TRANSCORTIN: A CORTICOSTEROID-BINDINGPROTEIN OF PLASMA*t

By W. ROYSLAUNWHITE, JR. AND AVERYA. SANDBERG

(From the Departments of Biochemistry Research and Medicine C, Roswell Park MemorialInstitute, Buffalo, N. Y.)

(Submitted for publication September 19, 1958; accepted October 30, 1958)

The general field of the binding of steroids andtheir conjugates to human plasma proteins hasrecently been reviewed by the authors (1). Evi-dence was presented that Fractions IV-1, IV-4 andV (obtained by the Cohn fractionation procedure)were mainly responsible for the binding, and hencetransport, of the unconjugated steroids. Thebinding of conjugates of steroids to the plasmaprotein was also shown to be considerable. Fur-thermore, evidence was obtained that cortisol andcorticosterone were avidly bound by a "special"protein in human plasma. This conclusion wasbased on the fact that cortisol and corticosteronewere bound to unfractionated human plasma muchmore strongly than determinations of binding toalcohol-fractionated human plasma proteins hadled us to expect. In fact, these two corticos-teroids were bound at least as strongly as estroneand estradiol to plasma diluted 1:4 or 1:40 [seeTable VII of (1) ]. The existence of such a"special" protein has been independently de-scribed by Daughaday (2-5). At first he at-tributed all the binding to albumin; that observedin Fraction IV-4 (a-globulin) was assigned to itscontent of albumin (6, 7). Later (2), using thesame techniques but employing a radioactive labelfor assay instead of a chemical determination, heconcluded that there was in the a-globulin fractionof human plasma from cortisone-treated patients(but not normal subjects) a protein low in ca-pacity but high in affinity for corticosteroids.This is in contrast to albumin which has a highcapacity but low affinity for corticosteroids (1).In subsequent papers (3-5) he concluded thatnormal subjects also possess corticosteroid-bind-ing protein.

* An abstract of part of this work appeared in the J.clin. Invest. 1958, 37, 928.

t This investigation has been supported in part by a

grant (A-1240) from the National Institutes of Health,United States Public Health Service.

Sometime ago Roberts and Szego (8) concludedthat Fraction III-0 was important in the bindingand transport of estrogens in human blood, a find-ing which was challenged by Bischoff, Staufferand Gray (9). In our studies we were unable todemonstrate the binding of estrogens to FractionsIII-0 and recent work by Bischoff and Stauffer(10) is in agreement with these findings, pointingto albumin as the major protein concerned withthe transport of estrogens.

In this paper we wish to present further datapointing to the existence of a special protein (orproteins) with great affinity for the binding ofcortisol and corticosterone. By analogy to trans-ferrin, the iron-binding protein of plasma, we havenamed the corticosteroid-binding protein "trans-cortin." Additionally, evidence will be presentedthat the concentration of this protein is con-siderably increased during the third trimester ofpregnancy. Data will also be shown which indi-cate the lack of such a protein for the binding ofestrogens in human plasma, from both normal andpregnant subjects.

METHODSAND MATERIALS

Equilibrium dialysis was performed according to meth-ods previously reported (1). Since previous studies in-dicated that undiluted plasma bound 99 per cent of theC"-corticosteroid added, dilution of plasma gave resultswhich made differences in the binding among the vari-ous steroids more pronounced and easier to quantitate.It was also established that the presence or absence ofstabilizers in human serum albumin (HSA) did not af-fect the binding of the various steroids to HSAor plasma,a finding observed by Bischoff and Stauffer for the estro-gens (10). Hence, for the majority of the experimentssalt-poor HSA supplied in solution by the American RedCross was used. Free electrophoresis showed the HSAto be homogenous. It was also shown that when plasmaand HSA were diluted with phosphate buffer (pH 7.4,0.05 M), the binding of cortisol did not differ from thatobserved when 0.15 M saline was used for dilution. Themethod, therefore, consisted of placing 2 ml. of plasma di-

