Jounsal ok Clinical InvestigationVol. 46, No. 5, 1967

Estimation of Rapidly Exchangeable Cellular Thyroxinefrom the Plasma Disappearance Curves of Simul-

taneously Administered Thyroxine-'3'I andAlbumin- 25I *

JACK H. OPPENHEIMER,t GERALDBERNSTEIN,4 ANDJULIAN HASEN-§

(from the Endocrine Research Laboratory, Division of Medicine, Montefiore Hospital andMedical Center, and the Department of Medicine, Albert Einstein College of Medicine,

Bronx, N. Y.)

Summary. A mathematical analysis of the plasma disappearance curves ofsimultaneously injected thyroxine-131I and albumin-1251 allows the develop-ment of simple formulas for estimating the pool size and transfer kinetics ofrapidly exchangeable intracellular thyroxine in man. Evidence is presentedthat the early distribution kinetics of albumin-125I can be used to represent theexpansion of the thyroxine-13I-plasma protein complex into the extracellularcompartment. Calculations indicate that approximately 37% of total bodyextrathyroidal thyroxine is within such exchangeable tissue stores. Theaverage cellular clearance of thyroxine is 42.7 ml per minute, a value far inexcess of the metabolic clearance of this hormone. Results of external mea-surements over the hepatic area and studies involving hepatic biopsies indi-cate that the liver is an important but probably not the exclusive componentof the intracellular compartment. The partition of thyroxine between cellu-lar and extracellular compartments is determined by the balance of tissueand plasma protein binding factors. The fractional transfer constants areinversely related to the strength of binding of each compartment and directlyproportional to the permeability characteristic of the hypothetical membraneseparating compartments. Appropriate numerical values for these factorsare assigned. An increased fractional entrance of thyroxine-""1I into the cel-lular compartment was noted in a patient with congenital decrease in themaximal binding capacity of thyroxine-binding globulin and in three patientsafter the infusion of 5,5-diphenylhydantoin. Decreased intracellular spaceand impaired permeability characteristics were observed in five patients withhepatic disease. Studies of the rate of entrance of thyroxine-'81I and albu-min-125I into the pleural effusion of a patient with congestive heart failuresuggested that transcapillary passage of thyroxine independent of its bindingprotein is not a predominant factor in the total distribution kinetics ofthyroxine-'31I. The thesis is advanced that the distribution of thyroxine,

*Submitted for publication July 18, 1966; accepted t Career Scientist of the Health Research Council ofJanuary 12, 1967. New York City.

Supported by U. S. Public Health Service grant NB- Address requests for reprints to Dr. Jack H. Oppen-03000 and contract U-1238 from the Health Research heimer, Endocrine Research Laboratory, Division ofCouncil of New York City. Medicine, Montefiore Hospital and Medical Center, 111

Presented in part at the Fifty-eighth Annual Meeting E.\210th St., Bronx, N. Y. 10467.of the American Society for Clinical Investigation, May 4: Postdoctoral fellow of the American Cancer Society.2, 1966, Atlantic City, N. J., and abstracted in J. clin. § U. S. Public Health Service postdoctoral researchInvest. 1966, 45, 1054. fellow.

762

ESTIMATION OF INTRACELLULARTHYROXINE

both within the extracellular compartment and between the extracellular andintracellular compartments, is accomplished largely by the carrier protein andthe direct transfer of thyroxine from one binding site to another. The con-cept of free thyroxine is reassessed in terms of this formulation.

Introduction

Albert and Keating (1) first demonstratedthat a variety of tissues in the rat rapidly accu-

mulate radioactivity after the intravenous injec-tion of thyroxine-_311. By external radioactivemeasurements in man, Myant and Pochin (2)showed that a significant fraction of administeredthyroxine-131I is concentrated in the hepatic area.

Hazelrig (3), utilizing rat liver perfusion studiesperformed by Flock and co-workers, indicatedthat hepatic thyroxine is in rapid equilibrium withhormone in blood. The reversible transfer ofthyroxine between liver and blood in the isolatedrat liver preparation has recently been consideredin detail by Gorman, Flock, Owen, and Paris (4).Pochin (5) and subsequently Cavalieri and Searle(6, 7) postulated a similar relationship in man

on the basis of an analysis of the kinetics of thy-roxine-131I accumulation by the liver and thesimultaneous disappearance of radioactivity fromplasma.

In the present communication, the distributionkinetics of thyroxine-131I are analyzed by com-

paring over a 4-hour period the plasma disap-pearance curves of simultaneously injected thy-roxine- 13lI and albumin-'25I. A model system isproposed that allows kinetic description of theexchange between thyroxine in the instantaneousvolume of distribution of albumin-125I and thy-roxine outside this volume. The compartment

comprising thyroxine outside the instantaneousvolume of albumin-125I distribution can be identi-fied, as a first approximation, with the intracellu-lar compartment of exchangeable hormone. Therole of cellular and extracellular factors in deter-mining the partition of thyroxine between thesecompartments is analyzed under a variety ofphysiological and pathological settings.

Methods

For the acute distribution studies 5 to 10 ,uc of thyrox-ine-mI (SA 30 to 40 mc per mg) was mixed with an

approximately equal amount of albumin-w'I (SA 0.25 to

1 ,uc per mg) in a solution of 1% human serum albumin.'Both thyroxine-'I and albumin-'I were obtained com-mercially 2 and utilized without further purification.The solution containing the two tracer materials was in-jected into an antecubital vein through the rubber seg-ment of an intravenous infusion set. During the fol-lowing 4 hours, serial blood samples were obtained fromthe contralateral arm through an indwelling catheter.Approximately 6 ml of blood was obtained at each timeand anticoagulated with heparin. Plasma samples wereassayed for 'I and '5I in a Packard autogamma spec-trometer as previously described (8). Since we foundthat over 95% of 1'1I and 'I was precipitable with tri-chloroacetic acid (TCA), whole plasma was assayedfor radioactivity rather than TCA precipitate unlessotherwise stated. The simplifying assumption was madein our calculations that all radioactivity representedeither thyroxine-'I or albumin-'I. Appropriate dilu-tions of the administered dose were assayed simultane-ously, and the plasma concentration of radioactivity wasexpressed as a per cent of the administered dose per literplasma. Calculations and the underlying theoreticalconsiderations are considered in detail in the followingsection.

In a single study, the rate of disappearance of thyrox-ine-lI and albumin-'I from blood was compared to therate of entrance of these materials into the pleural ef-fusion of a 65-year-old man with congestive heart failure.Ten-ml samples of pleural fluid were withdrawn throughan indwelling catheter at the same time that plasmasamples were obtained from the antecubital vein. Sinceradioactive iodide existing as a contaminant in the radio-active preparation would tend to diffuse rapidly and thuscontribute disproportionately to the radioactivity in theeffusion in the early phase of the study, all plasma andpleural effusion samples were precipitated with equalvolumes of 20%o TCA and washed twice with 1% TCAbefore radioactive assay.

The calculated rate of intracellular accumulation ofthyroxine was compared to the rate at which the liverconcentrated radioactivity. Fifty to 100 Ac thyroxine-''Iwas injected intravenously into four patients with normalliver function. Plasma samples were obtained at fre-quent intervals. A collimated scintillation probe (2-inchsodium iodide crystal) was placed in direct skin contactat the center of hepatic dullness as determined by manualpercussion. The external rate was determined over 1-minute intervals for the first 30 minutes after the in-jection, and for progressively longer intervals thereafter.The radioactivity detected by the probe can be consideredto rise from two sources, hepatic parenchyma and blood

1Albumisol, Merck Sharp & Dohme, West Point, Pa.2 Abbott Laboratories, North Chicago, Ill.

