Lipid Component Contributions to the Surface Activity of Meibomian Lipids

Abstract

Purpose.: Meibomian lipid films have very complex physical properties that enable them to be compressed and expanded without collapsing. These properties can be attributed to the self assembly of the individual components, mainly wax and cholesteryl esters (WE and CE). Here, the surface pressure properties of WEs and CEs films have been compared to evaluate their contributions to meibomian lipid films.

Methods.: Films of different WEs and CEs were spread on a Langmuir trough and their surface pressure area profiles were compared with a particular emphasis on the effects caused by the degree of saturation of the alkyl/alkene chains.

Results.: Fully saturated WEs and CEs formed unstable films that collapsed upon compression. Very unsaturated waxes and CEs tended to have two distinct phases, one that reflects interaction with the aqueous subphase, while the second appeared to be with the multilayered bulk of the lipid film. With aging of the films, the WEs tended to move off the surface into the bulk. When meibomian lipid films were seeded with large amounts of WEs, only minor changes could be seen unless the WE was very unsaturated.

Conclusions.: These data are consistent with meibomian lipid films having a surfactant layer with a complex bulk layer external to this. It is speculated that the bulk layer contains thermotropic smectic chiral liquid crystals of CEs that are interacting with the WEs. This structure would tend to prevent collapse of the meibomian lipids onto the ocular surface and be very tolerant of lipophilic contaminants.

Introduction

One of the main features associated with the lipid layer of the tear film is that it lowers surface tension of the tear film, and this assists spreading of the tears across the ocular surface. ^1–3 The ability of the lipid layer to withstand the stresses put on it during a blink cycle and resist collapse onto the ocular surface depends upon its structure and, therefore, its composition. Although there is variation in the data from different laboratories, overall their results show that meibomian lipids comprise approximately 90% cholesteryl and wax esters (CE; WE), 2% triglycerides, 4% omega hydroxy acyl fatty acids (OAHFA), and a small amount of free fatty acids (FA). ^4–10 Within these lipid groups, the acyl chains vary in length and saturation, and oleic acid (C18:1; 18 carbons with one double bond) is a very commonly found acyl chain. ^6,7,9 Approximately 80% of the waxes have a monounsaturated fatty acid chain, and this is oleic acid in 90% of cases. ⁹ The fatty alcohol always appears to be saturated and a very common WE is arachidyl oleate (note that the nomenclature for the WEs, Cxy:z-Cqr:s, indicates the alcohol chain first with xy carbons and z double bonds, and the acid component second with qr carbons and s double bonds; C20:0-C18:1). ⁹ With age it appears that the ratio of CEs to WEs (0.57) increases, whereas with meibomian gland disease, it (0.34) decreases. ⁸ In addition, infrared spectroscopy used to measure phase transitions (melting points) of meibomian lipids (not as a spread film) indicated that higher melting points were a feature of meibomian gland disease compared with age-matched normals. By correlating these findings with the phase transitions of WE and CE mixtures, it was inferred that the degree of saturation of WE had a disproportionate influence on increasing the melting point and led to more order in the meibomian lipid structure. ¹¹ It was also shown ⁸ that infant meibum had a relative low CE to WE ratio (0.24), and since infants have very stable tear films, the question arises as to how the components of meibomian lipids add to its stability. In order to understand the influence of particular components of meibomian lipids on the physical properties of the lipid layer, different individual lipid species were examined in terms of their surface activity and in some cases when mixed with meibomian lipids. Measuring the surface activity of meibomian lipids in vitro has been used by several groups and has been useful to glean information about the structure of the lipid layer. ^12–17  

Methods

Lipids (Table) were purchased from Nu-Check-Prep, Inc. (Elysian, MN). They were chosen because of their commercial availability, and so that the role of the different components of the esters could be compared. It is important to note that many of the WEs chosen had an unsaturated fatty alcohol component, and, therefore, are not a normal component of meibomian lipids. This strategy was used to determine if the desaturation of the alcohol or the acid part of the waxes was important. Therefore, some studies were done using meibomian lipid films seeded with waxes having an unsaturated fatty alcohol. 

