A comparison of the location in membranes of curcumin and curcumin-derived bivalent compounds with potential neuroprotective capacity for Alzheimer’s Disease

Abstract

Curcumin and two bivalent compounds, namely 17MD and 21MO, both obtained by conjugation of curcumin with a steroid molecule that acts as a membrane anchor, were comparatively studied. When incorporated into 1,2-dipalmitoyl-sn-glycero-3-phosphocholine the compounds showed a very limited solubility in the model membranes. Curcumin and the two bivalent compounds were also incorporated in membranes of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine and quenching the fluorescence of pure curcumin or of the curcumin moiety in the bivalent compounds by acrylamide it was seen that curcumin was accessible to this water soluble quencher but the molecule was somehow located in a hydrophobic environment. This was confirmed by quenching with doxyl-phosphatidylcholines, indicating that the curcumin moieties of 17MD and 21MO were in a more polar environment than pure curcumin itself. NOESY-MAS-¹HNMR analysis supports this notion by showing that the orientation of curcumin was parallel to the plane of the membrane surface close to C2 and C3 of the fatty acyl chains, while the curcumin moiety of 17MD and 21MO positioned close to the polar part of the membrane with the steroid moiety in the centre of the membrane. Molecular dynamics studies were in close agreement with the experimental results with respect to the likely proximity of the protons studied by NMR and show that 17MD and 21MO have a clear tendency to aggregate in a fluid membrane. The anchorage of the bivalent compounds to the membrane leaving the curcumin moiety near the polar part may be very important to facilitate the bioactivity of the curcumin moiety when used as anti-Alzheimer drugs.

Graphical Abstract

1. Introduction

Curcumin is a phenolic compound (diferuloylmethane, Figure 1) obtained from turmeric (Curcuma longa). Curcumin is a phytochemical with a variety of biological activities and has been used in traditional medicine in Asia, as a medical herb for its many beneficial properties such as anti-inflammatory, antioxidant, antimicrobial and anti-tumor, although the molecular mechanism of some of their actions are emerging nowadays [1] and many of them are still poorly known [2, 3].

Figure 1.

Structures of curcumin and of 17MD and 21MO, bivalent compounds derived from curcumin. The letters A to G label the protons that were analyzed in the NMR experiments.

Among these activities some are notable. For example, its anticancer effect [4, 5], which may be related to its phenolic and antioxidant effect. Curcumin has also been described as a chelator that can be used to chelate metals because of its diketone structure [6], which enables it to eliminate toxic metals. Very much investigated in recent times is the capacity of turmeric to prevent some neurological diseases such as Alzheimer and Parkinson, probably through to its anti-inflammatory and antioxidant properties [7–9]. It has been observed that curcumin may alter the peptide aggregation and thus it may slow down the formation and elongation of fibrils [10]. Also with respect to Parkinson disease, curcumin modulates the aggregation of α-synuclein [11] and therefore it may be efficient in the treatment of this disease.

Curcumin is an amphipathic molecule (Figure 1) and shows propensity to interact with membranes where it may exert its antioxidant effect. Several studies have shown the modulatory effects of curcumin on biological and model membranes such as membrane dynamics [12] with a membrane thinning effect [13, 14], restricting the mobility of lipids and inducing negative curvature [15], disordering the membrane and favoring the formation of non-lamellar structures [16]. It has been described that curcumin modulates the formation of lipid raft domains using model membranes and it induces fusion of lipid raft domains at extremely low concentrations through the alteration of the boundary between the ordered and disordered phases[17]. Curcumin promotes the formation of the highly curved inverted hexagonal phase, which may influence exocytotic and membrane fusion processes within the cell [15, 16].

It was shown that curcumin binds as monomer to phospholipid bilayers occupying two different locations, one at the surface of the membrane and another in a more hydrophobic environment [14]. Biophysical studies suggested that curcumin may end to different localities depending on its concentration in the DMPC membranes. It was also recognized that a surface location could be preferred by curcumin in thicker DOPC membranes [15].

