Journal of Clinical InvestigationVol. 46, No. 6, 1967

Fatty Acid Transport and Incorporation into HumanErythrocytes In Vitro *

RICHARDK. DONABEDIANANDARTHURKARMENt(From the Department of Radiological Science, The Johns Hopkins Medical Institutions,

Baltimore, Md., and the Collaborative Studies Section, National Heart Institute,Bethesda, Md.)

Summary. When human erythrocytes were incubated in vitro with 14C-la-beled free fatty acids bound to serum albumin, labeled fatty acids were in-corporated into erythrocyte triglycerides and phospholipids. The first stepin this reaction was the transfer of free fatty acids from the albumin to thecells. This transfer was rapid and reversible. The acids were distributedbetween albumin and cells according to the relative quantities of albumin andcells present. Each acid had a different distribution coefficient. At equi-librium, relatively larger fractions of the stearic and palmitic acids and smallerfractions of the oleic and linoleic were associated with the cells. All thesefatty acids were then slowly incorporated into phospholipids and triglycerides.The rate of incorporation of each was a function of its concentration in thecells, but larger fractions of the oleic and linolei, were incorporated than ofthe stearic, palmitic, myristic, or lauric. The two processes of transfer andincorporation thus had almost opposite selectivities for the different fattyacids. As a result, the fatty acids incorporated into triglycerides and phos-pholipids resembled in composition the fatty acids on the albumin except formoderately less stearic acid.

Introduction

Farquhar and Ahrens (1) have observed thatafter a change in the fatty acid composition of thediet, the fatty acid composition of human erythro-cyte (RBC) phospholipids changed. An increaseor decrease in the linoleic acid content of the dietproduced a similar but smaller change in the RBCcomposition. The change was complete and a newstable composition was reached after only 4 to 6weeks on the new diet. Because this time was shortcompared to the average life-span of the RBC, thechange in composition could not be explained en-tirely by the incorporation of different fatty acidsinto maturing RBC in the bone marrow. It

* Submitted for publication May 24, 1966; acceptedMarch 10, 1967.

Supported by U. S. Public Health Service grantGM11535.

tAddress requests for reprints to Dr. Arthur Karmen,Dept. of Radiological Science, The Johns Hopkins Medi-cal Institutions, 615 N. Wolfe St., Baltimore, Md. 21205.

seemed more likely that the fatty acids of mature,circulating RBCwere being replaced. A possiblemechanism might be the replacement of the fattyacids of the RBC phospholipids with fatty acidsfrom the plasma free fatty acid pool. Oliveira andVaughan (2) and Mulder, De Gier, and VanDeenen (3, 4) have observed that RBC in vitrocan incorporate FFA into cellular phospholipids.This paper describes a study of this in vitro reac-tion. It was designed particularly to provide moreinformation about the selectivity of human RBCfor different long chain fatty acids, which may beone of the determinants of RBC fatty acid com-position.

Methods

Red blood cells. Freshly drawn blood was mixed withheparin and centrifuged at 100,000 g for 30 minutes.The plasma and the upper third of the cell column wereremoved. The cells were then resuspended in 3 vol ofisotonic saline and recentrifuged, and the supernatantand top of the column were again removed. The leu-

1017

RICHARD K. DONABEDIANAND ARTHURKARMEN

kocyte count of the remainder varied between 400 and800 per mm' packed RBC.

Fatty acids. Fatty acids labeled with "C in the car-boxyl carbon were purchased.' Specific activities werebetween 12 and 38 mc per mmole. For removal of anylabeled short chain acid impurities, each fatty acid wasdissolved in 10 ml hexane: glacial acetic acid (1: 1, vol/vol), 1 ml water was added, and the aqueous phase wasremoved and discarded (5). The hexane phase was thenwashed with water to remove any remaining acetic acidand then shaken with an equal volume of 0.1 N potas-sium hydroxide in 50% ethanol (6) to separate the fattyacids from any fatty acid esters present. The ethanol-KOHphase, which contained the fatty acids, was sepa-rated, acidified with 1 N H2S04, and extracted threetimes with hexane. Gas-liquid chromatography (GLC)of the methyl esters prepared from each acid revealedthat the radioactivity emerged from the column with thesame retention time as the corresponding unlabeled ester.The relationship between radioactivity injected into theGLC column and the peak area on the GLC record ofradioactivity was the same for each of the fatty acids andequal to that of the standard. These fatty acids weretherefore pure within the limits of these methods.