384

TRANSCORTIN: A CORTICOSTEROID-BINDING PROTEIN

luted to 10 ml. with 0.15 Msaline, in 23/32" NoJax Visk-ing casing. The casing was placed in a 125 ml. Erlen-meyer flask containing 29 ml. of either 1 per cent HSA(diluted with saline) or saline and 1 ml. of radioactivesteroid dissolved in saline. The flasks were rocked slowlyfor 72 or more hours at 40 C. to assure equilibration ofthese steroids. Small amounts of penicillin were addedto each flask (10,000 units) following studies which indi-cated the antibiotic did not interfere with the binding ofany of the steroids studied.

Blood (30 to 50 ml.) was collected in 0.5 ml. of heparin(4 mg. per ml.) and the erythrocytes separated from theplasma (use of serum did not alter the results). Theblood samples were obtained in the late morning or earlyafternoon from normal pre-menopausal subjects andpregnant women. The plasma was used the day on

which the blood was drawn.C'4-labeled steroids were added in amounts ranging

from 0.25 to 0.5 pug. per flask (ca. 2,000 cpm). Carriersteroid was added dissolved in small amounts of ethanol,the concentration of the alcohol never exceeding 0.1 per

cent. Both the carrier and radioactive steroids were 99per cent or more pure as judged by infrared spectros-copy and/or paper chromatography. Deoxycorticosteroneacetate and cortisone acetate were converted to the freealcohols by incubation with acetyl cholinesterase or bychemical hydrolysis.

Following dialysis the contents of the cellophane bagand of the flask were separately extracted three timeswith two volumes of chloroform. The chloroform ex-

tracts were dried by air, plated on steel planchettes andcounted in a windowless gas flow counter. The weightof the samples was negligible and no correction was neces-

sary for self-absorption of radioactivity. The percentageof binding of steroids to protein was determined by thefollowing Equation 1):

1) %bound = 100 (I D VR)

where R and D are the amounts of radioactivity presentinside and outside the dialysis casing, respectively, and VRand VD are the corresponding volumes.

The recovery of the added C14-steroids ranged from85 to 100 per cent. No binding of the steroid occurred tothe glassware or to the dialysis casing. In several ex-

periments. it was shown that under the conditions of thedialysis the added C'-cortisol was recovered unchangedas judged by paper chromatography. The complete pro-

cedure was performed in triplicate and the results repre-

sent the average of the three determinations which seldomvaried more than 3 percentage points from the average.

Certain samples were submitted to continuous flowpaper electrophoresis on the Spinco Model CP apparatususing' Schleicher and Schuell No. 470 paper.' Two buffersystems were used: barbital buffer, 0.02 M, pH 8.6 andacetate buffer, 0.02 M, pH 5.2. A potential of .660 to 800V., was applied, producing a current -flow of 40 to 60 ma.

Usually 25'. ml. 'of. serum was equilibrated against, the buf-fer overnight and the small amount of sediment formedcentrifuged down before application'-of; the'serum to the

paper at the rate of ca. 1 ml. per hour. The volumes ofthe 32 fractions varied; each tube was diluted to 15 ml.with buffer. The optical density at 280 m/A was deter-mined using a Beckman Model DU spectrophotometer.Protein analysis was performed using a Spinco Model Rpaper electrophoresis apparatus. The binding of 4-4-cortisol was determined on each fraction by means ofequilibrium dialysis. For this purpose 5 ml. of eachfraction diluted with 5 ml. of saline was used. After di-alysis the samples containing barbital buffer were ad-justed to pH 9 to 10 to avoid the extraction of diethyl-barbituric acid, which, if present, gave significant weightto the samples.

The optical density was converted to concentration us-ing a separate factor, based on Kjeldahl determinationskindly performed by Mr. Kenneth Buchwald- .of this In-stitute, for each protein fraction. From this-the amountof binding per mg. of protein was calculated.