Fluorescence quenching experiments in egg yolk-based membranes suggested that curcumin is found in the hydrophobic palisade of the membrane but not far away from the lipid-water interface [18]. Using molecular dynamics simulation it was found that with both DPPC and DMPG membranes the maximum stabilization of the molecule was reached with the main axis of the molecule parallel to the membrane surface [19]. However another molecular dynamics study suggested that curcumin may adopt both a carpet model, i.e. curcumin located at the surface of the membrane and also a transbilayer disposition, depending on hydration, osmotic conditions or different types of solutes [20], hence some uncertainties exist with respect to the interaction of curcumin with membranes and the location of this phytochemical. Therefore, it is very important to understand mechanistically how curcumin interacts with membrane system and this knowledge will facilitate the design and development of liposomes-containing curcumin, which are widely used for therapeutic purposes

The ability of a water-soluble curcumin derivative to inhibit amyloid aggregation even in the presence of a lipid membrane has been shown previously [21]. Curcumin and other chelators have been used to suppress the metal-induced amyloid aggregation and toxicity in previous studies from Lim and coworkers [22–25].

Recently, a bivalent strategy has been developed to link curcumin with a steroid anchor moiety.[26–33]. The design of such bivalent compounds was intended to increase the accessibility of the “warhead”, curcumin moiety, to the lipid rafts domain of the membrane. Our previous studies demonstrated that curcumin linked with only a spacer without the steroid moiety showed protective activities, thus suggesting the importance of the steroid moiety in the design of bivalent compounds [28–33]. Our mechanistic studies also suggested the important roles of the steroid moiety and the spacer length on the interactions with the membrane system [26, 28]. Thus, in the report, we selected bivalent compounds with different steroid moieties and spacer length to further model their interactions with the membranes. The two different bivalent compounds studied were 17MD with a 17-atom spacer and diosgenin and 21MO with a 21-atom spacer and cholesterol. To observe if these differences affect to the interaction with the membrane and their membrane location we have used differential scanning calorimetry, fluorescence quenching and ¹HNMR-MAS. The results demonstrated that curcumin, 17MD, and 21MO form aggregates in DPPC fluid membranes and that pure curcumin is localized near the lipid-water interface in POPC bilayers while the curcumin moiety of 17MD and 21MO is mainly localized in a polar location.

2. Materials and Methods

2.1. Materials

1,2-dipalmitoyl-sn-phosphatidylcholine (DPPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1-palmitoyl-2-stearoyl-(5-doxyl)-sn-glycero-3-phosphocholine (5-doxyl-PC), 1-palmitoyl-2-stearoyl-(12-doxyl)-sn-glycero-3-phosphocholine (12-doxyl-PC) and 1-palmitoyl-2-stearoyl-(16-doxyl)-sn-glycero-3-phosphocholine (16-doxyl-PC) were obtained from Avanti Polar Lipids (Birmingham, Alabama, USA). Curcumin was obtained from Sigma-Aldrich (St. Louise, USA), 21MO and 17MD were synthesized as described [26]. Other chemicals used were from Sigma Chemical Co. (Madrid, Spain).

2.2. Methods

2.2.1. Sample preparation

The samples analyzed in this study were multilamellar liposomes (MLV) and large unilamellar liposomes (LUV) composed of DPPC or POPC in the absence or presence of appropriate concentrations of curcumin, 21MO or 17MD. All mixtures were prepared by combining phospholipids with curcumin or the curcumin-derived compounds, previously dissolved in chloroform, in appropriate proportions. The mixtures were then dried by a stream of nitrogen and finally with a high vacuum for 3 hours. Typically, MLVs were generated by hydrating the mixtures in an aqueous buffer containing 100 mM NaCl, 25 mM Hepes and pH 7.4 and then vortexing vigorously for 1 minute after heating the samples to 55 °C, repeating the operation 5 times. To prepare the LUVs, the MLVs generated as described above were extruded through two polycarbonate membranes 0.1 μm using an Avanti Polar Lipids mini-extruder.

2.2.2. DSC

In differential scanning calorimetry (DSC) experiments, DPPC MLVs were used in the absence and presence of different concentrations of curcumin, 21MO and 17MD. The total amount of phospholipids was 2 mg and it was hydrated in a volume of 600 μl buffer. The molar ratios of compound/DPPC analyzed were 0.025, 0.05, 0.1 and 0.2. Prior to each measurement, samples and the buffer used as reference were degassed for 6 minutes. The measurements were carried out in a Microcal VP scanning calorimeter (Microcal, Northampton, MA, USA) using the following experimental parameters: the initial temperature was 10 °C, the final temperature 60 °C and the increase of 1 °C/minute. The thermograms were recorded and processed by using the Microcal Origin 5.0 software supplied with the calorimeter. The curves were normalized to the real concentration of phospholipids in the cuvette and the baseline was subtracted from each of them. The enthalpy of the main phase transition and its start and end temperature were determined for each sample examined. The pre-transition temperature was also determined when present.

