Exp. Eye Res.(1990) 51. 113-118
Calibration JENS
of Measurements
TAARNH0J,
LORNE Department
SCHLECHT,
In Vivo of Fluorescein JAY
of Ophthalmology,
(Received 3 January
W. McLAREN
AND
in the Cornea
RICHARD
F. BRUBAKER
Mayo Clinic, Rochester, MN, U.S.A.
1989 and accepted in revised form 28 November
1989)
Quenching of fluorescence of fluorescein is not observed with broad field fluorophotometers. Fiuorophotometric equipment which measures the fluorescence in a tiny spot has, however, been reported to underestimate the molarity of fluorescein in the rabbit cornea1stroma by as much as a factor of two. In this experiment, quenching was measured in the rabbit cornea with two scanning fluorophotometers. The quenching was measured by four different techniques : (1) by elution of fluorescein, (2) by elution of albumin, (3) by polarization of fluorescence, and (4) by spectrofluorophotometry. It was estimated by all four methods that quenching in the living rabbit cornea with these instruments is approximately 20%. Taken together, the four experiments suggest that the quenching of Eluorescenceof fluorescein can be explained entirely on the basis of the interaction of fluorescein and albumin in the stroma. Key words:‘fluorophotometry : cornea ; fluorescein ; quenching ; rabbit eye. 2. Materials and Methods
1. Introduction
Clinical measurements of the flow of aqueous humor and the permeability of the cornea1 endothelium to fluorescein often require measurement of the molarity of fluorescein in the cornea and the anterior chamber of fluorophotometry. The accuracy of measurements in the anterior chamber is easy to confirm, but the accuracy in the cornea remains in question. Jonesand Maurice (1966) measured the fluorescence of the fluorescein-stained rabbit cornea in vitro in a broad field of illumination and found that the total fluorescencein the stained button did not change after the button was digested. Brubaker (1982), who carried out a similar experiment, concluded that broad-field fluorescence measurements of the cornea permitted accurate calculation of the total number of moles of fluorescein in the cornea. He concluded that this resulted from the fortuitious balance between the
effects of scattering, which would tend to augment fluorescence, and the effects of quenching, which would tend to diminish fluorescence. Araie and Maurice (198 5) have recently studied the accuracy of focal measurements of fluorescence in the stroma, now commonly employed in clinical fluoro-
photometry. They concluded that such measurements underestimate the molarity of fluorescein in the stroma
by a factor of two. They attribute this finding to quenching
of fluorescence by the stroma. Since the
correspondence between fluorescence and molarity is critical to some experiments in which fluorophores are used in tracer experiments, we have pursued Araie’s lead in studying this relationship in the cornea. * R.P.B. foreignresearchfellow 1987-8, fromthe Eye Department of UmeGUniversityHospital,Umel, Sweden. t For reprint requestsat: Mayo Clinic, 200 First Street. SW.. Rochester,MN 55905. U.S.A.
00144835/90/080113+06 Y
$03.00/O
Experiments of two kinds have been carried out in
order to measure separately errors of geometric origin and errors due to quenching. Both sources of error are presumed to be instrument-dependent and the latter to be wavelength-dependent as well. (1) Experiment to Calibrate Reduction of Apparent FluorescenceDue to Stromal Thickness To simulate the cornea, an artificial chamber was made with an anterior and posterior curvature of 7.3 mm. The depth of the chamber could be varied continuously from 0 to 10 mm (McLaren and Brubaker, 1988). Solutions
of sodium
fluorescein
50 ng ml-’
or
500 ng ml-l in phosphate buffered solution (PBS), pH 7.4, were prepared. Two fluorophotometers were tested. The first was the scanning ocular spectrofluorophotometer (SOSF) described by McLaren and Brubaker (1988). This instrument
excites fluorescence with a thin cone of
light which intersects the eye at an angle of 55” below the axis of the radiometer. The radiometer accepts light from a narrow cone the principal axis of which is parallel to the optic axis of the eye. The union of these two cones (‘focal diamond ‘) has nominal dimensions of 0.3 x 0.6 x 0.7 mm (height, width, depth). The second fluorophotometer was the Fluorotron Master (Coherent Medical Inc., Palo Alto, CA) fitted with an anterior segment adaptor, which excites and collects light from opposite sidesof an objective lens at a maximal angle of 28” while scanning along the optical axis of the eye. The ‘focal diamond’ of this instrument is the intersection between the excitation which and emission beams measures
0.05 x 0.05 x 0.85 mm (height, width, depth). The 0 1990 Academic Press Limited EER 51
J. TAARNH0J
114
0.5
I.0 Thicknoes
I.5 (mm)
FIG. 1. Mean values of spatial correction curve for the SOSF(curve fit by ‘eye’). The curve was used to correct in viva cornea1 measurements of fluorescence for the error caused by the thickness of the target compared to the sizeof the focal diamond of the fluorophotometer. A similar curve was used for the Fluorotron Master.
excitation wavelength is 420490 nm and emission wavelenth 530-630 nm. Both instruments were used to measure the fluorescenceof solutions of fluorescein of known molarity and pH as the thickness of the chamber was adjusted between 0.1 mm and 1.5 mm in steps of 01 mm. The ratio of fluorescence in the thin target to the fluorescence of the same solution in a l-cm cuvette was found to vary with the thickness of the chamber according to the curve shown in Fig. 1. (2) Experiments to Calibrate Quenchingof Fluorescence in the Stroma
Four kinds of experiments were conducted to measure stromal quenching. Nine animals were used in each of these four experiments: (a) elution of fluorescein from stained stroma, (b) elution of albumin from stroma, (c) measurement of polarization of fluorescence in stained stroma, and (d) measurement of excitation spectrum of fluorescencein stained stroma. (a) El&ion of fluorescein. Nine adult pigmented rabbits were studied. All measurements of fluorescence were made with both the SOSFand the Fluorotron Master. Late in the afternoon the animals were anesthetized (standard dose, 100 mg ketamine+ 15 mg xylazine), and the autofluorescence of each cornea measured. The thickness of both corneas was measured with an ultrasonic pachometer (Acutome, Inc.) calibrated for a velocity of 1640 msec-‘. No correction was applied for the thickness of the epithelium or endothelium. The
ET AL.