RESULTS

Earlier work (1) showed that undiluted plasmabound trace amounts of added C14-labeled cortisoland corticosterone essentially quantitatively.Therefore, in order to be able to demonstrate in-creases in extent of binding, dilution of the plasmawas necessary. Dialyses of diluted plasma (1: 5with 0.15 Msaline) against 0.15 Msaline showed,however, only minor differences in the binding ofthe various. classes of steroids. We thereforesought to cancel the effect of the albumin onbinding by dialyzing against a concentration ofalbumin equal to that in diluted plasma. Sincethe albumin concentration in the plasma is usu-ally approximately 5 Gm. per 100 ml., dialyses ofplasma diluted 1: 5 were performed against 1 Gm.per 100 ml. HSA.

TABLE I

Demonstration of corticosteroid-binding protein otherthan albumin in hmmanplasma

C14-Sterold %Bound*

Cortisol 89, 88, 82Corticosterone 72, 781 1-Deoxycorticosterone 77Cortisone 62'Pregnane-3,11,20-trione 63Progesterone 20, 15A4-Androsten-1 IjS-ol-3,17-dione 18Testosterone 11, 10Estrone 0, 0Estradiol 0, 0-Estriol 12

* Dialysis of plasma from clinically normal individualsdiluted 1:5, against 1 per cent HSA. Each number is theaverage of triplicate determinations done with plasma fromthree subjects.

385

W. ROY SLAUNWHITE, JR. AND AVERY A. SANDBERG

TABLE II

Demonstration of a corticosteroid-binding protein other than albumin in human plasma *

Dialyzed %Subject Condition against Steroidt Bound

Pe. Pregnant Saline F 95Pe. Third trimester 1%HSA F 94Cr. Third trimester Saline F 94Cr. Third trimester 1%HSA F 93Cr. Third trimester Saline B 93Fa. Third trimester Saline F 96Fa. Third trimester 1%HSA F 94Fa. Third trimester Saline B 94Ke. Third trimester Saline F 95Ke. Third trimester 1%HSA F 96Py. Third trimester Saline F 95Py. Third trimester 1%HSA F 94Wa. Normal female Saline Estriol 77Wa. Normal female 1%HSA Estriol 0

* Dialysis of plasma diluted 1: 5.t F, cortisol; B, corticosterone.

The results of a series of such experimentswith various steroids are shown in Tables I andII. Binding varies from zero for the estrogensto 80 to 90 per cent for cortisol (Table I). Thesame results were obtained for estrone and corti-sol by dialysis in the opposite direction. AsDaughaday (4) observed, slight alterations in thecortisol molecule decrease the ability of the mole-cule to bind to transcortin. Because the bindingof cortisol to transcortin is so much stronger thanthat to albumin, dialysis against albumin usuallylowered the amount of binding by only one ortwo percentage points as compared to dialysisagainst physiological saline (Table II). On theother hand, the value for estriol dropped to zero,indicating binding exclusively to albumin.

From Tables I and II it can be seen that corti-costerone binds nearly as effectively as cortisol.

TABLE III

Competition experiments with cortisol (F) andcorticosterone (B)

C14 Carrier % Carrier %Steroid F Bound* B Bound*

F 61sug. 44 1Ag. 49

3 22 3 315 15 5 30

B 861 62 1 673 59 3 615 57 5 59

* This is the per cent of the radioactive steroid bound topooled plasma proteins from normal subjects diluted 1:5and dialyzed against 1 per cent HSA.

Competition experiments were performed to de-termine if the sites of binding of the two steroidswere identical. The results shown in Table IIIdemonstrate that cortisol is more effective thancorticosterone in inhibiting the binding of radio-active cortisol whereas these two steroids arenearly equally effective in inhibiting the bindingof radioactive corticosterone.