Lipids were dissolved at 1 mg/mL in CHCl₃ (HPLC grade containing 0.5% ethanol; Sigma Chemical Co., Castle Hill, Australia) and a small volume was applied to the surface of an artificial tear buffer ¹⁸ in a small Langmuir trough (surface area 80 cm²; Nima Technologies, Coventry, UK). Ten minutes was allowed for the chloroform to evaporate before isocycles were begun and surface pressure measurements made using a Wilhelmy balance. Surface pressure was recorded as the surface area of the trough was cycled between 79 cm² and 16 cm². Volumes were titrated so that the initial take-off point for the particular lipid was close to maximum area. Isocycles were carried out at 20°C until the shape of the curve and maximum surface pressure did not change (equilibrium profile) and then the temperature was increased to 35°C and isocylces continued. After an equilibrium profile at the new temperature was achieved, the temperature was then cooled back to 20°C, and isocycles continued again until equilibrium. This sequence was carried out in order to compare the data with those obtained from meibomian lipids where it has been found that the film expands upon cooling from 35°C. ¹⁶ Some films were also spread initially at 35°C for comparison with films spread at 20°C and then heated to 35°C. 

In some cases, the lipids (4 μL of 1 mg/mL) were mixed with meibomian lipids doped with 0.5% of the fluorophore, 1-palmitoyl-2-(12-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl)-sn-glycero-3-phosphatidylcholine (20 μL of 1 mg/mL; NBD-DPPC; Avanti Polar Lipids, Tullamarine, Australia). The meibomian lipids were collected as previously described ¹⁶ from a single healthy male donor with no ocular disease or on any medications. This was carried out in accordance with the tenets of the Declaration of Helsinki regarding human ethics. These mixtures were spread on the subphase of the trough. Their surface activity profiles were measured as described above and micrographs were taken using an epifluorescent microscope (Leica, Wetzlar, Germany) and an Andor Ixon high-speed camera (Belfast, UK). 

Table. 

 

List of Different Molecules Tested. Some Sample Melting Points as Available in the Literature Are Also Indicated

Name	Lipid Type	Structure	Melting Point (°C)
*Palmityl stearate	WE	C16:0-C18:0*	59 19 17 19 , 19 21 , 20 11
*†Palmityl oleate		C16:0-C18:1*†
*Palmityl linoleate		C16:0-C18:2*
*†Stearyl oleate	WE	C18:0-C18:1*†	‡24 19 , 37 20 , 26 21
*Stearyl linoleate		C18:0-C18:2*
*Stearyl linolenate		C18:0-C18:3*
Oleyl stearate	WE	C18:1-C18:0	‡35 19 , 27 20 , −1 20 , −4 19 , 0.5 11
†Oleyl oleate		C18:1-C18:1†
Oleyl linoleate		C18:1-C18:2
Oleyl linolenate		C18:1-C18:3
Linoleyl stearate	WE	C18:2-C18:0
†Linoleyl oleate		C18:2-C18:1†
Linoleyl linoleate		C18:2-C18:2
Linoleyl linolenate		C18:2-C18:3
Linolenyl stearate	WE	C18:3-C18:0
†Linolenyl oleate		C18:3-C18:1†
Linolenyl linoleate		C18:3-C18:2
Linolenyl linolenate		C18:3-C18:3
*†Arachidyl oleate	WE	C20:0-C18:1*†	32 19 , 30 (own measurement)
*Arachidyl linoleate		C20:0-C18:2*
*Arachidyl linolenate		C20:0-C18:3*
Cholesteryl palmitate	CE	Chol-C16:0
Cholesteryl linoleate		Chol-C18:2
Cholesteryl lignocerate		Chol-C24:0
Cholesteryl nervonate		Chol-C24:1
Cholesteryl arachidonate		Chol-C20:4
Oleic acid	FA	C18:1
Linolenic acid		C18:3
Arachidonic acid		C20:4
Lignoceric acid		C24:0
Nervonic acid		C24:1

*  These groups of lipids have saturated alcohol components in the WEs.

†  These lipids have oleic acid as the acid component.

‡  Due to the discrepancy in the literature (references 17 and 19), we conducted our own measurements and confirm that the melting points are as per reference 19. The authors of reference 20 must have confused the names in reporting their results and conclusions.