763

OPPENHEIMER,BERNSTEIN, AND HASEN

(intercostal vessels, thoracic structures, and hepatic ves-

sels). Thus,

Ht =--Ct - Bt E[1]where Ht = counts per minute emanating from the he-patic parenchyma, Ct = counts per minute detected by theprobe from all sources, and Bt = counts per minute origi-nating from blood. Weassumed that Bt = Bo (Tt/To),where Bo is the contribution from blood at time t = 0, andTt and To are the concentrations of thyroxine-'1I inplasma at times t and 0, respectively. Since Bo can be ap-

proximated from the initial counting rate over the he-patic area and since Tt and To can be determined by mea-

surement of plasma radioactivity, the term Bt can be eval-uated and Equation 1 solved for Ht.

The effect of 5,5-diphenylhydantoin $ on the acute dis-tribution kinetics of thyroxine-"31I was evaluated in threepatients with central nervous system disease. A com-

bined dose of thyroxine-'I and albumin-'5I was injectedin the standard fashion described above. Three to fourhours after injection of the isotopes, 150 mg of sodiumdiphenylhydantoin dissolved in 5 ml of the diluent pro-

vided by the manufacturer (40% propylene glycol and10% ethanol adjusted to pH 12) was injected over a

3-minute interval into the rubber segment of a runningintravenous infusion set. Samples were obtained froman indwelling catheter in an antecubital vein of the con-

tralateral arm. During the infusion, blood pressure,

pulse, and electrocardiogram were monitored. No ad-verse reactions were encountered.

The distribution kinetics of thyroxine-binding preal-bumin (TBPA) and albumin were compared after thesimultaneous injection of 10 ,tc TBPA-'311 and 10 ,uc al-bumin-'"I. TBPA was isolated from whole serum as

previously described (9). After iodination with 131I bythe method of Greenwood, Hunter, and Glover (10), thepreparation was further purified by starch gel electro-phoresis (11).

To determine whether thyroxine-binding proteins are

present in the soluble fraction of liver homogenates, we

performed the following experiment on two occasions.Portions of diagnostic liver biopsies obtained at the timeof laparotomy were weighed, homogenized in 10 vol of 0.1Mphysiologic saline (assuming a tissue density of 1), andcentrifuged at 100,000 X g in a Spinco preparative ultra-centrifuge for 60 minutes. Serum samples were also ob-tained from the same patients at the time of surgery.

Serum was diluted in saline to the same extent as the liverbiopsies. Radioactive thyroxine (0.03 sAg per ml solution)was added to the diluted serum, to the supernatant of theliver homogenate, to a mixture of equal parts of dilutedserum and supernatant liver homogenate, and to bufferalone. All preparations were then subjected to conven-

tional paper electrophoresis (glycine acetate, pH 8.6), andthe distribution of radioactivity was assessed in a paper

strip scanner (12).To obtain a direct estimate of hepatic thyroxine up-

take, we gave four patients scheduled to undergo abdomi-nal laparotomy a combined dose of thyroxine-"31I and al-

3 Dilantin, Parke, Davis, Detroit, Mich.

bumin-'5I 3 to 4 hours before the anticipated time ofabdominal exposure. Serial plasma samples were ob-tained and analyzed in the standard manner. Portions ofdiagnostic biopsies obtained at the time of laparotomywere lightly blotted and weighed. The contribution ofthyroxine-'I in residual blood in the tissue specimen tothe observed 'I counting rate could be assessed from thecounting rate of albumin-"uI in the tissue and the ratio of"I to "I counting rates in a peripheral plasma sample

obtained at the time of the biopsy. Thus, the net tissueuptake of thyroxine-"~I could be calculated from thedifference in total counting rate of 181I of the sample andthe contribution from residual blood within the specimen.

Theoretical analysis of the acute distribution kinetics ofthyroxine-`L I. As a first approximation, the plasma dis-appearance curves of thyroxine-"31I and albumin-l'I canbe analyzed without reference to the effects of metabolictransformation of either substance for the first 4 hoursafter intravenous injection. If we assume single com-partmental kinetics and a normal half-life for thyroxineand albumin, less than 2% of the administered thyroxineand less than 1% of the administered albumin will becatabolized in 4 hours.4 Thus, the volumes of distribu-tion of thyroxine-'31I and of albumin-"2I at any time tduring the first 4 hours after the injection of the tracerscan be obtained from the reciprocal of their plasmaconcentrations.

In the Results section experimental data will be pre-sented to substantiate the following assumptions: 1) Thedistribution kinetics of the thyroxine-'=I-plasma proteincomplex can be represented by those of albumin-'I. 2)Mixing of thyroxine in the extracellular compartment islargely governed by the mixing of the carrier protein.3) The rapid escape of thyroxine-"1'I from the albumindistribution space after intravenous injection is due pre-dominantly to the intracellular accumulation of tracer.

Nt, the per cent of administered thyroxine-'I outsidethe albumin distribution space at any time t during thefirst 4 hours after injection, will be the product of theinstantaneous concentration of thyroxine-'I (Tt) ex-pressed as per cent of the dose per liter and the differ-ences in the volumes of distribution of thyroxine-'I andalbumin-"I expressed in liters (VT - VA). Since VT =

(100/Tt) and VA = (100/at), where at is the instan-taneous concentration of albumin-'I, it follows that

Nt = 100Tt[(1/Tt) - (1/at)] = 100 E1 - (Tt/at)]. [2]Analysis of the acute distribution kinetics of thyrox-

ine-19'I and albumin-'2I is complicated by the finding thatthe initial distribution volume of thyroxine is frequentlylarger than that of albumin as determined by extra-

4The urinary excretion of "~'I was measured in fourpatients after the intravenous injection of thyroxine-"sI.Thyroidal accumulation of '1I was blocked by the oraladministration of Lugol's solution. The average 4-hoururinary excretion was 3%o of the dose. Although thisvalue is higher than the calculated metabolic transforma-tion, the discrepancy can be attributed to the preferentialearly excretion of iodide-"3I contaminating the adminis-tered dose.

764

ESTIMATION OF INTRACELLULAR THYROXINE

polation to t = 0. It is unclear whether this discrepancyis the artifactual result of inadequacy in the extrapola-tion technique and the difficulties in frequent early plasmasampling or whether there is a true increase in the ini-tial thyroxine distribution volume.5 For the purposesof simplicity we have assumed, somewhat arbitrarily,that the initial differences in distribution volumes be-tween albumin and thyroxine are real but do not repre-sent the effects of intracellular accumulation. Thus, wehave defined It, the per cent of administered thyroxine-'3"Iaccumulated in the cellular compartment, as It = Nt - No,where No represents the initial value of Nt as determinedby extrapolation. If we substitute into Equation 2,

It = 100[(To/ao) - (Tt/at)]. [3]The per cent of thyroxine remaining in the extracellularcompartment at time t, Et, will then be 100 -It. Al-ternative and probably equally justified interpretations ofthe initial discrepancy would lead to results substantiallysimilar to those presented here since the volume differ-ence under consideration is relatively small (less than5%o).

Curves illustrating It as a function of time are presentedin the Results. It can be seen that It achieves a plateauvalue between 3 and 4 hours after the administration ofthe isotopes. The shape of the intracellular accumula-tion curve suggests that it can be described by an ex-ponential growth curve of the general form

t= Imx(1 - ext). [4]The adequacy of this function in describing the functionIt can be graphically tested. If we rearrange the termsof Equation 4,

Imax - It = Imaxe-)t. [5]

The linearity of (Imax - It) as a function of t on a semi-logarithmic plot with an intercept of Im.x on the ordinateis also illustrated in the following section. In all sub-jects and patients that we have studied to date Equation4 appears to describe the observed intracellular accumu-lation curve within the limits of experimental error.

The function described by Equation 4 is known torepresent the accumulation of isotope in one of thecompartments of a closed two compartmental system asis illustrated in Figure 1. The properties of a closedtwo compartmental model have previously been discussedby Solomon (14). Isotope is injected into compartment

5 One possible source of the discrepancy between theextrapolated initial distribution volume of thyroxine-"sIand albumin-'lI is the presence of contaminating radio-active iodide and of radioactive thyronines (3,5,3'-triiodo-thyronine and 3,5',3'-triiodothyronine) in the thyroxinepreparations used (13). Such contaminants, being lesstightly bound to plasma proteins than thyroxine, mayalso be responsible for a slight overestimation of thefractional rate of exit of thyroxine from the plasmacompartment. The significance of these contaminants isfurther discussed in connection with the pleural effusionstudy.

