- Email: [email protected]
Carboxyfluorescein. A dye for evaluating the corneal endothelial barrier function in vivo
Exp. Eye Res. (1986). 42, 141-150
Carboxyfluorescein. Endothelial
A Dye for Evaluating the Cornea1 Barrier Function In Vivo MAKATO
Division California
ARAIE
of Ophthalwaology, Stanford University School of MedicineF Stanford, 94305, U.S.A. and Deparbment of Ophthalmology: University of Tokyo School of Medicine, Tokyo, 113, Japan (Received 19 June 1985 and accepted 21 October 1985)
The relationship of the cornea-aqueous distribution ratio (rCB) and concentration in vitro was established for fluorescein and carboxyfluorescein. The value of rCBfor fluorescein was found to fall from 3.2OkO.25 (meanfs.~., n = 6) to 1.78 as the concentration of the free fluorescein in the bathing medium rose from 5.8 x lo-* to 5.9 x 10m5g ml-‘. For carboxyfluorescein, it remained unchanged over the same concentration range, and the average for total determinations was 1,29&0.16 (n = 20). The value of T,, for carboxyfluorescein determined in vivo was 1.62 + 0.23 (mean i s.D., n = 6) and the cornea1 endothelial permeability to carboxyfluorescein in normal rabbits was 3.31 kO.66 x 1OV cm min-’ (n = ll), w h’K h was 35 o/0 lower than that for fluorescein. Because of its lower endothelial permeability and a value of r,, which is unchanged over a wide range of concentration, carboxyfluorescein may be better suited for the in vivo evaluation of the barrier function of the cornea1 endothelium than fluorescein. Key words: fluorescein; carboxyfluoreseein; cornea-aqueous distribution ratio; cornea1 endothelial permeability.
1. Introduction The barrier that the endothelium presents to diffusion is one of the important factors for the maintenance of the normal hydration and transparency of the cornea (Mishima, 1982 ; Maurice, 1984). Fluorescein has been widely used to evaluate the in vivo state of the endothelial barrier and its endothelial permeability coefficient (Pa,) has been taken as its measured quantitative index (Anselmi, Bron and Maurice, 1968 ; Waltman and Kaufman, 1970; Mishima and Maurice, 1971; Mishima, 1982, Bourne and Brubaker, 1983; Tsuru, Araie, Matsubara and Tanishima, 1984; Minkowski et al., 1984). In most determinations of P,C for fluorescein, it is necessary to know its distribution ratio: rC&, between the cornea1 stroma and the aqueous (Mishima and Maurice, 1971; Mishima, 1982), and this has been assumed to be constant in the concentration range offluorescein encountered in the experiment. However, this basic assumption does not seem well verified; because there are conflicting accounts of the in vitro relationship ofr,, and the fluorescein concentration; in one case the value of r,, had been reported to show an almost three-fold change for concentrations of fluorescein in the bathing medium between 5 x lo-’ and 1 x lO-5 g ml-l (Mishima and Maurice, 1971), and in another that it remained unchanged over a greater range of concentrations (Nagataki, Brubaker and Dwight, 1983). It is possible, in any case, that fluorescein may not be the most advantageous test substance for this purpose. It can penetrate in appreciable quantities across cell membranes as demonstrated by a histologic study using quantitative fluorescence Please send reprint requests to Makoto Araie, M.D., Department of Ophthalmology. Tokyo School of Medicine, 7-3-1, Hong”, Bunkyo-ku, Tokyo, 113, Japan. 0014-%835/86/020141+10
$03.00/O
University
01986 Academic Press Inc. (London) Limited
of
142
M. ABAIE
microscopy (Grimes, Stone, Laties and Li, 1982) ; this is compatible with its relatively high octanol-buffer partition coefficient (Grimes et al., 1982) and the short half-time of fluorescein leakage from liposomes (Weinstein, Yoshikami, Henkart, Blumenthal and Hagins, 1977). Thus, the endothelial permeability to fluorescein may comprise not only the dye transfer through the paracellular spaces of the endothelial cell layer, but also that across its cell membranes, and a tracer which penetrates the cell membrane with more difficulty may be more sensitive to changes in the barrier function of the cornea1 endothelium. Accordingly, carboxyfluorescein, which has a similar fluorescence cha,racteristics to fluorescein (Weinstein et al., 1977) but is less lipid-soluble and does not penetrate cell membranes so well (Grimes et al., 1982), may be expected to be more suitable. In this report, the relationship of rCs and fluorescein concentration in vitro was re-examined and the studies were extended to earboxyfluoreseein. In addition, the was determined in normal rabbits. The results value of Pa, for carboxyfluorescein confirmed that it may be a better test molecule for the in vivo determination of the harrier function of the cornea1 endothelium. 2. Material and Methods Dyes Fluorescein (332 MW) prepared for injection as a sodium salt at a concentration of 5% was used (Fluorescite@, Alcon Lab. Inc.; Fort Worth, Texas). Reagent grade 5(6)carboxyfluorescein (376 MW) was obtained from Eastman Kodak (Rochester, N.Y.) and solubilized as sodium carboxyfluorescein by adding NaOH. The octanol-water partition coefficient (pH = 7.3) determined according to the method of Grimes et al. (1982) at a concentration of 10 PM were 0.61 for fluorescein and 00007 for carboxyfiuorescein and the pK determined according to the method of Grotte, Mattox and Brubaker (1985) were 6.4 for the former and 6.5 for the latter. The concentrations of the dyes were determined from standard curves prepa.red under the same conditions, using a slit-lamp fluorometer (Tokyo Opt. Co. Ltd., Tokyo, Japan). In vitro determination of r,, for$uo~escein ad carboxyjhorescein New Zealand Albino rabbits of either sex weighing 2.0-3.0 kg were killed immediately before the experiment by intravenous injection of 1.5 ml T61@ euthanasia solution (American Hoechst Co., Somerville, N.J.). The eyes were enucleated, the cornea1 epithelium scraped off and a circular or semicircular stromal disc of about 20 mg in weight was excised. The endothelium was left untouched. In order to prevent swelling of the stroma a bathing medium was prepared by dissolving 12 y0 (w/v) d ex t ran, average MW 71000 (Sigma Chemical Co., St. Louis, MO.) in phosphate-buffered saline at pH 7.4 (PBS) containing 0.125 M sodium chloride and 9033 M phosphate; this will be referred to as PBS-dextran solution. The stromal disc was blotted, weighed, and immersed in 106 ml of PBS-dextran solution containing fluorescein or carboxyfluorescein at, concentration of about 8 x lo-‘, 8 x IO-‘, 8 x iOm6 or 8 x lop5 g ml-’ at room temperature (245 i”C). After 2 hr, the stromal disc was taken out, blotted, weighed again and the dye in it was eluted into at least 100 volumes of PBS at 4°C for 48 hr. The concentration of dye in the fluid was determined fluorometrically and its concentration per unit volume of the stromal tissue was calculated. It was assumed that after 49 hr elution, the concentration in the swollen tissue was equal to that in the PBS, and the specific gravity of the stromal tissue was taken as 1.05 (Felchlin, 1926). In order to obtain an estimate of how quickly the dye is equilibrated with the stroma under these conditions, a 95 hr immersion was also carried out in the soiutions containing 6uorescein at concentrations of 8 x lo-’ and 8 x 1O-5 g ml-l. The free concentration of the dye in the PBS-dextran solution was determined by dialysis across cellulose acetate membrane (13000 MW cut; Spectrum Medical Industries Inc., Los Angeles, California) against PBS containing Auorescein at a concentration of 75% or
CARBOXYFLUORESCEIN
143