Results

It is important to note, for interpreting the results, that the system is not stable because energy is put into the system by compressing and expanding the spread film and also by subsequently heating the film. Therefore, there is an expectation that the structure of the film is reorganized and therefore, the pressure-area curves change as more energy is put into it. Therefore, isocycles were carried out until the system reached equilibrium at either 20°C or 35°C (i.e., no further changes in maximum surface pressure or shape of the pressure-area profile). 

Wax Esters

Overall, data from the pressure-area isocycles indicate that as isocycles continued, more and more WEs left the surface and moved into the outer layer. This is illustrated in Figure 1, where two examples of WEs are presented: oleyl oleate and stearyl linoleate. For both WEs during the first isocycle (dashed line), there was no surface pressure initially (surface area = 80 cm²), until the molecules at the surface began to interact as the surface area was decreased. The pressure then increased gradually, and then flattened. It is believed that the flattening occurs because some of the molecules that are originally on the surface leave the surface and, hence, there is no change in surface pressure. At the equilibrium isocycle (solid line; the curve does not change in subsequent isocycles), the take-off area had decreased, and the second phase of the curve had steepened to give a higher maximum surface pressure than the first isocycle. These features were seen whether the initial lipid film was spread at 20°C or at 35°C, and were a consistent feature of all of the waxes (Fig. 2; left). Taken together, these results indicate that the WEs appear to form a duplex film ²² with a monolayer forming at the buffer surface and a multilayered hydrophobic phase on top of this layer. This hydrophobic layer will be referred to as “the bulk at the air interface.” 

Figure 1. 

 

Pressure area curves comparing films of oleyl oleate (C18:1-C18:1) with stearyl linoleate (C18:0-C18:2) when spread at 20°C (left) and 35°C (right). The first isocyle and equilibrium isocycles are shown for each of the different conditions. At key points (where there are inflections in the curves) the average areas per molecule are indicated based on all molecules remaining on the surface.

Figure 2. 

 

Composite figures of isocycles of WEs. All isocycles are drawn to the same scale except for the insert in the bottom left. The left shows comparisons between first (dashed curve) and equilibrium (solid curve) isocycles at 20°C. The right shows differences between equilibrium isocycles of similar waxes (degree and position of unsaturation different). The isocycles on the right all sit on the same baseline (0 mN/m), but have been shifted on the y-axis for clarity, and the vertical order of the curves corresponds to the order of the text.

In addition, for films that were cooled from 35°C to 20°C, an increase in maximum surface pressure at the smallest surface area (Fig. 1) was observed. This increase in pressure upon cooling was similar to that found for meibomian lipid films. ¹⁶ Take-off area had also decreased, indicating that molecules had moved off the surface into the outer layer during the heating process. In Figure 1, the average area per molecule was based on the assumption that all molecules are on the surface are given for different parts of the curve. It appears that the more saturated the molecule is in its FA component, the less surface area it takes up at the different parts of the curves. This indicates a greater fluidity as the molecules are able to slide over each other more readily to lift from the surface. 

The length of the alkyl chain, and the degree and position of the desaturation (alcohol or acid) also made a difference to the pressure-area curves of the isocycles. In general, the more unsaturated (number of double bonds), irrespective of whether it was in the acid component or the alcohol component, the flatter the pressure-area curves (Fig. 2). This indicates that they are more fluid (less elastic) and compressible. The equilibrium curves of different waxes with oleic acid as the fatty acid component were compared (Fig. 2) because analysis of the waxes in meibomian lipids has shown that the vast majority of waxes in meibum have oleic acid as the acid component. ⁹ C16:0-C18:1 and C18:0-C18:1 appeared to be similar and more rigid based on the hysteresis seen in the pressure area isocycle, whereas the films having unsaturated chains in the alcohol component suggested more fluidity. 

Films of fully saturated waxes spread at 20°C were very stiff (elastic) as indicated by the very high maximum surface pressure obtained during the first isocycle (Fig. 2; C16:0-C18:0). On expansion, these films fractured as indicated by the surface pressure returning to zero very rapidly and the films collapsed, which was evident by the take-off point moving to a smaller surface area and the maximum surface pressure decreasing (Fig. 2; C16:0-C18:0). These films remained collapsed during heating to 35°C as indicated by the low maximum surface pressure. The maximum surface pressure increased slightly when cooled back to 20°C, but the area at take-off did not change, indicating that the film had become stiffer rather than molecules had moved back onto the surface (Fig. 2). Similar behavior was observed using C18:0-C18:0 (not shown). 