The binding curve of cortisol to plasma is shownin Figure 1. This was obtained by determiningthe binding of C14-cortisol with the addition of 0to 10 jug. of carrier cortisol. From the equilib-rium equation for association of cortisol withplasma protein (Equation 2),

2) K - (Pr C)(Pr) (C)the following equation can be derived:

1 1 13) (Pr-C) K (Prt) (C) + (Prt)where (Prt) is the total concentration of protein,(Pr) and (C) are the concentrations of unboundprotein and cortisol, respectively, and (Pr.C) isthe concentration of protein-bound cortisol. Aplot of 1/(Pr*C) versus 1/(C) will be a straightline if the equilibrium constant K is the same forall sites (homogeneity of sites). Deviation; of thebinding curves from linearity reflects the hetero-geneity of the sites for cortisol. From the inter-cept of the y axis, it was calculated that (Prt)was 3.2 x 10-7 M per L. and 1.3 x 10-7 M perL. for the pregnancy and normal plasmas, respec-

386

ThNSCORTTN' A CORTtCMSTtROID-BINDING PROtEt3

tively. These values are actually the concentrationof sites. The protein concentration cannot bedetermined unless n, the number of sites, isknown.

When one-half of the sites are occupied, i.e.,(Prt)/(Pr C)= 2, then Kay. = 1/(C). Thisvalue is 3.0 X 107 L. per M at 40 C. for bothcases. Since the corresponding value for HSAis 0.5 x 104 L. per M (1), it is apparent thattranscortin binds cortisol approximately 6,000times more firmly than does albumin.

It would be advantageous to have an easiermeasure of the concentration of transcortin. Ef-forts to devise a measure have been thwarted by

I;L

lxSt

4a21

50 100

the presence of endogenous corticosteroids boundstrongly enough to defy removal except underthose circumstances which denature proteins. Analternative is to determine the amount of addi-tional cortisol required to reduce the binding of atracer amount of cortisol to an arbitrary level, say50 per cent, or the decrease in binding producedby the addition of a fixed amount of cortisol, say1 pg. The latter appears to be a more sensitiveindex as well as a more practical one. It shouldbe realized, however, that this measures the dif-ference between two variables, the level of trans-cortin and the level of corticosteroids.

Table IV compares such values for normal and

0-of *

O-O PREGNdANTl 03rd TRUSTER

0-e NORMAL0- I . . . .. .. ..

.3 I i 2 i S e

F4 CORTISOL ADOE

200

FIG. 1. THE BINDING CURVEOF CORTISOLTo 0.3 ug. of C1-cortisol in 0.15 M saline was added 0 to 10 Ag. of carrier

cortisol. This was dialyzed against plasma diluted 1: 5 with saline. Theinsert shows the data plotted as per cent C14-cortisol bound vs. the amount ofcortisol present. Each point is an average of determinations on three to sixserums. Of more theoretical interest is the larger plot, which is calculatedfrom the smaller one. The reciprocal of the concentration of bound cortisol isplotted against the reciprocal of the concentration of unbound cortisol, bothexpressed as L. per MX 60C. The solid and dashed lines denote plasmas ob-tained from normal and pregnant (third trimester) subjects, respectively.

387

W. ROY SLAUNWHITE, JR. AND AVERY A. SANDBERG

TABLE IV

The decrease in binding of 0.3 pug. of C14-cortisol uponthe addition of 1.0 pg. of cortisol *

Decrease% in %

No. of Bound boundsub- before after

State jects 1 pg. 1 pg.

Normal 6 82 I1.5t 26 :1 6.0Pregnancy, third trimester 12 96 :1 2.0 13 4 3.7Pregnancy, second trimester 9 93 i 1.6 20 :1 3.6Pregnancy, first trimester 3 87 + 5.5 21 a1 1.7

* Dialysis of plasma diluted 1:5 with saline againsteither saline or 1 per cent HSA.

t Average 4 standard deviation.

for pregnant subjects in each trimester. Aswould be expected from Figure 1, plasma frompregnant women showed a progressively smallerdecrease in binding as pregnancy progressed thandid those from nonpregnant women. Since thecorticosteroid level in the plasma of pregnantwomen is known to increase with the duration ofpregnancy (11), we interpret this finding as anindication of increased transcortin levels duringthis period of pregnancy.

