Analysis of the Macular Pigment by HPLG Retinal Distribution and Age Study

Richard A. Done,* John T. Landrum.t Lilio Fernandez, f and Sara L. Tarsis*

High performance liquid chromatography (HPCL) has been employed to study the distribution throughout the human retina of zeaxanthin and lutein, the two major components of the macular pigment. Differences between individuals have also been studied with a view to uncovering possible age-related effects. Both pigments were detected in prenatal eyes (~20 weeks gestation) but did not form a visible yellow spot. Generally they were not easily discernible until about 6 months after birth. For 87 donors between the ages of 3 and 95, no dependence on age was observed in the quantity of either pigment. For ~90% of these, zeaxanthin was dominant. For the remaining 10%, as well as for the seven youngest donors, all below the age of 2, and in prenatal eyes, lutein was the major pigment. In individual retinas, the lutein:zeaxanthin ratio increased from an average of approximately 1:2.4 in the central 0-0.25 mm to over 2:1 in the periphery (8.7-12.2 mm). The variation in this ratio with eccentricity was linearly correlated with the corresponding rodxone ratio. A selective mechanism of uptake, which results in cones and rods preferentially acquiring zeaxanthin and lutein, respectively, could explain this correlation. Invest Ophthalmol Vis Sci 29:843-849, 1988

The role of the macular pigments may be two-fold: has been addressed in a number of psychophysical to improve visual acuity1 and to protect retinal tissue investigations.8"11 One such study,9 involving sub- against photodegradation.2 While feeding studies jects in the age range 10 to 90, uncovered wide varia- using monkeys demonstrate the dietary origin of the tions in the optical density of their pigment, but no macular pigments,3 there are no reports on the mech- significant dependence on age. Optical densities ob- anism of their uptake by the neural retina. Is the tained by psychophysical methods, such as sensitivity mechanism highly discriminating, or are the pig- measurements," however, rest upon the validity of a ments transported nonselectively into retinal cells? number of assumptions. (See Pease et al10 for review Our recent observations, that the isomeric dihy- as well as means of circumvention.) Chemical analy- droxy-carotenoids, zeaxanthin and lutein, constitute sis, on the other hand, is relatively straightforward the macular pigment,4 might favor the latter. Alterna- and, being independent of observer skill, may be used tively, a selective uptake mechanism might deliver equally reliably on all age groups, including prenatal. the two pigments to different target cells within the retina. To explore these possibilities, we have applied high performance liquid chromatography (HPLC) to Materials and Methods examine the variation in pigment density and composition with retinal location and age. The rationale Sample Preparation for this approach was that the dependence of the cell populations on retinal eccentricity5 and age6'7 might Frozen donor eyes were provided by the Florida be reflected in the macular pigments. Lions Eye Bank and the National Disease Research The variability in pigmentation among individ- Interchange, and were stored in sealed containers at uals, including the possible contributing factor of age, -100°C until needed. Fetal eyes, supplied through the latter organization, were shipped on ice and analyzed immediately. Dissection of the thawed eye was From the Departments of *Physics and fChemistry, Florida In- ternational University (The State University of Florida at Miami), performed in 0.9% saline solution. After the neural Miami, Florida. retina had been separated, it was trimmed by cutting Presented in part at the 1986 and 1987 meetings of ARVO, around the equator, and inspected for signs of dam- Sarasota, Florida. age such as holes, tears or separation of retinal layers. Supported by NIH/MBRS Grant RR-O82O5-O2A3. If such damage occurred in the retinal area intended Submitted for publication: June 30, 1987; accepted December 7, for analysis, the macula was put aside for future use. 1987. Reprint requests: Richard A. Bone, PhD, Department of Physics, About 40% of all retinas were found to be unsuitable Florida International University, Miami, FL 33199. for the present study.