E at t = 0. The defining equations of the system are

(dEt/dt) = -(kE)Et + (kl)lt, and

(dlt/dt) = -(kI)lt + (kE)Et,

[6]

[7]where Et and It represent the per cent of the inj ecteddose within compartments E and I at any time t, ks, isthe fractional transfer constant from compartment E tocompartment I, and k1 is the fractional transfer con-stant in the opposite direction.

Solution of Equations 6 and 7 will indicate that at anytime t

It = [lO0kE/(ki + kE)][1 -e (kl+kE)t],

andEt=100ki l1OOkE E-k~F t= ki + kE + k + kE[eIE]

Thus, at t = a, It will reach a maximal value of

Imox = lOOkE/(kE + ki),

and Et will reach a minimal value at equilibrium of

Emin = 100kI/(kE + ki).Moreover,

X = ki + kE.

[8]

[9]

[10]

[11]

[12]

The volume of compartment E can be defined as VE=Emin/Teq, where Em.n and Teq represent the per cent ofthe dose in the extracellular compartment and the simul-taneous plasma concentration of thyroxine-"'1I in theplasma when the two compartments come to equilibriumrelative to each other. Analogously, the volume of com-partment I can be defined as VI = Imax/Teq. Althoughthe volume of the intracellular compartment is a usefulconcept, no immediate anatomical significance should beassigned to it.6

6 Actually, the volumes of compartments E and I (notto be confused with the volumes of distribution of thy-roxine-J'I and albumin-'I) are expanding during theperiod of observation. Initially, the volume of com-partment E is that of the plasma volume, about 3.2 L.At the equilibrium time, the volume of E is 4.2 L. Asimilar relative increase occurs in compartment I. Nev-ertheless, the interchange of thyroxine between thesecompartments can be treated as though we were dealingwith two static compartments, the volumes of which weredetermined at equilibrium time. The reason for this liesin the fact that fractional exchange of thyroxine betweenthe two compartments is relatively independent of thecompartment size during the period of observation.

An alternative kinetic scheme would involve an un-restricted two compartmental model analogous to thatoriginally employed by Berson and Yalow (15) and morerecently by Cavalieri and Searle (7). In such a systeman interchange between the plasma compartment and com-partment I would occur, but no restrictions would beplaced on the movement of thyroxine out of the plasmacompartment into the interstitial compartment. Solu-tion of such a model would yield essentially similar re-sults. The fractional transfer constants from the plasmacompartment to the intracellular compartment kp would

765

OPPENHEIMER,BERNSTEIN, AND HASEN

The nature of the equilibrium time deserves comment.This is the theoretical point at which the flux of thy-roxine-J'I from the extracellular compartment to the in-tracellular compartment is equal to the flux in the op-posite direction. On the time scale used in these studies,the point is determined as the midpoint of the Im.plateau. At this point the specific activity of thyroxinein both compartments can be assumed to be equal. Sincethe concentration of unlabeled thyroxine can be deter-mined from the chemical protein-bound iodine (PBI),the exchangeable pools of thyroxine in both compart-ments can be calculated from the data at hand. Thus,

ZI= VI(r), [13]and

ZE = VE(r), [14]

P can be evaluated by equilibrium dialysis. In an equi-librium dialysis system in which serum is diluted by afactor of 150, P = 150/DF, where DF represents thedialyzable fraction (8). The over-all binding by theextracellular compartment can be obtained by multiplyingP and Vs, the volume of the extracellular compartmentat the equilibrium time. Thus, if we substitute into Equa-tion 15 and rearrange terms,

h = (150/DF)(VE)(kE). [16]

The value for kim can be determined as follows. Thefunction X can be determined graphically from a plot oflog (Imax - It) against t (Equation 5). From Equations10 and 11, it follows that

[17]

where Z, and ZE are the pool sizes of thyroxine in com-partments I and E, respectively, and r is the concentra-tion of circulating thyroxine in plasma. It is apparentthat if thyroxine-'I accumulation in the intracellularcompartment were followed for a longer period andplotted on a time scale of days rather than hours, theapparent plateau level of It would be reduced to a maxi-mal point with a subsequent downward deflection of thecurve representing the net effects of thyroxine metabo-lism and the further distribution of thyroxine into theextracellular compartment.

Data will be presented to indicate that the transfer ofthyroxine from the extracellular compartment to theintracellular compartment is dependent in part on thestrength of thyroxine binding by the plasma proteins.The model system proposed can be used to quantitate thiseffect of protein binding. Thus, the fractional transferconstant kE can be expressed as

kE = h/bE, [15]

where h is a permeability factor (liters per minute) de-pendent on the properties of the hypothetical compositemembrane separating compartments including its surfacearea and effective porosity7 and bE is a term (liters)representing over-all binding of thyroxine by plasmaproteins in the extracellular compartment. The strengthof thyroxine binding by plasma proteins can be represented

nby the expression P _ L k,(Mi - TPI), where ki is

the apparent association constant of the ith species of atotal of n binding sites, M4 the molar concentration of theith binding sites, and TP,, the molar concentration of theith thyroxine binding site complex. This expression is,in essence, a measure of the effective concentration ofunoccupied thyroxine binding sites in plasma. The value

simply be related to kim by the following relationship:

kE/kp = Va/VE, where Va is the plasma volume.

7This is not to suggest that the exchange between theextracellular and intracellular compartments is necessarilya passive phenomenon. The concept of effective porosityis used purely in an operational sense and can be repre-sented by a number of physical models any of which coulddepend on active cellular metabolism.

I,,/(100 - I..) = (kE/ki),and from Equation 12 and 17

kE = X(I../100),and

[18][19]kI = X[1 - (Imax/100)].

If we substitute into Equation 16,h = X(150/DF)(VE)( Imax/100). [20]

Lastly, since VE will vary with body size, it is convenientfor the purpose of comparisons to express the perme-ability factor per liter extracellular volume at equilibrium.Thus,

/ /I 7 _ {.1C^ /rAx /T-%E T /I B{ r,) i(h/VE) = (15UA/L))(Imaf/lUu). LZIJ

Analogously, the fractional transfer constant from theintracellular to the extracellular compartment, ki, can beexpressed as

kI = h/bi, [22]

where h is the same permeability factor used in Equa-tion 15 and numerically defined in Equation 20, and b1represents net intracellular binding. It should be em-phasized that the intracellular binding may differ mark-edly in terms of its physicochemical characteristics fromplasma protein binding. It may be useful to regard in-tracellular binding simply as the effective retardingmechanism serving to maintain thyroxine within the cel-lular compartment. Again, bi is the product of a con-centration and a volume factor. Since the volume of theintracellular compartment is expressed in terms of plasmaequivalents at the equilibrium time, the binding factorper unit volume is the same as that for plasma (P=150/DF).

bI = (150/DF)(Vi). E23]The final picture that emerges from this formulation

is that the partition of thyroxine between tissues andthe extracellular compartments will be the net resultantof the balance of measurable tissue and protein bindingfactors. An increased Imas could result either from in-creased tissue binding or decreased extracellular bindingor both. Conversely, a diminished Imax could representeither increased plasma protein binding or diminishedcellular binding or both. If we assume normal perme-ability characteristics of the hypothetical membrane, in-creased binding by extracellular proteins should result

766

ESTIMATION OF INTRACELLULARTHYROXINE

in a diminished k3 but not affect ki. Similarly, di-minished binding by plasma proteins should result in ahigh ki but leave k1 unchanged.