1.2.3. Fluorescence quenching experiments

LUVs formed by POPC at a concentration of 50 μM in the presence of curcumin, 21MO and 17MD with a molar ratio of 10:1 were used in fluorescence quenching measurements. Two different types of experimental assays were performed using as quenchers in one case acrylamide, and in the other n-doxyl-PC. Both sets of experiments were performed with a Fluoromax-3 fluorometer (Jobin Yvon, Longjumeau, France) at 25 °C using a quartz cuvette. The fluorescence analyzed was the intrinsic fluorescence of curcumin resulting from exciting the sample at 420 nm and recording the emission at 500 nm. The fluorescence values of each set of experiments were plotted as a function of the quenchers concentration and the K_D constants calculated according to the Stern-Volmer equation [34]. In all cases the data were acquired in triplicate and the mean value and standard deviation were determined.

In experiments with acrylamide, the quencher was added in the cuvette to the already prepared liposomes in increasing concentrations of 0.108 mM up to a maximum concentration of 0.756 mM. In the case of n-doxyl-PCs, quenchers were added in the appropriate proportions in the lipid mixtures in chloroform and LUVs were then prepared. The effect of n-doxyl-PCs on the fluorescence of curcumin was evaluated at increasing cuvette concentrations of 1.5 μM to a maximum of 9 μM.

1.2.4. ¹H NOESY MAS-NMR

POPC model membranes were used in ¹H NMR measurements. MLVs were prepared as described above using 20 mg phospholipid and an appropriate amount of curcumin, 21MO or 17MD to obtain a molar ratio of 10:1 (lipid/compound). Liposomes were generated in a volume of 80 μl buffer prepared using deuterated water and inserted directly into a 4mm zirconia MAS-NMR rotor. Measurements were made in a Bruker AVANCE 600 spectrometer under the conditions described previously [35]. The data processing and the estimation of the location of the different compounds were performed according to prior works [35, 36].

1.2.5. Molecular dynamics

The molecular structure of curcumin, 17MD and 21MO were constructed with PyMOL 2.1.0 [37]. MD simulations were done with GROMACS 5.0.7 software [38], and GROMOS 54A7 force field [39]. Topology file for POPC was described by Poger et al. [40,41] to be used in GROMOS 54A7 force field. Topology files for curcumin, 17MD and 21MO were obtained using the Automated Topology Builder (ATB) and Repository [42,43]. Membrane bilayers containing 64 POPC molecules, and 2500 water molecules simulated by the SPC model, with and without 6 curcumin, 17MD, or 21MO molecules per leaflet were built using Packmol software [44] with the monolayer’s normal [37] to the z-axis. Other conditions were used as previously described [45–49]. The last 60 ns from a total 200 ns run were used for all calculations using GROMACS algorithms. MD calculations were carried out with the aid of the Computational Service of the University of Murcia (Spain).

All systems were equilibrated in the isothermal-isobaric ensemble (NPT) at 298 K constant average temperature for 200 ns using the V-rescale temperature coupling method [56], and pressure was controlled semi-isotropically using the Berendsen pressure coupling method [57], both with coupling constant of 0.2 ps. Equilibration was followed by production runs of 200 ns using the Parrinello-Rahman barostat [60] with coupling constants of 1 ps. The last 60 ns from the production run were used for all calculations using GROMACS algorithms. MD calculations were carried out with the aid of the Computational Service of the University of Murcia (Spain).

2. Results

2.1. Curcumin, 17MD, and 21MO show fluid immiscibility with DPPC bilayers and 21MO also shows immiscibility with the gel phase.

We have used DPPC for the DSC experiments. This model membrane has been considered as the standard and has been extensively used to characterize the behavior of intrinsic molecules incorporated into the membrane. Despite that DPPC has limited presence in live cells, the model bilayers formed by DPPC can be compared with other molecules to predict the behavioral difference in a membrane system.