843

Downloaded from iovs.arvojournals.org on 10/02/2021 844 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / June 1988 Vol. 29

l_Microscope was employed, this being sufficiently large to include the largest macula lutea. By placing the retina on a spherical surface, selected to match as closely as possible the curvature of the retina, problems of tears, folds or stretching were virtually eliminated. For the very small prenatal eyes, however, this technique was not feasible. Instead, extracts from the entire neural retina were analyzed for the presence of carotenoids. Each annulus or disk of retinal tissue was trans- ferred to a glass tissue homogenizer, to which was added a known mass (~ 10 ng) of lutein monomethyl ether, as an internal standard, and about 2 to 3 ml of acetone for extraction of the carotenoids. This ether is readily prepared from lutein,12 and its HPLC retention time corresponds to a window in the retinal chromatogram. Furthermore, its similarity of structure to lutein and zeaxanthin should render it equally sensitive to decomposition during work-up and chromatography (see Results). The acetone solutions were centrifuged at low speed, filtered (0.2 nm nylon-mesh syringe-filters), and dried in 5 ml pear-drop flasks under a stream of pure nitrogen. With extracts from larger tissue samples, greasy contaminants were Fig. 1. Apparatus used for dissecting retinas into annuli concen- clearly visible on the walls of the flask. It was possible tric with the fovea. to preferentially dissolve the carotenoids by briefly swirling cold (-20°C) acetone around the prechilled flask and quickly transferring it to a clean flaskwher e A lucite sphere, approximately matching the cur- it was dried prior to injection on the HPLC. vature of the retina (1 inch diameter for adults), was placed in the saline-filled dissecting dish, and the ret- Quantification by HPLC ina maneuvered into position above it. Upon lifting the sphere from the solution, the retina could be HPLC was conducted on an LDC/Milton Roy sys- placed smoothly over it without folds or wrinkles. tem (LDC/Milton Roy, Riviera Beach, FL) including The sphere, with the retina uppermost, was then a Spectromonitor D UV-visible detector (0-0.001 seated in a rubber-lined ring in the apparatus shown absorbance units full-scale) set at 450 nm. At this in Figure 1. wavelength, absorption by zeaxanthin is maximum; that of lutein is ~92% of maximum. A 250 X 4.6 mm Co-axial with the ring was a low-power microscope column with C18 reversed-phase support (5 nm with cross-wires on which the operator centered the Spherisorb ODS1) was supplied by Keystone Scien- fovea by rotation of the sphere. In order to analyze tific (State College, PA). Essentially baseline separa- the distribution of macular pigments as a function of tion was achieved isocratically with an eluent com- retinal eccentricity, the microscope tube was replaced posed of 92% methanol and 8% water/acetonitrile in the upper aluminum block (Fig. 1) by a set of (3:1 v/v) and a flow rate of 1 ml min~'. spaced, concentric, tubular cutters, designed to slide telescopically relative to one another and to the upper block itself. By bearing down with these onto the Results sphere, the retina could be cut into six annuli, con- Age Study centric with the fovea. Only adult eyes were used for this part of the study. The ranges of linear surface A typical chromatogram, labeled a), representing distance, from the center of the fovea to the inner and the pigments found in the central portion of the ret- outer edges of the annuli, were 0-0.75, 0.75-1.6, ina (0-2.3 mm), is shown in Figure 2. The dominant 1.6-2.5, 2.5-5.8, 5.8-8.7, and 8.7-12.2 mm. A single zeaxanthin peak and the secondary lutein peak were cutter, 0-0.25 mm, was also available, but could not found in the great majority of samples analyzed, with be used in conjunction with this set. For the age the reversed situation occurring only rarely. The four study, a single cutter covering the range 0-2.3 mm minor peaks have not yet been identified; however,

Downloaded from iovs.arvojournals.org on 10/02/2021 No. 6 MACULAR PIGMENT: RETINAL DISTRIBUTION AND AGE STUDY / Done er ol. 845

Fig. 2. Representative chromatograms of pigments extracted from retinal tissue. L = lutein, Z = zeaxanthin, detector wavelength = 450 nm. Chromatogram a), for a 47-year-old donor, was obtained from a central disk (0-2.3 mm), as used in the age study. Chromato- grams b), c), and d), for a 65-year-old donor, illustrate 2.5-5.8mm the dramatic change in the lutein:zeaxanthin ratio with retinal eccentricity. For the 1.6-2.5 mm | Retinal sake of clarity, the internal distribution standard peak (retention study time ~24 min), has been omitted.

