction of the original peak and may arise from imperfect align- ment of the two spectra. In most cases, these reson- ance positions show closely spaced peaks of positive and negative sign, indicating a small difference in chemical shift between the two spectra. However, with cholesterol, the difference spectra using a 50-ms delay time clearly shows several positive peaks in the region 1–2 p.p.m. (Fig. 6, left). This indicates that in the presence of cholesterol, PtdIns(4,5)P2 allows a closer proximity of the Tyr side chain of the peptide with the methylene groups of the acyl chains of the lipid. This phenomenon is not observed in the absence of cholesterol (Fig. 6, right). However, for the samples without cholesterol, for the slice at 6.9 p.p.m., the difference spectra shows a decrease of intensity at the resonance position of the CH 2 groups, compared to the samples with cholesterol, and an increase of the peak intensity at the resonance posi- tion of the terminal methyl group at 1 p.p.m. This is particularly clear from the spectra using a 300-ms delay time (Fig. 7, lower right spectrum). This indi- cates the ortho protons of the Tyr side chain gain closer approach to the terminal methyl groups of the acyl chains, on a millisecond time scale, in the pres- ence of PtdIns(4,5)P2, but not cholesterol. It should be pointed out, however, that there could be contri- butions to the weaker signals in the difference spectra from cross-peaks between the aromatic protons and aliphatic protons of the peptide that are not well resolved in the 1D spectrum. Even if there was such a contribution, the results would still indicate that PtdIns(4,5)P2 affects the geometrical relationship between the peptide and lipid. Peptide-induced changes in the chemical shift of the carbon atoms as measured by 13 C direct polarization (DP) ⁄ MAS indicate that the peptide affects the chem- ical shift at many positions in the lipid molecule. Such shifts are usually interpreted in terms of ring- current effects caused by the aromatic group of the peptide. However, it is unlikely that similar ring- current effects could occur at both the glycerol C3 and terminal methyl group of the acyl chain in the absence of cholesterol or at the glycerol C2 and the cholesterol C18 in the presence of cholesterol (Table 4). We suggest that in addition to ring-current effects there are peptide-induced changes in lipid packing and interaction with water. It is known that dehydration will cause an upfield chemical shift of 13 C resonances [20]. Fig. 6. Calculated differences of spectra shown in Fig. 4 using 50 ms delay time. Difference of spectra with PtdIns(4,5)P2 minus the spectra of the same mixture without PtdIns(4,5)P2. Pairs of spectra are either from samples with cholesterol (+ cholesterol) or without cholesterol (– cholesterol). Resonance position of the slice is indicated on the graph. Fig. 7. Same as Fig. 6 but for data with 300 ms delay time. R. F. Epand et al. Cholesterol-dependent lipid rearrangement FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS 1797 Discussion In addition to electrostatic interactions with PtdIns(4, 5)P2, the NAP-22-peptide has two features that can contribute to its interaction with membranes. These features include membrane interactions of the N-ter- minal myristoyl group and the phenolic side chain of the Tyr residue, both of which are hydrophobic moiet- ies known to partition into membranes [21,22]. With regard to myristoylation of NAP-22, this post- translational modification has been found to be required for the interaction of this protein with mem- branes [23]. In addition, the protein has no hydropho- bic segment and its free energy of partitioning into membranes can be accounted for by the insertion of its myristoyl group [11]. Myristoylated proteins are often found to sequester to cholesterol-rich domains in bio- logical membranes. We suggest that this group contri- butes to the