carboxyfluorescein at a concentration of 80 ‘$$ of that in the PBS-de&ran solution for 20 hr at room temperature; the figures of 75 and 80 o/0 are the approximate free fractions for the dyes in the PBS-dextran solution, estimated by preliminary experiments. Other experiments confirmed that this time was long enough for free dye to equally distribute in the both compartments and that the penetration of dextran through the membrane used was negligible. At the end of the 20 hr, the dye concentrations in the both compartments were measured fluorometrically by comparison with standard curves obtained in PBS or PBSdextran solution. Four dialysis experiments were carried out at each concentration level of the dye. Finally, to estimate the amount of the dextran that diffused into the stromal tissue during immersion, a solution containing 1 0/0FITC-dextran, average MW 66000, (FD-70Sa, Sigma Chem. Co., St. Louis, MO.) and 11 y0 unlabeled dextran was prepared in PBS. Stromal discs were immersed in the solution under the conditions described above. After 05 or 2 hr, they were well blotted, weighed and eluted in 10.0 ml PBS for 4 days at 4°C. The amount of dextran that diffused into the stromal tissue during the immersion was estimated from the amount of FITGdextran that was recovered, on the assumption that dextran and FITGdextran diffuse into the tissue at the same rate. In vivo determination
of r,, for carboxyjuorescein
Fifteen microliters of 4 y0 carboxyfluorescein solution at physiological pH was injected into the posterior lens by means of a 30 gauge needle passing through the pars plana on 13 eyes. The animals were pigmented rabbits of either sex, weighing 2.1-2.9 kg anesthetized with 50 mg kg -l intramuscular ketamine. After 3 days, the apparent concentration of the dye in the central part of the stroma (F,) and in the anterior chamber (pa&)were measured with the slit lamp fluorometer; the actual size of the area measured being a vertical rectangle of @29 x @15 mm. In six of the eyes, F, and F, fell very slowly and remaining a,lmost unchanged over a period of 6 hr, they were used to determine ~~a.The animal was killed with T61@ solution, the eyes were enucleated, and the cornea1 epithelium scraped off immediately. Approximately 100 ~1 of the aqueous was withdrawn from the middle of the anterior chamber t’hrough a 28 gauge needle and the central part of the stroma, 8 mm in diameter, was trephined. The aqueous sample was weighed and diluted five times by weight in PBS, and the dye concentration in the aqueous (C,) was determined from the dilution factor and the fluorometric reading of the sample. The cornea1 disc was blotted and weighed and the dye was eluted in at least 100 volumes of PBS. The mean dye concentration in the stroma (C,) was calculated as previously described and the ratio of C, to C, was taken as the value of ~~a.
Endothelial permeability coeflicient for carboleyjIuorescein In 12 eyes of six pigmented rabbits, 15 ~1 of 0.6 y0 carboxyfluorescein solution of physiological pH and osmolarity was injected into the center of the vitreous through a 30 gauge needle under ketamine anesthesia. The penetration was made through the superior rectus muscle to avoid reflux. Measurements of Fs and Fa were carried out hourly from 2&25 hr after the injection, and the thickness of the cornea was measured at 24 half-hours using an optical pachometer (Maurice and Giardini, 1951). In order to calculate P,,, it is necessary to convert the apparent concentrations, Fs and F,, obtained uder the present conditions of measurement to the true ones: C, and C,. Relationships between F, and C, and F, and C, were determined as follows. Immediately after the last measurements, approximately 100 yl of the aqueous was withdrawn under topical anesthesia and Ca was determined as described above and compared with Fa. As F, and F, were changing rather rapidly in this case; the relationship between C, and Fs was studied in another 10 eyes of five pigmented rabbits to reduce the time lag between the in vivo measurements and the excision of the cornea. The dye solution was injected in a same manner and F, was measured at 24 hr. Immediately after, the animal was killed, the eyes enucleated, the cornea1 epithelium scraped off, the central stroma excised and Cs was determined as described above and compared with F, obtained just before death.
Dejkition
oj the symbols used
Definition Aqueous-cornea transfer coefficient Cornea-aqueous distribution ratio Endotbelial permeability coefficient Apparent concentration in the stroma (aqueous) calculated directly from the fluorometric reading Mean concentration in the stroma (aqueous)
The value of Pa, was calculated as follows. The dye exchange between the cornea1 stroma and the aqueous wa,s assumed to follow the equation (Mishima, 1981, 1982); de,_ dt
(1)
k -
‘.a’
where k,,,, is the aqueous-cornea transfer coefficient. When C, and C, change in parallel after a long time, equation 1 can be rewritten;
din
k
dt
When the term In C, changes at a constant rate of A> the value of k,,,, is given by; k
(3)
of 4 is ealcuiated from the slope of the regression line fitted to the Fs data plotted on a semi-logarithmic paper, assuming that, FSis proportional to c!,. C, and C, are calculated from B!! and Fa, The value of Pa, is given by (Mishima, 1982) ;
The value
(4) Pa, = kc.,, x (stromal thickness). ., where the stromal thickness is assumed to represent 90% of the total cornea1 thickness (Maurice, 1984). The definition of the symbols used are summarized in Table I. 3. Results In vitro determination
of Y,,
for $uorescei,n
and curboxy$uorescein
The weight of the stromal disc showed no significant change after immersion in the PE+dextran solution in any of the experiments. After 05hr soaking in the solution containing FITC-dextran, its mean concentration in the stroma was 18f3% (mean-&S.D., n = 4) of that in the bathing solution and 2 hr soaking it was 32 + 5 ‘$,. This indicates that the dextran was entering by ordinary diffusion (Carslaw and Jaeger, 1947) and that a considerable amount of unlabeled dextran had penetrated into the stroma in these experiments. If any dye is bound to the penetrating dextran molecules, it should be excluded from the calculation of rca. Then, vca is given by the ratio of the concentration of dye in the stroma not. bound to dextran, C,, to the concentration of free dye in the bathing solution, Cr, and is determined by;
cs> c;-o.32c1,
rca = G
cf
CARBOXYFLUORESCEIK
P
FluoresceIn
concentration
(g ml--‘)
1. The value of r,, and the corresponding concentration of free fluorescein in the bathing medium. Each point and bar indicates mean and S.D.in six corneas. For explanation of r,.., see Table I. FIG.
at 2 hr where CL is the total concentration of the dye in the stroma calculated directly from the dye concentration in the eluate and C, is the concentration of dextran-bound dye in the bathing solution. For bot,h dyes, the free fraction in the PBS-dextran solution, C,/(C,+ Cf), was the same at all the concentration levels tested, and averaged 074 +@Ol (mean I s.o., n = 16) for fluorescein and 0.79+0.03 for carboxyfluorescein, so that C,/C, was 0.35 and 0.26, respectively. The ratio, C’,/C,, after 2-hr immersion was taken to represent TV&,because as will be discussed later, equilibrium between the dye and the stroma is expected to be almost, attained at that time. The value of rca for fluorescein fell from 3.20 f 0.25 (mean f s.D., 12= 6) to 1%7-J0*06 as C, rose from 5% x lo-* to 5.9 x 1W5 g ml-l (Fig. 1). The value obtained at 5.7 x 10W6g ml-r is significantly lower than that obtained at 58 x 10m8 g ml-l (unpaired t-test, P < 0.01) and that obtained at’ 5.9 x 10-j g ml-r is significantly lower than any of those obtained at lower concentrations (P < 0.001). For carboxyfluorescein, the value changed from 1.31 f0.24 (mean+s.n.; n = 5) to 123+@05 over the same concentration range and the differences are not significant at all the concentration levels (P > 605) (Fig. 2). If the binding of the dye by t,he dextran in the stroma is less than that determined for the free solution, the values of rca could be up to about 3-5% greater for fluorescein and about 6% greater for carboxyiluorescein. After 05hr immersion, C,/C, for fiuorescein was 2.55 4 0.17 (mean f s.n., n = 6) at 5.8 x lo-’ g ml-l and 1.78 +@15 at 5.9 x 10V5 g ml-l; these averaged 89 y. of the 2 hr values. I77 viva determination
of rcft for
carboxy$uorescein
Three days after the injection of carboxyfluorescein into the lens, the cortex was found to be strongly but not evenly stained and was presumably controlling the supply of carboxyfluorescein to the anterior segment of the eye as described by Kaiser and
M. SRATE
146
Carboxyfluorescein
concentration
(g ml-‘)
FIG. 2. The values of T,, and the corresponding
concentration of free carboxyfluorescein in the bathing medium. Each noint and bar indicates mean and S.D. in five corneas. For explanation of rCs, see Table I.