0 - 2.3 mm A Age study

10 20 30 Time, minutes

their spectra indicate that they too are carotenoids masses of zeaxanthin and lutein varied considerably and it is conceivable that they are cis-trans isomers of among donors, as illustrated by the frequency distri- lutein or zeaxanthin.13 Their presence raises the ques- bution of Figure 3, neither this nor the composition tion of which pigments constitute the macular pig- of the pigment showed any significant variation with ment. In what follows, we restrict ourselves to the two age after 2 years. This can be seen in Figure 4, where major compounds, with the realization that in so doing, we may be slightly underestimating the total

quantity of what may legitimately be called the macu- 13- lar pigment. The masses of zeaxanthin and lutein in each sample were determined from their peak areas relative to 20-

that of the internal standard, the mass of which was SJOUOP known. The peak areas were suitably weighted by the corresponding extinction coefficients of the compounds at 450 nm. From HPLC analyses of mixtures o — of spectroscopically determined amounts of the three n\o- compounds, the accuracy of this method was found to be better than 96%. 5- For the age study, ten donors were available in each decade of life apart from the second and tenth, for 1 ^ 1 ^ 1 1 ' 10 20 30 40 50 60 70 80 90 which the numbers were eight and six, respectively. Lutein 4 zeaxanthin, ng For approximately 40% of the donors, results were obtained from both eyes and an average was re- Fig. 3. Frequency distribution showing the number of donors having a given mass of lutein and zeaxanthin in the central region corded. For the remainder, reliable data were ob- of the retina (0-2.3 mm). The average value was determined to be tained for one eye only. Although the combined 27 ± SD 11 ng.

Downloaded from iovs.arvojournals.org on 10/02/2021 846 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / June 1988 Vol. 29

size. With the same diameter cutter, tissue samples from smaller eyes would span a greater range of angular eccentricities. By recalculating lutein:zeaxanthin ratios within equal ranges of angular eccentricity, we have concluded that donors below the age of 2 are still significantly different from older donors. Also for donors less than a year old, the total amount of pigment was generally well below the average for older donors, resulting in an imperceptible yellow spot. In the case of prenatal eyes (17 to 22 weeks gestation), both lutein and zeaxanthin were detectable by HPLC (but not by visual inspection), again with the former in greater abundance. The lutein:zeaxanthin ratio for the entire neural retina was 2.51 ± SD 0.94 compared with 1.14 ± SD 0.34 for seven postnatal donors (mean age 61). The average abundance in the prenatal eye, of ca. 3 ng of combined carotenoids, was comparable, on the basis of mass per unit area, with that found in postnatal donors.

20 40 60 Retinal Distribution Age, years Three sample chromatograms, labeled b), c), and Fig. 4. Age distributions showing for each decade the total mass of lutein and zeaxanthin (lower) and the lutein/zeaxanthin ratio d), of the six derived from concentric annuli, are (upper) in the central region of the retina (0-2.3 mm). shown in Figure 2. The important feature to be noted is the change from zeaxanthin being the dominant pigment in the central 0 to 0.75 mm of the retina to lutein taking that role at eccentricities exceeding ap- the average results and associated standard deviations proximately 2.5 mm. The quantitative variation in are plotted for each decade. The only instance in the lutein:zeaxanthin ratio with eccentricity may be which we noticed any age effect was in the 0 to 2 age seen in Table 2, where the average results from seven group, for which the dominant pigment was consis- donors are given. The table includes the total mass of tently lutein. This can be seen in Table 1, the results pigment per unit area, a quantity which decreased obtained from donors in the first decade, where the from the macula to the peripheral retina by a factor of average lutein:zeaxanthin ratio for donors below age nearly 300. Also presented in the first row of Table 2 2 was 1.44 ± SD 0.16 compared with 0.77 ± SD 0.20 are the average results, obtained using the 0-0.25 mm for all other donors. We have taken into account the cutter, for six donors of average age 49. possibility that this might be a result of smaller eye A number of earlier eluting peaks were found in chromatogram d) (Fig. 2), as well as those obtained at greater eccentricities. Their apparent absence at Table 1. Macular pigment data for donors smaller eccentricities (chromatograms a), b), and c)) in the first decade is thought to be due largely to the correspondingly smaller areas of tissue available for pigment extrac- Age Lutein + zeaxanthin* Lutein/zeaxanthin* (years) (ng) (mass ratio) tion. The largest of the earlier peaks, and that immediately preceding it, have been examined spectro- o.ot 7.0 .59 scopically. Their spectra are characterized by a broad 0.4 4.5 .25 0.5 24.6 .41 peak at about 390 nm. 0.8 4.8 .67 1.3 66.2 .75 1.6 41.5 .24 Discussion 1.7 28.6 .49 3.0 25.1 0.67 7.0 41.0 0.53 HPLC has proven its effectiveness in our studies of 9.0 27.6 0.87 the macular pigment, beyond merely aiding in its identification.4 The results of our age study not only * Extracted from 4.7 mm diameter disk of tissue centered on the fovea. t One day old. reinforce psychophysical findings, that macular pig-