Human serum albumin was obtained commercially.2As supplied, it contained 0.3 limole fatty acid per 50 mgalbumin by GLC analysis using heptadecanoic acid as in-ternal standard (7). [Lauric (12: 0) = 2%; myristic(14:0) =3%; palmitic (16:0) =28%; stearic (18:0) =5%; oleic (18: 1) = 36%; linoleic (18: 2) = 19%;arachidonic (20: 4) = 7%.] The purified 14C-labeledfatty acids were bound to the albumin by dissolving themin 2 ml of heptane and then layering the heptane over20 ml of a solution containing albumin in buffer (0.061M in P04, 0.077 M NaCl, pH 7.4) (8) in a stopperedculture tube. The tube was then flushed with nitrogen andsealed, and the contents were allowed to equilibrate for 24hours. More than 90%o of the "C was transferred to thealbumin phase by this procedure. The heptane phasewas then removed and the "C fatty acids therein were re-covered. (We chose this method of binding fatty acidsto albumin in preference to the simpler method of add-ing the albumin to a dilute aqueous solution of the fattyacid salt to be certain that all the labeled fatty acid wasbound to albumin and not in suspension as particles ofundissolved soap.) The same method was used to trans-fer unlabeled fatty acids to albumin when the quantity orthe composition of the FFA on the albumin was varied.Cellulose strip electrophoresis of the albumin preparedby this method revealed a homogeneous band identical tothat from the original albumin. Liquid scintillationcounting of sections of the electrophoretic strip revealedno significant radioactivity except in the albumin band.

Incubation, extraction, and thin-layer chromatography(TLC). Except where otherwise indicated, RBC were

1New England Nuclear Corp., Boston, Mass. (lauric,myristic, palmitic, stearic), and Applied Science Labora-tories, State College, Pa. (linoleic).

2 Courtland Laboratories, Los Angeles, Calif.

incubated at 370 C in buffer containing albumin, labeledfatty acid, and in most instances sufficient additional al-bumin so that the molar ratio of FFA to albumin andthe FFA composition were not appreciably differentfrom that of the original albumin. At the end of the in-cubation period, the cells were centrifuged and resus-pended five times with approximately 10-ml portions ofisotonic saline. That this procedure removed the albu-min was demonstrated by using 'MI-labeled albumin as atracer. (No 'I radioactivity above background wasdetectable afterwards.) The erythrocyte lipids werethen extracted into chloroform: methanol (1:1, vol/vol)three times according to the method of Ways and Hana-han (9). Samples of the chloroform phase were evapo-rated to dryness and assayed for radioactivity by liquidscintillation counting. The remainder was evaporated toa small volume and analyzed by thin-layer chromatog-raphy on silica gel G plates using a solvent consistingof hexane, diethyl ether, methanol, and glacial aceticacid (225: 50: 7: 5 vol/vol). Standard lipids were chro-matographed on each plate. In this TLC system, phos-pholipids remain at the origin, FFA migrate with anRf of about 0.6, and cholesterol esters and triglyceridesmigrate close together with an Rf of about 0.9. Theplates were sprayed lightly with 2'-7'-dichlorofluoresceinand examined under ultraviolet light. Only free cho-lesterol and phospholipids could be visualized. The re-gions corresponding to the phospholipid, FFA, and tri-glyceride of the standard were isolated and scraped offthe plate. The remaining silica gel was then also scrapedoff. The radioactivity in the spot near the solvent frontfrom three separate RBC incubations was clearly iden-tified as triglyceride by chromatography in a less polarsystem. In subsequent analyses, this spot was analyzedwithout further subfractionation. The lipids were theneluted from the silica gel, the phospholipids with methanoland the FFA and triglycerides with chloroform. A sam-ple of each eluate was evaporated to dryness and as-sayed for radioactivity by liquid scintillation countingat approximately 78% efficiency. From 80 to 90% ofthe radioactivity applied to the plate was recovered inthe triglyceride, FFA, and phospholipid fractions. Eacheluate was then evaporated to dryness, and the residuewas incubated at 60° C for 16 to 20 hours with 2 ml of3% sulfuric acid in methanol (vol/vol) to convert thefatty acids to their methyl esters. After incubation, 2 mlof water was added, and the methyl esters were extractedinto light petroleum ether.3 The petroleum ether solu-tion of methyl esters was then evaporated almost to dry-ness, and a portion consisting of most of the sample wasinjected into the GLC column.

3 The completeness of transesterification of the fattyacids in phospholipids by this procedure was studied byWhyte (10). His results indicated that transesterificationwas essentially complete but that particular care had tobe exercised in the extraction of the methyl esters de-rived from phospholipid from the aqueous methanol toensure that extraction was quantitative. He found thatall the long chain esters were equally affected.

1018

FATTY ACID INCORPORATIONINTO HUMANERYTHROCYTES

Because of the possibility of selective loss of unsatu-rated acids during TLC, care was taken to minimize thetime that the lipids were permitted to remain on the TLCplate or on the silica gel exposed to air before they wereeluted. Many analyses of mixtures of triglycerides andFFA before and after they were subjected to TLC withthe procedures used here revealed no selective loss oflinoleic or any other acid.