1
I 0
I I
I
I
I
I
2
3
4
5
Time (hr) FIG. 3. The time coume of apparent carboxyfluorescein
concentration in the stroma (4; 0-O) and in the anterior chamber (Pa; @---a) after intravitreal injection of the dye; time 0 is 20 hr after the injection. The values obtained at time l-5 hr are normalized to that obtained at time 0. Each point and bar indicates mean and S.D. in 11 eyes.
Maurice (1964) for fluorescein. Those six eyes gave a very slow time change of E”, and F, after 3 days and the rate of decline during the experiment calculated from the slope of the regression line fitted to the data plotted on a semilogarithmic paper was -0~006f0014 hr-l (meanfsn., N = 6) for F, and -0*013+0.013 for F,. There is no significant difference between them (paired t-test, P > 0.05) and neither figure is significantly different from zero (P > 0.05). The value of r,, averaged 1.62 to.23 in a C, range between 1.2 x 10e7 and 2.8 x IO-” g ml-‘.
CARBOXYFLUORESCEIN
147
TABLE II
Summary
of endothelial
permeability
coeficient P ( x 1OV G min-L)
Cornea1 thickness (mm)
Rabbit
1
038
064
2
0.38 0.37 0.36 0.36 038 038 0.37 @37 0.39 039
040 070 054 0.54 0.43 @44 061 oY50 0.58 067
3.66 455 386 2.94 2.93 2.43 2.50 3.39 2.78 3-39 3.92
0.38 0.01
059 @ii
331 0.66
3 4 5 6 Mean S.D.
For explanations
Endothelial
of calculation
of k,,,, and Pa,, see Table I.
permeability
coeficient for carboxyjluoresceilb
In 11 out of 12 eyes, F, and Fa declined almost in parallel during the experiment and the rate of decline was 0142&O-037 h-l (mean+S.D., n = 11) for F, and @144+0.034 hr-l for F, and the ratio, FJF,, was 1.66kO.16. The thickness of the corneas was 0.38+0.01 mm. The time course of the average values of F, and F, is shown in Fig. 3, The ratio of CZ to F& was 1.05+0~08 (n = 12), which is not significantly different from unity (P > 0.05), and the correlation coefficient between Ca and F, was 0.999. No significant correlation was found between the value of C,/F, and Fa (P > 0.1). The ratio of C, to FS was 1.63f022 (n = 10): which is significantly different from unity (P < O.OOl), and the correlation coefficient between C, and FSwas 0.995. No significant correlation was found between the value of C,/F, and F, (P > 0.4). Accordingly, the relationships, C, = 1*6F, and C, = l.OF,, were assumed to hold in a concentration range encountered in the experiment. Inserting these values and rcB, 1.62, in equation (3) gave a value of 0.59 kO.11 for /Gc.ac (meanfS.D., n = 11) and a value of 3.31+0.66x lop4 cm min-l for P,,. The results of the calculation are summarized in Table II. 4. Discussion Recently, the in vivo value of rca for fluorescein in the rabbit eye was determined to be 4.0 in a C, range between 10e8 and lo-’ g ml-l (Araie and Maurice, 1985b), which is higher than the in vitro value obtained here, about 3.2, for a similar range dye concentration in the bathing medium. This is also found for carboxyfluorescein. Part of the difference might be overestimation of dextran binding in the stroma. The present method of determination of the in vivo rca for carboxyfluorescein requires that the concentration in the stroma during the experiment remains unchanged. However, its actual time course showed a small rate of change averaging -0.006 hr-1,
148
NI. ARAIE
which could result in an overestimate of rcB. This overestimate would be only of the order of a few percent from equation (3), since the value of k,.,, is the order 0fO.6 hr-‘. Stromal hydration is known to be dependent on pH and salt concentration in its bathing medium (lMaurice, 1984). Therefore, it might be possible that the dye interaction with the stromal matrix, which is reflected in the value of rcL, is also dependent on these factors. Two-hour immersion in the PBS-dextran solution, which contains only sodium hydrochloride and unphysiologically high phosphate, would cause a considerable change in the ionic environment in the stroma, because the diffusion of small ions should be much more rapid than those of the dyes used, and this may be responsible for some part of discrepancy in the values of rca obtained in vitro and in vivo. A 2hr soaking seems enough to reach equilibrium. If the endothelium is impermeable to the dye, its concentration at the endothelial surface of the stroma should reach within about 1-2 o/0of that at the stroma-bathing medium boundary when Dt/qz = 2, where D is the diffusion constant of the dye across the stroma, t is time and q is the thickness of the stroma (Carslaw and Jaeger, 1947). As q is the order of O-36 mm (Maurice, 1984), D must be 3-6 x iOe7 cm2 see-’ or more for this condition to be satisfied. A separate experiment demonstrated that D for fluorescein is the order of 6 x lo-’ cm2 set-l and that for carboxyfluorescein is rather greater (Araie and Maurice, 1985a). In accordance with this, the value of C,/C, obtained for fluorescein after 0.5hr soaking averaged 89 o/o of that obtained after 2 hr soaking. According to Carslaw and Jaeger (1947), if C,,lC, at 05 hris about 90 y0 of that at 2 hr, then the value at 2 hr should be very close to the value at equilibrium. Actually, the endothelium would be somewhat damaged in these experiments, so that the dye would also penetrate across this surface, speeding the approach to equilibrium. Two reports in the literature give figures for r,,, for fluorescein over a range of concentration. Mishima and Maurice (1971) using an in vitro technique similar to that described here, found a similar decline in the value of s,,; about 5 at 5 x 10e7 g ml-’ falling to 1.6 at 1 x 10m5 g ml-l. Nagataki et al. (1983) reported that it was 1.33 and showed no change in a concentration range between 1 x 10F7 and 1 x 10e5 g ml-r. These two results and the present one show considerable discrepancies in the value of rca at lower concentration of fluorescein and possible explanations a!re that (i) the bathing medium used is different; Mishima and Maurice (1971) used unbuffered physiological saline containing 4-5 y0 dextran, and Nagataki et al. (1983) used phosphate buffer at pH 7.2. As discussed above, this difference might influence the dye-stromal matrix interaction : (ii) Nagataki et al. (1983) used the apparent fluorescein concentration in the stroma calculated directly from the fluorometric readings (F’) in computing the value of rca. However, for fluorescein, C, might not be equal to F, (Araie and Maurice, 1985). In experiments using the topical application of dye to determine Pa,, the value of vca must be constant in a concentration range employed. Based on the present results with fluorescein, it seems that rcil is almost constant when the stromal concentration is 10W6g ml-1 or less, but not when it is higher. Thus, a high initial stromal concentration of fluorescein must be avoided in the in vivo determination of Pa,. On the other hand, the value of r,, determined in vitro for carboxyfluorescein, unlike that for Buorescein, showed no significant change over the concentration range tested. To calculate Pat, the ratio of the mean concentration in the stroma, C,, to the apparent measured one, Fs, under the present conditions of measurement, was determined and the obtained value, 1.6, was larger than unity. One of the reasons for this discrepancy