Downloaded from iovs.arvojournals.org on 10/02/2021 No. 6 MACULAR PIGMENT: RETINAL DISTRIBUTION AND AGE STUDY / Done er ol. 847

Table 2. Average macular pigment data and photoreceptor data f within different ranges of linear surface distance from the fovea Lutein + zeaxanthin Distance from per unit area Lutein/zeaxanthin Rod/cone\ fovea (mm) (ng mm~2) Optical densityX (mass ratio) (# ratio)

0-0.25 13.3 ±4.3* 0.33 ±0.11* 0.42 ± 0.04* 0.12 ±0.03 0-0.75 5.7 ± 1.5 0.14 ±0.04 0.50 ±0.10 1.3 ±0.1 0.75-1.6 2.9 ± 1.1 0.073 ± 0.028 0.73 ±0.11 5.5 ±0.1 1.6-2.5 0.81 ±0.25 0.020 ± 0.006 1.04 ±0.24 10.4 ±0.2 2.5-5.8 0.143 ± 0.027 0.0036 ± 0.0007 1.85 ±0.57 21.7 ±4.7 5.8-8.7 0.067 ± 0.024 0.0016 ± 0.0006 2.24 ± 0.47 26.6 ±5.1 8.7-12.2 0.047 ±0.018 0.0012 ± 0.0005 2.22 ±0.55 25.2 ±5.6

* Mean data for six donors. All other data in these columns are mean t Calculated from Osterberg's" data for a single 16-year-old donor eye. values for a different set of seven donors. t Calculated from the data in the second column.