Results

Evidence supporting the use of albumin-125I intracing the distribution of thyroxine-'111 intothe extracellular space

Distribution of the thyroxine-binding proteins.The plasma disappearance curves of simultane-ously injected TBPA-'31I and albumin-125I for thefirst 4 hours are illustrated in Figure 2. Similarresults were observed in a second study. Plasmalevels of the two iodinated proteins do not differappreciably during the period of observation. Thedata indicate that the acute distribution kineticsof TBPA and albumin are similar and consonantwith previous demonstrations that the space ofTBPA distribution is the same as that of serumalbumin (16, 17).

Unfortunately, purified TBG is not availablefor similar kinetic studies. Indirect techniques

. +

COMPARTMENTI

ki

COMPARTMENTE

kE

FIG. 1. SCHEMATICREPRESENTATIONOF MODELSYSTEMEMPLOYEDTO REPRESENTTHE INTERCHANGEOF THYROXINEBETWEENINTRACELLULAR (I) AND EXTRACELLULAR (E)COMPARTMENTS. During the 4 hours of observation, thesize of each compartment increases. The initial state isindicated by the broken line, and the direction of ex-

pansion by the broken arrows. The fractional inter-change (kic, ki) remains unaltered during the period ofexpansion. Mathematically, this model can be treatedas a quasistatic system with the dimensions of the finalcompartments represented by the solid lines.

.j50.

ig8 -t

t 20-

I0

Ilo

-J9- 6-

0 60 120 ISOTIME (MINUTES)

240

FIG. 2. PLASMA DISAPPEARANCE CURVES OF SIMUL-TANEOUSLYINJECTED THYROXINE-MI-BINDING PREALBUMIN( ) AND ALBUMIN-I (-- -).

were, therefore, applied to obtain informationabout the distribution of this protein. As will beshown subsequently, a large fraction of adminis-tered thyroxine-'31I leaving the plasma compart-ment enters the liver. If intrahepatic thyroxineexists as a thyroxine-TBG complex, one shouldbe able to detect TBG in the soluble fraction ofliver homogenates. Results of two separate ex-periments failed to disclose evidence of any strongsoluble intracellular binding protein capable ofcompeting effectively with plasma proteins forthyroxine. Thyroxine-181I added to the super-natant fraction migrated electrophoretically as asingle peak with a mobility identical to that ofthyroxine-131I alone. Moreover, when the super-natant fraction was mixed with an equivalent vol-ume of appropriately diluted serum, the mobilityand distribution of thyroxine-131I were identicalto those of an equivalent dilution of serum insaline. These results suggest that TBG is notpresent in the soluble fraction of liver homoge-nates.

Evaluation of the possible influence of differ-ential transcapillary migration of thyroxine andits binding proteins. Current estimates based onequilibrium dialysis of serum against aqueousbuffer have suggested that from 0.030 to 0.050%of total circulating thyroxine is unbound (8, 18-20). If this free fraction diffused very rapidlyacross the capillary endothelium, it would be the-oretically possible for thyroxine to distribute it-self in the extracellular space more rapidly thanits binding proteins. The kinetic analysis pro-posed in this paper assumes that such independentdiffusion is quantitatively unimportant. The ex-

767

OPPENHEIMER,BERNSTEIN, AND HASEN

perimental findings underlying this assumptionwill be presented.

Since interstitial fluid in man is generally un-available for sampling, the movement of simul-taneously injected thyroxine-131I and albumin-125Ifrom the plasma into the pleural effusion of a65-year-old man was studied. The pleural fluidhad an albumin concentration of 2.1 g and a globu-lin concentration of 1.7 g per 100 ml. The con-comitant plasma albumin and globulin concentra-tions were 3.4 and 3.6 g per 100 ml, respectively.The PBI concentration of the effusion was 3.4 ,ugper 100 ml, and that of the plasma was 6.1 ug per100 ml. It should be noted parenthetically thatthe ratio of PBI concentrations in pleural fluidand plasma, 0.56, approximates within experi-mental error the ratio of albumin concentrations,0.62. Also, the distribution of tracer thyrox-ine-131I among binding proteins after paper elec-trophoresis was identical in both fluids. Thesefindings suggest that the movement of thyroxine-binding proteins across this capillary bed is simi-lar to that of serum albumin.8

The concentration of thyroxine-131I and albu-min-125I in plasma and pleural fluid is illus-trated in Figure 3. At the end of 4 hours, thespecific activities of both tracers in the pleuralfluid were less than one-tenth that of plasma. Be-cause of the slow movement of both tracers theradioactivity in pleural fluid up to 40 minuteswas found to be predominantly derived from con-taminating iodide present in the original ship-ments of the tracers. Accordingly, all samples,both of pleural fluid and plasma were precipitatedwith TCA to remove the bulk of iodide. Never-theless, the problem of coprecipitation of iodidewith thyroxine (12) prevented accurate measure-ments until the 60-minute pleural sample.

An estimate of the relative fractional rate ofentrance of thyroxine-131I and albumin-125I intothe pleural fluid can be obtained from the follow-

8This suggestion, of course, is predicated on the as-sumption that the fractional removal rates of the variouscomponents in the pleural space (PBI, TBPA, TBG, andalbumin) are similar. Moreover, the pleural fluid shouldnot be considered necessarily representative of the ex-travascular fluid in general. In this connection, however,it is of interest to note that the ratio of the concentra-tions of PBI to protein in thoracic duct lymph is thesame as that in serum (7).

a N.PLASMA

1.0 -

t 0.5--J

0

0

0.1-

PLEURAL FLUID

L0 60 120

TIME (MINUTES)lo0 240

FIG. 3. CONCENTRATIONSOF THYROXINE-II (T4) ANDALBUMIN-'"I IN PLASMAANDPLEURALFLUID AFTER SIMUL-TANEOUSINJECTION OF TRACERS. All samples were pre-cipitated with trichloroacetic acid.

ing considerations. Assume that the movementof both tracers from the plasma compartment tothe pleural compartment obeys first-order ki-netics. Since the movement of both tracers isslow and the plasma-pleural system far fromequilibrium, it can be assumed that the netmovement of tracers out of the pleural fluid isnegligible during the interval under consideration.Thus,

(CT)t(Ve) = (kT, i)Va Ttdt, [24]

where (cT) t is the concentration of thyroxine-131Iin the pleural fluid at time t (per cent dose perliter); Ve, the volume of the pleural effusion(liters); kT,i, the fractional transfer constant fromthe plasma compartment into the pleural com-partment (minutes-'); V., the plasma volume(liters); and Tt, the concentration of thyroxine-"'I in plasma at time t (per cent dose per liter).Since values from 0 minutes to 60 minutes areunreliable, the function described by Equation 24is more appropriately evaluated between 60 and240 minutes and can be rewritten

f240[(CT)240 - (CT)60]Ve = (kT. i)Vas Ttdt.

60[25]

768

ESTIMATION OF INTRACELLULARTHYROXINE

Similarly,r240

[(Ca)240 - (Ca)60]Ve = (ka, i)Va atdt,J60

[26]where (ce) t is the concentration of albumin-'25Iin the pleural fluid at any time t; ka,i, the frac-tional transfer constant of albumin-125I from theplasma compartment to the pleural compartment;and at the concentration of albumin-125I in theplasma at any time t.

From Equations 20 and 21, it follows that240

kT, i (CT)240 - (CT)60 J0 [27]ka, i L(Ca)240- (Ca)60 i240 E27

J 0

Values for CT and ca are directly available, andthe integral expressions can be evaluated graphi-cally. The above ratio of the fractional transferconstants was determined to be 1.25. Thus, thepossibility exists that there is a 25% faster ingressof thyroxine into the pleural fluid and that thisdifference could be due to independent diffusionof hormone. It appears likely, however, that thecalculated ratio of 1.25 is an overestimation. Re-cent observations (13) have indicated that Ab-bott thyroxine-151I is contaminated with severalper cent radioiodinated 3,5,3'-L-triiodothyronineand 3,5',3'-L-triiodothyronine. These substancesare much less strongly bound to plasma proteinsthan thyroxine. Thus, material appearing in thepleural fluid in the early phase of the equilibriumperiod could contain a disproportionate contribu-tion from these iodinated contaminants.

