Fluorescein Binding to Normal Human Serum Proteins
`Demonstrated by Equilibrium Dialysis
`
`Weiye Li, MD, John H. Rockey, MD, PhD
`
`\s=b\The binding of fluorescein to normal
`human serum proteins in a physiologic
`solvent at 37 \s=deg\Cwas measured by equilib-
`rium dialysis. Human serum contained
`3.28 x 10-3M concentration fluorescein-
`binding sites, with an average association
`constant at 37 \s=deg\Cof 0.54 X 104M -1. The
`percentage of total fluorescein bound by
`human serum proteins ranged from 83%
`to 53% when the total
`fluorescein con-
`centration ranged from 6.9 X 10-5
`to
`6.2 X 10-3M.
`(Arch Ophthalmol 1982;100:484-487)
`
`TTUuorescein has been widely used as
`-* a tracer to test the permeability of
`for
`the blood-ocular barrier
`two
`decades.1-2 Although several studies
`have been carried out on the binding
`of fluorescein to isolated serum albu¬
`min, whether fluorescein is substan¬
`tially bound to any plasma protein
`after
`intravenous injection in man
`remains controversial.2"7
`Laurence5 initially demonstrated
`isolated bovine serum albu¬
`that
`min showed concentration-dependent
`binding of fluorescein by fluorescence
`polarization. He also reported that
`fluorescein fluores¬
`the intensity of
`cence dropped markedly concomitant
`with binding to bovine albumin (fluo-
`
`Accepted for publication April 29, 1981.
`From the Scheie Eye Institute, Department of
`Ophthalmology, University of Pennsylvania
`Medical School, Philadelphia.
`Reprint
`requests to Scheie Eye Institute,
`Myrin Circle, 51 N 39th St, Philadelphia, PA
`19104 (Dr Rockey).
`
`rescence quenching), and the fluores¬
`cein absorption spectrum was red
`shifted.5 Both observations are fur¬
`ther evidence that
`fluorescein does
`bind to bovine albumin. Andersson et
`al6 subsequently demonstrated fluo¬
`rescein binding to isolated bovine
`albumin by equilibrium dialysis at 5
`to 30 °C.
`Recently, however,
`Ianacone et al7
`have studied the binding of radioac¬
`tive fluorescein using polyacrylamide
`gel electrophoresis and gel filtration
`and have questioned whether fluores¬
`cein is bound in significant quantity
`to any plasma protein under physio¬
`logic conditions.
`the character of
`A knowledge of
`fluorescein binding to blood proteins
`under physiologic conditions is impor¬
`tant for an understanding of quanti¬
`fluorescence measurements,
`tative
`both in clinical diagnosis and in exper¬
`imental research. We therefore have
`reexamined this question using equi¬
`librium dialysis, a classic method
`firmly established on thermodynamic
`theory that
`is well suited to study
`ligand binding to proteins.814
`small
`Fluorescein binding to human serum
`proteins at 37 °C in a physiologic
`solvent was measured instead of bind¬
`ing to isolated bovine albumin at low
`temperatures. Substantial concentra¬
`tion-dependent binding of low affinity
`was observed throughout a wide range
`of fluorescein concentrations.
`MATERIALS AND METHODS
`Fresh blood was obtained from normal
`male volunteers. Serum was used imme-
`
`diately or stored at 4 °C until use. Serum
`proteins were diluted tenfold with Tyrode's
`solution (pH 7.4) for equilibrium dialysis.
`Fluorescein sodium (Funduscein) was
`diluted with Tyrode's solution to give con¬
`2,500 mg/L
`12.5
`centrations
`of
`to
`(0.33 X10"4 to 0.66 X10-2M).
`Equilibrium dialysis was performed
`with specially designed cells, each of which
`comprised two identical closed compart¬
`ments separated by a 23-mm (diameter)
`cellulose-disk dialysis membrane.1013 The
`cells were immersed in a constant-temper¬
`ature water bath at 37 °C and rotated at a
`speed of 5 rpm. Three groups of experi¬
`ments were carried out. In group 1, a given
`fluorescein solution (eg,
`concentration of
`12.5 mg/L) was dialyzed against Tyrode's
`solution for variations of dialysis time
`from 15 minutes to 50 hours to determine
`the time required to attain equilibrium of
`ligand diffusion across the dialysis mem¬
`In group 2, a series of concentra¬
`brane.
