Isoflurane decreases the membrane transition temperature, similar to previous observations [16
]. This indicates that isoflurane is more soluble in the liquid than the gel phase, and would be consistent with the concept that tail separation by headgroup intercalation of the anesthetic molecules facilitates liquefaction. Both dyes track this change, and, although the temperature resolution in these experiments is too low to allow analysis, it is clear that it is strongly dependent on isoflurane concentration.
Steady-state anisotropy represents the average limitation to non-axial movements (wobble of the rod-shaped molecule in a conical or hour-glass shaped cage) during the random interval between photon absorption and fluorescence emission [15
]. It can be used to estimate the free volume available to the probe, which in turn can be related to lateral pressure at the level of the probe [18
]. Currently, this relationship between fluorescence anisotropy and lateral pressure profile needs further refinement, especially through molecular dynamics simulations. The model system employed here and results should spur further definition of this relationship.
Fluorescence quenching experiments [19
] with dioleoylphosphatidylcholine bilayers have been used to estimate the position of DPH within the membrane. These experiments are likely to be relevant to DPPC in the liquid state based on similar bilayer thicknesses - phosphorus separation of 40.3 Å for dioleoylphosphatidylcholine [20
], and 39.2 Å at 50°C for DPPC [21
]. The quenching experiments revealed that the DPH molecule is positioned approximately parallel to the lipid molecules in the tail region of one leaflet of the bilayer, similar to simulation results obtained previously [22
], with its center 7.8 Å from the bilayer midplane. The terminal carbons are positioned 1.3 and 14.3 Å from the bilayer center. The central carbon of the DPH moiety in TMA-DPH molecules is located 3.1 Å closer to the bilayer surface, with the terminal carbon 4.4 Å from the center of the bilayer and the TMA N 18.7 Å from the center of the bilayer. These results are consistent with the molecular dynamics trajectories reported here: the dye has a strong propensity to orient in one leaflet with one end near the bilayer center and the other near the interface. Averaging among the three initialization cases, the N of TMA-DPH is located 17.4 Å from the bilayer center. Likewise, the terminal hydrogens near the bilayer center are in the same leaflet and within a few Å of the center for both dye molecules. Given the lengths of the molecules, the average tilt angles (17.5° [15
]), and the bilayer thickness, we estimate that TMA-DPH anisotropy reflects the lateral pressures in all three regions (headgroup, glycerol backbone, and tail), whereas DPH should be sensitive to the lateral pressure primarily in the tail region.
Our simulations address controversies that exist over the position of DPH in lipid bilayers [23
]. Neutron diffraction studies [24
] contradict the conclusion of the fluorescence quenching parallax studies [19
], indicating the DPH partitions into the center of the bilayer, between the two leaflets. However, for the neutron diffraction, it was necessary to use 50× higher dye:lipid ratio than is usually used in fluorescence experiments like those reported. Given the alacrity with which sandwiched dye molecules abandon the center of the bilayer in our simulations, it seems likely that virtually no dye is sandwiched between leaflets at low mole ratios. The high dye concentrations used, (and perhaps configurational effects of high headgroup spacing due to the 75% humidity conditions used [25
]) may lead to the formation of a separated dye phase in the bilayer interior. Although some dye molecules can be positioned parallel to the lipid surface [26
], these should only contribute modestly to the fluorescence anisotropy, being but a small component of the fluorescence amplitude, perhaps in part due to aqueous quenching [28
Isoflurane increased the anisotropy of TMA-DPH in the liquid-crystalline phase, implying reduced fluctuations in the available cone angle, possibly due to increased lateral pressure. For DPH, the reduction of anisotropy caused by isoflurane indicates increased dye mobility in the lipid tail region. This decrease was observed at both low and high temperatures, indicating that, although the anesthetic must be more miscible with the liquid phase of the membrane (based on the leftward shift of the melting temperature), it must still partition sufficiently into the gel phase to perturb motion constraints on DPH. Presumably, formation of segregated complexes among isoflurane molecules or other heterogeneities account for the ability of the anesthetic to alter DPH anisotropy even though its miscibility in the DPPC gel phase is low. The phenomenon of lipophilic membrane contaminants altering both the phase transition and physical behavior of the phospholipid at temperature extremes has been observed previously for molecules such as fatty acids and lysophospholipids [29
]. We suggest that the effect of isoflurane on the overall anisotropy may be due to separation of the headgroups by intercalated anesthetic molecules [31
] and relief of lateral pressure in the tail region. Therefore, the increased anisotropy for TMA-DPH must be due to increased lateral pressure in the headgroup region, as would be expected for preferential absorption of anesthetic molecules to that region. The lateral pressure is negative (i.e. is lateral tension) at the level of the glycerol backbone due to the propensity of water molecules to flee the hydrophobic core. It is positive in the bilayer interior due to crowding imposed by the high-tension surface above and the loss of configurational entropy of the partially oriented tails. Extensive simulations done with a pure DPPC bilayer membrane in pure water at 323 K [32
] yield the symmetrized lateral pressure profile redrawn in Figure as a diagram to illustrate the size and positions of the dye molecules in the monolayers. The high tension trough between 1.2 and 1.9 nm from bilayer center represents the interface between the hydrophilic bath and the hydrophobic bilayer interior and may attract anesthetic molecules, which then modulate the profile and significantly affect dye oscillations.