CARBOXYFLCOR:%SCEIN
149
is possibly the quenching of the fluorescence of carboxyfluorescein by the stromal tissue. An underestimate of Fs due to some artifacts during the measurement might also be responsible for a part of it. However, these factors should not affect the calculation of P,,, since F, was converted to C, from the results of the normalization experiment carried out under identical conditions to the main P,, experiment. The value of P,, for carboxyfluorescein obtained in the normal rabbit eye was 3.3 x lo-” cm mm-l, and 35 o/0 smaller than that for fluorescein, 5.1 x 10d4 cm min-l (Araie and Maurice, 198513). The molecular weight of carboxyfluorescein, 376, is 13 y’ larger than that of fluorescein, 332. However, it seems unlikely that a 13 o/0 increase in the molecular weight would cause 35% decrease in the endothelial permeability, if penetration was through water-filled pores. Pa, for mannitol (182 MW) is reported to be 5.5 x IO-* cm min-l, being rather close to the value for fluorescein (Kim, Green, Mart,inez and Paton, 1971; Maurice, 1984), which suggests that fluorescein is moving more rapidly than is appropriate for its molecular weight. Active transport of fluoreseein by the cornea1 endothelium has not been reported and its relatively higher value of Pa, may be explained by assuming that it can penetrate the cell membrane in its passage across the endothelial cell layer, as opposed to carboxyfluorescein of which passage is restricted to the paracellular space. This interpretation is compatible with the histologic findings obtained in the ciliary and iris epithelium under fluorescence microscope (Grimes et al., 1982) and its relatively high lipid solubility (Weinst,ein et al., 1977). On the other hand, the value of P,, for carboxyfluorescein determined here agrees with that for sucrose, 3.3-3.9 x lop4 cm min-I : (Mishima and Trenberth, 1968; Kim et al., 1971; Riley, 1977; Maurice, 1984), and their molecular weights are similar, which suggests that these substances share the same route in the passage across the endothelium. Thus, carboxyfluorescein appears to have two advantages over fluorescein as a tracer dye for the in vivo evaluation of the barrier function of the endothelium: (i) the value of r,, does not change over a wide range of concentration. (ii) It does not penetrat.e the endothelial cell layer across the cell membrane and its endothelial permeability may be expected to reflect the state of the intercellular junctional complexes more sensitively. Recently: it has been shown that carboxyfluorescein is less readily converted to a fluorescent metabolite than fluorescein after systemic administration (McLaren and Brubaker, 1985). When P,, is determined by the systemic administration of a tracer, this would also make its use advantageous.
ACKNOWLEDGMENTS The author is grateful to Prof. David M. Maurice for his advice during the course of this study. This work was supported by IVIH Grant EY 04863.
REFERENCES Araie,
M. and Maurice, D. MM. (1985a). Transcorneal transfer of fluorophores with different lipid solutility. Invest. O@thalmoZ. Via. Sci. (Suppl.) 26, :!. Araie. M. and Maurice, D. M. (1985b). A re-evaluation of cornea1 endothelial permeability to Auorescein. Exp. Eye Res. 41, 383-390 Bnselmi. P., Bron, A. J and Maurice, D. M. (1968). Action of drugs on the aqueous flow in man measured by fluorophotometry. Exp. Eye Res. 7: 48’7-96. Bourne, W. M. and Brubaker, R. F. (1983). Decreasedendothelial permeability in transplanted corneas. Am. J. Ophthalmol. 96, 362-7.
150
M. ARAIE
Carslaw, H. S. and Jaeger, J. G. (1947). Conduction of Heat in Solids. Pp. 82-5. Clarendon Press : Oxford. Felchlin, M. (1926). Versuche zur Ermittlung des spezifischen Gewichts der verschiedenen Augenmedien mittels einer neuen Methode. Arch. Ophthalmol. 117, 325-42. Grimes, P. A., Stone, R. A., Laties, A. M. and Li, W. (1982). Carboxyfluorescein. A probe of the blood-ocular barriers with lower membrane permeability than fluorescein. Arch. Ophthalmol. 100, 6359. Grotte, D., Mattox, V. and Brubaker, R. F. (1985). Fluorescent, physiological and pharmacokinetic properties of fluorescein glucuronide. Exp. Eye Res. 40, 23-33. Kaiser; R. J. and Maurice, D. M. (1964). The diffusion of fluorescein in the lens. Exp. Eye Res. 3, 15665. Kim, J. H., Green, K., Martinez, M. and Paton, D. (1971). Solute permeability ofthe cornea] endothelium and Descemet’s membrane. Exp. Eye Res. 12, 231-8. Maurice, D. M. (1984). The cornea and sclera. In The Eye Vol. lb, 3rd edn. (Ed, Davson, H.). Pp. i-158. Academic Press: London. Maurice, D. M. and Giardini, A. A. (1951). A simple optical apparatus for measuring the cornea1 thickness and the average thickness of the human cornea. Br. J. Ophthalmol. 35; 16977. McLaren, J. F. and Brubaker, R. F. (1985). Lack of a spectral shift in fluorescence following systemic administration of carboxyfluorescein. Invest. Ophthalmol. i/i:s. Sci. (Suppl.) 26, 192. Minkowsi, J. S., Bartels, S. P., Delori, F. C., Lee, S. R., Kenyon, K. R. and Neufeld, A. H. (1984). Cornea1 endothelial function and structure following cryo-injury in the rabbit. Invest. Ophthalmol. Vi’is. Sci. 25, 1416-25. Xshima, S. (1981). Clinical pharmaeokinetics of the eye. Invest. Ophthalmol. Vis. Sci. 21, 504-41. Mishima; S. (1982). Clinical investigations on the cornea1 endothelium. Am. J. OphthaZmol. 93, l-29. Mishima, S. and Maurice, D. M. (1971). In vivo determination of the endothelial permeability to fluorescein. Acta Sot. Ophthalmol. Jpn 75, 23G-43. Mishima, S. and Trenberth, S. M. (1968). Permeability of the cornea1 endothelium to nonelectrolytes. Iwest. Ophthalmol. 7, 34-43. Nagataki, S., Brubaker, R. F. and Dwight, A. (1983). The diffusion of fluorescein in the stroma of rabbit cornea. Exp. Eye Res. 36, 765-71. Riley, M. V. (1977). A study of the transfer of amino acids across the endothelium of the rabbit cornea. Exp. Eye Res. 24, 35-44. Tsuru, T., Araie, M., Matsubara, M. and Tanishima, T. (1984). Endothelial wound-healing of monkey cornea : fluorophotometric and specular microscopic studies. Jpn J. Ophthalmol. 28, 105-25. Waltman, S. and Kaufman, H. E. (1970). In vivo studies of human cornea1 endothelial permeability. Am. J. Ophthalmol. 70, 45-7. Weinstein, J. N., Yoshikami, S., Henkart, P., Blumenthal, R. and Hagins, W. A. (1977). Liposomeeell interaction : transfer and intracellular release of a trapped fluorescent marker. Science 195, 489-92.