`tions of fluorescein were dialyzed against
`Tyrode's solution alone to determine the
`amount of dye bound by the cellulose mem¬
`brane as a function of the free fluorescein
`concentration at equilibrium. In group 3,
`the series of concentrations of fluorescein
`solutions were dialyzed against
`the 1:10
`diluted serum. The dialysis was terminated
`after 50 hours in experimental groups 2
`and 3. All cells in the three groups were
`run in duplicate. The optical density of the
`fluorescein in the nonprotein compartment
`was measured at 490 nm with a spectro-
`photometer (Zeiss PMQ II) using a 10-mm
`path-length cell. The molar extinction
`coefficient (e) of fluorescein at 490 nm was
`determined to be 8.6 X 10". The data of the
`group 3 experiments (dialysis of fluores¬
`cein against serum proteins) were cor¬
`rected for ligand binding by the cellulose
`dialysis membranes with the data of
`experimental
`2. Protein-bound
`group
`
`Downloaded From: http://archopht.jamanetwork.com/ by a University of Kansas HQ User on 09/19/2016
`
`CIPLA EXHIBIT 1007
`Page 1 of 4
`
`

`

`100 -
`
`Time, hr
`Fig 1.—Time required to attain equilibrium of fluorescein diffusion
`across dialysis membrane in absence of serum proteins. Percent of
`total fluorescein present in two compartments of equilibrium dialysis
`cell (upper curve, initial fluorescein solution side; lower curve, initial
`solvent side) is plotted against dialysis time.
`
`Log F,
`Fig 2.—Concentration-dependent binding of fluorescein by equilib¬
`rium dialysis cell membrane, determined in absence of serum
`proteins. Logarithm^ of membrane-bound fluorescein (log Fmb)
`is
`plotted as function of the logarithm^ of free fluorescein concentra¬
`tion (log F,) at equilibrium. Data were fitted to straight line by method
`of least squares.
`
`—
`
`Binding of fluorescein by human serum proteins at 37 °C,
`Fig 3.
`measured by equilibrium dialysis. Logarithm^ of protein-bound
`fluorescein concentration (corrected for fluorescein bound to the
`dialysis membrane), log F„,
`is plotted as function of logarithm,0 of
`free fluorescein concentration, log F,, at equilibrium. Experimental
`curve (solid line), measured with serum protein diluted 1:10 in
`Tyrode's solution, has been corrected for dilution (binding by
`undiluted serum proteins) in upper curve (broken line).
`
`—
`
`Equilibrium dialysis data of Fig 3 presented in terms of
`Fig 4.
`is amount of fluorescein bound by protein
`Scatchard plot where r
`least squares
`and c is free fluorescein concentration. Method of
`was used to obtain best-fit straight line.
`
`Downloaded From: http://archopht.jamanetwork.com/ by a University of Kansas HQ User on 09/19/2016
`
`CIPLA EXHIBIT 1007
`Page 2 of 4
`
`

`

`fluorescein (Fb) was obtained from the
`following relationship:
`Fmb
`Ff
`Fb = F,
`where F, is the total concentration of fluo¬
`rescein, Fr is the concentration of free fluo¬
`rescein at equilibrium, and Fmb is the fluo¬
`rescein bound by the dialysis cell mem¬
`brane.
`
`—
`
`-
`
`RESULTS
`
`The time required to obtain equilib¬
`fluorescein diffusion across
`rium of
`the dialysis membrane separating the
`two compartments of the equilibrium
`dialysis cell, determined in the group
`1 experiments, was 50 hours (Fig 1).
`Therefore, all equilibrium dialysis
`fluorescein against
`experiments of
`serum proteins were continued for 50
`hours or longer. The amount of fluo¬
`rescein bound by the dialysis mem¬
`brane, plotted against
`the free fluo¬
`rescein concentration at equilibrium,
`determined in the group 2 experi¬
`ments, is shown in Fig 2. The binding
`to the cellulose membrane was con¬
`centration dependent. This curve was
`in calculating protein-bound
`used
`fluorescein (Fb) from the experimen¬
`tal group 3 results, where the total
`fluorescein (Ft) and free fluorescein
`(Ff) concentrations at equilibrium
`were known.
`In Fig 3, the logarithm of the pro¬
`tein-bound fluorescein concentration
`(logio Fb), corrected for the dye bound
`to the artificial membrane, is plotted
`that of
`the free fluorescein
`against
`the
`concentration. The flatness of
`the highest
`binding curve at
`free
`fluorescein concentrations indicated
`that protein fluorescein-binding site
`saturation had been obtained (Fig 3).
`The experimental data were deter¬
`mined using tenfold-diluted human
`serum proteins. To approximate the in
`the second (upper)
`vivo situation,
`curve of Fig 3 shows the data replot-
`ted for binding to undiluted serum
`proteins.
`simply
`This
`is
`curve
`obtained because the free fluorescein
`concentration (Ff) determines only
`the fraction of available binding sites
`occupied by fluorescein; the amount of
`fluorescein bound at a given free
`fluorescein concentration is a simple
`linear function of the concentration of
`available binding sites (eg, proteins)
`(see below, Fig 4).