Even if we assume that the ratio of the frac-tional transfer constants is correct, calculationsbased on this figure would suggest that the effectsof independent transcapillary diffusion of thyrox-ine are negligible in terms of the over-all dis-tribution kinetics of thyroxine. In this patientinitial clearance of albumin-'25I from plasma, es-timated from the graphically determined initialfractional exit rate and the plasma volume, is 6.24ml per minute. Similarly, the total clearance ofthyroxine from the plasma in this patient is 37.6ml per minute. If independent thyroxine diffusionis 25 % of the albumin clearance, the clearance ofthyroxine from plasma by means of independentdiffusion is 1.56 ml per minute. Thus, independent

diffusion would constitute only 4.1% [= (1.56/37.6) X 100] of the total thyroxine clearance.Since the 25%o figure is probably an overestima-tion, as pointed out above, the effects of inde-pendent diffusion are minimal.

This analysis is based on the assumption thatpermeability characteristics of the pleural capil-lary bed in the patient studied are typical ofother capillary beds that allow only slow trans-capillary passage of protein. The fact that tracerthyroxine is cleared initially at a sixfold greaterrate than tracer albumin from the plasma com-partment can readily be attributed to the morerapid rate of equilibration of protein across se-lected capillary beds such as the hepatic sinusoids(21). A small pool of rapidly exchangeable pro-tein in such beds would not greatly influence theapparent initial disappearance rate of albumin-'3,Ifrom the plasma but would serve to facilitate arapid and direct exchange of thyroxine fromplasma protein to cellular receptor sites.

The transport of thyroxine from the inter-stitial space into the cells would thus not be lim-ited by the diffusion of thyroxine in an aqueousmedium or by the transcellular movement ofplasma binding proteins.

Comparison of the rates of intracellular andhepatic accumulation of thyroxine-131!. Results

< 40 -/251-ALBUMIN

20THIMYROXINE

- '

jI° :

w 40 -

4 0 - --HP@ trxn t't

D0 D

|J120 ,;z 5

12000

1000 E

500 z

300

200

IUX

0 20 40 60 80 100 20 140 160 180TIME AFTER INJECTION ( n n )

FIG. 4. COMPARISONOF RATES OF HEPATIC UPTAKEANDINTRACELLULAR ACCUMULATIONOF THYROXINE-'I. He-patic uptake was calculated from the observed externalcounting rate over the liver with an appropriate correc-tion for the falling levels of blood thyroxine detected bythe probe (see text). Intracellular accumulation wascalculated from the plasma disappearance curves of thy-roxine-'I and albumin-'I.

769

OPPENHEIMER,BERNSTEIN, AND HASEN

illustrated in Figure 4 are representative of one

of four studies in which the intracellular and

hepatic accumulations of thyroxine-131I were si-

multaneously determined. Since the absolute

fraction of hormone taken up by the liver cannot

be readily determined from external measure-

ments alone, only the rates of these processes can

be meaningfully compared. The close correspond-ence of these rates suggests either that the liver

is the predominant component of the intracellular

compartment or that the kinetic characteristics of

nonhepatic components resemble those of the

liver. Under any circumstances these findings

lend credence to the basic assumptions underlying

our analysis.

Results of the analysis (Table I)

The results of a representative study in a nor-

mal subject are illustrated in Figure 5. Here,

initial distribution volume of thyroxine as deter-

mined by simple extrapolation is approximately

4.5%o greater than the initial distribution volume

of albumin. Thyroxine rapidly leaves the vascu-

lar compartment, but by 200 minutes after in'jec-tion, the fractional rates of egress of thyroxine-

"'I~ and albumin-'1251I are similar. In the 4-hour

period, the fall of albumin-'25I is 1 5.5%7 of the

initial concentration, whereas the decline in thy-

roxine- 1311I during the same interval is 62%o.

The calculated per cent of the dose in the intra,-

cellular compartment rises to a plateau value of

51.1%7. The linearity of a semilogarithmic plot

D.S.

w

0

0l

0-

14-

Ia

104O-

C. S. M.AkN.

MN.

0In

I I I

0 120

TIME (MINUTES)

I

IS0 240

FIG. 5. ]REPRESENTATIVE STUDY IN A NORMALSUBJECT

(M.N.). It = intracellular accumulation, calculated from

the plasma disappearance curves of thyroxine-'I and

of (Imax. It) as a function of time, which forms

the basis of the two compartmental analysis, is

illustrated in Figure 6. Kinetic data on six nor-,

mal subjects are summarized in Table I.

WS. F0.G R. P.

I -I -I I

0 50 250 50 250 50 250 50 950 250 50 250 50 150 250

TIME (minutes)

FIG. 6. LINEAR RELATIONSHIP OF LOG (I... It) AS A FUNCTIONOF TIME IN THE GROUPOF SIX NORMALSUBJECTS.

The intracellular thyroxine-"1I accumulation (I) is plotted on the same scale. A = It; 0 = I... It.

770

ESTIMATION OF INTRACELLULARTHYROXINE

. .

r- OA m OO)C0000 0 4N N m C4 m

N 0 '0 0o 00

0 0c_ 0t-

X 4-- - -M

U)) U) lo W) W0

C co co O c

co co

C4t00

_ Uo U) O

'* V 0- 0 b U)

O O U)

6...1

3tR-

e

04

4).

theo

-k

0d 0 00 00 Cl-_ - - C

\ U) _ 0%

t U) U)

-

+ 0

_'0 q 00

*- 0. *0

'0 0%\0\

0 0

-. 0 .l .)0 U)i0 _U)_ _-

00 Un NC0oN 00

N U)00I- _- CS _

" )0 N

- oen ON o enN _

_ ; _ _; to ob UU e o

o U)0 0o-.

M 0%t- '40 'O

co~co

- - O - - -

Xo ov U) 4oo

in -- -" -

U) )Ct

m W

. . . .

0o In 0o

CR0) '0

% co m f in o) W

U) 0 ..0~ 0

u) CL 0 0 0 co

A) _ U) ao

0% o _ _ 0%

)0 .e

0 Uo. 0. .

o0 a a u

C)

0 -_ ro i 0m. I U )

Ik Eo0

z t) Z; C# A, cn (n44 C004 u

771

N

N

W

IN

.0

0Sl~a

E., M

t

._3e

K0

¢.

fU)0)

F4*tb

.0

'k

P4

mU

0t%itb

11

$a

Cl

Cl

Uo

mNO

Cll

0

Cl

co

0

0

m

'0

0

0

ei

oqm4

U)

0

0)-

> vY-

00)

o0 __

0~0>

5 g4,NOX

N

jO)

0o 0)St

H 0 n0J @

0 ' oXS0=

EE iEd

< oY.0

.0 =00

g DX' 0a.. 0~o

.04)000.0

0

H

- .

s

0U.W4b

(-m ¢-, Q '2 5 -

772 OPPENHEIMER,BERNSTEIN, AND HASEN

TABLE II

Thyroxine-"'I content of liver biopsies

Esti- Esti-Esti- mated mated

mated* %Dose/g hepatic %I.LxPatient Age Sex Weight liver wt liver T/Pt uptake I..$ in liver

years kg kg % %SB 60 F 59.1 1.18 .0439 3.29 51.9 63.4 81.8ZF§ 35 F 51.8 1.04 .0115 0.46 11.9 27.4 43.5AM 26 F 60.4 1.21 .0275 1.91 33.2 47.9 69.4JR 77 M 78.2 1.56 .0276 3.01 43.2 52.3 82.6

* Liver weight estimated to be 2%of body weight.t T/P = tissue/plasma ratio = per cent dose per gram liver/per cent dose per milliliter plasma at time of biopsy.T I, = maximal fraction of administered dose thyroxine-1311 calculated to be within intracellular compartment.§ Fatty infiltration, indicated by liver biopsy, may account for relatively low I.. and hepatic uptake.