The effects of curcumin, 17MD, and 21MO on the phase transition behavior of DPPC share some similarities, even if these molecules show clear structural differences. The three compounds produced the dissappearance of the pretransition that appeared at 33 °C in pure DPPC (from $L_{{\beta }^{\prime }}$ to $P_{{\beta }^{\prime }}$ phases), already at a low intrinsic molecule molar fraction of 0.025 (Figure 2, A, B and C). At higher concentrations it was observed for the three intrinsic molecules that the main transition (from $P_{{\beta }^{\prime }}$ to L_α) that appeared at 41.5 °C in pure DPPC was widened as the concentration of these molecules increased and the transition onset progressively decreased at the same time (Figure 2, A, B and C).

Figure 2.

DSC thermograms for samples of pure DPPC and DPPC incorporating different concentrations of curcumin (A), 21MO (B) and 17MD (C). The effect on enthalpy change (ΔH) of increasing concentration of the intrinsic molecule / phospholipid molar ratio is represented in (D). The graphs shown in (E) represent the changes in the beginning (continuous lines) and end (dotted lines) temperatures of the main DPPC phase transition vs. the increasing concentration of curcumin, 17MD or 21 MO in the model membrane.

Figure 2D shows that ΔH of the DPPC main transition decreased at the concentration of the three molecules increased. The plot of the onset and completion temperatures shown in Figure 2E, which is a partial phase diagram, provides insights about the mode of disposition of the different molecules in the DPPC bilayer. In the case of curcumin and 17MD the temperature corresponding to the onset decreased progressively until the maximum concentration of intrinsic molecules of 0.2 molar fraction was reached, indicating a decrease in the molecular order of the DPPC membrane because of the interposition of the intrinsic molecules between the phospholipid acyl chains. It can also be seen in Figure 2E that in the case of 21MO the decrease of the onset temperature reached a maximum at a 0.05 molar fraction and at higher concentrations not further decrease was observed. This indicates that 21MO has a very limited solubility with DPPC in the gel phase producing domains rich in 21MO as the molar fraction was 0.05 or higher. In the case of the completion temperature there occurred an initial decrease at the lowest concentrations of 0.025 but at higher concentrations these temperatures showed an increase and the temperature of the completions were the same than that of pure DPPC. This indicates that at moderate and high concentrations of intrinsic molecules a separation of these molecules happened with the formation of domains rich in these intrinsic molecules with therefore low solubility of these molecules in the fluid DPPC bilayers. Taking together these DSC results of curcumin and the two bivalent compounds suggest that the spacer length and the anchor moiety can influence the way how they organize and interact with the DPPC bilayer.

2.2. The intrinsic fluorescence quenching of curcumin and the curcumin moiety in 17MD and 21MO indicates that the chromanol ring of 17MD and 21MO is located at a more polar environment than pure curcumin

Indications on the location of a fluorescent molecule within a membrane can be obtained by determining the quenching that specific compounds have on the emission intensity. In this study, two strategies were applied using quenchers with significantly different characteristics and which together can give complementary information. In one case the quencher used was acrylamide whose property of not being able to penetrate the membrane provides information on the distance of the fluorophore from the surface of the bilayer [50]. On the contrary, the n-doxyl-PCs, used in the second case, are quenchers that are inserted and arranged to form part of the hydrophobic palisade. In this second case the efficiency of the quenching on the fluorescence of the fluorophore, provides information on its proximity to the position where the doxyl is linked to the stearic acid chain and therefore at what depth of the membrane the chromanolic ring of the curcumin or the part of the curcumin of 17MD and 21MO is positioned [51–55]. As shown in Figure 3 where the Stern-Volmer plot with a POPC/fluorescent compound molar ratio of 10:1 is represented and the respective K_D values are reported, the efficiency of quenching follows the order 17MD>21MO>curcumin, which is indicative of the accessibility of acrylamide to fluorophore and therefore the proximity of this to the hydrophilic part of the membrane.

Figure 3.

Representation of the effect of acrylamide quenching on the fluorescence of the chromanol ring of curcumin. F₀ is the initial fluorescence without quencher, F is the fluorescence measured at increasing acrylamide concentrations. The F₀/F values were plotted as a function of the quencher concentration and the K_D constants were calculated according to the Stern-Volmer equation. The liposomes used were POPC LUVs and the measurements performed at 25 °C.