Because transcortin exhibits electrophoretic mo-bility, does not dialyze, and generally exhibits thecharacteristics expected of a protein, it has beenassumed that it is a protein. As an additionalcheck on the validity of this assumption, a sampleof partially purified transcortin was digested over-night at 370 C. with trypsin at pH 7.5. Theability of transcortin to bind cortisol was com-pletely lost.

The preceding dialysis was performed in thecold room. To answer questions regarding the

TABLE V

Effect of temperature on the binding ability andstability of transcortin *

%Bound

Expt. Expt. Expt.Procedure 1 2 3

Equilibrium dialysis at 40 C. 85 90Equilibrium dialysis at 370 C. 61 73Stand at 370 C. for 72 hrs.; then equilib-

rium dialysis at 40 C. 87 88Stand at 370 C. for 72 hrs.- then cortisol

added and equilibrium dialysis at 40 C. 87 87 92Stand at 40 C. for 72 hrs.; then cortisol

added and equilibrium dialysis at 40 C. 86 94

* Two ml. of normal plasma diluted to 10 ml. with salinedialyzed against 30 ml. of 0.15 M NaCl and 0.31pg. ofC4Lcortisol.

i2-~~~~~~~~~-0.3

2 - -2 U620 2 2

10 N~~~~~~~~~~~~~~~~~~~~~~~~~~.2

N~~~~~~FATO NO.

25M.O NRA UMNPAMIN0.0 ARIA

BUFFER, P 8.6, 660VT A0290

4 8 1 6 20 4 2 3

FRACTION NO.

FIG. 2. CONTINuous FLow PAPERELECTROPHORESISOF25 ML. OF NORMALHUMANPLASMAIN 0.02 M BARBiTAI.BUFFER, PH 8.6, 660 VOLTS, CA. 50 MA.

The right-hand ordinate is the ratio of bound to un-bound cortisol per mg. of protein.

effect of temperature on binding and stability oftranscortin, the experiments outlined in Table Vwere performed on normal plasma. Elevationof the temperature produced a definite decrease inthe amount of binding of cortisol, indicating apositive enthalpy of binding. This effect was notobserved in the binding of cortisol and othersteroids to human serum albumin (12). Trans-cortin is stable in vitro at 370 C. for three days,for no decrease in binding was observed uponreturn of the system to 40 C. as compared tothe original determination or to a system whichhad remained at 40 C. throughout the experiment.Similar experiments on plasma from pregnantsubjects led to the same conclusion.

Pooled blood-bank plasmas, collected in acid-citrate-dextrose (ACD) solution and stored in arefrigerator at 40 C., for various ages (fresh tofour weeks old) showed essentially the same bind-ing capacity. Freezing of these plasmas, or anyother plasma, at -100C. for several weeks didnot affect their binding capacity for cortisol.

Continuous flow paper electrophoresis effectsa considerable purification of transcortin anddemonstrates that it is quite acidic. In Figure 2are shown the results of electrophoresis of normalserum in barbital buffer at pH 8.6. The trans-cortin has migrated as an a-globulin. In Figure3 is shown the result of electrophoresis of normalserum in acetate buffer at pH 5.2. It should benoted that the serum in this case was applied inthe center instead of the left quarter point. The,f- and y-globulins have moved toward the cathode

388

TRANSCORTIN: A CORTICOSTEROID-BINDING PROTEIN

while the a-globulins, albumin and transcortinhave moved toward the anode. Transcortin hasthe mobility of an a-globulin. This result is inaccord with that of Daughaday (5).