cholesterol modulation of the membrane interaction of the NAP-22-peptide. We have directly tested the role of the Tyr residue in the membrane interactions of the NAP-22-peptide by comparing it with a myristoylated peptide in which the sole Tyr residue was substituted with Leu. The NAP- 22-peptide is more effective in sequestering cholesterol than is the Y11L mutant. From the DSC results, this is indicated by fact that the NAP-22-peptide is able to promote the formation of a greater cholesterol- depleted domain as shown by the higher enthalpy of the SOPC transition in the presence of this peptide compared with the Y11L-NAP-22-peptide, both in the presence and absence of PtdIns(4,5)P2 (Table 1). In addition, in mixtures of SOPC ⁄ cholesterol (60 : 40) the NAP-22-peptide induces the formation of more anhy- drous cholesterol crystals than the Y11L mutant (Table 2). We suggest that these crystals form because cholesterol surpasses its solubility limit in the mem- brane in cholesterol-rich domains whose formation is promoted by the peptides. It should also be pointed out that any cholesterol that is directly bound to a peptide would be less likely to form crystals. However, the amount of cholesterol is much larger than the amount of peptide, so that most of the cholesterol in these domains will not be binding directly to the pep- tide. The Y11L-NAP-22-peptide is slightly less effective than the NAP-22-peptide in sequestering Bodipy- TMR-PI(4,5)P2 in the presence of cholesterol (Fig. 3), but more dramatic is that the cholesterol dependence of Bodipy-TMR-PI(4,5)P2 sequestering is almost com- pletely eliminated. Tyr is an essential element in the CRAC motif, suggested to be responsible for choles- terol recognition [24]. Although the NAP-peptide does not have other elements required for a CRAC motif, the sole presence of an aromatic residue may be a con- tributing factor for cholesterol interaction. We have previously shown that the aromatic side chains of the Table 4. Peptide-induced 13 C chemical shift differences of lipid resonances. Data show the chemical shift differences in p.p.m. for the indi- cated lipid mixture between the pure lipid and lipid with 10 mol% NAP-22 peptide. Cholesterol present in equimolar ratio with POPC and PtdIns(4,5)P2 as 0.2 mol%. ND, Not determined because of poor resolution of the peak. Assignment Chemical shift (p.p.m.) POPC POPC + PtdIns(4,5)P2 POPC ⁄ cholesterol (1 : 1) POPC ⁄ cholesterol (1 : 1) + PtdIns(4,5)P2 Acyl C ¼ O 174 0.07 0.03 ) 0.10 ) 0.07 Acyl C ¼ C 130.0 0.07 0.04 ) 0.07 ) 0.05 Acyl C ¼ C 129.6 0.08 0.05 0.05 0.04 Glycerol C2 71 0.04 0.00 ) 0.15 ) 0.11 Choline b 67 0.04 0.04 ) 0.02 ) 0.02 Glycerol C3 64 0.08 0.10 0.05 ND Glycerol C1 63 0.05 0.00 0.00 ND Choline a 60 0.06 0.03 ) 0.01 ) 0.04 Cholesterol C14 ⁄ 17 57 – – ) 0.03 0.00 Quaternary CH 3 54.5 0.03 0.01 ) 0.04 ) 0.03 Cholesterol C9 51 – – ND 0.03 Cholesterol C13 ⁄ C4 43 – – ) 0.03 ) 0.01 Cholesterol C10 37 – – 0.02 0.00 Acyl C2 35 0.06 0.05 ) 0.06 0.01 Cholesterol C25 28.5 – – ) 0.06 ) 0.04 Cholesterol C19 20 – – ) 0.04 ) 0.04 Cholesterol C21 19.5 – – ) 0.05 ) 0.01 Acyl terminal methyl 14 0.08 0.07 0.03 ) 0.02 Cholesterol C18 13 – – ) 0.14 ) 0.15 Cholesterol-dependent lipid rearrangement R. F. Epand et al. 1798 FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS peptide N-acetyl-LWYIK-amide can interact with the A ring of cholesterol [25]. The Y11L-NAP-22-peptide is also less potent in inducing the formation of choles- terol clusters than is the NAP-22-peptide. This is indi- cated by the observation that no cholesterol crystallites are observed with SOPC and 40 mol% cholesterol in the presence of Y11L-NAP-22-peptide, while they do form in the presence of the NAP-22 peptide. In addi- tion, there is no evidence for the formation of a choles- terol-depleted phase with the Y11L-NAP-22-peptide, which would result in a more cooperative chain melt- ing transition of SOPC with higher