In general, having a saturated acid component made the films stiffer. This is most obviously seen by comparing the saturated alcohol component (C18:0-C18:1; melting point [mp] 24°C) with equivalent saturated acid component (C18:1-C18:0; mp 37°C; Fig. 2). The latter behaves like a saturated wax at 20°C and its film collapsed similar to that of C16:0-C18:0 and C18:0-C18:0. 

Such stiffness in the film seems to come from the saturated fatty acid component exemplified by using C20:0 as the alcohol component (arachidyl WEs; Fig. 3A). In this case, the compression part of the isocycle was similar to those seen for the lower molecular weight WEs (Fig. 2). However, the expansion part of the isocycle showed a large hysteresis indicating that these molecules exerted a stronger resistance to being pulled apart after they had been squeezed together, and then the film fractured as it expanded. Such behavior could be explained by superposition of the molecules, such as an alignment of the saturated components caused by stronger van der Waals forces between these WEs. With the arachidyl WE films, the movement of take off to smaller areas with increasing isocycles is consistent with molecules being pushed off the surface as was seen with other WEs. Heating of the arachidyl wax films to 35°C showed a remarkable difference from the equivalent stearyl (C18:0) wax films (Fig. 3B). The pressure area isocycles of the archidyl wax films were smooth at 35°C, whereas for stearyl wax films they were distinctly biphasic. When cooled, the pressure area isocycle curves for arachidyl wax films again assumed the complex curves as seen before heating (Fig. 3C), whereas there was little difference in the curves for stearyl waxes on cooling except that take off was at a smaller surface area and the maximum pressure had increased (Fig. 3D): this would be expected as taking thermal energy out of the system makes the molecules move closer together on average and the molecules are less mobile. A tentative explanation is that the arachidyl waxes at 35°C form a liquid layer above and resting on the aqueous surface that does not interact with the aqueous surface and, hence, little change in surface pressure is seen on compression. The stearyl waxes, being slightly more hydrophilic, perhaps are still able to interact with the surface to a minor degree. 

Figure 3. 

 

Comparison of isocycles of films made from arachidyl waxes (C20:0-C18:N) and stearyl waxes (C18:0-C18:N). The first and equilibrium isocycles at 20°C of arachidyl waxes (A) are compared with the equilibrium isocycles of arachidyl and stearyl waxes at 35°C (B). When cooled back to 20°C the arachidyl wax films reverted back to a complex expanded isocycle (C), whereas the stearyl waxes showed isocycles consistent with the molecules occupying less space and being less mobile, consistent with the expectations of cooling (D).

Cholesteryl Esters

Films of CEs with a saturated acyl group (chol-C16:0 or chol-C24:0) appeared to be very stiff and collapsed on the first isocycle (Fig. 4) similar to the saturated WEs (e.g., palmityl stearate; Fig. 2). Films of unsaturated CEs had different pressure area profiles (Fig. 4) according to the chain length and degree of desaturation (chol-C18:2; chol-C20:4; chol-C24:1). Films of cholesteryl linoleate (chol-C18:2) and cholesteryl arachidonate (chol-C20:4) gave little surface pressure on the first isocycle, but this slowly changed resulting in a slightly higher surface pressure at equilibrium in the case of chol-C18:2, or substantially higher surface pressure in the case of chol-C20:4. There was little hysteresis seen for either films. Films of cholesteryl nervonate (chol-C24:1) were very fluid, and we have shown previously ²³ that they formed a type of oil slick, despite cholesteryl nervonate being solid at 20°C. They gave a peculiar pressure area profile, in that they did not exhibit much surface pressure, exhibited substantial hysteresis, and did not collapse. This appeared to be a hybrid between the saturated films and the unsaturated films. Heating the films to 35°C caused no change to the saturated ester (not shown), enhanced the profile of chol-C18:2, gave minimal change to chol-C20:4, and decreased hysteresis of chol-C24-1 (Fig. 4). 