Results of hepatic biopsies obtained approxi-mately 4 hours after the injection are indicatedin Table II. If we assume that the weight ofthe liver is 2%o of the body weight, the thyroxine-131I content of liver is on the average approxi-mately 70% of the calculated Imax. The impor-

DH..42 30-U)4a. 20--JLxJUl)°0-

50-

20-

wU)00

4-4

10-

5-

3-

2-

tant contribution of the liver to intracellular ac-cumulation of thyroxine-131I is also indicated byresults of studies in five patients with liver diseasedue to Laennec's cirrhosis or acute hepatitis(Table I). Figure 7 illustrates the character-

istic changes in the plasma disappearance curveof thyroxine-131I. The calculated intracellularaccumulation of thyroxine (Imax) was markedly

251- ALBUMIN41 Q.

THYROX/NE

80-

50-

IniU) -0

4-4NORMALRANGE

T. H.

o 60 120TIME (MINUTES)

180 240

FIG. 7. PLASMA DISAPPEARANCEAND CALCULATED IN-

TRACELLULARACCUMULATIONOF THYROXINE-I IN A PA-

TIENT (D.H.) WITH SEVERE LIVER FAILURE DUE TO

LAENNEC'S CIRRHOSIS. Range of intracellular accumu-lation in normal subjects is indicated by crosshatchedarea. Nqte the relatively small splay between thyroxine-'I and albumin-MI curves in comparison to similar

curves in the normal subject illustrated in Figure 5.

10-

a60 120 180

TIME (MINUTES)

FIG. 8. PLASMA DISAPPEARANCEAND CALCULATEDIN-

TRACELLULAR ACCUMULATION OF THYROXINE-"11I IN A

PATIENT (T.H.) WITH A CONGENITALDEPRESSION IN THE

MAXIMAL BINDING CAPACITY OF THYROXINE-BINDING

GLOBULIN. Note the wide splay between thyroxine-'Iand albumin-'`I curves in comparison to that in thenormal subject illustrated in Figure 5. Normal range

of intracellular uptake is illustrated by crosshatched area.

240

ESTIMATION OF INTRACELLULARTHYROXINE

DIPHENYL-HYDANTOIN

IAM - .40- AL&- r I&'-' 1.vmgI

t201'

60

20-

1..I..I

0 80 160 240 320 400

TIME AFrER INVJECTIONF ISOTOPEStMINUTES)

FIG. 9. EFFECT OF ACUTE INFUSION OF 150 MG DI-

PHENYLHYDANTOINON THE PLASMA DISAPPEARANCE OF

THYROXINE-'I AND THE CALCULATEDINTRACELLULAR AC-

CUMULATION CURVE.

reduced in this group of patients. The fractionaltransfer constant from extracellular to cellularcompartments, kE, was low, although the frac-tional transfer constant in the opposite directionwas normal. Since the net binding by serum

proteins as measured by equilibrium dialysis was

either normal or diminished, the reduced frac-tional entrance into the cellular compartmentmust be attributed to a change in the perme-

ability function (h/VE). Such changes in perme-

ability can be interpreted either in terms of alter-ations in the effective porosity of the plasma mem-

brane or in terms of disruption of a metabolicallydependent transport system for thyroxine. Con-sonant with these observations, the calculated in-tracellular pool of thyroxine was also reduced.

The effect of alterations in the plasma thyrox-ine-binding proteins is illustrated by a study ina 17-year-old female with a genetically deter-mined reduction in the maximal binding capacityof TBG (Table I, Figure 8).9 The concentration

9 Kindly referred to our attention by Dr. B. Segal.

of thyroxine-131I falls precipitously after injection,leading to the calculated increase in kE and I.a..The values for k, and the permeability factorh/VE are within the normal range. Although thecalculated extracellular pool of thyroxine at theequilibrium time is reduced, the calculated intra-cellular pool is normal. Thus, the observedchanges in the partition of tracer can be attributedexclusively to an alteration in the maximal bind-ing capacity of TBG.

The role of binding proteins in regulating thepartition of -thyroxine-131I between extracellularand cellular compartments is also apparent fromthe effects of the intravenous injection of diphenyl-hydantoin, a drug known to displace thyroxinefrom TBG and lower the serum PBI (22-24).Studies were carried out in three patients. Di-phenylhydantoin (150 mg) was infused when theintracellular accumulation curve had reached aplateau value. Results in a representative studyare indicated in Figure 9. Within minutes afterthe injection, there was a downward deflection ofthe plasma thyroxine-'31I curve. No change inthe albumin-125I curve was noted. The calculatedintracellular accumulation rose from 46 to 52%oof the administered dose.

Discussion

A number of assumptions underlying this anal-ysis deserve special consideration. In interpretingthe results of the plasma disappearance curvesand in estimating thyroxine accumulated in thehepatic samples, we have tacitly assumed thatserum albumin is predominantly extracellular.Although intracellular albumin (or material im-munologically related to albumin) has been dem-onstrated by immunofluorescence (25, 26), thisis not considered to be quantitatively significantin relation to the mass of extracellular protein(27). Moreover, radioautographic studies car-ried out in the rat during the initial distributionperiod of albumin and thyroxine revealed no in-tracellular localization of albumin-125I but diffusecytoplasmic distribution of thyroxine-'25I (28).This observation justifies, in part, the use of theterm "intracellular," since selective concentrationof thyroxine-125I in the interstitium or on thesurface of cell membranes could not be demon-strated.

773

OPPENHEIMER,BERNSTEIN, AND HASEN

Wehave also assumed that the distribution ofTBG does not differ significantly from that ofalbumin and TBPA. Direct proof of this as-sumption must await the purification of TBG sothat the kinetics of labeled TBGcan be measured.Nevertheless, our failure to find TBG in thesoluble fraction of liver homogenate serves largelyto exclude TBG as the transport vehicle acrossthe hepatic cell membrane. If TBGwere indeedresponsible for the intracellular accumulation ofthyroxine, one would expect to find at least twoto three times the concentration of TBG in liveras in plasma, based on the relative concentrationsof thyroxine in liver and plasma. Our data donot exclude the possibility that TBG may bestrongly associated with particulate cellular ele-ments, but in such an event it would be difficultto assign to intracellular TBG the transport func-tion that would be required by a rapid two wayexchange across the cell membrane. Moreover,other studies have failed to demonstrate greaterthyroxine binding by particulate elements thanby the cell sap of hepatic cells (29, 30). Lastly,the argument previously advanced by Cavalieriand Searle (6) strongly militates against any pos-sible role of TBG in transcellular transport. Inpatients with congenitally low TBG one wouldnot expect increased hepatic uptake of adminis-tered tracer thyroxine if TBG were indeed re-sponsible for transcellular thyroxine transport.

The importance of the liver in the extravasculardistribution of thyroxine is illustrated by thediminished intracellular thyroxine in patients withliver disease, by the close correspondence of therates of intracellular and hepatic uptake of thy-roxine- 31I, and by the liver biopsy data. Esti-mates based on the normal histological specimenslisted in Table II (i.e., excluding ZF) suggestthat approximately 78%o of the calculated intra-cellular thyroxine is intrahepatic. If we assumea normal intracellular volume of thyroxine dis-tribution of 4.5 L, our estimate of intracellularhepatic thyroxine will be 3.5 L. This figure is inexcellent agreement with the recent calculationsof Cavalieri and Searle (7) for extravascularhepatic thyroxine space, 3.8 L ± 0.5 (SD). Thiscorrespondence serves to strengthen the under-lying assumptions made in our analysis and toconfirm our radioautographic data suggesting

only negligible concentration of thyroxine in in-terstitial hepatic spaces.10

Since 78% of the calculated intracellular thy-roxine space can be attributed to cellular hepaticuptake, the question arises as to the partition ofthe remaining 227o. Preliminary data obtainedfrom experimental animals (rabbit, dog, rat) sug-gest that the kidney also makes a significant con-tribution to the net intracellular thyroxine com-partment. In a patient with hepatic cirrhosis anda diminished thyroxine-'31I uptake by the liver,the renal outlines were clearly demonstrated byexternal radioactive scanning 4 hours after theadministration of thyroxine-131I. In normal sub-jects, the kidneys are not seen under these cir-cumstances, probably because of the interferencefrom hepatic concentration of radioactivity. In-sufficient data are available to make a quantitativeestimate of renal thyroxine uptake in man or toassess the contribution of other tissues. The tech-nical limitations in our study previously cited alsodo not allow us to rigorously exclude the possi-bility that a small proportion of the remaining22%o of radioactivity is in fact associated withplasma proteins outside the instantaneous albu-min-'25I distribution space.