Carboxyfluorescein. Endothelial
A Dye for Evaluating the Cornea1 Barrier Function In Vivo MAKATO
Division California
ARAIE
of Ophthalwaology, Stanford University School of MedicineF Stanford, 94305, U.S.A. and Deparbment of Ophthalmology: University of Tokyo School of Medicine, Tokyo, 113, Japan (Received 19 June 1985 and accepted 21 October 1985)
The relationship of the cornea-aqueous distribution ratio (rCB) and concentration in vitro was established for fluorescein and carboxyfluorescein. The value of rCBfor fluorescein was found to fall from 3.2OkO.25 (meanfs.~., n = 6) to 1.78 as the concentration of the free fluorescein in the bathing medium rose from 5.8 x lo-* to 5.9 x 10m5g ml-‘. For carboxyfluorescein, it remained unchanged over the same concentration range, and the average for total determinations was 1,29&0.16 (n = 20). The value of T,, for carboxyfluorescein determined in vivo was 1.62 + 0.23 (mean i s.D., n = 6) and the cornea1 endothelial permeability to carboxyfluorescein in normal rabbits was 3.31 kO.66 x 1OV cm min-’ (n = ll), w h’K h was 35 o/0 lower than that for fluorescein. Because of its lower endothelial permeability and a value of r,, which is unchanged over a wide range of concentration, carboxyfluorescein may be better suited for the in vivo evaluation of the barrier function of the cornea1 endothelium than fluorescein. Key words: fluorescein; carboxyfluoreseein; cornea-aqueous distribution ratio; cornea1 endothelial permeability.
1. Introduction The barrier that the endothelium presents to diffusion is one of the important factors for the maintenance of the normal hydration and transparency of the cornea (Mishima, 1982 ; Maurice, 1984). Fluorescein has been widely used to evaluate the in vivo state of the endothelial barrier and its endothelial permeability coefficient (Pa,) has been taken as its measured quantitative index (Anselmi, Bron and Maurice, 1968 ; Waltman and Kaufman, 1970; Mishima and Maurice, 1971; Mishima, 1982, Bourne and Brubaker, 1983; Tsuru, Araie, Matsubara and Tanishima, 1984; Minkowski et al., 1984). In most determinations of P,C for fluorescein, it is necessary to know its distribution ratio: rC&, between the cornea1 stroma and the aqueous (Mishima and Maurice, 1971; Mishima, 1982), and this has been assumed to be constant in the concentration range offluorescein encountered in the experiment. However, this basic assumption does not seem well verified; because there are conflicting accounts of the in vitro relationship ofr,, and the fluorescein concentration; in one case the value of r,, had been reported to show an almost three-fold change for concentrations of fluorescein in the bathing medium between 5 x lo-’ and 1 x lO-5 g ml-l (Mishima and Maurice, 1971), and in another that it remained unchanged over a greater range of concentrations (Nagataki, Brubaker and Dwight, 1983). It is possible, in any case, that fluorescein may not be the most advantageous test substance for this purpose. It can penetrate in appreciable quantities across cell membranes as demonstrated by a histologic study using quantitative fluorescence Please send reprint requests to Makoto Araie, M.D., Department of Ophthalmology. Tokyo School of Medicine, 7-3-1, Hong”, Bunkyo-ku, Tokyo, 113, Japan. 0014-%835/86/020141+10
$03.00/O
University
01986 Academic Press Inc. (London) Limited
of
142
M. ABAIE
microscopy (Grimes, Stone, Laties and Li, 1982) ; this is compatible with its relatively high octanol-buffer partition coefficient (Grimes et al., 1982) and the short half-time of fluorescein leakage from liposomes (Weinstein, Yoshikami, Henkart, Blumenthal and Hagins, 1977). Thus, the endothelial permeability to fluorescein may comprise not only the dye transfer through the paracellular spaces of the endothelial cell layer, but also that across its cell membranes, and a tracer which penetrates the cell membrane with more difficulty may be more sensitive to changes in the barrier function of the cornea1 endothelium. Accordingly, carboxyfluorescein, which has a similar fluorescence cha,racteristics to fluorescein (Weinstein et al., 1977) but is less lipid-soluble and does not penetrate cell membranes so well (Grimes et al., 1982), may be expected to be more suitable. In this report, the relationship of rCs and fluorescein concentration in vitro was re-examined and the studies were extended to earboxyfluoreseein. In addition, the was determined in normal rabbits. The results value of Pa, for carboxyfluorescein confirmed that it may be a better test molecule for the in vivo determination of the harrier function of the cornea1 endothelium. 2. Material and Methods Dyes Fluorescein (332 MW) prepared for injection as a sodium salt at a concentration of 5% was used (Fluorescite@, Alcon Lab. Inc.; Fort Worth, Texas). Reagent grade 5(6)carboxyfluorescein (376 MW) was obtained from Eastman Kodak (Rochester, N.Y.) and solubilized as sodium carboxyfluorescein by adding NaOH. The octanol-water partition coefficient (pH = 7.3) determined according to the method of Grimes et al. (1982) at a concentration of 10 PM were 0.61 for fluorescein and 00007 for carboxyfiuorescein and the pK determined according to the method of Grotte, Mattox and Brubaker (1985) were 6.4 for the former and 6.5 for the latter. The concentrations of the dyes were determined from standard curves prepa.red under the same conditions, using a slit-lamp fluorometer (Tokyo Opt. Co. Ltd., Tokyo, Japan). In vitro determination of r,, for$uo~escein ad carboxyjhorescein New Zealand Albino rabbits of either sex weighing 2.0-3.0 kg were killed immediately before the experiment by intravenous injection of 1.5 ml T61@ euthanasia solution (American Hoechst Co., Somerville, N.J.). The eyes were enucleated, the cornea1 epithelium scraped off and a circular or semicircular stromal disc of about 20 mg in weight was excised. The endothelium was left untouched. In order to prevent swelling of the stroma a bathing medium was prepared by dissolving 12 y0 (w/v) d ex t ran, average MW 71000 (Sigma Chemical Co., St. Louis, MO.) in phosphate-buffered saline at pH 7.4 (PBS) containing 0.125 M sodium chloride and 9033 M phosphate; this will be referred to as PBS-dextran solution. The stromal disc was blotted, weighed, and immersed in 106 ml of PBS-dextran solution containing fluorescein or carboxyfluorescein at, concentration of about 8 x lo-‘, 8 x IO-‘, 8 x iOm6 or 8 x lop5 g ml-’ at room temperature (245 i”C). After 2 hr, the stromal disc was taken out, blotted, weighed again and the dye in it was eluted into at least 100 volumes of PBS at 4°C for 48 hr. The concentration of dye in the fluid was determined fluorometrically and its concentration per unit volume of the stromal tissue was calculated. It was assumed that after 49 hr elution, the concentration in the swollen tissue was equal to that in the PBS, and the specific gravity of the stromal tissue was taken as 1.05 (Felchlin, 1926). In order to obtain an estimate of how quickly the dye is equilibrated with the stroma under these conditions, a 95 hr immersion was also carried out in the soiutions containing 6uorescein at concentrations of 8 x lo-’ and 8 x 1O-5 g ml-l. The free concentration of the dye in the PBS-dextran solution was determined by dialysis across cellulose acetate membrane (13000 MW cut; Spectrum Medical Industries Inc., Los Angeles, California) against PBS containing Auorescein at a concentration of 75% or
CARBOXYFLUORESCEIN
143
carboxyfluorescein at a concentration of 80 ‘$$ of that in the PBS-de&ran solution for 20 hr at room temperature; the figures of 75 and 80 o/0 are the approximate free fractions for the dyes in the PBS-dextran solution, estimated by preliminary experiments. Other experiments confirmed that this time was long enough for free dye to equally distribute in the both compartments and that the penetration of dextran through the membrane used was negligible. At the end of the 20 hr, the dye concentrations in the both compartments were measured fluorometrically by comparison with standard curves obtained in PBS or PBSdextran solution. Four dialysis experiments were carried out at each concentration level of the dye. Finally, to estimate the amount of the dextran that diffused into the stromal tissue during immersion, a solution containing 1 0/0FITC-dextran, average MW 66000, (FD-70Sa, Sigma Chem. Co., St. Louis, MO.) and 11 y0 unlabeled dextran was prepared in PBS. Stromal discs were immersed in the solution under the conditions described above. After 05 or 2 hr, they were well blotted, weighed and eluted in 10.0 ml PBS for 4 days at 4°C. The amount of dextran that diffused into the stromal tissue during the immersion was estimated from the amount of FITGdextran that was recovered, on the assumption that dextran and FITGdextran diffuse into the tissue at the same rate. In vivo determination