dye was applied to one eye as six drops of a 2 % sodium fluorescein solution with 3 min between each drop. The eyelids were taped closed between the drops. The following day in the morning the stromal fluorescence (F,) was measured in each eye with the SOSF (excitation wavelength 490 nm, emission 540 nm) and the Fluorotron Master, and the time of measurement noted. The reading was corrected for thickness of the cornea, as determined in the first experiment, and for autofluorescence. The animal was killed with an i.v. infusion of 6 ml pentobarbital (65 mg ml-‘). The epithelium was removed, a central cornea1 button was excised with an S-mm trephine, and the time noted. The fluorescence which was measured in vivo was corrected for the loss of fluorescence up to the time of the excision, usually less than 15 min, assuming a loss of 0.32 % min-‘. The cornea1button was blotted, weighed on a Cahn 7500 microbalance, and transferred to a vial containing 1 ml of PBS. After 48 hr elution at 4% during which
the vial was shaken twice daily, the fluorescence was measured in the eluate. To correct for incomplete elution caused by binding of fluorescein within the cornea1 button and loss of fluorescence due to degradation during the 48-hr elution, the unstained fellow cornea was equilibrated under identical conditions in a PBSsolution containing a known amount of fluorescein. After 48 hr equilibration the fluorescence of the control solution was measured. The loss of fluorescence from the control solution was usedto correct the measured fluorescence of the eluate of the stained cornea (see Fig. 2). This corrected value, C,, was considered to be the molarity of fluorescein in the stained cornea in vivo. The ratio F,/C,, of time-corrected in vivo fluorescence to molarity, was calculated as a measure of quenching. (b) Elution of albumin. The cornea1 buttons from experiment 1 were left in the solutions for an additional 5 days. These eluates were analyzed for total protein by binding to anazolene sodium (Coo-
Control cornea
I
Cornea without dye,valume Cco “C’ 0‘
G-cof -. Cd
1000 pl v’
Y’
Cornea with fluorescein
Cornea with iye, volume V’ At 48 hr the total mass in the system is M+=C, .(I000 pl +r.~) and C, = $
FIG. 2. On the left side of Fig. 2 we have shown how the unstained cornea from the fellow eye was used to calculate the distribution ratio of fluorescein between the eluting solution and the cornea1 button. On the right side the fluorescein elution process is illustrated and the equations used to calculate the true unquenched fluorescence in the cornea. Nine experiments were done.
QUENCHING
OF FLUORESCEIN
IN THE STROMA
115 TABLE
Summary
results offour diRerent methods of measuring F,/C, (n = 9 for each experiment)
Experiment number
2(a)
I
Technique
UC, Mean + S.D.
Fluorotron Master SOSF Fluorotron Master SOSF SOSF SOF
0.81 kO.24 0.78 + 0.28 0.86 0.82 0.82 &-0.02 0.78 k 0.02
SOF SOF
074+0.04 0.72 f 0.03
SOSF SOSF
083 kO.03 0.79kO.02
Riutefluorescein (corrected for cornea1 thickness) (corrected for stromal thickness, 0.3 5 mm)
2(b)
Elute albumin
2(c)
Polarization of fluorescence (i) Direct calibration (ii) Albumin calibration Red shift (i) Direct calibration (ii) Albumin calibration
2(d)
Fluorophotometer
massie blue). The remainder of the eluate was concentrated approximately ten times by ultrailltration through a micropore filter (Centricon 10, Micron, Division of W. R. Grace and Co.). The eluate was placed on a cellulose acetate plate, and the proteins separated by electrophoresis. The percent of total protein due to albumin was calculated from the area of Coomassie blue staining of the albumin peak in comparison to all other areas of the electrophoretic pattern. The concentration of albumin in the eluate was then calculated from the total protein and the percent albumin. The concentration of albumin in the living stroma was calculated from the volume of the unswollen cornea1 button and the total mass of albumin eluted. This concentration was regarded’as the lower limit since some of the albumin may have remained in the swollen cornea1 button and will not have been present in the eluate. The ratio F,/C, was calculated from the albumin concentration on the basis of the following assumptions: (1) rabbit albumin has only one type of binding site for fluorescein, (2) the concentration in the stroma of albumin-fluorescein binding sites is much greater than the concentration of fluorescein in the stroma, (3) quenching of fluorescence of fluorescein in the stroma due to all causes besidesalbumin is negligible in comparison to albumin’s effect, and (4) the greatest part of albumin by 7 days is in the eluate rather than the swollen stromal button. The dissociation constant I& of the system of rabbit albumin and fluorescein has been determined to be 2.3 x 10e4M and the concentration of fluorescein binding sites in rabbit plasma found to be 1.6 x 1O-3M (Brubaker et al., 1982). The concentration of fluorescein binding sites in the stroma, B, based on the assumptions given above, can then be determined from the ratio of the concentration of albumin in the stroma to that in plasma.
[albumin in stroma] B = 1.6 x 1O-3M [albumin in plasma] ’
The ratio of bound fluorescein in the stroma, C,, to unbound fluorescein in the stroma, C,, for any given total concentration of fluorescein in the stroma, C,, (which meets assumption 2) can be calculated from the law of mass action.
c,/c, = p. The concentrations of unbound and bound fluorescein can then be calculated.
c, = W(WC,) + 11 c, = c,- Cf. The intensity of fluorescence of fluorescein at concentration C, under these conditions in the stroma can then be calculated. To do so requires knowledge of the molar fluorescent intensity ratio of the bound form of fluorescein to the molar fluorescent intensity of the unbound form, I,/&. This ratio is wavelength dependent. For the scanning ocular spectrofluorophotometer (SOSF)&/II is 0.69 and for the scanning ocular fluorophotometer, it is 0.61. This ratio has been determined by a linearized graph of the fluorescence of fluorescein in a series of albumin solutions of increasing concentration (Brubaker et al., 1982). The fluorescence in the stroma, F,, is then given by The quenching can then be calculated as before, WCS. The accuracy of FJC, is not critically dependent on the accuracy of the constants B, KD and I,/&. A 20% error in either B or KD or a 10% error in I,/I, results in approximately a 1 y0error in the calculated value of F,. (c) Polarization of fluorescence - (i) Calculation from Dundliker equation. When the in vivo fluorophotometric
measurements for the elution experiments were made, the polarization of fluorescence in the central cornea of these nine animals was measured with the SOF as 9-2
J. TAARNH0J
116
describedby Herman, McLaren, and Brubaker( 1988). The ratio of bound fluorescein to unbound fluorescein, C,/C,, can be calculated from the observed polarization of fluorescence in the stroma, p (Brubaker et al., 1982).