enthalpy (Table 1 and Fig. 1). Peptides with cationic clusters, even simple oligo- mers of Lysine, will sequester the polyanionic PtdIns(4,5)P2 [12,26–28]. The unique feature of the NAP-22 peptide is that this clustering of PtdIns(4,5)P2 is strongly dependent on the presence of cholesterol [13]. A well studied peptide that does not require cho- lesterol for sequestering PtdIns(4,5)P2 is the MARCKS peptide [12]. There are several differences between the MARCKS peptide and the NAP-22-peptide. The MARCKS peptide has 13 positive charges compared to only seven cationic residues for the NAP-22-peptide. As a consequence, electrostatic interactions alone will provide a stronger driving force for the MARCKS peptide to sequester PtdIns(4,5)P2, compared with the NAP-22-peptide. Although the MARCKS protein, like NAP-22, is N-terminally myristoylated, the longest cluster of five Lys residues in MARCKS begins at resi- due 86, far removed from the amino-terminal myris- toyl group. Also the model MARCKS peptide is not myristoylated, unlike the peptides used in the present work. With regard to aromatic residues, the MARCKS peptide has five Phe residues while the NAP-22-peptide has only one Tyr. In the case of MARCKS peptide, the major cross-peak between the aromatic resonance of the peptide and the lipid protons is with the methy- lene peak [19], while in the case of the NAP-22-peptide there is a more intense cross-peak with the terminal methyl group of the acyl chain, particularly when cho- lesterol is present (Fig. 5). The depth of insertion is not greatly altered when all but two of the Phe resi- dues of the MARCKS peptide are replaced with Ala [19]. However, when all five Phe residues are replaced with Ala, spin label studies indicate less penetration of the peptide into the membrane [29]. Nevertheless, this Ala substituted peptide has only somewhat diminished ability to sequester PtdIns(4,5)P2. This is not that dif- ferent from the effects of removal of the Tyr residue from the NAP-22 peptide when studied in membranes containing cholesterol. However for membranes devoid of cholesterol, the Y11L-NAP-22-peptide has greater activity in sequestering PtdIns(4,5)P2 than the unmodi- fied NAP-22 peptide. We suggest that the ability of peptides to form domains of PtdIns(4,5)P2 is a conse- quence of the combined interactions of the cationic cluster of amino acid residues and the insertion of hydrophobic amino acids into the membrane. In some cases, the insertion of groups that promote the forma- tion of cholesterol-rich domains will result in the pref- erential sequestering of PtdIns(4,5)P2 into one of the domains. This would be a mechanism additional to the direct electrostatic interaction between the peptide and PtdIns(4,5)P2. When electrostatic interactions predominate, there is sequestering of PtdIns(4,5)P2, independently of the nat- ure of the surrounding lipid. However, when the elec- trostatic interactions are reduced, as it is in NAP-22 compared with the MARCKS peptide, then sequester- ing of PtdIns(4,5)P2 is also affected by the insertion of hydrophobic moieties into the membrane that change the depth of burial of the peptide, the orientation of the peptide with respect to the membrane, and the lat- eral distribution of lipids into domains through hydro- phobic interactions. These hydrophobic interactions alone are insufficient in the case of the Y11L-NAP-22- peptide to modulate the sequestering of cholesterol. In the case of NAP-22, the combined interactions of the myristoyl group, the Tyr side chain and the cationic cluster in the peptide, result in a cholesterol-dependent sequestering of PtnIns(4,5)P2 into domains. There is also a structural aspect that makes NAP-22 unusual. Many proteins are N-terminally myristoylated [30] but only a few have in addition, clusters of cationic residues comprised of four or more Lys or Arg residues in sequence. One of the few examples we have found