Figure 5 Lateral pressure profile in DPPC as reported previously , with the positions of DPH (left) and TMA-DPH (right), taken from fluorescence quenching parallax analysis and confirmed here and in  by simulation, accurately superimposed.
These results can be contrasted and compared to two related studies. First, an extensive study of the effects of several volatile anesthetics on DPH and TMA-DPH anisotropy in red blood cell ghosts showed that all anesthetics caused anisotropy to decline in a dose-dependent fashion, not only for DPH, but also for TMA-DPH [13
]. It must be acknowledged, therefore, that the crossover reported here may be unique to DPPC, or perhaps more generally to purified alkyl-tail lipids. Nevertheless, we note that sphingomyelin, a common component of rafts in neuronal cell membranes, has two alkyl tails like DPPC (as opposed to one alkene and one alkyl tail commonly found in natural non-raft membranes), so the crossover phenomenon may be pertinent to modulation of lateral pressure profiles in lipid rafts. As a side note, we point out that the results of Norman et al. [13
] were interpreted as evidence against microviscosity changes playing a role in ion channel modulation by anesthetics. This was due to the lack of uniformity in anisotropy changes at MAC, the mean alveolar concentration required to immobilize half of human patients, for different anesthetics. However, in retrospect, it is noteworthy that in many cases there were abrupt shifts in the curve or the curvature of anisotropy vs. anesthetic concentration at or near MAC for one dye or the other, which may suggest exaggerated effects of individual anesthetics on lateral pressure profile near their MAC values. Such changes could affect the splay in helix bundle membrane proteins.
On the other hand, the results were reproducible and our interpretation from the TMA-DPH anisotropy changes in DPPC that isoflurane increases pressure (i.e. reduces the negative pressure, or the tension) in the glycerol backbone region while decreasing pressure in the tail region is consistent with results of excimer fluorescence. Using vesicles doped with di-pyrenyl phosphatidylcholine having tails of different lengths, 4, 6, 8, or 10 carbons [14
], it was observed that pyrenyl groups held close to the membrane surface showed lower excimer formation upon addition of DOPE, while pyrenyl groups attached beyond the cis
double bonds showed increased excimer formation upon addition of DOPE. This was interpreted to mean that DOPE changed the lateral pressure profile, decreasing pressure in the headgroup/backbone region and increasing pressure in the distal tail region. However, in that approach, different tail lengths changed both pyrenyl depth and confinement. The results presented here suggest that lateral pressure, and not just confinement, contributed to the DOPE effect in excimer formation.
More broadly, this project is intended to complement more general explorations into how anesthetics and other small amphiphilic compounds might influence ion channels and other membrane proteins indirectly by perturbing the pressure profile or other characteristics of lipid membranes. Previously studied examples include effects of microscopic curvature on alamethecin channel kinetics [33
], effects of numerous classes of amphiphiles (e.g. lysolecithin [34
], alcohols [35
], and halothane [36
]) on gramicidin channel properties, effects of short chain alcohols and general anesthetics on potassium channel conductance properties [37
], effects of small organic acids on monolayer properties, tadpole immobilization, and glycine receptors [38
], and effects of ketamine on measured bilayer thickness and simulated lateral pressure profile [39
]. All of these studies indicate that amphiphiles can influence membrane mechanics, and possible membrane protein structure and function, when found in the bath at the 0.1-1 mM level, similar to MAC levels for volatile anesthetics.
It must be noted that the results reported here are limited in scope and utilize high isoflurane concentrations, and are therefore preliminary in nature. Only one lipid species was analyzed and both the durations of the molecular dynamics simulations and the equilibration periods for the anisotropy experiments are modest. In particular, the flexibility of the DPH molecules in the simulations requires more extensive simulations for complete analysis. For instance, simulations on the order of 20-50 ns would allow consideration of the full range of angles available to DPH during the excited state and therefore direct verification of the anisotropy under the various experimental conditions.
In the future, the effect of lipid species on the crossover effect should be examined more closely, as crossover was not observed in erythrocyte ghosts [13
], and expansion of the time domain for both the anisotropy experiment and the simulations should be examined. A complete free energy profile for rotations of the dye in the leaflet and during passage through the inter-leaflet region would be interesting, particularly for DPH, which is not anchored by a charged group at one end like TMA-DPH. The impact of isoflurane and other volatile anesthetics on the lateral pressure profile and on the dynamics of the fluorophores could be explored by simulation. Preliminary work in our lab indicates that these will be fruitful directions.