The Effect of Ambient Ozone on Unsaturated Tear Film Wax Esters

Abstract

Purpose: Tear film lipid layer (TFLL) is constantly exposed to reactive ozone in the surrounding air, which may have detrimental effects on ocular health. Behenyl oleate (BO), a representative tear film wax ester, was used to study the reaction with ozone at the air-water interface.

Methods: Time-dependent changes in mean molecular area of BO monolayers were measured at different ozone concentrations and surface pressures. In addition, the effect of ascorbic acid on the reaction rate was determined. Reaction was followed using thin-layer chromatography and reaction products were identified using liquid chromatography-electrospray ionization mass spectrometry (LC-MS). Tear fluid samples from healthy subjects were analyzed with LC-MS for any ozonolysis reaction products.

Results: Behenyl oleate was found to undergo rapid ozonolysis at the air-water interface at normal indoor ozone concentrations. The reaction was observed as an initial expansion followed by a contraction of the film area. Ascorbic acid was found to decrease the rate of ozonolysis. Main reaction products were identified as behenyl 9-oxononanoate and behenyl 8-(5-octyl-1,2,4-trioxolan-3-yl)octanoate. Similar ozonolysis products were not detected in the tear fluid samples.

Conclusions: At the air-water interface, unsaturated wax esters react readily with ozone in ambient air. However, no signs of ozonolysis products were found in the tear fluid. This is most likely due to the antioxidant systems present in tear fluid. Last, the results show that ozonolysis needs to be controlled in future surface chemistry studies on tear film lipids.

Ozone (O₃) is one of the major oxidants of the lower atmosphere and is created in a set of complex photochemical reactions involving oxygen, methane, nitrogen oxides, volatile organic compounds, and carbon monoxide.¹ The main health effect of ozone exposure is lung irritation.² Because ozone is too reactive to penetrate far into the lung tissue, it has been suggested that the toxic effects of ozone are initiated by the reaction with unsaturated lipids in the lung-lining fluid of the alveoli.³ The products of this reaction, namely aldehydes and hydrogen peroxide, have been suggested to cause toxic effects in lung tissue and at extrapulmonary sites.³ Rapid removal of aldehydes is essential to maintain the health of epithelial cells.⁴ 

Tear film bears many similarities with lung-lining fluid of the alveoli, as these two structures are the only air-water interfaces in the human body.⁵ Being in constant contact with surrounding air, tear film is also subject to similar ozone stress as the alveolar fluid, and therefore it is possible that increased ozone exposure may lead to ocular irritation through a similar mechanism. Interestingly, exposure to increased ozone concentrations has indeed been shown to cause dry eye symptoms in mice, although their aqueous tear production was not affected.⁶ In humans, elevated ozone levels in ambient air have been associated with eye irritation, but this effect has been considered to be due to other oxidants associated with ozone, such as aldehydes and peroxyacetyl nitrate.² 

The purpose of this study was to evaluate the effects of ozone exposure in ambient air to the lipid component of the tear film. Ozone readily reacts with double bonds in hydrocarbon chains. Because wax esters (WEs) are the most abundant unsaturated component of tear film lipid layer (TFLL),^7–11 in the current study we observed the ozonolysis of behenyl oleate (BO), a representative wax ester, at the air-water interface under ambient conditions. Langmuir film techniques coupled with Brewster angle microscopy (BAM) were used to monitor the changes in the film properties, and thin-layer chromatography (TLC) and liquid chromatography–mass spectrometry (LC-MS) were used to monitor the changes in the composition of the films. In addition, tear fluid samples were analyzed using LC-MS to identify possible ozonolysis products of the human TFLL. 

Behenyl oleate was found to readily react with ambient ozone at the air-water interface to form potentially harmful aldehyde and ozonide products. No ozonolysis products were found in tear fluid samples of healthy subjects, likely due to the antioxidant mechanisms present in tear fluid. 

Materials and Methods

Behenyl oleate was obtained from Sigma-Aldrich Corp. (St. Louis, MO, USA), dissolved in chloroform, and stored at −20°C until used. Ascorbic acid was obtained from Sigma-Aldrich Corp. 