is the membrane fusion protein, p15, of baboon reovirus that is both myristoylated and has a cluster of four cationic residues [31]. Two other examples we have discsussed earlier are MARCKS and NAP-22. The structural difference between these two proteins is that the cationic cluster of NAP-22 is close to the myristoyl group at the amino terminus. This is not the case for MARCKS. Since myristoylation is a factor that causes proteins to sequester into raft domains, it would seem a priori more likely that sequestering of PtnIns(4,5)P2 would be coupled to translocation to a cholesterol-rich domain for NAP-22 than for MARCKS, as is found. The rearrangement of PtdIns(4,5)P2 and cholesterol in a membrane caused by the presence of NAP-22 pro- vides a mechanism by which this protein can affect the actin cytoskeleton. PtdIns(4,5)P2 plays an important role in the attachment of the cytoskeleton to the plasma membrane as well as affecting actin dynamics [32]. Since NAP-22 causes the sequestering of both R. F. Epand et al. Cholesterol-dependent lipid rearrangement FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS 1799 cholesterol and PtnIns(4,5)P2 into domains, we suggest that the protein recruits more PtnIns(4,5)P2 into raft- like domains. This will result in an increase in the interactions between the cytoskeleton and plasma membrane occurring at rafts and hence the rearrange- ment of the spatial distribution of the cytoskeleton. In neurons, several proteins including NAP-22, GAP-43 and MARCKS, affect the efficiency of raft dependent signaling [33]. Both the kinase that catalyses the syn- thesis of PtdIns(4,5)P2 [34] as well as the phosphatase that degrades it [35], affect cytoskeletal organization. NAP-22 together with related proteins, function to enhance the accumulation and assembly of PtdIns(4, 5)P2-rich raft domains [36]. During neuronal develop- ment, axonal elongation and branching are regulated by the activity of PI(4)P5 kinase [37], an enzyme that synthesizes PtdIns(4,5)P2. Thus, the amount and distri- bution of PtdIns(4,5)P2 will regulate cytoskeletal dynamics, which in turn will affect neuronal growth and development. CAP-23 accumulates in the neuronal growth cone and has a marked effect on the rearrange- ment of the actin cytoskeleton [38]. An early conse- quence of CAP-23 accumulation is an increase in dynamic actin structures and the loss of more stable actin filaments such as stress fibers. We can use this simplified system to identify certain molecular interactions that we can suggest form the basis for events that are observed at the cellular level. In this work we use a 19 amino acid lipopeptide cor- responding to the amino terminus of NAP-22. With this peptide, the consequences of the rearrangements of PtdIns(4,5)P2 we observe by fluorescence or by DSC are significantly greater than we observe with the intact protein. It is known that there are N-terminal fragments of NAP-22 present in cells [2]. Furthermore, a construct composed of the N-terminal segment of CAP-23 and containing 40 amino acids arranges in a punctate pattern on the cell surface and is associated with the cytoskeleton. Like the full length protein, this short construct produces marked changes in cell mor- phology but unlike the full length protein, it does not produce blebbing [38]. It has been estimated that PtnIns(4,5)P2 comprises 0.3–1.5% of the phospholipid of the plasma membrane of mammalian cells [12]. If dissolved in the total cell volume, this amount of PtnIns(4,5)P2 would have a concentration in the range 2–30 lm, although the PtnIns(4,5)P2 varies consider- ably among cell types and is particularly low in some cells [39]. Nevertheless, our use of 0.2 mol% PtnIns(4,5)P2 in the model membranes is within the physiological range. In comparison, in the developing brain NAP-22 comprises 0.4–0.8% of the total protein, corresponding to a concentration of 20–40 lm [40]. Thus, there are comparable