Figure 4. 

 

Comparison of pressure area isocycles of films made from CEs. The left side shows the first and equilibrium isocycles at 20°C of the different CEs and the right side shows the appearance of pressure area isocycles after the films were heated to 35°C. These isocycles (right) have been separated vertically for clarity. All isocycles are drawn at the same scale.

Fatty Acids

When the unsaturated fatty acids were applied to the surface, they were initially very surface active and then their films gave complex isocycles (Fig. 5). The surface pressures gradually reduced and became simpler. We believe that the free fatty acids are forming micelles that disappear into the aqueous subphase, rather than remaining on the surface as a film. 

Figure 5. 

 

Comparison of the first and equilibrium pressure area isocycles at 20°C of films made from unsaturated fatty acids in order of the text. All isocycles are drawn at the same scale and for the one fatty acid, the isocycles have not been separated vertically (i.e., there is a decrease in surface pressure as isocycles continue after the first isocycle).

Mixtures of Wax Esters with Meibomian Lipids

For these experiments, very high ratios of WEs to meibomian lipids were chosen (1 in 5 by mass) on the basis of trying to ensure a noticeable effect if there were to be one. Two of the waxes (C18:0-C18:1 vs. C18:1-C18:0) were chosen to compare the effect of having the double bond in the alcohol component versus the acid component the effect. The effect of lengthening the alcohol chain (C20:0-C18:1 vs. C18:0-C18:1), and the effect of a very unsaturated WE (C18:3-C18:3) were also tested. Furthermore, C18:0-C18:1 and C20:0-C18:1 are naturally found in meibomian lipids and it was of interest to see if changing their ratios by a marked amount had a noticeable effect on the meibomian lipid pressure-area isocycle profile. 

When different waxes were mixed with meibomian lipids (Fig. 6), there was a noticeable change in the profile in its second phase (the part of the curve at surface areas < approximately 40 cm²). Irrespective of the wax, the mixed films had the characteristic of pure meibomian lipid films in that: the films did not collapse (compared with pure arachidyl oleate; Fig. 3A); and the equilibrium isocycle at 20°C had a higher maximum pressure than the first isocycle (shown only for oleyl stearate; Fig. 6). On heating to 35°C, the isocylces became much smoother with little hysteresis, had a higher take-off pressure, and lower maximum surface pressure. On cooling back to 20°C the films became quite expanded, thus, showing a much greater maximum surface pressure (shown only for oleyl stearate; Fig. 6). For comparative purposes the equilibrium isocycles at 20°C are shown on the bottom of Figure 6 and include an equilibrium isocycle for pure meibomian lipids (right only, for clarity). For both stearyl oleate and linolenyl oleate mixed with meibomian lipids, the maximum surface pressure was the same as that for meibomian lipids alone (horizontal line on the bottom of Fig. 6). For meibomian lipids with arachidyl oleate or oleyl stearate, the maximum surface pressures for the equilibrium cycles at 20° were higher than for meibomian lipids alone and highest for the arachidyl oleate mixture. This is consistent with pure arachidyl oleate and oleyl stearate films forming being very rigid and collapsing during isocycles (Figs. 2, C18:1-C18:0, and 3A, C20:0-C18:1). Again, these profiles tend to indicate a duplex film with surfactant molecules (possibly OHAFAs)in the meibomian lipids interacting with each other to form the compression part of the curves from 80 cm² to 40 cm² (curves were almost identical irrespective of which lipid they were mixed with) and then a different shaped curve for the different mixtures as the upper layers of the duplex film containing the seeded waxes interacted with each other 40 cm² to 16 cm². 

Figure 6. 

 

Comparison of equilibrium isocycles of different WEs mixed with meibomian lipids at 20°C and 35°C. For oleyl stearate, the first isocycle at 20°C and the equilibrium isocycle after having been cooled from 35°C to 20°C are also shown. The bottom compares the equilibrium isocycles of the different WE-meibomian lipid mixtures overlaid, and exploded. An equilibrium isocycle of meibomian lipids alone is also shown in the exploded view, and the line in the overlay represents the equilibrium maximum surface pressure of meibomian lipids alone.