c,/c, = m KP-PMPtl--Pn In this equation, p1is the polarization of fluorescence of unbound fluorescein, and pb is the polarization of fluorescence when it is bound to rabbit albumin. FJC, can be obtained from C&/C, and I,,/& as stated in method (ii). Important assumptions of this method are: (1) albumin is the principal substance which binds to and alters the polarization of fluorescence of fluorescein, (2) the viscosity of water in the stroma is insticient to cause polarization of fluorescence of unbound fluorescein, (3) few if any fluorescein binding sites on stromal albumin are occupied by competing substances, (4) the milieu of the stroma as regards polarization of fluorescence is very similar to the simple system employed for calibration, and (5) the concentration in the stroma of albumin-fluorescein binding sites is much greater than the concentration of fluorescein in the stroma. (c) Polarization offluorescence - (ii) Direct calibration of albumin concentration from polarization. The observed polarization can be used to measure the concentration of albumin directly from a calibration curve (Herman et al., 1988). A calibration curve of polarization versus concentration of albumin was made (Fig. 3). To mimic the binding of fluorescein in the cornea, rabbit aqueous humor was used as the calibrating medium to which rabbit albumin (fraction V) was added incrementally to produce solutions of increasing concentration up to 5 g dl-I. As more protein was added, the pH was observed to decline slightly and was titrated back to 7.59 with 1 N NaOH. The polarization of fluorescence
of each albumin solution was measured at room temperature and was plotted against the albumin concentration. The albumin concentration in the living cornea was determined from the in vivo measurements of polarization and the calibration curve. The corresponding F,/C, value was calculated as described under experiment 2. Unlike experiment 2, experiment 3 does not depend on complete elution of albumin since the polarization measurement was made in vivo. (d) Excitation spectrum - (i) Direct calibrution of quenching from spectral shift. When fluorescein is bound
to protein, the peak of excitation is shifted towards longer wavelengths (red shift). At the same time the intensity of steady-state fluorescence is reduced (quenching). With the SOSFthe excitation spectrum and fluorescence (F) of a known concentration of fluorescein (C) was measured in a phosphate buffered solution contained in a cuvette at pH 7.4. The same measurements were made in rabbit aqueous humor to which increasing amounts of rabbit albumin (fraction V) were added as described in experiment 3. The excitation wavelength was varied from 450 nm to 520 nm and the emission wavelength set at 540 nm. The red shift produced by albumin was calculated as described by McLaren, Taarnhoj and Brubaker (1988). The excitation curve for fluorescein in buffer was displaced along the wavelength axis until it provided a best fit determined by the method of least squares to the data points for fluorescein in albumin. Quenching was expressedas the ratio of the observed intensity of fluorescence to the unquenched intensity, the ratio F/C. A calibration curve for F/C (unitless) (Fig. 4) was constructed by plotting F/C vs. red shift (in nm). The red shift was measured in vivo with the same instrument in nine rabbit corneas stained with fluorescein. The calibration curve derived from rabbit aqueous humor was employed to convert red shift to c/C,. If the fluids of the stroma in vivo are similar to aqueous humor in their interaction with the fluI.0
./.-’
.
m
.
ET AL.
/
/
0.9 c, c
/
0.6
0.1
0.5 [Albumin]
I.0
5.0
(g dl-‘)
FIG. 3. Calibration curve for converting the measured polarization value to albumin concentration.
0
5
IO
15
Red shift (nm)
FIG. 4. Calibration curve for converting the measured red shift value to quenching.
QUENCHING
OF FLUORESCEIN
117
IN THE STROMA
5.0 -
When the stromal thickness is used, the corresponding values for F,/C, are 086 and O-82.
.t /.
i =6 -T T.E
I.0 -
/”
In experiment 2(b) the concentration of elutable protein in the cornea was 1.20 + 0.12 g dll’, of which 43 f 7% was albumin, giving an albumin concentration of 0.52 g dl-‘. The calculated ratios F,/C,. on the basis of the albumin concentration, were 0.82 f 0.02 (I,/& = 0.69) for the SOSFand 0.78 + 0.02 (I,/& = 0.61) for the SOF.
./’
0.5.i
e3 4
Experiment 2(b). Elution of Albumin.
de
/ O” 0.05
./* 7
0.01 I 0
I 5
I IO
I 15
Red shift (nm)
FIG.5. Calibrationcurve for convertingthe measuredred shift to albumin concentration. orescein-albumin system, then the F/C derived from the red shift in aqueous humor can be used as an estimate of quenching in vivo, F,/C,. (d) Excitation spectrum - (ii) CaIcwfationof albumin concentration from spectrul shift. The measurements of red shift in fluorescein-aqueous humor solutions containing known amounts of rabbit albumin mentioned above were also used to construct a calibration curve for albumin by plotting red shift against albumin concentration (Fig. 5). The albumin concentration in the living cornea could then be derived from the in vivo measurement of spectral shift. The FJC, ratio was then calculated from the albumin concentration so derived, as explained in experiment 2. 3. Results
Experiment I The results of experiment 1 are shown in Fig. 1 for the SOSF.The values are the means of two measurements. The curve was fitted by eye. Since the thickness of the rabbit cornea in this experiment was 0.3 7 k 0.02 ( + s.D.), the uncorrected SOSFreading underestimates the fluorescence of the rabbit cornea by a factor of 2.04. The corresponding value for the Fluorotron Master was 2.44. If fluorescein is largely located in the stroma rather than the epithelium, the correction should be based on stromal thickness. If a figure of 0.35 mm is used (Maurice and Watson, 1965), the two correction factors are 2.17 and 2.56. The results of the elution method are dependent on which thickness is used to determine this correction, the correct value probably lying somewhere between the two. Experiment 2(a). Elution of Fluorescein. In the elution experiment when the cornea1 thickness is used for the spatial correction, the F,/C, ratio was found to be 0.81 kO.24 (meanf SD.) for the Fluorotron Master and 0.78 f0.28 for the SOSF.
Experiment 2(c). Polarization of Fluorescence.