of r,, for carboxyjuorescein
Fifteen microliters of 4 y0 carboxyfluorescein solution at physiological pH was injected into the posterior lens by means of a 30 gauge needle passing through the pars plana on 13 eyes. The animals were pigmented rabbits of either sex, weighing 2.1-2.9 kg anesthetized with 50 mg kg -l intramuscular ketamine. After 3 days, the apparent concentration of the dye in the central part of the stroma (F,) and in the anterior chamber (pa&)were measured with the slit lamp fluorometer; the actual size of the area measured being a vertical rectangle of @29 x @15 mm. In six of the eyes, F, and F, fell very slowly and remaining a,lmost unchanged over a period of 6 hr, they were used to determine ~~a.The animal was killed with T61@ solution, the eyes were enucleated, and the cornea1 epithelium scraped off immediately. Approximately 100 ~1 of the aqueous was withdrawn from the middle of the anterior chamber t’hrough a 28 gauge needle and the central part of the stroma, 8 mm in diameter, was trephined. The aqueous sample was weighed and diluted five times by weight in PBS, and the dye concentration in the aqueous (C,) was determined from the dilution factor and the fluorometric reading of the sample. The cornea1 disc was blotted and weighed and the dye was eluted in at least 100 volumes of PBS. The mean dye concentration in the stroma (C,) was calculated as previously described and the ratio of C, to C, was taken as the value of ~~a.
Endothelial permeability coeflicient for carboleyjIuorescein In 12 eyes of six pigmented rabbits, 15 ~1 of 0.6 y0 carboxyfluorescein solution of physiological pH and osmolarity was injected into the center of the vitreous through a 30 gauge needle under ketamine anesthesia. The penetration was made through the superior rectus muscle to avoid reflux. Measurements of Fs and Fa were carried out hourly from 2&25 hr after the injection, and the thickness of the cornea was measured at 24 half-hours using an optical pachometer (Maurice and Giardini, 1951). In order to calculate P,,, it is necessary to convert the apparent concentrations, Fs and F,, obtained uder the present conditions of measurement to the true ones: C, and C,. Relationships between F, and C, and F, and C, were determined as follows. Immediately after the last measurements, approximately 100 yl of the aqueous was withdrawn under topical anesthesia and Ca was determined as described above and compared with Fa. As F, and F, were changing rather rapidly in this case; the relationship between C, and Fs was studied in another 10 eyes of five pigmented rabbits to reduce the time lag between the in vivo measurements and the excision of the cornea. The dye solution was injected in a same manner and F, was measured at 24 hr. Immediately after, the animal was killed, the eyes enucleated, the cornea1 epithelium scraped off, the central stroma excised and Cs was determined as described above and compared with F, obtained just before death.
Dejkition
oj the symbols used
Definition Aqueous-cornea transfer coefficient Cornea-aqueous distribution ratio Endotbelial permeability coefficient Apparent concentration in the stroma (aqueous) calculated directly from the fluorometric reading Mean concentration in the stroma (aqueous)
The value of Pa, was calculated as follows. The dye exchange between the cornea1 stroma and the aqueous wa,s assumed to follow the equation (Mishima, 1981, 1982); de,_ dt
(1)
k -
‘.a’
where k,,,, is the aqueous-cornea transfer coefficient. When C, and C, change in parallel after a long time, equation 1 can be rewritten;
din
k
dt
When the term In C, changes at a constant rate of A> the value of k,,,, is given by; k
(3)
of 4 is ealcuiated from the slope of the regression line fitted to the Fs data plotted on a semi-logarithmic paper, assuming that, FSis proportional to c!,. C, and C, are calculated from B!! and Fa, The value of Pa, is given by (Mishima, 1982) ;
The value
(4) Pa, = kc.,, x (stromal thickness). ., where the stromal thickness is assumed to represent 90% of the total cornea1 thickness (Maurice, 1984). The definition of the symbols used are summarized in Table I. 3. Results In vitro determination
of Y,,
for $uorescei,n
and curboxy$uorescein
The weight of the stromal disc showed no significant change after immersion in the PE+dextran solution in any of the experiments. After 05hr soaking in the solution containing FITC-dextran, its mean concentration in the stroma was 18f3% (mean-&S.D., n = 4) of that in the bathing solution and 2 hr soaking it was 32 + 5 ‘$,. This indicates that the dextran was entering by ordinary diffusion (Carslaw and Jaeger, 1947) and that a considerable amount of unlabeled dextran had penetrated into the stroma in these experiments. If any dye is bound to the penetrating dextran molecules, it should be excluded from the calculation of rca. Then, vca is given by the ratio of the concentration of dye in the stroma not. bound to dextran, C,, to the concentration of free dye in the bathing solution, Cr, and is determined by;
cs> c;-o.32c1,
rca = G
cf
CARBOXYFLUORESCEIK
P
FluoresceIn
concentration
(g ml--‘)
1. The value of r,, and the corresponding concentration of free fluorescein in the bathing medium. Each point and bar indicates mean and S.D.in six corneas. For explanation of r,.., see Table I. FIG.
at 2 hr where CL is the total concentration of the dye in the stroma calculated directly from the dye concentration in the eluate and C, is the concentration of dextran-bound dye in the bathing solution. For bot,h dyes, the free fraction in the PBS-dextran solution, C,/(C,+ Cf), was the same at all the concentration levels tested, and averaged 074 +@Ol (mean I s.o., n = 16) for fluorescein and 0.79+0.03 for carboxyfluorescein, so that C,/C, was 0.35 and 0.26, respectively. The ratio, C’,/C,, after 2-hr immersion was taken to represent TV&,because as will be discussed later, equilibrium between the dye and the stroma is expected to be almost, attained at that time. The value of rca for fluorescein fell from 3.20 f 0.25 (mean f s.D., 12= 6) to 1%7-J0*06 as C, rose from 5% x lo-* to 5.9 x 1W5 g ml-l (Fig. 1). The value obtained at 5.7 x 10W6g ml-r is significantly lower than that obtained at 58 x 10m8 g ml-l (unpaired t-test, P < 0.01) and that obtained at’ 5.9 x 10-j g ml-r is significantly lower than any of those obtained at lower concentrations (P < 0.001). For carboxyfluorescein, the value changed from 1.31 f0.24 (mean+s.n.; n = 5) to 123+@05 over the same concentration range and the differences are not significant at all the concentration levels (P > 605) (Fig. 2). If the binding of the dye by t,he dextran in the stroma is less than that determined for the free solution, the values of rca could be up to about 3-5% greater for fluorescein and about 6% greater for carboxyiluorescein. After 05hr immersion, C,/C, for fiuorescein was 2.55 4 0.17 (mean f s.n., n = 6) at 5.8 x lo-’ g ml-l and 1.78 +@15 at 5.9 x 10V5 g ml-l; these averaged 89 y. of the 2 hr values. I77 viva determination
of rcft for
carboxy$uorescein
Three days after the injection of carboxyfluorescein into the lens, the cortex was found to be strongly but not evenly stained and was presumably controlling the supply of carboxyfluorescein to the anterior segment of the eye as described by Kaiser and
M. SRATE
146
Carboxyfluorescein
concentration
(g ml-‘)
FIG. 2. The values of T,, and the corresponding
concentration of free carboxyfluorescein in the bathing medium. Each noint and bar indicates mean and S.D. in five corneas. For explanation of rCs, see Table I.