mentation shows no significant variation from age 10 the results. For the three references noted above, to 90,9 but have extended the age range to include average peak optical densities were reported to be newborn and fetal eyes. Thus we are able to answer 0.39,9 0.77,10 and 0.53.8 By interpolating between the the outstanding question of Nussbaum et al14 con- data in Table 2, we have estimated the corrections to cerning the presence or absence of the macular pig- these figures to be about +14%, +3%, and +3%, re- ment in the newborn. In all of our studies, we have spectively. Therefore eccentricities of 7° or more never encountered a retina in which zeaxanthin and would appear to be appropriate locations in the retina lutein were not present. This is not to suggest that the where absorption by the macular pigment can be as- pigments were always observable by visual inspec- sumed to be practically zero. In psychophysical stud- tion. In prenatal retinas, a yellow spot was never visi- ies, however, increasing eccentricity generally in- ble. In postnatal cases, the pigments were discernible volves added difficulty to the visual task which is only if their combined masses in the macula exceeded being performed, and a compromise must be sought. about 5 ng. The smallest cutter used in our distribution study In many instances, reliable samples were obtained had a radius of 0.25 mm, corresponding in the adult from both left and right eyes of the donor, thereby eye, to a visual field of approximately 1.8° in diame- permitting a comparison to be made. Of the 36 ter. Within this region, the average optical density donors in this category, the percentage difference in was calculated to be 0.33 ± SD 0.11 (see Table 2). total pigment ranged from 0 to 67% with an average This may be compared with the somewhat higher value of 29 ± SD 19%. In contrast, Snodderly15 re- values obtained in psychophysical studies, in which a ported that among ten squirrel monkeys, pigment smaller visual field (eg, 1 ° in diameter) is often used. density distributions in both eyes were very similar. The results are consistent with the radial density gra- As far as the lutein:zeaxanthin ratios are concerned, dient which has been observed by microspectropho- we found a higher consistency between left and right tometry within this central region.1617 eyes, with differences ranging from 0 to 43% and av- Perhaps the most thought provoking discovery eraging 13 ± SD 10%. emerging from this investigation is the increase in the Chromatography has also proven a valuable tool in lutein:zeaxanthin ratio with eccentricity, a variation, allowing us to quantify the macular pigment as a incidentally, found in preliminary studies to be du- function of retinal eccentricity. In studies involving plicated in nonhuman primates (M. mulatta, M. fas- spectral sensitivity measurements to assess the optical cicularis, C. ethiops). Had these pigments maintained density of the pigment, results are usually calculated a constant ratio throughout the retina (as well as be- by assuming negligible absorption by the pigment at tween both eyes), it might have been assumed that some eccentric location. Eccentricities of 5°, 7°, and this merely reflected their relative abundance in the 8° have been used,8"10 for example, corresponding to diet. It might have been argued further that the simi- linear surface distances of approximately 1.5,2.0 and larity in structure of these isomers rendered them in- 2.3 mm. The validity of this assumption may be ex- distinguishable to the uptake mechanism responsible amined by calculating the variation in the average for their presence in the retinal tissue. However, the optical density of the pigment from the areal densities dramatic change in the ratio with eccentricity and, to given in Table 2. The third column in this table gives a lesser extent, the differences often observed between

Downloaded from iovs.arvojournals.org on 10/02/2021 848 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / June 1988 Vol. 29

ratios (Fig. 5) yielded a linear correlation (r = 0.998) 2.5-, at the >99.9% confidence level, sufficiently high to warrant further investigation. It should be noted, however, that for the first data point, representing 2.0- very nearly the rod-free area, the lutein:zeaxanthin ratio is still finite. This suggests that lutein and zeaxanthin may not be associated exclusively with rods 8 1.5- o and cones respectively. In addition, the slope of the 0) N graph would suggest that the amount of zeaxanthin c associated with each cone may be considerably

0.5- The predominance of zeaxanthin over lutein in the central 0-2.3 mm of the retinas of the majority of donors aged 3 and above is clearly evident from Fig- ure 4 and Table 1. For about 10% of these donors, the 10 15 20 25 30 situation was reversed, as also was the case in postna- Rod:cone tal eyes from donors below the age of 2 and in prena- Fig. 5. Linear relationship between the average lutein:zeaxanthin tal eyes. In light of our speculation above, there exists ratio and the corresponding rodxone ratio within different ranges of linear surface distance from the fovea. The rodxone data is due the possibility that the rodxone ratios for these to Osterberg." The straight line is a least squares fityieldin g a linear groups are higher than in the normal adult. A com- correlation at the >99.9% confidence level. parison of this rodxone ratio between very young and adult eyes is a problem which should be addressed, assuming the correlation with the lutein:zeaxanthin left and right eyes, effectively nullifies these argu- ratio, revealed in the present study, can be shown to ments and suggests, in addition, that each pigment be truly causal. may be associated with a specific cell. From microspectrophotometry studies,1617 we Key words: macular pigment, zeaxanthin, lutein, age, know that a major portion of the macular pigment, at cones, rods least in the macaque, is seen in the Henle fibers, a layer which consists largely of the internal fibers of Acknowledgments photoreceptor cells. This observation finds support in 18 The authors gratefully acknowledge the National Disease studies of Haidinger's polarization brushes. A possi- Research Interchange and the Florida Lions Eye Bank for ble interpretation of the decrease in pigmentation supplying human donor eyes. We also thank the Bowman with eccentricity is that it is simply related to the Gray School of Medicine for supplying monkey eyes corresponding change in dimensions, particularly a through their Tissue Request Program, supported by NIH decrease in length, of these fibers. We have consid- Grant #RR00919. J. Martinez provided valuable technical assistance. ered the possibility that lutein and zeaxanthin may be associated primarily with each of the two photoreceptor types, rods and cones, respectively. Owing to this References variation in receptor geometry with eccentricity, it 1. Reading VM and Weale RA: Macular pigment and chromatic would be inappropriate to explore this possibility by aberration. J Opt Soc Am 64:231, 1974. examining, for example, the extent to which cone 2. Kirshfeld K: Carotenoid pigments: Their possible role in pro- density and the areal density of zeaxanthin are lin- tecting against photooxidation in eyes and photoreceptor cells. early correlated. On the other hand, the geometrical Proc R Soc Lond B216:71, 1982. 3. Malinow MR, Feeney-Burns L, Peterson LH, Klein ML, and factor may be largely eliminated by comparing the Neuringer M: Diet-related macular anomalies in monkeys. In- average luteinrzeaxanthin ratio as a function of ec- vest Ophthalmol Vis Sci 19:857, 1980. centricity with the corresponding rod:cone ratio. This 4. Bone RA, Landrum JT, and Tarsis SL: Preliminary identifica- comparison has been made using 0sterberg's19 rod/ tion of the human macular pigment. Vision Res 25:1531, 1985. cone data (See Table 2) for a 16-year-old donor. Our 5. Farber DB, Flannery JG, Lolley RN, and Bok D: Distribution own macular pigment data were obtained from older patterns of photoreceptors, protein, and cyclic nucleotides in donors whose receptor populations were therefore the human retina. Invest Ophthalmol Vis Sci 26:1558, 1985. probably different.6 In spite of this, a graph of the 6. Gartner S and Henkind P: Aging and degeneration of the