amounts of PtnIns(4,5)P2 and NAP-22 in the cell and the ratio is within the range used in our work. Since NAP-22 binds to PtnIns(4,5)P2 by nonspecific electrostatic interactions, one molecule of NAP-22 can promote the formation of a domain of many molecules of PtnIns(4,5)P2 [27]. Thus not all of the NAP-22 has to be bound to PtnIns(4,5)P2 in order for a major fraction of this lipid to be sequestered into a domain. This is different from proteins with specific folded domains that bind PtnIns(4,5)P2 in a stoichiometric fashion [41]. The greater potency of the N-terminal peptide in forming domains would suggest that membrane lipid domain formation may be facilitated by proteolytic processing of NAP-22. Myristoylated proteins interacting with membranes through both electrostatic interactions as well as insertion of a myristoyl group, can be dissoci- ated from the membrane by proteolytic cleavage [42]. It is possible that this is an example of the opposite, i.e. proteolytic cleavage would cause increased seques- tration to the membrane by removing the anionic por- tion of the protein that would repel anionic lipids. The pI of rat NAP-22 is only 4.5. Another indication of the importance of the amino terminal fragment of NAP-22 is that the first 21 amino acids are invariant among NAP-22 of several mammalian species and this segment differs by only one residue with chicken NAP-22 (CAP-23). Thus both cholesterol and PtdIns(4,5)P2 affect the location of the NAP-22-peptide in a bilayer. The lipo- peptide has little capability of inducing phase separ- ation in mixtures of SOPC and cholesterol, but with addition of PtdIns(4,5)P2 there is a cholesterol- dependent separation into a cholesterol enriched and a cholesterol-depleted domain. This segregation is repre- sented in the drawing in Fig. 8 (not drawn to scale). These results demonstrate how sensitive the interaction Fig. 8. Schematic representation of the domain enrichment caused by the peptide (red) in the presence of cholesterol (blue) and PtdIns(4,5)P2 (green). The other lipid headgroups are presented in grey and the acyl chains in orange. The clustering of charges in the peptide permits interaction with the negative charges on the head- group of PtdIns(4,5)P2 concomitantly resulting in the redistribution of cholesterol. Cholesterol-dependent lipid rearrangement R. F. Epand et al. 1800 FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS of even small peptides with membranes is to the lipid composition of the membrane. Experimental procedures Materials The synthetic lipopeptide with the sequence: myristoyl- GGKLSKKKKGYNVNDEKAK-amide, corresponding to the 19 amino terminal residues of NAP-22, as well as a variant of this lipopeptide, Y11L were purchased from Bio- Source International (Hopkinton, MA, USA). Phospho- lipids and cholesterol were purchased from Avanti Polar Lipids (Alabaster, AL, USA). PtdIns(4,5)P2 was purified from porcine brain. Bodipy-TMR-PI(4,5)P2 was purchased from Molecular Probes (Eugene, OR, USA). Preparation of samples for DSC and NMR experiments Lipid components were codissolved in chloroform ⁄ meth- anol (2 : 1, v ⁄ v). For samples containing peptide, an ali- quot of a solution of the peptide in methanol was added to the lipid solution in chloroform ⁄ methanol. The amount of peptide used was monitored by the absorbance at 280 nm using an extinction coefficient calculated from the amino acid composition [43]. The solvent was rapidly evaporated at 30 °C under a stream of nitrogen with constant rotation of a test tube to avoid separation of lipid components [12] and to deposit a uniform film of lipid over the bottom third of the tube. Last traces of solvent were removed by placing the tube under high vacuum for at least 2 h. The lipid film was then hydrated with 20 mm Pipes, 1 mm EDTA, 150 mm NaCl with 0.002% NaN 3 , pH 7.40 and suspended by intermittent vortexing and heating to 50 °C over a period of 