The polarization of fluorescence in experiment 2(c, ii) was found to be 0.21 f0.039 (unitless) in the center of the cornea. The FJC, value calculated from the polarization was 0.74 f 0.04. In Fig. 3 the calibration curve used in experiment 2(c, ii) is shown. When plotted on a log/log scale there is an almost linear relationship between polarization of fluorescence and concentration of albumin in the range of interest. The albumin concentration of the cornea, derived with the assistance of the calibration curve and the measured polarization, was 0.9 5 + 0.28 g dl-‘. The corresponding F,.C, value, calculated as explained under experiment 2(c), is 0.72 f0.03. Experiment 2(d). Excitation spectrum.
The two calibration curves used in experiment 2(d, i and ii) are shown in Figs 4 and 5. The red shift measured in the center of the cornea was 10.52 + 1.08 nm, and the corresponding &/C, value derived from the curve was 0.83 + 0.03. The red shift was also used to calculate the concentration of albumin in the central cornea. The albumin concentration by this method was 0.8 6 + 0.30 g cl-‘. This value was, in turn, used to calculate F,/C, as done in experiment 2(b). The calculated value was 0.79 + 0.02. 4. Discussion In the elution experiment we found a F,/C, ratio of 0.81-0.86 for the Fluorotron and 0.78-0.82 for the SOSF.If the fluorescence of fluorescein were the same in the stroma as in the buffered standard solutions, a ratio of 1.00 would be expected. The observed F,/C, ratio suggests that the molarity of fluorescein in the stroma is underestimated by approximately 15-20 % by these two instruments. It is noteworthy that the underestimate cannot be explained merely by the fact that the cornea is thin. A correction was applied to take into account its geometry. Had no geometric correction been applied, the FJC, ratio would have been much less, and we would have erroneously concluded that quenching of
J. TAARNHBJ
118
fluorescence was much greater than it actually is. What causesthe observedquenching ?For the reasons given below, the data are consistent with the idea that the phenomenon is due to the well-known interaction between fluorescein and albumin. When fluorescein interacts with serum albumin, its excitation and emission spectra shift (Delori, Castany and Webb, 1978), and the peak intensity of fluorescencediminishes. If fluorescence is being measured at fixed wavelengths near the excitation and emission peaks of the fluorophore, these effectswill reduce the intensity of the observed signal. The excitation and emission spectra obtained from fluorescein in the cornea in vivo are shifted and reduced in intensity quite like the spectra observed from dilute solutions of plasma in aqueous humor. Small shifts in pH near neutrality also alter the intensity of fluorescein’s fluorescence but hardly affect the excitation peak (Thomas et al., 1990). We believe the shifts observed in the living cornea are due primarily to interaction betweep fluorescein and albumin. When fluorescein binds to albumin, its emission is polarized (Laurence, 1952). In experiment 2(c), we observed polarization of fluorescence from the cornea but not the anterior chamber. We believe that the polarization is due primarily to the binding of fluorescein to albumin in the stroma. The albumin concentrations calculated from the different methods used in these experiments are somewhat higher than reported by Maurice and Watson (1965). They found a gradient of albumin concentration across the radial profile of the cornea. The limbus was approximately one-third the plasma (approximately 0.75 g dl-‘) and the center approximately one-tenth the plasma (approximately 0.25 g dl-‘). In the elution experiment, where the excised button would have included both central and peripheral portions, we calculated a concentration of 0.52 g dl-‘, in good agreement with the data of Maurice and Watson. In the polarization experiment, we obtained a figure of 0.95 g dl-‘. We do not know why this technique yielded a higher estimate of albumin. Perhaps the viscosity of water in the stroma is high enough to cause some polarization of unbound fluorescein. The spectral shift experiment also yielded a higher estimate of albumin of 0.86 g dl-I. Again, no reason is obvious unless the conditions in our calibrating medium differ significantly from those in the stroma.
ET AL.
Despite the variations in the concentration of albumin calculated by the different methods, the agreement in the calculated F,/C, is good. It would seemreasonable to make allowance for this quenching effect in the stroma when attempting to deduce molarity from fluorescence. In addition, we conclude that the principal reason for this quenching is the interaction of fluorescein and albumin in the stroma.
Acknowledgments
This work wassupportedin part by NIH grant EY00634, Researchto Prevent Blindness,Inc., New York, NY, the MayoFoundation,Rochester,MN, KronprinsessanMargaretha’s Arbetsngmd fijr Synskadade,Sweden, and Carmen och Bertil Regners Fond fijr forskning inom omrddet i)gonsjukdomar, Sweden.
References Araie, M. and Maurice, D. M. (1985). A re-evaluation of cornea1endothelial permeability to fluorescein. Exp. Eye Res. 41. 383-90. Brubaker, R. F. (1982). The flow of aqueous humor in the human eye. Trans. Am. Ophthahnol. Sot. 80, 391474. Brubaker, R. F., Penniston, J. T., Grotte, D. A. and Nagataki, S. (1982). Measurement of fluorescein binding in human plasma using fluorescence polarization. Arch. OphthaZmoZ. 100, 625-30. Delori, F. C.. Castany, M. A. and Webb, R. H. (1978). Fluorescence characteristics of sodium fluorescein in plasma and whole blood. Exp. Eye Res. 27, 417-2 5. Herman, D. C., McLaren, J. W. and Brubaker. R. F. (1988). A method of determining concentration of albumin in the living eye. Invest. Ophthalmol. Iris. Sci. 29, 133-7. Jones, R. F. and Maurice, D. M. (1966). New methods of measuring the rate of aqueous flow in man with fluorescein. Exp. Eye Res. 5. 208-20. Laurence, D. J. R. (1952). A study of the adsorption of dyes on bovine serum albumin by the method of polarization of fluorescence. Biochem.I. 51, 168-80. McLaren, J. W. and Brubaker. R. F. (1988). A scanning ocular spectrofluorophotometer. Invest. Ophthalmol. Vis. Sci. 29, 1285-93. McLaren, J. W., TaarnhBj, J. and Brubaker, R. F. (1988). Spectral changes of fluorescein in the anterior segment. Invest. OphthalmoI. Vis. Sci. 29 (ARVO Suppl.). 88. Maurice, D. M. and Watson, P. G. (1965). The distribution and movement of serum albumin in the cornea. Exp. Eye Res.4, 355-63. Thomas, J. V., Brimijoin, M. R., Neault, T. R. and Brubaker, R. F. (1990). The fluorescent indicator pyranine is suitable for measuring stromal and camera1pH in vivo. Exp. Eye Res. 50. 241-9.