Micrographs comparing the appearance of meibomian lipids mixed with oleyl stearate or stearyl oleate indicated that oleyl stearate mixed less well with the meibomian lipids than stearyl oleate (Fig. 7). This is indicated by the dark patches in the films with oleyl stearate micrographs compared with gray patches in films containing stearyl oleate. At high pressures, both 20°C and 35°C, the films containing stearyl oleate were very amorphous, whereas films with oleyl stearate still had some dark patches at 20°C and a speckled appearance at 35°C. The speckled appearance indicates that the film was less homogeneous compared with films containing stearyl oleate. Once cooled back to 20°C, both films were relatively homogeneous, but at very high pressures, some patches could be seen in both mixed films. Some evidence for a duplex film comes from the micrographs for films containing oleyl stearate, which can be seen in Figure 7 at 6.6 mN/m and 7.5 mN/m during isocycle 3. These micrographs are of the films when the isocycle curve becomes very steep at relatively small surface areas (< 30 cm²). The dark patch seen in both micrographs is compressed to a smaller size, which could indicate that the molecules on the upper layer of duplex film are being compressed together as the surface area decreases. 

Figure 7. 

 

Micrographs comparing the appearance of meibomian lipids mixed with oleyl stearate or stearyl oleate. In each box, the surface pressure during compression of the film and the isocycle is given (e.g., 8.8 iso 3 means that the pressure was 8.8 mN/m during compression in isocycle 3). In some cases, the films are shown on expansion (opening), and in other cases at 35°C or having been cooled from 35°C to 20°C. For oleyl stearate at pressures 6.6 mN/m the, film was moving slowly and the same dark area can be seen at 7.5 mN/m. It is essentially the same shape but compressed in size. Scale bar equals 100 μm.

Films containing arachidyl oleate or linolenyl linolenate at 20°C (Fig. 8) were similar in appearance to those containing stearyl oleate, giving gray patches in the film, but these patches appeared to be smaller in the films containing linolenyl linolenate. Once heated and at high pressure, films with arachidyl oleate seemed to be similar to those of oleyl stearate and showed a patchiness indicating the films were not homogeneous. This contrasts with the films containing linolenyl linolenate, which formed a homogeneous film at these pressures. Notable in the films containing linolenyl linolenate was the appearance of small dark circular regions (possibly lipid lenses) particularly noticeable at high pressures and once the films had been cooled back to 20°C. Similar to oleyl stearate, some patches could be seen to be compressed at moderate pressures in the linolenyl linoleate films. For the arachidyl oleate films, once cooled back to 20°C, there were some very dark areas indicating that there were perhaps zones containing mainly the WE. 

Figure 8. 

 

Micrographs comparing the appearance of meibomian lipids mixed with arachidyl oleatete or linolenyl linolenate. In each box, the surface pressure during compression of the film and the isocycle is given. In some cases, the films are shown on expansion (opening), and in other cases at 35°C or having been cooled from 35°C to 20°C. For linolenyl linolenate at pressures 15.1 mN/m the, film was moving slowly and the same darker area can be seen at 17.2 mN/m (arrows). It is essentially the same shape but compressed in size. Scale bar equals 100 μm.

A general observation was that for films containing oleyl stearate or arachidyl oleate, the films appeared to fracture on expansion (Figs. 7, 8, opening), which would account for the increased hysteresis in these films. Another general observation for all mixtures with meibomian lipids was that the films were very mobile at low pressures and at 35°C. This is expected when there is a low density of molecules on the surface or there is a lot of thermal energy in the system. 