`The maximal and minimal fractions
`fluorescein bound by
`the total
`of
`undiluted serum proteins at 37 °C
`throughout the range of free fluores¬
`cein concentrations examined are giv¬
`en in the Table. The molar concentra¬
`fluorescein-binding sites in
`tion of
`undiluted human serum, determined
`
`Percentage of Total Fluorescein Bound by Undiluted Serum Proteins at Lowest
`and Highest Free Fluorescein Concentrations Measured
`
`Fluorescein, M
`
`Ratio F„/F,
`Free
`Bound
`Total
`(F,)_(FO_(Fj)_X 100, %
`6.9 X 105_1.2 X 10~5_5.7 10~5_83
`3.3 10"3
`6.2 X 10"3
`53
`2.9 X 10"3
`
`from Fig 3, was 3.28 X 10"3M.
`Binding data of Fig 3 were replotted
`in Fig 4 in the form of a Scatchard
`plot,1112·14 r/c vs r, where r is the
`amount of protein-bound fluorescein
`free
`is the concentration of
`and c
`fluorescein. The maximum value for
`r (n) was arbitrarily chosen as 2, so
`that the association constant (K0) is
`given by the reciprocal of the free-dye
`(half of
`the
`concentration at r = 1
`the proteins were
`binding sites of
`occupied by fluorescein), as obtained
`from the relationship K0 = r(n — r)_1
`c-i io,i4 The average intrinsic associa¬
`(K0) of human serum
`tion constant
`proteins in a physiologic solvent at 37
`°C for fluorescein was 0.54 X WM"1.
`
`COMMENT
`
`The present results of equilibrium
`dialysis demonstrate that in a physio¬
`logic solvent at 37 °C, substantial
`fluorescein to human
`binding of
`serum proteins occurs throughout a
`wide range of fluorescein concentra¬
`tions, even though the association is
`of low affinity. The appropriateness of
`equilibrium dialysis for evaluating
`fluorescein binding under physiologic
`conditions has been questioned by
`Ianacone et al7 because of
`the long
`time required to obtain equilibrium.
`This misunderstanding has occurred
`because the time required for free
`ligand to reach equilibrium across the
`artificial dialysis membrane (eg, 50
`hours) has not been differentiated
`from the forward rate constant for
`fluorescein-protein association,
`ob¬
`tained when fluorescein is added
`directly to a protein solution, which is
`limited only by the rate of diffusion of
`fluorescein in free solution.812 The
`semipermeable
`dialysis
`artificial
`membrane is only used so that an
`unambiguous measure of
`the free
`ligand concentration may be obtained.
`The technique is entirely appropriate
`for measurements of equilibrium
`thermodynamic parameters of asso¬
`rapidly interacting sys¬
`ciation of
`tems.8"14
`filtration and polyacrylamide
`Gel
`gel electrophoresis,7 however, may not
`be appropriate techniques to measure
`of
`associations
`fluorescein-protein
`
`low affinity. In both techniques, fluo¬
`rescein-protein complexes are contin¬
`uously being separated from free
`fluorescein because of their differen¬
`tial migration rates. This factor is
`filtration,
`particularly true in gel
`where fluorescein-protein complexes,
`because of their larger size, are con¬
`tinuously being transferred into sol¬
`vent that contains no free fluorescein.
`This process is equivalent to dialyzing
`complexes
`fluorescein-protein
`the
`against an infinite volume of solvent
`and is an effective way to dissociate
`ligand-protein complexes of low affin¬
`ity. These techniques will not measure
`the extent of binding of fluorescein to
`blood proteins that occurs in the pres¬
`free
`ence of high concentrations of
`fluorescein.
`fluorescein-
`The concentration of
`binding sites in serum was deter¬
`mined to be 3.28 X 10-3M. The normal
`concentration of albumin in human
`Isolated
`serum is
`0.65 X 10"3M.15
`serum albumin has
`been
`bovine
`reported to have three binding sites
`per molecule.5 The ratio for bound
`fluorescein to human serum albumin
`of 5 (3.28:0.65) indicates either that
`human serum albumin has five fluo¬
`rescein-binding sites per molecule or,
`more likely, that other serum proteins
`low-affinity IgG immunoglobu-
`(eg,
`lins) also bind fluorescein under phys¬
`iologic conditions. A knowledge of the
`fluorescein binding by blood proteins
`of substantially different size and dif¬
`fusion rates would be of importance
`for a complete picture in quantitative
`studies of blood-ocular permeability
`changes.
`A knowledge of the extent of fluo¬
`rescein binding to proteins at differ¬
`free fluorescein concentrations
`ent
`also is necessary when blood fluores¬
`cein levels are measured fluorometri-
`cally, since these data must be cor¬
`rected for the fluorescence quenching
`the bound fluorescein if a true
`of
`measure of total fluorescein content is
`to be obtained by this method.