Calibration JENS
of Measurements
TAARNH0J,
LORNE Department
SCHLECHT,
In Vivo of Fluorescein JAY
of Ophthalmology,
(Received 3 January
W. McLAREN
AND
in the Cornea
RICHARD
F. BRUBAKER
Mayo Clinic, Rochester, MN, U.S.A.
1989 and accepted in revised form 28 November
1989)
Quenching of fluorescence of fluorescein is not observed with broad field fluorophotometers. Fiuorophotometric equipment which measures the fluorescence in a tiny spot has, however, been reported to underestimate the molarity of fluorescein in the rabbit cornea1stroma by as much as a factor of two. In this experiment, quenching was measured in the rabbit cornea with two scanning fluorophotometers. The quenching was measured by four different techniques : (1) by elution of fluorescein, (2) by elution of albumin, (3) by polarization of fluorescence, and (4) by spectrofluorophotometry. It was estimated by all four methods that quenching in the living rabbit cornea with these instruments is approximately 20%. Taken together, the four experiments suggest that the quenching of Eluorescenceof fluorescein can be explained entirely on the basis of the interaction of fluorescein and albumin in the stroma. Key words:‘fluorophotometry : cornea ; fluorescein ; quenching ; rabbit eye. 2. Materials and Methods
1. Introduction
Clinical measurements of the flow of aqueous humor and the permeability of the cornea1 endothelium to fluorescein often require measurement of the molarity of fluorescein in the cornea and the anterior chamber of fluorophotometry. The accuracy of measurements in the anterior chamber is easy to confirm, but the accuracy in the cornea remains in question. Jonesand Maurice (1966) measured the fluorescence of the fluorescein-stained rabbit cornea in vitro in a broad field of illumination and found that the total fluorescencein the stained button did not change after the button was digested. Brubaker (1982), who carried out a similar experiment, concluded that broad-field fluorescence measurements of the cornea permitted accurate calculation of the total number of moles of fluorescein in the cornea. He concluded that this resulted from the fortuitious balance between the
effects of scattering, which would tend to augment fluorescence, and the effects of quenching, which would tend to diminish fluorescence. Araie and Maurice (198 5) have recently studied the accuracy of focal measurements of fluorescence in the stroma, now commonly employed in clinical fluoro-
photometry. They concluded that such measurements underestimate the molarity of fluorescein in the stroma
by a factor of two. They attribute this finding to quenching
of fluorescence by the stroma. Since the
correspondence between fluorescence and molarity is critical to some experiments in which fluorophores are used in tracer experiments, we have pursued Araie’s lead in studying this relationship in the cornea. * R.P.B. foreignresearchfellow 1987-8, fromthe Eye Department of UmeGUniversityHospital,Umel, Sweden. t For reprint requestsat: Mayo Clinic, 200 First Street. SW.. Rochester,MN 55905. U.S.A.
00144835/90/080113+06 Y
$03.00/O
Experiments of two kinds have been carried out in
order to measure separately errors of geometric origin and errors due to quenching. Both sources of error are presumed to be instrument-dependent and the latter to be wavelength-dependent as well. (1) Experiment to Calibrate Reduction of Apparent FluorescenceDue to Stromal Thickness To simulate the cornea, an artificial chamber was made with an anterior and posterior curvature of 7.3 mm. The depth of the chamber could be varied continuously from 0 to 10 mm (McLaren and Brubaker, 1988). Solutions
of sodium
fluorescein
50 ng ml-’
or
500 ng ml-l in phosphate buffered solution (PBS), pH 7.4, were prepared. Two fluorophotometers were tested. The first was the scanning ocular spectrofluorophotometer (SOSF) described by McLaren and Brubaker (1988). This instrument
excites fluorescence with a thin cone of
light which intersects the eye at an angle of 55” below the axis of the radiometer. The radiometer accepts light from a narrow cone the principal axis of which is parallel to the optic axis of the eye. The union of these two cones (‘focal diamond ‘) has nominal dimensions of 0.3 x 0.6 x 0.7 mm (height, width, depth). The second fluorophotometer was the Fluorotron Master (Coherent Medical Inc., Palo Alto, CA) fitted with an anterior segment adaptor, which excites and collects light from opposite sidesof an objective lens at a maximal angle of 28” while scanning along the optical axis of the eye. The ‘focal diamond’ of this instrument is the intersection between the excitation which and emission beams measures
0.05 x 0.05 x 0.85 mm (height, width, depth). The 0 1990 Academic Press Limited EER 51
J. TAARNH0J
114
0.5
I.0 Thicknoes
I.5 (mm)
FIG. 1. Mean values of spatial correction curve for the SOSF(curve fit by ‘eye’). The curve was used to correct in viva cornea1 measurements of fluorescence for the error caused by the thickness of the target compared to the sizeof the focal diamond of the fluorophotometer. A similar curve was used for the Fluorotron Master.
excitation wavelength is 420490 nm and emission wavelenth 530-630 nm. Both instruments were used to measure the fluorescenceof solutions of fluorescein of known molarity and pH as the thickness of the chamber was adjusted between 0.1 mm and 1.5 mm in steps of 01 mm. The ratio of fluorescence in the thin target to the fluorescence of the same solution in a l-cm cuvette was found to vary with the thickness of the chamber according to the curve shown in Fig. 1. (2) Experiments to Calibrate Quenchingof Fluorescence in the Stroma
Four kinds of experiments were conducted to measure stromal quenching. Nine animals were used in each of these four experiments: (a) elution of fluorescein from stained stroma, (b) elution of albumin from stroma, (c) measurement of polarization of fluorescence in stained stroma, and (d) measurement of excitation spectrum of fluorescencein stained stroma. (a) El&ion of fluorescein. Nine adult pigmented rabbits were studied. All measurements of fluorescence were made with both the SOSFand the Fluorotron Master. Late in the afternoon the animals were anesthetized (standard dose, 100 mg ketamine+ 15 mg xylazine), and the autofluorescence of each cornea measured. The thickness of both corneas was measured with an ultrasonic pachometer (Acutome, Inc.) calibrated for a velocity of 1640 msec-‘. No correction was applied for the thickness of the epithelium or endothelium. The
ET AL.