Discussion

A notable feature of this study is that meibomian lipid films are quite tolerant of very large changes in their composition. This is indicated by the fact that mixing large proportions of the waxes with the meibomian lipids caused little change to their pressure area isocycle profiles. In general, the effect of adding different lipid types to meibomian lipid films was that the more unsaturated the component (particularly the acid component) and the longer the saturated carbon chain (alcohol or acid), the more rigid the film. It is also evident from the properties of arachidyl oleate, which has been shown to comprise approximately 4.3% of WEs in meibomian gland secretions, ⁹ that the film needs to have the stiffness associated with such waxes. This stiffness would be further imbued by saturated waxes that we have not tested here such as C20:0-C16:0 (0.5% of total WEs), and unsaturated waxes with oleate as the fatty acid component esterified to longer chain saturated alcohol components such as C24-29:0-C18:1 WEs (approximately 70% of total WEs). ⁹ At the same time, meibomian lipid films must remain fluid, and in this case, it appears that having the unsaturated fat in the acid rather than the alcohol component is important, as the films with oleyl stearate are much stiffer than films of stearyl oleate. This is intuitively surprising because the only difference in these molecules (stearyl oleate and oleyl stearate) is the relationship between the ester bond and the double bond and, hence, an expectation that there would be little difference. However, it appears that meibomian WEs have evolved to have this particular feature. 

An important observation was that the use of linolenyl linolenate mixed with meibomian lipids appeared to form lipid droplets in the film. This needs to be considered when developing lipid-based eye formulations. It seems that using unsaturated lipids might severely disrupt the lipid layer rather than stabilize it. A similar expectation would be of fatty acids as all of the fatty acids appeared to form micelles and move off the surface. 

Pure and mixed films of CEs have been a focus of studies investigating their roles in the formation of fatty plaque in atherosclerosis. ^24–26 In such studies, spread films of saturated waxes were stiff and collapsed (they did not show pressure area curves), and spread films containing unsaturated esters, cholesteryl linoleate, and cholesteryl arachidonate, ²⁴ showed similar characteristics to those here: there was virtually no surface pressure on initial application, but after aging of the films, marked surface pressure was observed for cholesteryl arachidonate films, and marked, but less surface pressure for cholesteryl linoleate. Only isotherms (the compression part of the isocycle only) were presented and so it was not possible to compare the hysteresis of these films on expansion. They believed that the increase in surface pressure through aging was due to auto-oxidation of the double bonds in the ester chain. We believe that this is highly unlikely because in oxidation experiments that we have done (data not shown here), auto-oxidation is far too slow to account for the changes seen (an increase in surface pressure occurs from the first isocycle onwards). In addition, if auto-oxidation were causing an increase in pressure, ²⁴ it would also be seen in the WE film–pressure area isocycles where an increase in surface pressure would be reflected in an increase in initial take-off area (the area where the pressure first rises above the baseline). This was not the case, but instead the take off moved to the left (smaller surface areas), which is consistent with very unsaturated WEs moving off the surface into the outer phase (bulk). Therefore, a rearrangement, rather than chemical modification, of the molecules is much more likely. When first applied, it is likely that the molecules are very adherent to each other and, hence, form lipid lenses that are pushed into each other on compression (no change in surface pressure), but with time more molecules interact with the aqueous surface via the ester bonds. We have observed this directly in films of cholesteryl nervonate. ²³ This idea of the cholesterol esters interacting through their ester bond is supported by investigations of mixed monolayers of cholesterol esters with the triglyceride (triolein). ²⁷  

An interesting consideration in meibomian lipids is that the fatty acid component of CEs can contain up to 30 C-atoms and typically is monounsaturated. ⁶ The data from pure CE films presented here indicate that when the fatty acid component is unsaturated the films collapse, whereas being monounsaturated the films are stiff and resist collapse, but fragment easily once the film is expanded (indicated by the strong hysteresis; Fig. 4). Using polyunsaturated fats as the acid moiety, leads to very fluid films. We have previously mixed cholesteryl nervonate with meibomian lipids and found that it initially mixed well, but eventually was removed into the upper layer of the film (the bulk) where it seemed to form clumps. ²³ In this case the films were observed using high resolution color microscopy which enabled discernment of interference colors (thicknesses) and lenses. This technique might be usefully applied in future studies of mixtures of the WE and meibomian lipid films, particularly to resolve if lenses were formed when linolenyl linolenate was mixed with meibomian lipids. 