Film Area Measurements

Wax esters were applied to the surface of a KSV Mini trough (Helsinki, Finland) in 5 mM chloroform solution and immediately compressed to the target surface pressure at a speed of 10 to 20 Ų molecule⁻¹ min⁻¹. Phosphate-buffered saline was used as the subphase in all experiments unless otherwise indicated. Surface pressure was maintained constant and the change in trough area with time was measured. Temperature of the trough was maintained at 36 ± 1°C during the experiments using a Lauda ECO E4 thermostat (Lauda-Königshofen, Germany). Measurements were performed at different surface pressures and ground level ozone concentrations. A KSV NIMA microBAM (Helsinki, Finland) was used to record images of the films at 2-minute intervals. 

For the LC-MS and TLC, lipids were collected after the experiment by compressing the lipid film to a small (approximately 5 cm²) area using the barriers and a polytetrafluoroethylene (PTFE) plate. The lipids were then swept from the surface using a polyvinylidene difluoride (PVDF) filter. Both the PVDF filter and the PTFE plate were immersed and stirred in 2:1 chloroform:methanol to dissolve the lipids, and the solution was then stored at −80°C until analyzed. 

Tear Fluid Collection

The study was conducted according to the Declaration of Helsinki and approved by the Ethical Committee of the Helsinki-Uusimaa Hospital District. A written informed consent was obtained from each subject. Tear samples were collected from the lower conjunctival sac of five healthy volunteers (age 25–35 years) using 5-μL microcapillaries under a biomicroscope in three separate sessions as described previously.¹² After emptying, the microcapillary tips were briefly dipped to 2:1 chloroform:methanol solutions to dissolve any lipids that adhered to the capillaries. Tear film lipids were extracted from a total of 165 μL of tear fluid as described previously.¹² The solution used for washing microcapillary tips was combined with the extracted lipids. Samples were stored at −80°C and analyzed using LC-MS as described below. 

Liquid Chromatography–Mass Spectrometry

The liquid chromatographic analysis was performed with a Hewlett-Packard Series 1100 liquid chromatograph (Palo Alto, CA, USA) coupled with an Esquire 3000 plus ion trap mass spectrometer (Bruker Daltonics, Billerica, MA, USA) via electrospray ionization (ESI). Chromolith HighResolution RP-18 endcapped column (50 mm, 4.6 mm i.d.; Merck, Darmstadt, Germany) was used for analysis in isocratic mode with methanol, modified with 5% chloroform and 2% ammonium acetate. Higher percentages of chloroform, which would speed up the analysis, could not be used due to low tolerance of monolith toward halogenated solvents. Thus, analysis time was 60 minutes for BO samples and 180 minutes for tear fluid samples with flow rate of 0.5 mL min⁻¹. Nebulizer (N₂) pressure was 30 psi, drying gas (N₂) flow and temperature were 10 L min⁻¹ and 300°C, respectively. Standard samples containing BO, stearyl oleate, myristyl oleate, cholesteryl oleate, tripalmitin, trimyristin, trilaurin, tricaprin, tricaprylin, eggPC, and lyso-eggPC were used to identify the major lipids in tear fluid samples. Samples were dried under a gentle stream of nitrogen and dissolved in 1 mL of mobile phase. For standard samples, 10 μL was injected, whereas for tear samples also 20-μL injection was tried. Scan range was 100 to 750 amu for BO samples and 100 to 1200 amu for tear fluid samples. AutoMSⁿ from base peak chromatograms was used for tandem MS experiments. 

Thin-Layer Chromatography

Behenyl oleate was spread on the water surface and maintained at a constant surface pressure of 0.5 mN/m for the duration of 0, 10, 30, 60, or 120 minutes before collecting the lipid film as described above. An additional sample was maintained at a constant surface pressure of 3.4 mN/m for 120 minutes before collection. The samples were analyzed using TLC as described previously for nonpolar lipids.¹² The intensity of the spots was quantitated using ImageJ (http://imagej.nih.gov/ij/; provided in the public domain by the National Institutes of Health, Bethesda, MD, USA).¹³ Oxidation product standards are not available and therefore spot intensities were scaled according to the number of carbon atoms in each molecule to obtain a better approximation of the molar concentration in the samples. 