Figure 4 shows the effect of the quencher on the intrinsic fluorescence of the different compounds studied depending on the position of the doxyl molecule along the acylic chain. From the results obtained it can be seen that the quenching efficiency is in all cases greater for 5-doxyl-PC, then for 12-doxyl-PC and lastly for 16-doxyl-PC. This suggests a fluorophore localization closer to the lipid-water interface. The figure also shows the K_D values of the three compounds studied, which are very similar to each other.

Figure 4.

Representation of the effect of 5-doxyl-PC, 12-doxyl-PC and 16-doxyl-PC quenching on the fluorescence of the chromanol ring of curcumin (top panel), 21MO (middle panel) and 17MD (middle panel). F₀ is the initial fluorescence without quencher, F is the fluorescence measured at increasing quencher concentrations. The F₀/F values were plotted as a function of n-doxyl-PC concentration and the K_D constants were calculated according to the Stern-Volmer equation. The liposomes used were POPC LUVs and the measurements performed at 25 °C

Finally, it is interesting to note that in both the experimental approaches used and for all the compounds, the Stern-Vomer plots are linear, denoting in all cases that it is dynamic quenching in which there is only one homogeneously distributed fluorophore population [34].

2.3. ¹H-NMR and¹H NOESY MAS-NMR studies demonstrated that curcumin is located with its main axis perpendicular to the normal to the membrane and embedded in the upper part of the hydrophobic bilayer

In this study, we employed the model membranes of POPC because of its presence in biological membranes and formation of fluid membranes at 25 °C. Another reason to use POPC is that 2D-¹H-NMR NOESY will allow the analysis to relate the cross relaxation rates of protons of our intrinsic molecules with protons bound to the phospholipid molecules and POPC possesses a double bond providing a reference located at the middle of the fatty acyl chain, which is useful for the topographical studies carried out in this work. Using ¹H-NMR-MAS, the location of curcumin, 21MO and 17MD in POPC membranes was studied. Figure 1S (Supplementary Material) shows the ¹H-MAS NMR 1D spectra of POPC bilayers to which curcumin or the bivalent compounds have been incorporated at a 10:1 POPC to curcumin or derivatives molar ratio).

In the presence of these intrinsic molecules all signals corresponding to POPC were shifted upfield (Figure 5). These shifts are attributed to the ring current of the aromatic groups of curcumin and it is possible to correlate bigger shifts with greater closeness to the aromatic groups. As shown in Figure 5–1, in the case of curcumin the biggest shifts are seen for protons bound to C3 and C2 of the fatty acyl chains. It can be concluded that the aromatic rings are close to the first carbons of the fatty acyl chains and thus not far away from the polar groups of the phospholipids. In the case of 17MD (Figure 5–2) and 21MO (21MO) the biggest shifts also correspond to C3 and C2 but the differences with respect to the shifts suffered by protons bound to other POPC carbons are not so big as in the case of curcumin.

Figure 5.

Chemical shifts induced to protons of POPC along the long axis of the molecule from the center of the membrane to the polar group induced by the incorporation of curcumin (panel 1), 17MD (panel 2) or 21MO (Panel 3). The shifts were calculated in the spectra depicted in Figure 1S taking as reference those of pure phospholipids (not shown). Cross relaxation rates obtained from the ¹H-NMR NOESY spectrum of POPC/curcumin (panels 4 to 7). Cross-relaxation rates corresponding to the protons bound to the different POPC groups along the long axis of the molecule from the center of the membrane to the polar group (shown in ordinates) with respect to the curcumin carbons identified in each panel (A to E). Mean values ±standard deviations (5 determinations).

To shed more light on the location of the three intrinsic molecules studied 2D-NOESY measurements were employed to establish a correlation between given protons of these molecules as labelled in Figure 1 and protons bound to POPC by the measurement of the cross-peaks.

Figure 2S (Supplementary Material) shows the 2D-NOESY spectrum of a POPC/curcumin spectrum. Curcumin shows five resonances clearly different from the phospholipids, which are within the framing that correspond to the five protons labelled in Figure 1. These groups show crosspeaks with most phospholipid groups although of dissimilar size.

Crosspeaks related to groups A, B, C, D and E of curcumin are labelled in Figure 5 (panels 4 to 7). The interaction strength of curcumin protons with protons that belong to POPC can be deduced form crosspeaks volumes and considerable information about the relative location of curcumin and POPC can be obtained from a quantitative analysis of cross-relaxation rates [56].