For films of WEs alone, they appeared to interact with the surface to some degree and this is probably through the most hydrophilic component, the ester bond. Such an interaction was proposed by Adam ²⁸ after investigating the effect of changing the length of the alcohol chain component. In terms of general observations, he found that an alcohol chain of 4 C-atoms, the thermal energy of the system had to be lowered, and the lateral force on the molecules increased to have the alcohol tail align in the subphase. For longer carbon chains, he modeled the ester bond sitting on the subphase surface with both alkyl chains bending off this and directed off the surface towards the air. The effect was eloquently described as a liquid expanded state such “that the molecules are held to the water by their oxygen containing groups, which also attract one another laterally, providing a good deal of the lateral adhesion between the molecules; the long chains are probably whipping about in violent agitation, causing the molecules to occupy more space on the surface than they do in the condensed state.” ²⁸ Because of this “violent agitation” of the alkyl chains he found that octyl palmitate (C8:0-C16:0) occupied more space than expected at take-off 83Ų/molecule, which is greater than that reported here (e.g., stearyl linoleate [C18:0-C18:2]), which had an average area of 49Ų/molecule (Fig. 1). This can be explained by his model, ²⁸ which shows that the longer chains fold more towards each other to form a very acute angle between them at the ester bond (i.e., they take up less space on the surface). However, it is also possible that not all molecules are on the surface but multilayered, as evidenced in a study using attenuated total reflectance spectroscopy using a mixture of oleyl oleate and stearyl palmitate as a model. ²⁹ In this study, no shift of the ester carbonyl band absorption frequency of the infrared spectra of the WE mix exposed to an aqueous buffer could be seen, which might be indicative of the fact that the molecules are organized in a bulk rather than interacting with the surface of the water phase. 

Compared with meibomain lipid films, the pure wax films did not show the characteristics of meibomian lipid films when they were heated and then cooled. Although the isocycles had a greater maximum surface pressure after cooling, the take-off area had decreased (Fig. 3C). This suggests that molecules have moved off the surface during heating and reorganized into a superficial lipid layer (the bulk), which would be characteristic of a duplex film. This contrasts with meibomian lipid films where there are presumably polar molecules at the surface, which do not re-organize, and so there is no apparent shift of take off when films are cooled from 35°C to 20°C (i.e., the surfactant molecules remain on the surface at all temperatures). 

Although these data cannot be compared directly with the lipid layer of the tear film (in vivo), they are informative. Our data are consistent with the idea described by Borchman et al. ¹¹ that the more saturated the WEs, then the more ordered (rigid) the film. However, Borchman et al. ¹¹ concluded this from phase transition evaluation of bulk meibomian lipids. In our case, we were looking at spread films and when meibomian lipids were mixed with different WEs, the meibomian lipid films were very tolerant of the added WEs as measured by pressure area isocycles (i.e., the degree of saturation [or unsaturation] had minimal effect). Overall, these data are consistent with the WEs in meibomian lipids being mainly in the outer layer of a duplex film. While the ester bond of the waxes can interact with a polar surface, for example, the water interface, or the surfactant molecules (probably OHAFAs), as soon as energy is put into the system through blinking, the waxes would move off the surface. Also, off the surface would be CEs, which are likely to form thermotropic smectic chiral liquid crystals. ^27,30 This means that being liquid crystals (mesophase), they have local order, that is, the cholesterol rings all align in approximately the same direction and this gives liquid qualities that allow flow, but solid qualities, which would resist collapse of the film onto the ocular surface. It also means that they are layered (smectic) in a spirally and tilting manner as one moves from one layer to the next (chiral), and if enough heat is added to the system, it will become liquid (thermotropic). ³¹  

In addition, despite the results presented here being limited to commercially available wax and CEs, they do give insight into the importance of the structure of the waxes and CEs to the physicochemical properties of the meibomian lipid films. The data suggest that using eye drops containing very fluid fats or waxes might indeed be deleterious to the lipid layer, rather than enhancing its performance, for example, if the conjecture about the formation of smectic crystals proves to be true, then cholesteryl arachidonate, which does not form liquid crystals, may disrupt the lipid layer if used in eye drop formulations, whereas oleoyl, linoleoyl, and linolenoyl esters of cholesterol, which do form smectic liquid crystals, might enhance the performance of the lipid layer. 

Footnotes

 Supported by the Australian Government Cooperative Research Centres Vision Research Grant.

Footnotes

 Disclosure: B.S. Schuett, Allergan (F), Alcon (F); T.J. Millar, Allergan (F), Alcon (F)