2 min under argon. Samples used for NMR analysis were hydrated with the same buffer made in 2 H 2 O and adjusted to a pH meter reading of 7.0 (pD ¼ 7.4) and incubated at least 24 h at 4 °C to allow conversion of any anhydrous cholesterol crystals to the monohydrate form. For the NMR measurements, the samples were first spun in an Eppendorf centrifuge at room temperature. The resulting hydrated pellet was transferred to a 4 mm zirco- nia rotor with the 12-lL Kel-F insert, attempting to pack the maximal amount of lipid into the rotor while keeping it wet. DSC Measurements were made using a Nano Differential Scan- ning Calorimeter (Calorimetry Sciences Corporation, American Fork, UT, USA). The scan rate was 2 ° CÆmin )1 and there was a delay of 5 min between sequential scans in a series to allow for thermal equilibration. The features of the design of this instrument have been described [44]. DSC curves were analyzed by using the fitting program, DA-2, provided by Microcal Inc. (Northampton, MA, USA) and plotted with origin, version 5.0. Preparation of LUV for fluorescence spectroscopy A solution of POPC and 0.1 mol% Bodipy-TMR-PI(4,5)P2 with or without 40 mol% cholesterol was prepared in chlo- roform ⁄ methanol (2 : 1) and the lipid deposited on the walls of a glass test tube by solvent evaporation with a stream of nitrogen gas. Last traces of solvent were then removed by evaporation for 2 h under vacuum. Films were hydrated with a 10 mm Hepes buffer pH 7.4 containing 1mm EDTA and 140 mm NaCl. The lipid suspensions were further processed by five cycles of freezing and thawing, fol- lowed by 10 passes through two stacked 0.1 lm polycar- bonate filters, using a Lipex extruder [45], to convert the lipid suspension to LUVs. The content of lipid phospho- rous was determined by the method of Ames [46]. Fluorescence quenching Fluorescence measurements were made in silanized glass cuvettes containing 2 mL of the appropriate buffer, at 25 °C, under constant stirring with Teflon magnets. An amount of LUVs were added to the cuvette and then titra- ted with successive additions of small aliquots of peptide solution, using silanized Eppendorf tips. Peptide solutions were made in the appropriate buffer and the peptide con- centration was quantified by absorbance at 275 nm. Peptide solutions were kept in silanized containers at 4 °C until used. The excitation and emission monochromators were set at 542 nm and 571 nm, respectively, with a 500-nm cut-off fil- ter in the emission path. The excitation and emission band- pass slits were set at 4 nm. Cuvettes were maintained in the dark with the shutters closed between additions of peptide; the shutter was toggled only at the beginning of the record- ing of each emission scan, to prevent photobleaching of the probe. Two independent determinations were performed with each batch of LUVs. The corresponding set of titra- tion curves with buffer not containing peptide were subtrac- ted from the titration with peptide. 1 H NOESY MAS/NMR High resolution MAS spectra were acquired using a spin- ning speed of 5.5 kHz in a Bruker AV 500 NMR spectro- meter. Probe temperature was 24 ± 1 °C. The 2D NOESY spectra were obtained using delay times of 50 and 300 ms. Resonances were assigned based on reports of phosphat- idylcholine [18], cholesterol [47] and amino acid residues [48]. R. F. Epand et al. Cholesterol-dependent lipid rearrangement FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS 1801 [...]