`The finding that a substantial frac¬
`tion of the total injected fluorescein is
`bound at low affinity by human blood
`proteins in a physiologic solvent at
`37 °C is consistent with clinical obser¬
`vations. The laminar flow seen in reti-
`
`Downloaded From: http://archopht.jamanetwork.com/ by a University of Kansas HQ User on 09/19/2016
`
`CIPLA EXHIBIT 1007
`Page 3 of 4
`
`

`

`nal vessels2·16 in all likelihood reflects
`fluorescein bound to proteins, as the
`protein-bound fluorescein diffusion
`rate will be much slower than that of
`free fluorescein. The rapid changes in
`the volumes of distribution of fluores¬
`ob¬
`cein and 131I-labeled albumin,
`served when both were injected simul¬
`taneously,16 would reflect
`the low
`fluorescein
`association constant of
`
`binding.
`A complete understanding of quan¬
`titative fluorescein angiography of
`the retina and vitreous fluorometry
`will require a detailed knowledge of
`the contribution of fluorescein bind¬
`ing by blood constituents during the
`changing concentration ratios occur¬
`ring intravascularly and extravascu-
`larly after the initial injection of the
`
`fluorescein. The present
`bolus of
`experimental results may be used to
`calculate the contribution of fluores¬
`cein binding by human serum proteins
`in such studies.
`
`This study was supported by an unrestricted
`grant from Research to Prevent Blindness, Ine,
`New York, and by
`the Harry and Edith
`Hubschman Research Fund.
`
`References
`
`1. Novotmy HR, Alvis DL: A method of photo-
`graphing fluorescence in circulating blood in the
`human retina. Circulation 1961;24:82-86.
`2. Wessing A: Biochemical properties of fluo-
`rescein, in Fluorescein Angiography of the Reti-
`na. St Louis, CV Mosby Co, 1969, pp 13-14.
`3. Lange K, Boyd LJ: Use of
`fluorescein
`method in establishment of diagnosis and prog-
`nosis of peripheral vascular diseases. Arch
`Intern Med 1944;74:175-184.
`4. Hodge JY, Dollery CT: Retinal soft exu-
`dates: A clinical study by colour and fluorescence
`photography. Q J Med 1964;33:117-130.
`5. Laurence DJR: A study of the absorption of
`dyes on bovine serum albumin by the method of
`polarization of
`fluorescence. Biochem J 1952;
`51:168-180.
`6. Andersson LO, Rehnstr\l=o"\mA, Eaker DL:
`Studies on 'non-specific' binding: The nature of
`
`the binding of fluorescein to bovine serum albu-
`min. Eur J Biochem 1971;20:371-380.
`Ianacone DC, Felberg NT, Federman JL:
`7.
`Tritiated fluorescein binding to normal human
`plasma proteins. Arch Ophthalmol 1980;98:1643\x=req-\
`1645.
`8. Karush F, Sonenberg M:
`Interaction of
`homologous alkyl sulfates with bovine serum
`albumin. J Am Chem Soc 1949;71:1369-1376.
`9. Haschemeyer RH, Haschemeyer AE: Meth-
`od based on charge or charge distribution,
`in
`Haschemeyer RH, Haschemeyer AE (eds): Pro-
`teins: A Guide to Study by Physical and Chemical
`Methods. New York, John Wiley & Sons Inc, 1973,
`pp 254-268.
`10. Karush F: The interaction of purified anti-
`body with optically isomeric haptens. J Am Chem
`Soc 1956;78:5519-5526.
`11. Klotz I: The application of the law of mass
`
`action to binding by proteins: Interactions with
`calcium. Arch Biochem 1946;9:109-117.
`12. Scatchard G, Scheinberg IH, Armstrong
`SH: Physical chemistry of protein solutions: IV.
`The combination of human serum albumin with
`chloride ion. J Am Chem Soc 1950;72:535-540.
`13. Rockey JH: Equine antihapten antibody:
`The subunits and fragments of anti-\g=b\-lactoside
`antibody. J Exp Med 1967;125:249-275.
`14. Wu WH, Rockey JH: Antivasopressin anti-
`body: Characterization of high-affinity rabbit
`antibody with limited association constant heter-
`ogeneity. Biochemistry 1969;8:2719-2728.
`15. Guyton AC (ed): Textbook ofMedical Phys-
`iology. Philadelphia, WB Saunders Co, 1976, pp
`393-394.
`16. Dollery CT, Hodge JV, Engel M: Studies of
`the retinal circulation with fluorescein. Br Med J
`1962;2:1210-1215.
`
`Downloaded From: http://archopht.jamanetwork.com/ by a University of Kansas HQ User on 09/19/2016
`
`CIPLA EXHIBIT 1007
`Page 4 of 4
`
`