dye was applied to one eye as six drops of a 2 % sodium fluorescein solution with 3 min between each drop. The eyelids were taped closed between the drops. The following day in the morning the stromal fluorescence (F,) was measured in each eye with the SOSF (excitation wavelength 490 nm, emission 540 nm) and the Fluorotron Master, and the time of measurement noted. The reading was corrected for thickness of the cornea, as determined in the first experiment, and for autofluorescence. The animal was killed with an i.v. infusion of 6 ml pentobarbital (65 mg ml-‘). The epithelium was removed, a central cornea1 button was excised with an S-mm trephine, and the time noted. The fluorescence which was measured in vivo was corrected for the loss of fluorescence up to the time of the excision, usually less than 15 min, assuming a loss of 0.32 % min-‘. The cornea1button was blotted, weighed on a Cahn 7500 microbalance, and transferred to a vial containing 1 ml of PBS. After 48 hr elution at 4% during which
the vial was shaken twice daily, the fluorescence was measured in the eluate. To correct for incomplete elution caused by binding of fluorescein within the cornea1 button and loss of fluorescence due to degradation during the 48-hr elution, the unstained fellow cornea was equilibrated under identical conditions in a PBSsolution containing a known amount of fluorescein. After 48 hr equilibration the fluorescence of the control solution was measured. The loss of fluorescence from the control solution was usedto correct the measured fluorescence of the eluate of the stained cornea (see Fig. 2). This corrected value, C,, was considered to be the molarity of fluorescein in the stained cornea in vivo. The ratio F,/C,, of time-corrected in vivo fluorescence to molarity, was calculated as a measure of quenching. (b) Elution of albumin. The cornea1 buttons from experiment 1 were left in the solutions for an additional 5 days. These eluates were analyzed for total protein by binding to anazolene sodium (Coo-
Control cornea
I
Cornea without dye,valume Cco “C’ 0‘
G-cof -. Cd
1000 pl v’
Y’
Cornea with fluorescein
Cornea with iye, volume V’ At 48 hr the total mass in the system is M+=C, .(I000 pl +r.~) and C, = $
FIG. 2. On the left side of Fig. 2 we have shown how the unstained cornea from the fellow eye was used to calculate the distribution ratio of fluorescein between the eluting solution and the cornea1 button. On the right side the fluorescein elution process is illustrated and the equations used to calculate the true unquenched fluorescence in the cornea. Nine experiments were done.
QUENCHING
OF FLUORESCEIN
IN THE STROMA
115 TABLE
Summary
results offour diRerent methods of measuring F,/C, (n = 9 for each experiment)
Experiment number
2(a)
I
Technique
UC, Mean + S.D.
Fluorotron Master SOSF Fluorotron Master SOSF SOSF SOF
0.81 kO.24 0.78 + 0.28 0.86 0.82 0.82 &-0.02 0.78 k 0.02
SOF SOF
074+0.04 0.72 f 0.03
SOSF SOSF
083 kO.03 0.79kO.02
Riutefluorescein (corrected for cornea1 thickness) (corrected for stromal thickness, 0.3 5 mm)
2(b)
Elute albumin
2(c)
Polarization of fluorescence (i) Direct calibration (ii) Albumin calibration Red shift (i) Direct calibration (ii) Albumin calibration
2(d)
Fluorophotometer
massie blue). The remainder of the eluate was concentrated approximately ten times by ultrailltration through a micropore filter (Centricon 10, Micron, Division of W. R. Grace and Co.). The eluate was placed on a cellulose acetate plate, and the proteins separated by electrophoresis. The percent of total protein due to albumin was calculated from the area of Coomassie blue staining of the albumin peak in comparison to all other areas of the electrophoretic pattern. The concentration of albumin in the eluate was then calculated from the total protein and the percent albumin. The concentration of albumin in the living stroma was calculated from the volume of the unswollen cornea1 button and the total mass of albumin eluted. This concentration was regarded’as the lower limit since some of the albumin may have remained in the swollen cornea1 button and will not have been present in the eluate. The ratio F,/C, was calculated from the albumin concentration on the basis of the following assumptions: (1) rabbit albumin has only one type of binding site for fluorescein, (2) the concentration in the stroma of albumin-fluorescein binding sites is much greater than the concentration of fluorescein in the stroma, (3) quenching of fluorescence of fluorescein in the stroma due to all causes besidesalbumin is negligible in comparison to albumin’s effect, and (4) the greatest part of albumin by 7 days is in the eluate rather than the swollen stromal button. The dissociation constant I& of the system of rabbit albumin and fluorescein has been determined to be 2.3 x 10e4M and the concentration of fluorescein binding sites in rabbit plasma found to be 1.6 x 1O-3M (Brubaker et al., 1982). The concentration of fluorescein binding sites in the stroma, B, based on the assumptions given above, can then be determined from the ratio of the concentration of albumin in the stroma to that in plasma.
[albumin in stroma] B = 1.6 x 1O-3M [albumin in plasma] ’
The ratio of bound fluorescein in the stroma, C,, to unbound fluorescein in the stroma, C,, for any given total concentration of fluorescein in the stroma, C,, (which meets assumption 2) can be calculated from the law of mass action.
c,/c, = p. The concentrations of unbound and bound fluorescein can then be calculated.
c, = W(WC,) + 11 c, = c,- Cf. The intensity of fluorescence of fluorescein at concentration C, under these conditions in the stroma can then be calculated. To do so requires knowledge of the molar fluorescent intensity ratio of the bound form of fluorescein to the molar fluorescent intensity of the unbound form, I,/&. This ratio is wavelength dependent. For the scanning ocular spectrofluorophotometer (SOSF)&/II is 0.69 and for the scanning ocular fluorophotometer, it is 0.61. This ratio has been determined by a linearized graph of the fluorescence of fluorescein in a series of albumin solutions of increasing concentration (Brubaker et al., 1982). The fluorescence in the stroma, F,, is then given by The quenching can then be calculated as before, WCS. The accuracy of FJC, is not critically dependent on the accuracy of the constants B, KD and I,/&. A 20% error in either B or KD or a 10% error in I,/I, results in approximately a 1 y0error in the calculated value of F,. (c) Polarization of fluorescence - (i) Calculation from Dundliker equation. When the in vivo fluorophotometric
measurements for the elution experiments were made, the polarization of fluorescence in the central cornea of these nine animals was measured with the SOF as 9-2
J. TAARNH0J
116
describedby Herman, McLaren, and Brubaker( 1988). The ratio of bound fluorescein to unbound fluorescein, C,/C,, can be calculated from the observed polarization of fluorescence in the stroma, p (Brubaker et al., 1982).