Ozone Concentration

Ozone concentration data were obtained from the Finnish Meteorological Institute Web service (http://www.ilmanlaatu.fi/index.php). Ozone concentrations reported in this study represent an average of outdoor ground-level ozone concentrations measured at the two closest air-quality monitoring sites in Helsinki (Kallio 2 and Mannerheimintie). The sites are located in a similar urban environment 2.5 km and 3.0 km away from the site where the lipid layer measurements were performed, respectively. Ozone concentration was averaged over the duration of each measurement. Each measurement was repeated at least three times and the presented ozone concentrations indicate the mean and SD of ozone concentration during separate experiments. As a control, the measurements were repeated within a plastic enclosure (volume 80 L). Dry air was passed through an Ozone Solutions ODS-3P ozone destruct unit (Hull, IA, USA) and into the enclosure at a rate of 76 L/min in order to maintain an ozone-free atmosphere inside the enclosure. 

Results

Film Area Measurements

Monolayers of BO maintained at a constant surface pressure of 0.5 mN/m displayed time-dependent changes in mean molecular surface area (MMA), which depended on the concentration of ground-level ozone (Fig. 1). An initial increase in MMA of the film was observed, followed by a decrease in MMA. A large peak of 58 Ų/molecule was observed within 10 minutes at high ozone concentrations (62 ± 3 μg/m³). The peak appeared later and peak magnitude decreased with decreasing ozone concentration. At low ozone concentrations (11 ± 6 μg/m³) a wide peak of 29 Ų/molecule was observed at approximately 50 minutes. When the lipid layer was maintained in an ozone-free atmosphere, the surface area remained constant at 28 Ų/molecule for 70 minutes. Initially, a low-intensity reflection was observed in BAM images regardless of ozone level, indicating a thin, fluid film (Fig. 1a). At high ozone concentrations, solid domains appeared in BAM images approximately 10 minutes after the beginning of the experiment and their size and number increased over time (see Figs. 1b, 1c). When ground-level ozone concentration was low (< 20 μg/m³), no change was observed in BAM images within the duration of the measurement (Figs. 1d, 1e). 

Figure 1

 

Effect of ozone concentration on the stability of BO monolayers at 0.5 mN/m surface pressure. Mean molecular area of the film with the corresponding BAM images. Shaded areas in the graph represent SEM. The position of the letters represents the time at which the BAM images were captured. Dark regions in BAM images (a–e) are fluid areas of the film and light regions are condensed domains. Scale bar: 500 μm.

The effect of surface pressure on BO monolayers in high ozone concentrations (63 ± 13 μg/m³) is shown in Figure 2. The initial surface area decreased with increasing surface pressure and a peak in MMA was observed in at 0.5 to 2.0 mN/m surface pressures. At 3.4 mN/m, MMA of the film remained stable at 18 Ų/molecule for 120 minutes, and in BAM the film had a similar solid appearance, as has been reported for behenyl palmitoleate previously.¹⁴ No change was observed in the appearance of the solid film during the experiment (Supplementary Fig. S1). 

Figure 2

 

Effect of high ozone concentration (63 ± 13 μg/m³) on BO monolayer area at different surface pressures; MMA is presented as mean ± SEM.

The effect of antioxidants on the ozonolysis rate was studied by adding ascorbic acid to the subphase beneath the wax ester film (Fig. 3). The addition of 5 mM ascorbic acid did not prevent the increase in MMA, but the peak in MMA appeared later, similar to lower ozone concentrations (Fig. 1), indicating a reduction in the ozonolysis rate. 

Figure 3

 

Effect of subphase ascorbic acid on the ozonolysis rate of BO at the air-water interface; MMA is presented as mean ± SEM. Ozone concentration during the experiments was 42 ± 11 μg/m³.