A higher probability of contact between protons of POPC and protons of curcumin can be induced from biggest cross-relaxation rates. The phospholipid groups depicted in the ordinates are ordered as a function of their location from the most hydrophobic position close to the center of the bilayer, as it is the case of the terminal methyl of the fatty acyl chains to the membrane polar surface. It is interesting that the biggest cross-relaxation rates were observed for the three intrinsic molecules for protons bound to carbons located in the half of the acyl chain in the vicinity of the polar interface as it is the case for C3 and C2, indicating that these molecules are not far away from the polar groups of the phospholipids but nevertheless slightly immersed in hydrophobic part of the membrane.

In the case of 17MD, it can be seen in Figure 6 (panels 1 to 7) that protons corresponding to the curcumin moiety (A, B, C, D and E) are localized in the proximity of the membrane-water interface, with the biggest cross-correlation rates occurring with G1 and C2 and also with alpha especially in the case of protons B (Figure 6–2), C (Figure 6–3) and D (Figure 6–4). This shows that the curcumin moiety is closer to the polar part of the membrane than curcumin itself and this result agrees with the acrylamide quenching results. Proton F is bound to the diosgenin moiety and Figure 6–6 shows that its maximum cross correlation is with the CH3, i.e. methyl terminal of the fatty acid chains, indicating a location in the center of the bilayer. Proton G that it is located on the spacer between the diosgenin and the curcumin moieties has its maximum correlation with the protons bound to beta followed by C2 and C3 of POPC and thus in the membrane-water interface (Figure 6–7).

Figure 6.

Cross relaxation rates obtained from the ¹H-NMR NOESY spectrum of POPC/17MD (panels 1 to 7). Cross-relaxation rates corresponding to the protons bound to the different POPC groups along the long axis of the molecule from the center of the membrane to the polar group (shown in ordinates) with respect to the curcumin carbons identified in each panel. Mean values ±standard deviations (5 determinations). Cross relaxation rates obtained from the ¹H-NMR NOESY spectrum of POPC/21MO are shown in panels 8 to 14. Cross-relaxation rates corresponding to the protons bound to the different POPC groups along the long axis of the molecule from the center of the membrane to the polar group (shown in ordinates) with respect to the curcumin carbons identified in each panel. Mean values ±standard deviations (5 determinations).

Figure 6 shows that, for protons A (Figure 6–8), B (Figure 6–9), C (Figure 6–10), D (Figure 6–11) and E (Figure 6–12) of 21MO, which belong to the curcumin moiety, the highest cross-relaxation rates are observed with protons of POPC localized at the water-lipid interface, but for protons A, C and E the maximum values are found for G1, with C2 and C3 also showing high rates. This indicates that this curcumin moiety is closer to the membrane surface than in the case of curcumin itself, this observation concords well with the fluorescence quenching results shown above using acrylamide.

On the other hand, proton F (Figure 6–13), which is located in the cholesterol moiety, has a clear maximum of correlation with the CH3 groups of the fatty acyl chains of POPC, indicating that it is close to the center of the bilayer. Proton G (Figure 6–14), however, which is located on the link between cholesterol and curcumin moieties has a maximum correlation rate with C2 suggesting that this link is near the surface of the membrane, as it could be expected given its proximity to a polar amide bond.

2.3. Molecular Dynamics Simulations

Molecular dynamics simulation is a useful technique to probe the structure and dynamics of different types of molecules inserted in phospholipid model membranes [57–59]. In particular, the integration of molecular dynamics with NOESY-NMR provides information about the dynamics and location of macromolecules [60, 61]⁠. Therefore, we conducted molecular dynamics simulations to obtain further evidence on the location of the molecules under study in the POPC membrane bilayer.

To corroborate the proper equilibration of the simulations, the area per lipid was evaluated (Supplementary Material, Figure 3S). A good stability of the area per lipid was observed in all cases during the production run, indicating that equilibrium was reached in all systems. Pure POPC exhibited an average area per lipid of 0.71 ± 0.01 nm² which is in good agreement with the experimental values of 0.698 nm² [62] and 0.705 nm² [63] and theoretical values ranging from 0.65–0.70 nm² [64–66].