Downloaded from iovs.arvojournals.org on 10/02/2021 No. 6 AAACULAR PIGMENT: RETINAL DISTRIBUTION AND AGE STUDY / Done er ol. 849

human macula: 1. Outer nuclear layer and photoreceptors. Br 14. Nussbaum JJ, Pruett RC, and Delori FC: Historic perspectives. J Ophthalmol 65:23, 1981. Macular yellow pigment: The first 200 years. Retina 1:296, 7. Yuodelis C and Hendrickson A: A qualitative and quantitative 1981. analysis of the human fovea during development. Vision Res 15. Snodderly M: Macular pigment screening density profiles in 26:847, 1986. the retinas of squirrel monkeys (Saimiri sciureus). ARVO Ab- 8. Bone RA and Sparrock JMB: Comparison of macular pigment stracts. Invest Ophthalmol Vis Sci 28(Suppl):263, 1987. densities in human eyes. Vision Res 11:1057, 1971. 16. Snodderly DM, Brown PK, Delori FC, and Auran JD: The 9. Werner JS, Donnelly SK, and Kliegl R: Aging and human macular pigment: I. Absorbance spectra, localization, and dis- macular pigment density. Vision Res 27:257, 1987. crimination from other yellow pigments in primate retinas. 10. Pease PL, Adams AJ, and Nuccio E: Optical density of human Invest Ophthalmol Vis Sci 25:660, 1984. macular pigment. Vision Res 27:705, 1987. 11. Stark WS: Photopic sensitivities to ultraviolet and visible 17. Snodderly DM, Auran JD, and Delori FC: The macular pig- wavelengths and the effects of the macular pigments in human ment: II. Spatial distribution in primate retinas. Invest Oph- aphakic observers. Curr Eye Res 6:631, 1987. thalmol Vis Sci 25:674, 1984. 12. Jensen SL and Hertzberg S: Selective preparation of the lutein 18. Bone RA and Landrum JT: Macular pigment in Henle fiber monomethyl ethers. Acta Chem Scand 20:1703, 1966. membranes: A model for Haidinger's brushes. Vision Res 13. Braumann E and Grimme HL: Reversed-phase high-perfor- 24:103, 1984. mance liquid chromatography of chlorophylls and carot- 19. 0sterberg G: Topography of the layer of rods and cones in the enoids. Biochim Biophys Acta 637:8. 1981. human retina. Acta Ophthalmol 6:1, 1935.

Downloaded from iovs.arvojournals.org on 10/02/2021