Liquid Chromatography–Mass Spectrometry

Total ion chromatograms of BO samples maintained at a surface pressure of 0.5 mN/m for 0, 60, and 120 minutes are shown in Figure 4 with the mass spectra corresponding to the peaks. Three major components were detected: (1) BO (mass 590.6 amu) had a retention time of 42.6 ± 0.4 minutes and was detected as (M + H)⁺, (M + NH₄)⁺, and (M + Na)⁺ ions. Fragmentation around the ester moiety was observed in collision-induced dissociation (CID) (Supplementary Fig. S2A), similar to that observed previously for WEs.¹⁵ (2) A more polar component was observed as a double peak with retention times of 23.2 ± 0.2 minutes and 25.2 ± 0.2 minutes. This molecule was detected as (M + NH₄)⁺, (M + Na)⁺, (M + K)⁺, and (M + 59)⁺ ions and had a molecular mass of 638.5 amu. The identity of the 59-amu adduct is unknown, but it has been also previously detected in ESI-MS with methanol- and acetic acid–containing eluent.¹⁶ The 48-amu increase in mass compared with BO indicated that this compound was an ozonide. In addition, CID showed characteristic fragmentation around the 1,2,4-trioxolane ring¹⁷ (Supplementary Fig. S2B), confirming the identity of this compound as a BO ozonide (BOoz). Same fragments were detected for both peaks, indicating that the ozonide is present as cis- and trans-isomers, as has been previously detected for secondary ozonides.¹⁷ (3) An even more polar component was detected with a retention time of 5.3 ± 0.1 minutes and mass of 480.4 amu. Multiple adducts were detected, including (M + H)⁺, (M + 15)⁺, (M + NH₄)⁺, (M + MeOH + Na)⁺, and (M + MeOH + 59)⁺. This mass corresponds to a typical ozonolysis product, behenyl 9-oxononanoate (Boxno). The (M + 15)⁺ ion is possibly an oxocarbenium ion formed in a reaction with methanol in the ionization process as has been reported previously for aromatic aldehydes.¹⁸ In CID, fragmentation around the ester moiety was observed, identical to BO (Supplementary Fig. S2C). In addition to the three main peaks, several peaks with smaller retention times were observed (Supplementary Fig. S3). However, these peaks were ignored, as they were also observed in a blank sample and were putatively identified as sodium acetate clusters, phthalates, and Irganox-antioxidants likely originating from the solvents and containers used for collecting and handling the samples. 

Figure 4

 

Liquid chromatography–MS (positive ESI) total ion chromatograms of BO samples held at the air-water interface at 0.5 mN/m surface pressure for 0 to 120 minutes. A mass spectrum and a proposed structure are presented for each detected compound. Ozone concentration was 27 ± 8 μg/m³ during the experiments.

The integrated intensity of the BO ions decreased by 90% during 120 minutes and intensities of BOoz and Boxno ions increased by 11- and 17-fold, respectively, demonstrating almost complete conversion of BO into BOoz and Boxno. 

Thin-Layer Chromatography

Thin-layer chromatography of BO collected from the air-water interface after different incubation times at either 0.5 mN/m or 3.4 mN/m surface pressure revealed the same major compounds that were detected in the LC-MS analysis (see Supplementary Fig. S4 for original TLC plate). Cis- and trans-isomers of BOoz were visible as a distinct double spot also on TLC analysis, as has been previously observed for methyl ester ozonides.¹⁹ Densitometric analysis of the TLC plates is shown in Figure 5A. At a surface pressure of 0.5 mN/m, a gradual decrease in the amount of BO was observed alongside an increase in the amount of BOoz and Boxno. At a surface pressure of 3.4 mN/m, conversion of BO into Boxno and BOoz also occurred, but the relative proportions of the products were different (Fig. 5B). While at 0.5 mN/m, the ratio of Boxno:BOoz was approximately 3:2, at 3.4mN/m, the ratio was 1:5. Thin-layer chromatography analysis also showed minor spots not detected in the LC-MS analysis, including a second, more polar double spot, possibly a diester ozonide and a spot in the fatty acid region. However, as the intensity of these spots was less than 3% of total intensity in all of the experiments, their identification was beyond the scope of this study. 

Figure 5

 

(A) Densitometric analysis of the TLC measurements of BO samples held at the air-water interface at 0.5 mN/m surface pressure for 0 to 120 minutes. (B) Composition of BO samples after 120 minutes at either 0.5 mN/m or 3.4 mN/m surface pressure. Ozone concentration during the measurements was 37 ± 9 μg/m³.