... into membrane microdomains of live cells Science 296, 913–916 6 Maekawa S, Kumanogoh H, Funatsu N, Takei N, Inoue K, Endo Y, Hamada K & Sokawa Y (1997) Identification of NAP-22 and GAP-43 (neuromodulin) as major protein components in a Triton insoluble low density fraction of rat brain Biochim Biophys Acta 1323, 1–5 7 Terashita A, Funatsu N, Umeda M, Shimada Y, OhnoIwashita Y, Epand RM & Maekawa S (2002)... CAP23 and GAP43 in actin regulation, neurite outgrowth, and anatomical plasticity J Cell Biol 149, 1443–1454 2 Zakharov VV, Capony JP, Derancourt J, Kropolova ES, Novitskaya VA, Bogdanova MN & Mosevitsky MI (2003) Natural N-terminal fragments of brain abundant myristoylated protein BASP1 Biochim Biophys Acta General Subj 1622, 14–19 3 Widmer F & Caroni P (1990) Identification, localization, and primary... (4,5) P (2) rafts EMBO J 20, 4332–4336 Hernandez-Deviez DJ, Roth MG, Casanova JE & Wilson JM (2004) ARNO and ARF6 regulate axonal elongation and branching through downstream activation of phosphatidylinositol 4-phosphate 5-kinase alpha Mol Biol Cell 15, 111–120 Wiederkehr A, Staple J & Caroni P (1997) The motilityassociated proteins GAP-43, MARCKS, and CAP-23 share unique targeting and surface activity-inducing... NL, Maekawa S, Epand RM & Jacobson K (2003) Binding of NAP-22, a calmodulin-binding neuronal protein, to raft-like domains in model membranes Biochemistry 42, 4780– 4786 Maekawa S, Sato C, Kitajima K, Funatsu N, Kumanogoh H & Sokawa Y (1999) Cholesterol-dependent localization of NAP-22 on a neuronal membrane microdomain (raft) J Biol Chem 274, 21369–21374 Epand RM, Maekawa S, Yip CM & Epand RF (2001)... (4,5) P-2 into membrane domains by the N-terminal fragment of NAP-22 (neuronal axonal myristoylated membrane protein of 22 kDa) Biochem J 379, 527–532 Pike LJ, Han X, Chung KN & Gross RW (2002) Lipid rafts are enriched in arachidonic acid and plasmenylethanolamine and their composition is independent of caveolin-1 expression: a quantitative electrospray ionization ⁄ mass spectrometric analysis Biochemistry... Nasuhoglu C, Feng S, Mao J, Yamamoto M, Yin HL, Earnest S, Barylko B, Albanesi JP & Hilgemann DW (2002) Nonradioactive analysis of phosphatidylinositides and other anionic phospholipids by anion-exchange highperformance liquid chromatography with suppressed conductivity detection Anal Biochem 301, 243–254 Maekawa S, Murofushi H & Nakamura S (1994) Inhibitory effect of calmodulin on phosphorylation of. .. formation of cholesterol-rich domains Biochemistry 40, 10514–10521 Gambhir A, Hangyas-Mihalyne G, Zaitseva I, Cafiso DS, Wang J, Murray D, Pentyala SN, Smith SO & McLaughlin S (2004) Electrostatic sequestration of PIP2 on phospholipid membranes by basic ⁄ aromatic regions of proteins Biophys J 86, 2188–2207 Epand RM, Vuong P, Yip CM, Maekawa S & Epand RF (2004) Cholesterol-dependent partitioning of Ptdlns... Epand RF, Maekawa S & Epand RM (2003) Specificity of membrane binding of the neuronal protein NAP-22 J Membr Biol 193, 171–176 Li H, Yao Z, Degenhardt B, Teper G & Papadopoulos V (2001) Cholesterol binding at the cholesterol recognition ⁄ interaction amino acid consensus (CRAC) of the peripheral-type benzodiazepine receptor and inhibition of steroidogenesis by an HIV TAT-CRAC peptide Proc Natl Acad... of Peptides with Basic and Aromatic Residues to Bilayer Membranes: Phenylalanine in the myristoylated Alanine-rich C kinase substrate effector domain penetrates into the hydrophobic core of the bilayer J Biol Chem 278, 21459–21466 Kimura T, Okamura E, Matubayasi N, Asami K & Nakahara M (2004) NMR study on the binding of FEBS Journal 272 (2005) 1792–1803 ª 2005 FEBS R F Epand et al 21 22 23 24 25 26 27... USA 98, 1267–1272 Epand RM, Sayer BG & Epand RF (2003) Peptideinduced formation of cholesterol-rich domains Biochemistry 42, 14677–14689 McLaughlin S, Wang J, Gambhir A & Murray D (2002) PIP (2) and proteins: interactions, organization, and information flow Annu Rev Biophys Biomol Struct 31, 151–175 Wang J, Gambhir A, McLaughlin S & Murray D (2004) A computational model for the electrostatic sequestration . Induction of raft-like domains by a myristoylated NAP-22 peptide and its Tyr mutant Raquel F. Epand 1 , Brian G. Sayer 2 and Richard M. Epand 1,2 1. amino terminal fragment of NAP-22 is that the first 21 amino acids are invariant among NAP-22 of several mammalian species and this segment differs by only