c,/c, = m KP-PMPtl--Pn In this equation, p1is the polarization of fluorescence of unbound fluorescein, and pb is the polarization of fluorescence when it is bound to rabbit albumin. FJC, can be obtained from C&/C, and I,,/& as stated in method (ii). Important assumptions of this method are: (1) albumin is the principal substance which binds to and alters the polarization of fluorescence of fluorescein, (2) the viscosity of water in the stroma is insticient to cause polarization of fluorescence of unbound fluorescein, (3) few if any fluorescein binding sites on stromal albumin are occupied by competing substances, (4) the milieu of the stroma as regards polarization of fluorescence is very similar to the simple system employed for calibration, and (5) the concentration in the stroma of albumin-fluorescein binding sites is much greater than the concentration of fluorescein in the stroma. (c) Polarization offluorescence - (ii) Direct calibration of albumin concentration from polarization. The observed polarization can be used to measure the concentration of albumin directly from a calibration curve (Herman et al., 1988). A calibration curve of polarization versus concentration of albumin was made (Fig. 3). To mimic the binding of fluorescein in the cornea, rabbit aqueous humor was used as the calibrating medium to which rabbit albumin (fraction V) was added incrementally to produce solutions of increasing concentration up to 5 g dl-I. As more protein was added, the pH was observed to decline slightly and was titrated back to 7.59 with 1 N NaOH. The polarization of fluorescence
of each albumin solution was measured at room temperature and was plotted against the albumin concentration. The albumin concentration in the living cornea was determined from the in vivo measurements of polarization and the calibration curve. The corresponding F,/C, value was calculated as described under experiment 2. Unlike experiment 2, experiment 3 does not depend on complete elution of albumin since the polarization measurement was made in vivo. (d) Excitation spectrum - (i) Direct calibrution of quenching from spectral shift. When fluorescein is bound
to protein, the peak of excitation is shifted towards longer wavelengths (red shift). At the same time the intensity of steady-state fluorescence is reduced (quenching). With the SOSFthe excitation spectrum and fluorescence (F) of a known concentration of fluorescein (C) was measured in a phosphate buffered solution contained in a cuvette at pH 7.4. The same measurements were made in rabbit aqueous humor to which increasing amounts of rabbit albumin (fraction V) were added as described in experiment 3. The excitation wavelength was varied from 450 nm to 520 nm and the emission wavelength set at 540 nm. The red shift produced by albumin was calculated as described by McLaren, Taarnhoj and Brubaker (1988). The excitation curve for fluorescein in buffer was displaced along the wavelength axis until it provided a best fit determined by the method of least squares to the data points for fluorescein in albumin. Quenching was expressedas the ratio of the observed intensity of fluorescence to the unquenched intensity, the ratio F/C. A calibration curve for F/C (unitless) (Fig. 4) was constructed by plotting F/C vs. red shift (in nm). The red shift was measured in vivo with the same instrument in nine rabbit corneas stained with fluorescein. The calibration curve derived from rabbit aqueous humor was employed to convert red shift to c/C,. If the fluids of the stroma in vivo are similar to aqueous humor in their interaction with the fluI.0
./.-’
.
m
.
ET AL.
/
/
0.9 c, c
/
0.6
0.1
0.5 [Albumin]
I.0
5.0
(g dl-‘)
FIG. 3. Calibration curve for converting the measured polarization value to albumin concentration.
0
5
IO
15
Red shift (nm)
FIG. 4. Calibration curve for converting the measured red shift value to quenching.
QUENCHING
OF FLUORESCEIN
117
IN THE STROMA
5.0 -
When the stromal thickness is used, the corresponding values for F,/C, are 086 and O-82.
.t /.
i =6 -T T.E
I.0 -
/”
In experiment 2(b) the concentration of elutable protein in the cornea was 1.20 + 0.12 g dll’, of which 43 f 7% was albumin, giving an albumin concentration of 0.52 g dl-‘. The calculated ratios F,/C,. on the basis of the albumin concentration, were 0.82 f 0.02 (I,/& = 0.69) for the SOSFand 0.78 + 0.02 (I,/& = 0.61) for the SOF.
./’
0.5.i
e3 4
Experiment 2(b). Elution of Albumin.
de
/ O” 0.05
./* 7
0.01 I 0
I 5
I IO
I 15
Red shift (nm)
FIG.5. Calibrationcurve for convertingthe measuredred shift to albumin concentration. orescein-albumin system, then the F/C derived from the red shift in aqueous humor can be used as an estimate of quenching in vivo, F,/C,. (d) Excitation spectrum - (ii) CaIcwfationof albumin concentration from spectrul shift. The measurements of red shift in fluorescein-aqueous humor solutions containing known amounts of rabbit albumin mentioned above were also used to construct a calibration curve for albumin by plotting red shift against albumin concentration (Fig. 5). The albumin concentration in the living cornea could then be derived from the in vivo measurement of spectral shift. The FJC, ratio was then calculated from the albumin concentration so derived, as explained in experiment 2. 3. Results
Experiment I The results of experiment 1 are shown in Fig. 1 for the SOSF.The values are the means of two measurements. The curve was fitted by eye. Since the thickness of the rabbit cornea in this experiment was 0.3 7 k 0.02 ( + s.D.), the uncorrected SOSFreading underestimates the fluorescence of the rabbit cornea by a factor of 2.04. The corresponding value for the Fluorotron Master was 2.44. If fluorescein is largely located in the stroma rather than the epithelium, the correction should be based on stromal thickness. If a figure of 0.35 mm is used (Maurice and Watson, 1965), the two correction factors are 2.17 and 2.56. The results of the elution method are dependent on which thickness is used to determine this correction, the correct value probably lying somewhere between the two. Experiment 2(a). Elution of Fluorescein. In the elution experiment when the cornea1 thickness is used for the spatial correction, the F,/C, ratio was found to be 0.81 kO.24 (meanf SD.) for the Fluorotron Master and 0.78 f0.28 for the SOSF.
Experiment 2(c). Polarization of Fluorescence.