Tear Fluid Analysis

Identified ions from the tear fluid samples are presented in the Figure 6. Major lipid classes detected in the tear fluid samples in positive ion mode were WEs, cholesteryl esters (CE), triglycerides (TG), and lysophospholipids. In negative ion mode, fatty acids and (O-acyl)-omega-hydroxy fatty acids (OAHFA)²⁰ were detected. Number of carbons in detected WEs varied from 38 to 48 and number of double bonds from 0 to 4. Highest intensities were detected for 44:2, 43:1, 44:1, 42:1, 42:2, and 46:2 WEs. Detected cholesteryl ester chain lengths were from 18 to 24 carbons with 0 to 3 double bonds. Longer CEs were not detected due to their very long retention times (>3 hours) with the system used. Total number of TG carbons varied from 36 to 56 with a total of 0 to 5 double bonds. Notably, only phosphatidylcholine (PC) phospholipids were analyzed, and major PCs found in the tear fluid were 16:0, 18:0, and 18:1 lyso-PCs. Diacyl PCs were not detected in significant amounts. Chain lengths of the detected fatty acids varied from 12 to 34 with 0 to 3 double bonds. Total number of carbons in detected OAHFAs ranged from 40 to 54 with 0 to 4 double bonds. Both ozonolysis products of BO were clearly separated from all major tear film lipids (Fig. 6A), and therefore any similar ozonolysis products would be easily separated from other tear film lipids. The MS data were analyzed for analogous ozonolysis products of all the detected WEs with one or two double bonds and CEs with one double bond in the mass/retention time range shown in Figure 6A. No ions corresponding to the ozonides (+48 amu) were detected. Some ions corresponding to the aldehyde products (−110 amu) were detected with retention times of 3 to 25 minutes. However, these ions coeluted with higher mass ions, which appeared as +44 amu series, suggesting that they are likely polymer contamination, possibly from the containers used in handling the samples. Therefore, no ions with significant intensity could be reliably identified as ozonide or aldehyde products of WE or CE in the tear fluid samples. 

Figure 6

 

Main lipid compounds identified in tear fluid samples by LC-MS analysis. Retention times of BOoz and Boxno are shown for comparison. Symbols are separated by one carbon unless indicated otherwise. Symbol shape indicates the number of double bonds. Symbol size represents detected relative intensity. (A) Positive ion mode. Dashed lines indicate the analysis range used to search for ozonolysis products. (B) Negative ion mode.

Discussion

The results presented here demonstrate that BO decomposes at the air-water interface in the presence of normal levels of ground-level ozone (Fig. 1). Similar results were also obtained for other monounsaturated WEs, including behenyl palmitoleate, stearyl oleate, and lauryl oleate (data not shown). The main products formed in the decomposition of BO in ambient air were found to be BOoz and Boxno (Fig. 4), which are the primary products expected from direct ozonolysis. In addition, when ozone was removed from the air using an ozone annihilator, no decomposition was observed (Supplementary Fig. S5). Therefore, it is apparent that direct ozonolysis is the main decomposition mechanism for BO at the ambient air-water interface. This is consistent with earlier results showing that oleic acid and unsaturated phospholipids only undergo autoxidation at the air-water interface in the presence of O₃ but not in the presence of pure O₂.^21,22 

Figure 7A presents the ozonolysis reactions leading to the formation of Boxno and BOoz.²³ The first step is the reaction of ozone with the BO double bond to create a very unstable molozonide. Molozonide rapidly decomposes into a Criegee intermediate (CI) and an aldehyde. The aldehyde and CI can then react with each other to rearrange into a more stable ozonide with 1,2,4-trioxolane ring (BOoz). The short-chain compounds created by the decomposition of the molozonide (nonanal and CI-1) are volatile and slightly soluble in water, so instead of rearranging into a more stable ozonide, they may leave the film instead of reacting with Boxno or a long-chain CI (CI-2). In the presence of water, CIs have been shown to react with water to produce hydroxyhydroperoxides (HHPs), which then decompose to form an aldehyde and H₂O₂.²⁴ The CI-1 and CI-2 are therefore likely to react with the water of the subphase, which leads to the formation of nonanal and Boxno. 

Figure 7

 

(A) Proposed reaction scheme for the ozonolysis of BO at the air-water interface. (B) A tentative model of the reaction kinetics used to explain the changes in MMA (see Discussion for details).

When maintained at a surface pressure of 0.5 mN/m, BO forms a mostly fluid monolayer, in which the likely conformation is that hydrocarbon chains are disordered and the ester moiety is facing toward the water phase. When surface pressure is increased, film begins to assume a condensed state, in which BO likely forms a tightly packed lattice in an extended conformation, as has been previously reported for behenyl palmitoleate.¹⁴ 