The rapid exit of thyroxine from the albumindistribution volume clearly cannot be attributedto the effects of metabolic transformation of thy-roxine. The average normal cellular clearance ofthyroxine (VEkE) is 42.7 ml per minute. If weassume an average thyroxine distribution volumein these patients of 12.7 L based on a meanweight of 79.8 kg (31) and a thyroxine half-lifeof 6.0 days, the metabolic clearance can be esti-mated to be approximately 1.0 ml per minute. Itfollows that thyroxine must return from the intra-cellular compartment without having undergoneeffective metabolic transformation. This conclu-sion is in essential agreement with the findingsof others (4, 5, 7).

10Rapid equilibrium of albumin-'I between the vascu-lar and interstitial spaces of the liver has been established(21). Calculations based on the three normal hepatic bi-opsy specimens listed in Table II indicate that the aver-age total thyroxine distribution volume per gram of wettissue is 2.88 ml plasma. The average albumin distribu-tion volume in the same specimens was 0.14 ml plasma.Thus 20.5 (=2.88/0.14) times as much thyroxine is as-sociated with the cellular components as with plasma pro-tein' in the vascular and extravascular spaces.

774

ESTIMATION OF INTRACELLULAR THYROXINE

The fraction of administered thyroxine-131Iwithin the intracellular compartment (Imax) wasevaluated at a time between 3 and 4 hours afterinjection of the isotopes, when the intracellularand extracellular compartments were in equilib-rium relative to each other. At this time, how-ever, albumin-125I and presumably the thyroxine-131I-plasma protein complex are not yet completelydistributed in the total extravascular protein space.Several days are required for complete distributionequilibrium of isotopic albumin. The proportionof total body thyroxine in the rapidly exchange-able tissue pools may thus be less than Ima,,. Thepartition of unlabeled thyroxine between cellu-lar and extracellular thyroxine can be estimatedfrom the following considerations. If we assumethat the ratio of extravascular to intravascularalbumin as determined by multicompartmentalanalysis is 1.33 (32), the final average albumin-125I distribution volume in our series of normalsubjects can be estimated to be 7.53 L. Thus,the average distribution volume of thyroxine inthese subjects, estimated from the sum of thefinal albumin-125I distribution space and the intra-cellular distribution space of thyroxine, is 12.0 L.This is in excellent agreement with the predictedvalue of 12.7 L for the thyroxine distributionspace calculated for an average weight of 79.8 kgfrom the empirical formula proposed by Oddie,Meade, and Fisher (31). Our findings would,therefore, suggest that 37.3% (= 4.47/12.00) oftotal body thyroxine is situated within rapidlyexchangeable tissue stores. Thus, in the normalgroup, the average estimated total body pool ofthyroxine is 885 ug, of which 329 pg is intracellu-lar. Further data supporting the validity of ourestimate of intracellular thyroxine were derivedfrom two additional studies in which the totaldistribution volume of thyroxine, estimated byextrapolation to t = 0 (33),11 was compared tothe value representing the sum of the total albu-min distribution volume obtained by the equilib-rium time method (35) and the intracellularthyroxine distribution volume. In one patient the

"1The extrapolation method along with other singlecompartmental analyses currently used for approximatingthe thyroxine distribution volume may result in an over-estimation of the space. As pointed out by Rall, Robbins,and Lewallen (34), however, this overestimation prob-ably does not exceed 10% of the true value.

extrapolated thyroxine volume was 11.6 L,whereas the sum of the albumin and intracellularspaces was 11.2 L. In the other patient, the cor-responding values were 8.8 and 9.2 L.

The results of our studies strongly support oneof our underlying assumptions, namely, that thedistribution of labeled thyroxine into the extra-cellular space is largely, if not completely, deter-mined by the distribution kinetics of the carrierproteins. Diffusion of thyroxine independent ofits carrier proteins does not appear to make a sig-nificant contribution to the kinetics of extravascu-lar distribution. Thyroxine appears to leave thevascular tree in association with its carrier proteinsthrough openings in the endothelial membrane.The mechanism of such transcapillary movementof protein has been extensively discussed by May-erson, Wolfram, Shirley, and Wasserman (21).

The transfer of thyroxine from the extracellularto the cellular compartment probably also occursas the consequence of an interaction betweenplasma protein and membrane receptor sites.Thus, the concept of free thyroxine may not benecessary in describing the acute distributionkinetics of the hormone. Measurement of thedialysis fraction (or per cent free thyroxine) byequilibrium dialysis, however, provides an indexinversely related to the over-all strength of plasmaprotein binding. The absolute concentration offree thyroxine, determined from the product ofthe dialyzable fraction and the total thyroxineconcentration in plasma, would then be propor-tional to absolute rate of unidirectional transfer.

Robbins and Rall (36) have postulated that theturnover rate of thyroxine is proportional to theconcentration of free thyroxine in plasma. Sincethe metabolism of thyroxine is related to cellularfunction, it would be of interest to determine in fu-ture studies whether the turnover of thyroxine isin fact more closely related to the exchangeableintracellular pool. Weshould anticipate that pri-mary changes in plasma protein binding wouldresult in similarly directed and proportionalshifts in free thyroxine. and the intracellular pool.On the other hand, primary changes in cellularbinding would produce oppositely directed changesin free thyroxine and the intracellular pool.

The nature of intracellular binding and thecharacter of the reversible thyroxine transportacross the cell surface have not been established.

775

OPPENHEIMER,BERNSTEIN, AND HASEN

It is known that subcellular fractions can bindthyroxine (29, 30). To what extent such bind-ing contributes to the influx of thyroxine in vivoand to what extent active metabolic processesare required for the bidirectional movement ofthe hormone must be determined in future studies.

Lennon, Engbring, and Engstrom (37) firstreported the slowed disappearance of intrave-nously injected thyroxine-1311 in patients withliver disease. An increased fractional disappear-ance of thyroxine--31I in patients with hyperthy-roidism was also noted by these authors. Similarobservations have been made in our laboratoryand can be attributed, in our opinion, to the di-minished binding of thyroxine by plasma proteinsin this disease.

The diminished hepatic thyroxine space anddecreased hepatic clearance in patients with liverdisease described in the earlier communication byCavalieri and Searle (6) correspond well withthe relative changes in intracellular distributionand clearance of thyroxine reported here. Simi-larly, the alterations in hepatic distribution ki-netics in two subjects with congenitally low TBGvalues recently reported by the same authors (7)correspond with similar shifts noted in our pa-tient. It is apparent that cellular function must beconsidered in conjunction with the concentrationand the affinities of the thyroxine-binding plasmaproteins in assessing the extrathyroidal factorsthat govern the concentrations of circulating thy-roxine.

The methods described in this paper for esti-mating intracellular thyroxine and characterizingthe distribution kinetics of this hormone are tech-nically simple since they require only the mea-surement of serial plasma samples and the appli-cation of uncomplicated formulas. If the inves-tigator wishes to confine his attention to the mea-surement of the function Imax,, the procedure canbe even further simplied since the number ofplasma samples required can be significantlyreduced.

AcknowledgmentsWe thank Dr. A. Aaron Yalow for helpful discussions

and for his critical review of this manuscript. Mr. Mo-desto Martinez, Mr. Frank Martinez, and Mr. RubenAlmazan rendered expert technical assistance. Thesecretarial assistance of Miss Joan Tomes is gratefullyacknowledged.