Fractions 22 through 26 inclusive (Figure 3)were pooled after withdrawal of one-third of thevolume of each tube for determination of binding.After dialysis against two changes of distilledwater to remove buffer, lyophilization yielded 10mg. of dry residue, indicating a 150-fold proteinconcentration over the original plasma. Furtherpurification and characterization are being con-

tinued.Transcortin is also present in varying degrees

in the blood of animals (Table VI). The dog ap-

pears to have very little transcortin while rab-

bits, guinea pigs, rats and perhaps alligators havesubstantial concentrations.

DISCUSSION

It is apparent from the data of our experimentsand from those published by Daughaday (4) thathuman and some animal plasmas contain a protein(or proteins) with a great avidity for the bind-ing of cortisol. We have named this proteintranscortin with the full realization that it may

actually constitute a system of several proteinsrather than a single one and that under abnormalconditions the changes in the binding capacity forcortisol may not be related necessarily to changesin transcortin concentrations. Nevertheless, we

believe that the name simplifies communication

11

0

0

6- X X ALBMN TRANSCORTIN

14 -* 0.D. a1.4O--OlPr-/C) /mg. PROTEIN

12.o-o 25ok0OOOOwO~~~~~w a8

D4 z

1.2 -

_ T _ - T - -_I4 9 12 16 20 24 asFRACTION NO.

FIG. 3. CONTINUOUS FLOw PAPER ELECTROPHORESISOF 25 ML. OF NORMALHUMANPLASMAIN 0.02 ACETATEBUFFER, PH 5.2, 800 VOLTS, 40 TO 60 MA.

The right-hand ordinate is the ratio of bound to unboundcortisol per mg. of protein.

TABLE VI

Binding of cortisol to animal plasma *

Amount Guineaof carrier Rabbit Dog pig Rat Alligator

0 Ag. 91t 39 83 86t 932 60 20 70 635 48 12 61 49

10 36 52 4020 8 38

* Plasma diluted 1:5 dialyzed against saline.t Corticosterone, 82 per cent.t Corticosterone, 75 per cent.

and does not commit any particular protein (gly-coprotein, lipoprotein, and so forth) as being thecortisol-binding protein until rigid proof of theidentity is established. In agreement with Daugh-aday (5) we believe transcortin to be associatedwith the a-globulin area in the electrophoreticseparations.

Since no specific method exists for determiningthe concentration of transcortin, recourse wasmade in our experiments, as well as in those ofDaughaday (4, 5), to the ability of human plasmato bind cortisol in vitro during equilibrium dialy-sis. The estimate of the transcortin binding ca-pacity under these conditions is only an approxi-mation. The reasons for this statement are asfollows. Although transcortin binds cortisol mostavidly, it appears that it can also bind corticos-terone, 1 1,8-hydroxyandrostenedione and othersteroids to a lesser extent. In addition, it wouldbe difficult to deny, until definite evidence is pre-sented, that transcortin does not bind conjugatesand metabolites, both unknown and known, ofmany other steroids. Since the determination ofthe concentration of these various steroid meta-bolites is at the moment impossible, it is apparentwhy the determination of the binding capacity ofhuman plasma for cortisol remains an approxi-mation. Nevertheless, the reproducibility andconstancy of the binding capacity of transcortinunder the conditions of our experiments offer atpresent the best method for the quantitative esti-mation of transcortin concentration.

The results of the competition experiments arein accord with the hypothesis that there are twotypes of sites at which corticosterone and cortisolmay be bound. For example, at one type of sitecortisol may be bound strongly and corticosteroneweakly so that the former would have a greater

389

W. ROY SLAUNWHITE, JR. AND AVERY A. SANDBERG

inhibiting effect on the binding of C14-cortisol.At the second type of site both steroids wouldthen be bound with an intermediate strength;otherwise, the first type of site would bind nearlyall the cortisol.