1
I 0
I I
I
I
I
I
2
3
4
5
Time (hr) FIG. 3. The time coume of apparent carboxyfluorescein
concentration in the stroma (4; 0-O) and in the anterior chamber (Pa; @---a) after intravitreal injection of the dye; time 0 is 20 hr after the injection. The values obtained at time l-5 hr are normalized to that obtained at time 0. Each point and bar indicates mean and S.D. in 11 eyes.
Maurice (1964) for fluorescein. Those six eyes gave a very slow time change of E”, and F, after 3 days and the rate of decline during the experiment calculated from the slope of the regression line fitted to the data plotted on a semilogarithmic paper was -0~006f0014 hr-l (meanfsn., N = 6) for F, and -0*013+0.013 for F,. There is no significant difference between them (paired t-test, P > 0.05) and neither figure is significantly different from zero (P > 0.05). The value of r,, averaged 1.62 to.23 in a C, range between 1.2 x 10e7 and 2.8 x IO-” g ml-‘.
CARBOXYFLUORESCEIN
147
TABLE II
Summary
of endothelial
permeability
coeficient P ( x 1OV G min-L)
Cornea1 thickness (mm)
Rabbit
1
038
064
2
0.38 0.37 0.36 0.36 038 038 0.37 @37 0.39 039
040 070 054 0.54 0.43 @44 061 oY50 0.58 067
3.66 455 386 2.94 2.93 2.43 2.50 3.39 2.78 3-39 3.92
0.38 0.01
059 @ii
331 0.66
3 4 5 6 Mean S.D.
For explanations
Endothelial
of calculation
of k,,,, and Pa,, see Table I.
permeability
coeficient for carboxyjluoresceilb
In 11 out of 12 eyes, F, and Fa declined almost in parallel during the experiment and the rate of decline was 0142&O-037 h-l (mean+S.D., n = 11) for F, and @144+0.034 hr-l for F, and the ratio, FJF,, was 1.66kO.16. The thickness of the corneas was 0.38+0.01 mm. The time course of the average values of F, and F, is shown in Fig. 3, The ratio of CZ to F& was 1.05+0~08 (n = 12), which is not significantly different from unity (P > 0.05), and the correlation coefficient between Ca and F, was 0.999. No significant correlation was found between the value of C,/F, and Fa (P > 0.1). The ratio of C, to FS was 1.63f022 (n = 10): which is significantly different from unity (P < O.OOl), and the correlation coefficient between C, and FSwas 0.995. No significant correlation was found between the value of C,/F, and F, (P > 0.4). Accordingly, the relationships, C, = 1*6F, and C, = l.OF,, were assumed to hold in a concentration range encountered in the experiment. Inserting these values and rcB, 1.62, in equation (3) gave a value of 0.59 kO.11 for /Gc.ac (meanfS.D., n = 11) and a value of 3.31+0.66x lop4 cm min-l for P,,. The results of the calculation are summarized in Table II. 4. Discussion Recently, the in vivo value of rca for fluorescein in the rabbit eye was determined to be 4.0 in a C, range between 10e8 and lo-’ g ml-l (Araie and Maurice, 1985b), which is higher than the in vitro value obtained here, about 3.2, for a similar range dye concentration in the bathing medium. This is also found for carboxyfluorescein. Part of the difference might be overestimation of dextran binding in the stroma. The present method of determination of the in vivo rca for carboxyfluorescein requires that the concentration in the stroma during the experiment remains unchanged. However, its actual time course showed a small rate of change averaging -0.006 hr-1,
148
NI. ARAIE
which could result in an overestimate of rcB. This overestimate would be only of the order of a few percent from equation (3), since the value of k,.,, is the order 0fO.6 hr-‘. Stromal hydration is known to be dependent on pH and salt concentration in its bathing medium (lMaurice, 1984). Therefore, it might be possible that the dye interaction with the stromal matrix, which is reflected in the value of rcL, is also dependent on these factors. Two-hour immersion in the PBS-dextran solution, which contains only sodium hydrochloride and unphysiologically high phosphate, would cause a considerable change in the ionic environment in the stroma, because the diffusion of small ions should be much more rapid than those of the dyes used, and this may be responsible for some part of discrepancy in the values of rca obtained in vitro and in vivo. A 2hr soaking seems enough to reach equilibrium. If the endothelium is impermeable to the dye, its concentration at the endothelial surface of the stroma should reach within about 1-2 o/0of that at the stroma-bathing medium boundary when Dt/qz = 2, where D is the diffusion constant of the dye across the stroma, t is time and q is the thickness of the stroma (Carslaw and Jaeger, 1947). As q is the order of O-36 mm (Maurice, 1984), D must be 3-6 x iOe7 cm2 see-’ or more for this condition to be satisfied. A separate experiment demonstrated that D for fluorescein is the order of 6 x lo-’ cm2 set-l and that for carboxyfluorescein is rather greater (Araie and Maurice, 1985a). In accordance with this, the value of C,/C, obtained for fluorescein after 0.5hr soaking averaged 89 o/o of that obtained after 2 hr soaking. According to Carslaw and Jaeger (1947), if C,,lC, at 05 hris about 90 y0 of that at 2 hr, then the value at 2 hr should be very close to the value at equilibrium. Actually, the endothelium would be somewhat damaged in these experiments, so that the dye would also penetrate across this surface, speeding the approach to equilibrium. Two reports in the literature give figures for r,,, for fluorescein over a range of concentration. Mishima and Maurice (1971) using an in vitro technique similar to that described here, found a similar decline in the value of s,,; about 5 at 5 x 10e7 g ml-’ falling to 1.6 at 1 x 10m5 g ml-l. Nagataki et al. (1983) reported that it was 1.33 and showed no change in a concentration range between 1 x 10F7 and 1 x 10e5 g ml-r. These two results and the present one show considerable discrepancies in the value of rca at lower concentration of fluorescein and possible explanations a!re that (i) the bathing medium used is different; Mishima and Maurice (1971) used unbuffered physiological saline containing 4-5 y0 dextran, and Nagataki et al. (1983) used phosphate buffer at pH 7.2. As discussed above, this difference might influence the dye-stromal matrix interaction : (ii) Nagataki et al. (1983) used the apparent fluorescein concentration in the stroma calculated directly from the fluorometric readings (F’) in computing the value of rca. However, for fluorescein, C, might not be equal to F, (Araie and Maurice, 1985). In experiments using the topical application of dye to determine Pa,, the value of vca must be constant in a concentration range employed. Based on the present results with fluorescein, it seems that rcil is almost constant when the stromal concentration is 10W6g ml-1 or less, but not when it is higher. Thus, a high initial stromal concentration of fluorescein must be avoided in the in vivo determination of Pa,. On the other hand, the value of r,, determined in vitro for carboxyfluorescein, unlike that for Buorescein, showed no significant change over the concentration range tested. To calculate Pat, the ratio of the mean concentration in the stroma, C,, to the apparent measured one, Fs, under the present conditions of measurement, was determined and the obtained value, 1.6, was larger than unity. One of the reasons for this discrepancy