Figure 7B depicts a qualitative model based on the reactions shown in Figure 7A, which can be used to explain the observed surface area changes (Figs. 1, 2) and compositional changes (Fig. 5A). The abundance of all detected compounds changed monotonically, unlikely to cause two-phased changes in surface area. Therefore, the initial expansion of the film is likely due to the initially rapid formation of CIs (CI-1 and CI-2), hydroxyhydroperoxides (HHP-1 and HHP-2), and nonanal, which are the most polar of the products and would favor the expansion and spreading of the film. Criegee intermediates and hydroxyhydroperoxides are removed from the film as they react to from Boxno, BOoz, and nonanal. Nonanal is likely to leave the film due to evaporation or dissolution to the subphase. Only BO, BOoz, and Boxno, which favor the formation of a solid film, remain in the film. This leads to the contraction phase, which is accompanied with the formation of solid, needle-like domains, as observed with BAM (Figure 1). The intensity of these solid domains is lower compared with solid BO films, likely because of the high abundance of shorter Boxno. 

At 3.4 mN/m surface pressure, no change in film area or appearance is observed. However, according to the TLC results (Fig. 5B), ozonolysis still occurs at a similar rate, but results in different distribution of products. At this surface pressure, BO molecules are packed tightly in an ordered lattice and therefore the CIs are more confined and likely to react back to form an ozonide rather than leave the lattice to react with water. Therefore, the formation of BOoz is favored over other ozonolysis products in high surface pressures. This also explains the seemingly stable appearance of the film in high surface pressures, as BOoz is similar in structure to BO and might retain a similar conformation, maintaining the film properties and appearance. 

According to clinical studies, the turnover rate of the TFLL is approximately as slow as the reaction rates estimated for BO in this study (approximately 1%/min).²⁵ Therefore, it was expected that ozonolysis products would accumulate also in the tear film. However, no appreciable amounts of ozonolysis products were detected in the tear fluid samples of the small number of healthy subjects participating in this study. Some specific molecular organization may occur in TFLL, which shields the double bonds of TFLL lipids from the effect of ozone. However, such effect was not observed in tightly packed BO films. Tear film is also known to contain water soluble antioxidants like ascorbate, urate and glutathione,^26,27 antioxidant proteins like superoxide dismutase,²⁸ and lipid soluble antioxidants like vitamin E,²⁹ squalene,³⁰ and plasmalogens.¹² It is likely that the antioxidant systems in tear fluid protect the TFLL from ozonolysis. To demonstrate this effect, 5mM ascorbic acid was added to the subphase, which significantly reduced the ozonolysis rate of BO (Fig. 3). 

There are at least two possible mechanisms, by which ozonolysis of the TFLL can cause detrimental effects for the ocular surface. First, some ozonolysis products may disturb the organization of the TFLL and destabilize the tear film. Second, some potentially toxic ozonolysis products like nonanal or other aldehydes may diffuse to the corneal epithelium or other eye tissues. Normal, healthy eyes are protected from both of these mechanisms by the antioxidant system described above and the detoxification enzymes of the cornea like aldehyde dehydrogenase 3A1, which makes up 5% to 50% of soluble corneal protein.³¹ It is likely that the high ozone concentrations, which have been shown to cause dry eye symptoms in mice,⁶ overload these protective systems, causing damage to the ocular surface. High ozone concentrations like this occur in highly polluted urban areas,¹ where also the prevalence of dry eye symptoms is increased.^32–34 

The results presented here also have practical implications for measurements with tear film lipid model systems. Nonpolar tear film lipids do not readily spread to the air-water interface, especially if the temperature is too low.^14,35 Due to this, an approach of using multiple compression-expansion cycles has been adopted in studies using Langmuir films.^30,35–51 However, this approach relies on the assumption that the lipid films remain chemically stable for the duration of the experiment, which is not the case for unsaturated WEs in ambient air. Also, measurements on unsaturated CEs have shown time-dependent changes in the isotherms due to autoxidation.⁵² Therefore, the chemical stability of the film should be ensured when lengthy monolayer studies are conducted so as to avoid confusing changes in film organization with changes in the composition of the film. 

Acknowledgments

Supported by the Finnish Eye Foundation, the Finnish Eye and Tissue bank Foundation, the Sigrid Juselius Foundation, the State Subsidiary System (EVO), the Evald and Hilda Nissi Foundation, and the Magnus Ehrnrooth Foundation. 

Disclosure: R.O. Paananen, None; A.H. Rantamäki, None; J. Parshintsev, None; J.M. Holopainen, Croma Pharma (C, S), Alcon (S), Allergan (S), Santen (S)