Using undiluted plasma, Daughaday (13) failedto show any differences between the binding ofnormal and pregnant subjects. On the otherhand, with the use of diluted plasma and the ad-dition of carrier cortisol we were able to show adefinitely increased transcortin capacity in theplasmas of pregnant women during the thirdtrimester. Daughaday's failure (13) to show dif-ferences between normal and pregnant plasmaswas due to the high equilibrium constant of trans-cortin combined with the use of a high proteinconcentration. The increased levels of trans-cortin during the third trimester of pregnancy areeven more pronounced in light of the possibilitythat the greatly increased levels of other steroids(progesterone and its derivatives, corticosteroids,and so forth) infringe on the capacity of transcortinto bind the added labeled and carrier cortisol.The identity of equilibrium constants for bothnormal and pregnant plasma indicates that noqualitative changes in transcortin have occurredas a result of pregnancy. In agreement withDaughaday (4) we believe that at so-called physio-logical concentrations of cortisol in plasma it isbound preponderantly by transcortin. It shouldbe pointed out, however, that equilibrium mustexist between transcortin-bound cortisol, cortisolpresent in solution and albumin-bound cortisol.Thus transcortin, like albumin, would functionin the transport of cortisol through the vascularsystem.

Although some of the concepts to be presentedare merely heuristic, we believe that they can beof help in furthering the identification and studyof the significance of transcortin. We proposethat transcortin-bound cortisol may be for allpractical purposes biologically inactive, whereasthat cortisol which is not bound to transcortin isphysiologically active. Thus, when the cortisolconcentration exceeds the transcortin capacity,marked physiological effects will appear. This isseen following the administration of pharmacologi-cal doses of cortisone or cortisol or during in-creased adrenocortical activity (stress, adreno-corticotropic hormone [ACTH] administration).

The lack of real evidence of hyperadrenocorticismduring pregnancy, in spite of the elevated concen-trations of cortisol, can best be explained on thebasis of increased transcortin concentrations. Onthe other hand, it is possible that in some patientswith Cushing's syndrome having normal levels ofcorticosteroids in the plasma, the disease may bedue* to the decreased concentrations of trans-cortin, resulting in an increased concentration ofnontranscortin-bound cortisol with resultant in-creased physiological effects. The levels of trans-cortin in this and in other conditions in whichadrenocortical function have been implicated as afactor are now under study in our laboratory. Itwill also be interesting to ascertain the role oftranscortin in the regulation of ACTHsecretionand whether transcortin-bound cortisol is meta-bolized in a fashion similar to that which is notbound to transcortin. These studies await thepurification and ultimate isolation of transcortin,so that it can be administered with or withoutcortisol.

Our failure to demonstrate a protein-bindingsystem, akin to transcortin, for the estrogens cor-roborates the findings of Bischoff and associates(9, 10) and indicates the lack of existence of aspecial protein for the binding of these steroids aspostulated by Roberts and Szego (8). Theseauthors indicated that Fraction III-0 of Cohnwas mainly responsible for the binding of theestrogens. Our data and those of Bischoff andco-workers (9, 10) point towards albumin as theprotein system primarily responsible for the bind-ing and transport of estrogens. Indeed, the dataof the present study indicates that albumin playsa paramount role in the binding, and hence prob-ably transport, of such steroids as progesterone,testosterone and some of the other androgens.

SUMMARY

There is in human plasma of normal and preg-nant subjects, as well as that of a number of ani-mals, a protein having a high affinity for cortisol(K = 3.0 x 107 L. per M) but which is present inlow concentrations (1.3 x 107 M per L.). Wepropose to call this protein transcortin. Competi-tion experiments showed that cortisol and corti-costerone are bound at two types of sites. Atone, cortisol is definitely bound more strongly

390

TRANSCORTIN: A CORTICOSTEROID-BINDING PROTEIN

than corticosterone, while at the second, they areapproximately equal in strength of binding. Es-trogens, androgens and progesterone are boundweakly or not at all by transcortin. A method ofmeasuring transcortin concentration is described.Using this method it was found that transcortinlevels increase during pregnancy. A speculativehypothesis was advanced to explain the physio-logical significance of transcortin. Basic to thishypothesis is the assumption that corticosteroidsbound to transcortin are biologically inactive. Theconcentration of corticosteroids in the plasmamust, therefore, exceed the effective concentrationof transcortin before physiological, and especiallypharmacological, effects of the corticosteroidsbecome evident. Transcortin is stable for at leastone week in plasma or in dilute solution at either4 or 370 C. It is destroyed by incubation withtrypsin. Continuous flow paper electrophoresis atpH 5.2 separates transcortin from the bulk of theplasma proteins, effecting a 150-fold purification.It is believed to be an a-globulin.