References1. Albert, A., and F. R. Keating, Jr. The role of the

gastrointestinal tract, including the liver, in themetabolism of radiothyroxine.. Endocrinology 1952,51, 427.

2. Myant, N. B., and E. E. Pochin. The metabolism ofradiothyroxine in man. Clin. Sci. 1950, 9, 421.

3. Hazelrig, J. B. The impact of high speed automatedcomputation on mathematical models. Proc.Mayo Clin. 1964, 39, 841.

4. Gorman, C. A., E. V. Flock, C. A. Owen, Jr., and J.Paris. Factors affecting exchange of thyroidhormones between liver and blood. Endocrinology1966, 79, 391.

5. Pochin, E. E. U. S. Atomic Energy Commissionreport TID-7678. Oak Ridge Institute of NuclearStudies, 1964, p. 413.

6. Cavalieri, R. R., and G. L. Searle. The role of theliver in the distribution of "31I-thyroxine in man inCurrent Topics in Thyroid Research, Proceedingsof the Fifth International Thyroid Conference, C.Cassano and M. Andreoli, Eds. New York, Aca-demic Press, 1965, pp. 336-45.

7. Cavalieri, R. R., and G. L. Searle. The kinetics ofdistribution between plasma and liver of 131I-la-beled L-thyroxine in man: observations of subjectswith normal and decreased serum thyroxine-bind-ing globulin. J. clin. Invest. 1966, 45, 939.

8. Oppenheimer, J. H., and M. I. Surks. Determinationof free thyroxine in human serum: a theoreticaland experimental analysis. J. clin. Endocr. 1964,24, 785.

9. Oppenheimer, J. H., M. I. Surks, J. C. Smith, andR. Squef. Isolation and characterization of humanthyroxine-binding prealbumin. J. biol. Chem. 1965,240, 173.

10. Greenwood, F. C., W. M. Hunter, and J. S. Glover.The preparation of "'I-labelled human growthhormone of high specific radioactivity. Biochem.J. 1963, 89, 114.

11. Smith, J. C., G. Bernstein, M. I. Surks, and J. H.Oppenheimer. A method of preparative starch-gelelectrophoresis. Biochim. biophys. Acta (Amst.)1966, 115, 81.

12. Oppenheimer, J. H., R. Squef, M. I. Surks, and H.Hauer. Binding of thyroxine by serum proteinsevaluated by equilibrium dialysis and electropho-retic techniques. Alterations in nonthyroidal ill-ness. J. clin. Invest. 1963, 42, 1769.

13. Volpert, E., M. Martinez, and J. H. Oppenheimer.Radioiodinated impurities in commercial prepara-tions of 'I-thyroxine and their effect on the mea-surement of free thyroxine by equilibrium dialysis.J. clin. Endocr. 1967, 27, 421.

14. Solomon, A. K. Equations for tracer experiments.J. clin. Invest. 1955, 28, 1297.

15. Berson, S. A., and R. S. Yalow. The iodide trappingand binding functions of the thyroid. J. clin. In-vest. 1955, 34, 186.

776

ESTIMATION OF INTRACELLULARTHYROXINE

16. Oppenheimer, J. H., M. I. Surks, G. Bernstein, andJ. C. Smith. Metabolism of iodine-131-labeled thy-roxine-binding prealbumin in man. Science 1965,149, 748.

17. Socolow, E. L., K. A. Woeber, R. A. Purdy, M. T.Holloway, and S. H. Ingbar. Preparation of I1""labeled human serum prealbumin and its metabo-lism in normal and sick patients. J. clin. Invest.1965, 44, 1600.

18. Schussler, G. C., and J. E. Plager. Free thyroxinein human serum (abstract). Clin. Res. 1965, 13,248.

19. Ingbar, S. H., L. E. Braverman, N. A. Dawber, andG. Y. Lee. A new method for measuring freethyroid hormone in human serum and an analysisof the factors that influence its concentration. J.clin. Invest. 1965, 44, 1679.

20. Sterling, K., and M. A. Brenner. Free thyroxine inhuman serum: simplified measurement with theaid of magnesium precipitation. J. clin. Invest.1966, 45, 153.

21. Mayerson, H. S., C. G. Wolfram, H. H. Shirley, Jr.,and K. Wasserman. Regional differences in capil-lary permeability. Amer. J. Physiol. 1960, 198,155.

22. Oppenheimer, J. H., L. V. Fisher, K. M. Nelson, andJ. W. Jailer. Depression of the serum protein-bound iodine level by diphenylbydantoin. J. clin.Endocr. 1961, 21, 252.

23. Wolff, J., M. E. Standaert, and J. E. Rall. Thyrox-ine-displacement from serum proteins and depres-sion of serum protein-bound iodine by certaindrugs. J. clin. Invest. 1961, 40, 1373.

24. Oppenheimer, J. H., and R. R. Tavernetti. Studies onthe thyroxine-diphenylhydantoin interaction: ef-fect of 5,5'-diphenylhydantoin on the displacementof L-thyroxine from thyroxine-binding globulin(TBG). Endocrinology 1962, 71, 496.

25. Coons, A. H. The localization of antigen in tissuecells by means of fluorescein-labeled antibody inThe Nature and Significance of the Antibody Re-sponse, A. M. Pappenheimer, Ed. New York,Columbia University Press, 1953, p. 200.

26. Gitlin, D., B. H. Landing, and A. Whipple. Thelocalization of homologous plasma proteins in thetissues of young human beings as demonstrated

with fluorescent antibodies. J. exp. Med. 1953, 97,163.

27. Gitlin, D., and C. A. Janeway. Some isotopic stud-ies on the distribution and metabolism of plasmaproteins in Advances in Biological and MedicalPhysics, C. A. Tobias and J. H. Lawrence, Eds.New York, Academic Press, 1960, vol. 7, p. 249.

28. Sutton, C., and J. H. Oppenheimer. Unpublishedobservations.

29. Tata, J. R., L. Ernster, and E. M. Suranyi. Interac-tion between thyroid hormones and cellular con-stituents. I. Binding to isolated subcellular particlesand sub-particulate fractions. Biochim. biophys.Acta (Amst.) 1962, 60, 461.

30. Heninger, R. W., F. C. Larson, and E. C. Albright.Intracellular distribution of iodine-containing com-pounds in rat liver, kidney and heart. Endocrinol-ogy 1966, 78, 61.

31. Oddie, T. H., J. H. Meade, Jr., and D. A. Fisher.An analysis of published data on thyroxine turn-over in human subjects. J. clin. Endocr. 1966, 26,425.

32. Beeken, W. L., W. Volwiler, P. D. Goldsworthy,L. E. Garby, W. R. Reynolds, R. Stogsdill, andR. S. Stemler. Studies of I"~-albumin catabolismand distribution in normal young male adults. J.clin. Invest. 1962, 41, 1312.

33. Sterling, K., J. C. Lashof, and E. B. Man. Disap-pearance from serum of I`81-labeled L-thyroxineand L-triiodothyronine in euthyroid subjects. J.clin. Invest. 1954, 33, 103.

34. Rall, J. E., J. Robbins, and C. G. Lewallen. The thy-roid in The Hormones, G. Pincus, V. K. Thiman,and E. B. Astwood, Eds. New York, AcademicPress, 1964, vol. 5, p. 320.

35. Pearson, J. D., N. Veall, and H. Vetter. A practicalmethod for plasma albumin turnover studies.Strahlentherapie Sonderbaende 1958, 38, 290.

36. Robbins, J., and J. E. Rall. The interaction of thy-roid hormones and protein in biological fluids.Recent Progr. Hormone Res. 1957, 13, 161.

37. Lennon, E. J., N. H. Engbring, and W. W. Eng-strom. Studies of the rate of disappearance oflabeled thyroxine from the intravascular compart-ment. J. clin. Invest. 1961, 40, 996.

777