As shown in Figure 7 (left panels), the averaged mass density along the z-axis of the simulated system (Figure 7, right panels) locates curcumin in the polar phase of the bilayer, just below the phosphate groups of POPC, but with fully overlapping bands (Figure 7, upper right panel). 17MD orients itself all along the lipid layers although three main peaks are shown at the center of the membrane and just below the POPC phosphate groups, indicating the location of the curcumin group of 17MD (Figure 7, middle right panel). 21MO is centered at the center of the bilayer but expanding throughout the POPC backbone with a slight overlap with the phosphate groups. The curcumin group of 21MO is distributed mainly in the center of the bilayer and just below the phosphate groups (Figure 7, lower right panel). Cluster formation can be observed in the center of the bilayer for 21MO (Figure 7, lower left panel), consistent with the DSC result, indicating the formation of domains rich in 21MO in a DPPC membrane (Figure 2).

Figure 7.

Representative snapshot of POPC bilayers with curcumin, 17MD or 21MO (left panel). Water molecules are depicted in red and white sticks, POPC carbon backbone in orange, and curcumin derivatives carbon backbones in green sticks. The mass density profiles along z-axis of the membrane (perpendicular to the layer membrane) are shown in the right panels for phosphorous atom from POPC (red curves), curcumin, 17MD and 21MO (black curves), and curcumin group of 17MD and 21MO (green curves). The z-axis are scaled to their corresponding snapshots.

In order to obtain parameters from molecular dynamics to simulate the results obtained from NMR experiments (Figures 5 and 6), the number of contacts between the same atom groups used in NMR was calculated within 0.6 nm to include most type of interactions (Figure 8). The number of contacts was normalized to the total number of atoms present in the atom groups involved. We can assume the number of contacts as a measure of the probability to encounter both groups in close contact (0.6 nm), thus the number of contacts is interpreted as a measure of proximity, similarly to the chemical shifts and cross-relaxation rates in the NMR experiments.

Figure 8.

Number of contacts between selected atoms from curcumin, 17MD and 21MO, and selected atoms from POPC (see Figure 1). Number of contacts within 0.6 nm was calculated with mindist function of gromacs and normalized to the total number of atoms (abscissa). The same color scheme is used in all panels.

It can be observed that the maximum number of contacts between A to E atoms (curcumin group) was found for C3 and C2 atoms of POPC, indicating that the curcumin moiety of all compounds studied is located close to the polar phase of the bilayer (Figure 8). It can be seen, however, that curcumin group of 17MD and 21MO is more spread in the lipid layer than the curcumin compound. F atom of the sterol group of 17MD and 21MO is clearly close to the terminal methyl group of POPC indicating that the sterol group would be located in the center of the bilayer. G atom corresponds to a carbon atom located on the link between the sterol and the curcumin moieties. In 17MD and 21MO molecules this atom would be located mainly closer to C3 and C2 atoms of POPC, although in the case of 21MO it would also be distributed from C3 atom to the methyl terminal (Figure 8, right panel). These results agree quite well with the experimental results obtained from chemical shifts (Figure 5) and cross-relaxation rates (Figures 5 and 6) of ¹H-NMR-MAS experiments.

The presence of curcumin increases the average area per lipid from 0.71 nm2 to 0.75 nm2 (Supplementary Material Figure 3S). This effect could be explained by the presence of curcumin in the surface of the membrane which would increase the POPC polar headgroup separation and, in turn, would increase the surface area per POPC molecule. Conversely, the presence of 21MO and 17MO mainly in the POPC hydrocarbon tail region, decreases the area per lipid to about 0.67 nm². These results would be in agreement with the distribution of the molecules in the membrane previously proposed in this work.

3. Discussion

To compare the organization and localization, curcumin and two bivalent compounds that contain the curcumin moiety, 17MD and 21MO, were studied by incorporating them in phospholipid membranes. The DSC results showed that all three intrinsic molecules showed a poor miscibility with DPPC, especially in the fluid phase. In the gel phase, there is also a poor immiscibility observed with DPPC, especially for 21MO. These results concord well with the conclusions reached after the molecular dynamics simulation that suggest aggregation in the POPC bilayer at 25 °C. Fluid immiscibilities originated from the incorporation of intrinsic molecules in membranes have been observed previously from DSC experiments, e.g., vitamins K [67, 68] and vitamin E [69].