CARBOXYFLCOR:%SCEIN
149
is possibly the quenching of the fluorescence of carboxyfluorescein by the stromal tissue. An underestimate of Fs due to some artifacts during the measurement might also be responsible for a part of it. However, these factors should not affect the calculation of P,,, since F, was converted to C, from the results of the normalization experiment carried out under identical conditions to the main P,, experiment. The value of P,, for carboxyfluorescein obtained in the normal rabbit eye was 3.3 x lo-” cm mm-l, and 35 o/0 smaller than that for fluorescein, 5.1 x 10d4 cm min-l (Araie and Maurice, 198513). The molecular weight of carboxyfluorescein, 376, is 13 y’ larger than that of fluorescein, 332. However, it seems unlikely that a 13 o/0 increase in the molecular weight would cause 35% decrease in the endothelial permeability, if penetration was through water-filled pores. Pa, for mannitol (182 MW) is reported to be 5.5 x IO-* cm min-l, being rather close to the value for fluorescein (Kim, Green, Mart,inez and Paton, 1971; Maurice, 1984), which suggests that fluorescein is moving more rapidly than is appropriate for its molecular weight. Active transport of fluoreseein by the cornea1 endothelium has not been reported and its relatively higher value of Pa, may be explained by assuming that it can penetrate the cell membrane in its passage across the endothelial cell layer, as opposed to carboxyfluorescein of which passage is restricted to the paracellular space. This interpretation is compatible with the histologic findings obtained in the ciliary and iris epithelium under fluorescence microscope (Grimes et al., 1982) and its relatively high lipid solubility (Weinst,ein et al., 1977). On the other hand, the value of P,, for carboxyfluorescein determined here agrees with that for sucrose, 3.3-3.9 x lop4 cm min-I : (Mishima and Trenberth, 1968; Kim et al., 1971; Riley, 1977; Maurice, 1984), and their molecular weights are similar, which suggests that these substances share the same route in the passage across the endothelium. Thus, carboxyfluorescein appears to have two advantages over fluorescein as a tracer dye for the in vivo evaluation of the barrier function of the endothelium: (i) the value of r,, does not change over a wide range of concentration. (ii) It does not penetrat.e the endothelial cell layer across the cell membrane and its endothelial permeability may be expected to reflect the state of the intercellular junctional complexes more sensitively. Recently: it has been shown that carboxyfluorescein is less readily converted to a fluorescent metabolite than fluorescein after systemic administration (McLaren and Brubaker, 1985). When P,, is determined by the systemic administration of a tracer, this would also make its use advantageous.
ACKNOWLEDGMENTS The author is grateful to Prof. David M. Maurice for his advice during the course of this study. This work was supported by IVIH Grant EY 04863.
REFERENCES Araie,
M. and Maurice, D. MM. (1985a). Transcorneal transfer of fluorophores with different lipid solutility. Invest. O@thalmoZ. Via. Sci. (Suppl.) 26, :!. Araie. M. and Maurice, D. M. (1985b). A re-evaluation of cornea1 endothelial permeability to Auorescein. Exp. Eye Res. 41, 383-390 Bnselmi. P., Bron, A. J and Maurice, D. M. (1968). Action of drugs on the aqueous flow in man measured by fluorophotometry. Exp. Eye Res. 7: 48’7-96. Bourne, W. M. and Brubaker, R. F. (1983). Decreasedendothelial permeability in transplanted corneas. Am. J. Ophthalmol. 96, 362-7.
150
M. ARAIE
Carslaw, H. S. and Jaeger, J. G. (1947). Conduction of Heat in Solids. Pp. 82-5. Clarendon Press : Oxford. Felchlin, M. (1926). Versuche zur Ermittlung des spezifischen Gewichts der verschiedenen Augenmedien mittels einer neuen Methode. Arch. Ophthalmol. 117, 325-42. Grimes, P. A., Stone, R. A., Laties, A. M. and Li, W. (1982). Carboxyfluorescein. A probe of the blood-ocular barriers with lower membrane permeability than fluorescein. Arch. Ophthalmol. 100, 6359. Grotte, D., Mattox, V. and Brubaker, R. F. (1985). Fluorescent, physiological and pharmacokinetic properties of fluorescein glucuronide. Exp. Eye Res. 40, 23-33. Kaiser; R. J. and Maurice, D. M. (1964). The diffusion of fluorescein in the lens. Exp. Eye Res. 3, 15665. Kim, J. H., Green, K., Martinez, M. and Paton, D. (1971). Solute permeability ofthe cornea] endothelium and Descemet’s membrane. Exp. Eye Res. 12, 231-8. Maurice, D. M. (1984). The cornea and sclera. In The Eye Vol. lb, 3rd edn. (Ed, Davson, H.). Pp. i-158. Academic Press: London. Maurice, D. M. and Giardini, A. A. (1951). A simple optical apparatus for measuring the cornea1 thickness and the average thickness of the human cornea. Br. J. Ophthalmol. 35; 16977. McLaren, J. F. and Brubaker, R. F. (1985). Lack of a spectral shift in fluorescence following systemic administration of carboxyfluorescein. Invest. Ophthalmol. i/i:s. Sci. (Suppl.) 26, 192. Minkowsi, J. S., Bartels, S. P., Delori, F. C., Lee, S. R., Kenyon, K. R. and Neufeld, A. H. (1984). Cornea1 endothelial function and structure following cryo-injury in the rabbit. Invest. Ophthalmol. Vi’is. Sci. 25, 1416-25. Xshima, S. (1981). Clinical pharmaeokinetics of the eye. Invest. Ophthalmol. Vis. Sci. 21, 504-41. Mishima; S. (1982). Clinical investigations on the cornea1 endothelium. Am. J. OphthaZmol. 93, l-29. Mishima, S. and Maurice, D. M. (1971). In vivo determination of the endothelial permeability to fluorescein. Acta Sot. Ophthalmol. Jpn 75, 23G-43. Mishima, S. and Trenberth, S. M. (1968). Permeability of the cornea1 endothelium to nonelectrolytes. Iwest. Ophthalmol. 7, 34-43. Nagataki, S., Brubaker, R. F. and Dwight, A. (1983). The diffusion of fluorescein in the stroma of rabbit cornea. Exp. Eye Res. 36, 765-71. Riley, M. V. (1977). A study of the transfer of amino acids across the endothelium of the rabbit cornea. Exp. Eye Res. 24, 35-44. Tsuru, T., Araie, M., Matsubara, M. and Tanishima, T. (1984). Endothelial wound-healing of monkey cornea : fluorophotometric and specular microscopic studies. Jpn J. Ophthalmol. 28, 105-25. Waltman, S. and Kaufman, H. E. (1970). In vivo studies of human cornea1 endothelial permeability. Am. J. Ophthalmol. 70, 45-7. Weinstein, J. N., Yoshikami, S., Henkart, P., Blumenthal, R. and Hagins, W. A. (1977). Liposomeeell interaction : transfer and intracellular release of a trapped fluorescent marker. Science 195, 489-92.