The polarization of fluorescence in experiment 2(c, ii) was found to be 0.21 f0.039 (unitless) in the center of the cornea. The FJC, value calculated from the polarization was 0.74 f 0.04. In Fig. 3 the calibration curve used in experiment 2(c, ii) is shown. When plotted on a log/log scale there is an almost linear relationship between polarization of fluorescence and concentration of albumin in the range of interest. The albumin concentration of the cornea, derived with the assistance of the calibration curve and the measured polarization, was 0.9 5 + 0.28 g dl-‘. The corresponding F,.C, value, calculated as explained under experiment 2(c), is 0.72 f0.03. Experiment 2(d). Excitation spectrum.
The two calibration curves used in experiment 2(d, i and ii) are shown in Figs 4 and 5. The red shift measured in the center of the cornea was 10.52 + 1.08 nm, and the corresponding &/C, value derived from the curve was 0.83 + 0.03. The red shift was also used to calculate the concentration of albumin in the central cornea. The albumin concentration by this method was 0.8 6 + 0.30 g cl-‘. This value was, in turn, used to calculate F,/C, as done in experiment 2(b). The calculated value was 0.79 + 0.02. 4. Discussion In the elution experiment we found a F,/C, ratio of 0.81-0.86 for the Fluorotron and 0.78-0.82 for the SOSF.If the fluorescence of fluorescein were the same in the stroma as in the buffered standard solutions, a ratio of 1.00 would be expected. The observed F,/C, ratio suggests that the molarity of fluorescein in the stroma is underestimated by approximately 15-20 % by these two instruments. It is noteworthy that the underestimate cannot be explained merely by the fact that the cornea is thin. A correction was applied to take into account its geometry. Had no geometric correction been applied, the FJC, ratio would have been much less, and we would have erroneously concluded that quenching of
J. TAARNHBJ
118
fluorescence was much greater than it actually is. What causesthe observedquenching ?For the reasons given below, the data are consistent with the idea that the phenomenon is due to the well-known interaction between fluorescein and albumin. When fluorescein interacts with serum albumin, its excitation and emission spectra shift (Delori, Castany and Webb, 1978), and the peak intensity of fluorescencediminishes. If fluorescence is being measured at fixed wavelengths near the excitation and emission peaks of the fluorophore, these effectswill reduce the intensity of the observed signal. The excitation and emission spectra obtained from fluorescein in the cornea in vivo are shifted and reduced in intensity quite like the spectra observed from dilute solutions of plasma in aqueous humor. Small shifts in pH near neutrality also alter the intensity of fluorescein’s fluorescence but hardly affect the excitation peak (Thomas et al., 1990). We believe the shifts observed in the living cornea are due primarily to interaction betweep fluorescein and albumin. When fluorescein binds to albumin, its emission is polarized (Laurence, 1952). In experiment 2(c), we observed polarization of fluorescence from the cornea but not the anterior chamber. We believe that the polarization is due primarily to the binding of fluorescein to albumin in the stroma. The albumin concentrations calculated from the different methods used in these experiments are somewhat higher than reported by Maurice and Watson (1965). They found a gradient of albumin concentration across the radial profile of the cornea. The limbus was approximately one-third the plasma (approximately 0.75 g dl-‘) and the center approximately one-tenth the plasma (approximately 0.25 g dl-‘). In the elution experiment, where the excised button would have included both central and peripheral portions, we calculated a concentration of 0.52 g dl-‘, in good agreement with the data of Maurice and Watson. In the polarization experiment, we obtained a figure of 0.95 g dl-‘. We do not know why this technique yielded a higher estimate of albumin. Perhaps the viscosity of water in the stroma is high enough to cause some polarization of unbound fluorescein. The spectral shift experiment also yielded a higher estimate of albumin of 0.86 g dl-I. Again, no reason is obvious unless the conditions in our calibrating medium differ significantly from those in the stroma.
ET AL.
Despite the variations in the concentration of albumin calculated by the different methods, the agreement in the calculated F,/C, is good. It would seemreasonable to make allowance for this quenching effect in the stroma when attempting to deduce molarity from fluorescence. In addition, we conclude that the principal reason for this quenching is the interaction of fluorescein and albumin in the stroma.
Acknowledgments
This work wassupportedin part by NIH grant EY00634, Researchto Prevent Blindness,Inc., New York, NY, the MayoFoundation,Rochester,MN, KronprinsessanMargaretha’s Arbetsngmd fijr Synskadade,Sweden, and Carmen och Bertil Regners Fond fijr forskning inom omrddet i)gonsjukdomar, Sweden.
References Araie, M. and Maurice, D. M. (1985). A re-evaluation of cornea1endothelial permeability to fluorescein. Exp. Eye Res. 41. 383-90. Brubaker, R. F. (1982). The flow of aqueous humor in the human eye. Trans. Am. Ophthahnol. Sot. 80, 391474. Brubaker, R. F., Penniston, J. T., Grotte, D. A. and Nagataki, S. (1982). Measurement of fluorescein binding in human plasma using fluorescence polarization. Arch. OphthaZmoZ. 100, 625-30. Delori, F. C.. Castany, M. A. and Webb, R. H. (1978). Fluorescence characteristics of sodium fluorescein in plasma and whole blood. Exp. Eye Res. 27, 417-2 5. Herman, D. C., McLaren, J. W. and Brubaker. R. F. (1988). A method of determining concentration of albumin in the living eye. Invest. Ophthalmol. Iris. Sci. 29, 133-7. Jones, R. F. and Maurice, D. M. (1966). New methods of measuring the rate of aqueous flow in man with fluorescein. Exp. Eye Res. 5. 208-20. Laurence, D. J. R. (1952). A study of the adsorption of dyes on bovine serum albumin by the method of polarization of fluorescence. Biochem.I. 51, 168-80. McLaren, J. W. and Brubaker. R. F. (1988). A scanning ocular spectrofluorophotometer. Invest. Ophthalmol. Vis. Sci. 29, 1285-93. McLaren, J. W., TaarnhBj, J. and Brubaker, R. F. (1988). Spectral changes of fluorescein in the anterior segment. Invest. OphthalmoI. Vis. Sci. 29 (ARVO Suppl.). 88. Maurice, D. M. and Watson, P. G. (1965). The distribution and movement of serum albumin in the cornea. Exp. Eye Res.4, 355-63. Thomas, J. V., Brimijoin, M. R., Neault, T. R. and Brubaker, R. F. (1990). The fluorescent indicator pyranine is suitable for measuring stromal and camera1pH in vivo. Exp. Eye Res. 50. 241-9.