The fluorescence quenching experiments showed that the studied compounds are located near the membrane surface but within the hydrophobic part. It also indicated that the curcumin moieties of 21MO and 17MD are more accessible for the quencher than curcumin itself and this result agrees well with the ¹H-NMR NOESY results and also with the molecular dynamics predictions. Quenching with doxyl-PCs also revealed that curcumin and the curcumin moieties of 17MD and 21MO are close to the membrane surface.

The observed change in the chemical shifts of the phospholipids might be caused by the aromatic rings within the compounds. The results suggested that curcumin is predominantly close to the surface of the membrane. Given the symmetric structure of curcumin, we expect its parallel alignment with the membrane surface. The results of 17MD and 21MO however suggested a broader distribution for the aromatic rings of the curcumin moiety. Although this is different from the results of ¹H-NMR NOESY, fluorescence quenching experiments, and the molecular dynamics simulations, nonetheless, these measurements indicate that the preferred location of curcumin aromatic rings is near carbons C3 and G1. The discrepancies might be because the induced chemical shifts technique is less sensitive to discriminate between close and distant atoms or groups of atoms.

The preferred location of curcumin in the membrane is within the hydrophobic palisade but near the lipid-water interface, and the main axis of the molecule parallels to the membrane surface. Our results agree well with the conclusions obtained from molecular dynamics studies predicting a carpet disposition of curcumin [20], but there are studies that argue a perpendicular alignment with the plane of the membrane [19]. It is notable that the curcumin moieties of the bivalent compounds are located closer to the polar part of the membrane, in comparison with pure curcumin, thus facilitating the activity of curcumin with respect to its interaction with other molecules located outside of the membrane.

Our recent studies demonstrated the difference of 17MD and 21MO in stimulating neuritic outgrowth of N2a cells [26]. The could be due to the presence of diosgenin in 17MD and it has been described that diosgenin and its analogues may favor axonic outgrowth and cognitive improvement activities [70, 71]. Our results indicated a similar location for both 17MD and 21MO that is deep to the hydrophobic part of the membrane. Both compounds also adopt a similar organization within the membrane with low miscibility with the phospholipids. Therefore, our studies support the notion that the biological difference of 17MD and 21MO is due to the functional difference of the steroid anchors.

We have used in our study two different types of phosphatidylcholines, these being the most widely used phospholipid types since they easily form bilayers. One of them is fully saturated as DPPC and the other unsaturated as POPC. Both phospholipids are used s models and POPC has the advantage for the Noesy-NMR studies of possessing a double bond, which is useful for the topographical studies. It would be interesting to explore if the location of these compounds may vary in a membrane in which other phospholipids are included with negative charges and it would be very interesting in the case of lipid rafts. It has been described that curcumin presents interesting interactions with the lipids responsible of forming rafts, modulating their formation and modifying the interface between raft- and non-rafts lipids [17].

It should be mentioned that some limitations exit for our results. Curcumin and the derivatives studied here have a very poor solubility in water and this limits the administration of these compounds. Additionally curcumin, although is generally well tolerated and it presents a well-established safety may have in exceptional cases and with big doses some side effects such as diarrhea, head-ache or rash [72]. It should be also considered the limitations of the reported results since the in-vitro conditions employed in this study need not be realistic for biological situation to address the in-vivo activities of curcumin and other compounds.

In summary our study shows that curcumin, 17MD, and 21MO, all exhibit a poor miscibility with the fluid membrane forming domains rich in these compounds. They predominantly localize to the lipid-water interface, with a slight preference to a more polar location for 17MD and 21MO. The carpet disposition is the preferred orientation with the main axis of the curcumin perpendicular to the normal to the membrane plane. The membrane anchors of 17MD and 21MO are deeply immersed in the membrane.

Highlights.

• Curcumin adopts a carpet disposition in the membrane; the curcumin moiety of bivalent compounds is found in a more polar environment than pure curcumin; curcumin and bivalent compounds show a poor miscibility with phospholipids.

Abbreviations:

DPPC

1,2-dipalmitoyl-sn-phosphatidylcholine

POPC

1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine

5-doxyl-PC

1-palmitoyl-2-stearoyl-(5- doxyl)-sn-glycero-3-phosphocholine

12-doxyl-PC

1-palmitoyl-2-stearoyl-(12-doxyl)-sn-glycero-3-phosphocholine

16-doxyl-PC

1-palmitoyl-2-stearoyl-(16-doxyl)- sn-glycero-3-phosphocholine