ACKNOWLEDGMENTS

Wewish to acknowledge the cooperation of Dr. RobertPatterson of the Buffalo Medical Group, who supplied uswith plasma from pregnant subjects; the generous dona-tion of human serum albumin by the American Red Crossand of C'4-labeled corticosteroids by the EndocrinologyStudy Section of the National Institute of Health; and thevaluable technical assistance of Miss Jean Rae Schubarg,Mr. Lawrence Beecher and Mr. Elek Karsay.

ADDENDUM

After submission of our manuscript an interesting articleby Dr. Ian E. Bush (Ciba Foundation Colloquia on Endo-crinology, 1957, 11, 263) on "The Physicochemical Stateof Cortisol in Blood" came to our attention. This ar-ticle describes experiments pointing to the existence of acortisol-binding protein which becomes saturated at theupper level of the normal range for plasma cortisol con-centrations.

REFERENCES1. Sandberg, A. A., Slaunwhite, W. R., Jr., and Antoni-

ades, H. N. The binding of steroids and steroidconjugates to human plasma proteins. Rec. Progr.Hormone Res. 1957, 13, 209.

2. Daughaday, W. H. Evidence for two corticosteroidbinding systems in human plasma (abstract). J.Lab. clin. Med. 1956, 48, 799.

3. Daughaday, W. H. Corticosteroid binding by aplasma alpha globulin (abstract). J. clin. Invest.1957, 36, 881.

4. Daughaday, W. H. Binding of corticosteroids byplasma proteins. III. The binding of corticosteroidand related hormones by human plasma and plasmaprotein fractions as measured by equilibrium dialy-sis. J. clin. Invest. 1958, 37, 511.

5. Daughaday, W. H. Binding of corticosteroids byplasma proteins. IV. The electrophoretic demon-stration of corticosteroid binding globulin. J. clin.Invest. 1958, 37, 519.

6. Daughaday, W. H. Binding of corticosteroids byplasma proteins. I. Dialysis equilibrium and renalclearance studies. J. clin. Invest. 1956, 35, 1428.

7. Daughaday, W. H. Binding of corticosteroids byplasma proteins. II. Paper electrophoresis andequilibrium paper eletcrophoresis. J. clin. Invest.1956, 35, 1434.

8. Roberts, S., and Szego, C. M. The nature of circu-lating estrogen: Lipoprotein-bound estrogen inhuman plasma. Endocrinology 1946, 39, 183.

9. Bischoff, F., Stauffer, R. D., and Gray, C. L. Physico-chemical state of the circulating steroids. Amer.J. Physiol. 1954, 177, 65.

10. Bischoff, F., and Stauffer, R. D. Orientation of cir-culating human estrogens by albumin. Amer. J.Physiol. 1957, 191, 313.

11. Migeon, C. J., Bertrand, J., Wall, P. E., Stempfel,R. S., and Prystowsky, H. Metabolism and pla-cental transmission of cortisol during pregnancy,near term in Ciba Foundation Colloquia on Endo-crinology. Boston, Little, Brown, 1957, vol. 11,p. 338.

12. Slaunwhite, W. R., Jr., Sandberg, A. A., and Antoni-ades, H. N. Further observations on the bindingof steroids and steroid conjugates to human plasmaproteins. In preparation.

13. Daughaday, W. H. Binding of corticosteroids byplasma proteins. V. Corticosteroid-binding globu-lin activity in normal human beings and in certaindisease states. Arch. intern. Med. 1958, 101, 286.

391