Gas-liquid chromatography. The methyl esters wereanalyzed by GLC at 1900 C in a Packard model 801 gaschromatograph 4 on coiled glass columns 6 feet longcontaining Chromosorb Wcoated with ethylene glycolsuccinate polyester, 12% by weight. The effluent of thecolumn was split 6: 1. The smaller fraction was di-rected to a hydrogen flame ionization detector, the largerto a combustion tube filled with cupric oxide, which washeated to 7000 C. The effluent of the combustion tubewas passed through a tube containing magnesium per-chlorate to remove water and then through either of tworadiation detection systems. In the first, suitable quan-tities of additional argon, carbon dioxide, or both wereadded to make a total flow of 200 ml per minute consist-ing of about 90% argon, 10% carbon dioxide. The com-bined gases were then passed through a 50-ml gas flowproportional counter with a thin window (Packard model210 Flow Window Detector) 4 (11, 12). In the second,the effluent of the combustion tube was delivered to aflow-through anthracene crystal scintillation detector(Packard Flow Detector)4 (13). The counting rate wasrecorded on a strip chart recorder, and the distribution ofradioactivity in the sample was calculated from the -areasunder the peaks of counting rate vs. time.

Plan of experiments. The effects of time of incuba-tion, concentration of FFA, and addition of cofactorson the incorporation of fatty acids into RBC phospho-lipids and triglycerides were studied using palmitate-"Cas the only "C-labeled tracer. The transfer of the vari-ous FFA from albumin to RBC and the incorporationof these acids into RBC lipids were studied by incubatingthe RBC with albumin to which a mixture of six "C-labeled acids was bound. In each of the latter experi-ments, the albumin-FFA both before and after incuba-tion were eXtracted and analyzed at the same time as theRBC lipids. From the distribution of "C among thefatty acids of each lipid, the comparative or relativefractional uptakes of the different fatty acids from al-bumin-FFA or RBC-FFA, considered "precursors," intoRBC phospholipids and triglycerides, considered "prod-ucts," were calculated. The basis for these calculationswas as follows. As in most experiments in which iso-topes are used, it was assumed that the "C-labeled acidswere not distinguishable from their 'C counterparts.The fractional incorporation or uptake of an acid in atime interval was then given by the ratio of "C in thatfatty acid in the product to the "C in that fatty acid inthe precursor. For comparing fractional uptakes, thecontribution of each fatty acid to the total radioactivityin each product, in per cent, was divided by its con-

4 Packard Instrument Co., Downers Grove, Ill.

tribution in the precursor. These ratios were then "nor-malized" by dividing by the ratio for palmitic acid. Theresulting numbers, the relative fractional uptakes, werethus derived solely from the distribution of radioactivity.The `C fatty acid composition of the precursor and prod-uct did not enter into the calculation. It will be shownthat the relative fractional uptakes of different fatty acidsdid not vary when the ratios of the quantities of fattyacid present were changed. The usefulness of relativefractional uptakes as measures of the selectivity of theRBC for different fatty acids is based primarily on thisobservation.

Results

Time course of incorporation of fatty acid intophospholipids and triglycerides. WhenRBCwereincubated with albumin to which 14carbon-labeledpalmitate was bound, the radioactivity incorpo-rated into phospholipids and into triglycerides in-creased steadily for 4 hours, after which furtherincorporation was minimal (Figure 1). In thisexperiment, the rate of incorporation of radio-activity into phospholipid was greater than intotriglyceride; in others, the reverse was true. Thequantity of 14C in the RBC-FFA fraction was thesame in all samples taken.

30-

(I)az

W0I--Ia-

CL

25-

20-

15 -

10-

5-

PL

I I I I I I

1 2 3 4 5 6

HOURSOF INCUBATIONFIG. 1. TIME COURSEOF INCORPORATIONOF PALMITATE-

"C INTO RED BLOOD CELL (RBC) PHOSPHOLIPIDS (PL)AND TRIGLYCERIDES (TG). The ordinate of the Figureis counts per minute per milliliter RBC. The incubationmixture consisted of 10 ml washed RBC, 10 ml buffer,70 mg albumin (0.6 j.mole FFA), and 5 ,uc (8.8 X 10'cpm) palmitic acid-"C. The FFA fraction of each RBCsample contained 200,000 cpm per ml RBC.

1019

RICHARD K. DONABEDIANANDARTHURKARMEN

TABLE I

Effect of glucose and cofactors on fatty acid incorporationinto phospholipids (PL) and triglycerides (TG)

in red blood cells (RBC) stored in albuminovernight before incubation*

14C incorporated

Added nutrients No added nutrientsTime of

incubation PL TG PL TG

hours cPm1 786 207 600 1313 1,808 320 911 1405 4,183 508 2,209 136

* Incubation mixtures contained 2.74 X 106 cpm 14C-labeled fatty acids (12:0, 14:0, 16:0, 18:0, 18:1, 18:2),50 mg albumin, 4 ml buffer, and 3 ml washed RBC. Inone mixture the buffer contained, in addition, 3 mgglucose, 0.2 jsmole coenzyme A, 2 jumoles ATP, and 3umoles magnesium chloride (2).

Effect of glucose and cofactors on fatty acid in-corporation into phospholipids and triglycerides.Freshly drawn RBC incorporated palmitate-14Cinto triglycerides and phospholipids at the samerate in mixtures containing added ATP, coenzymeA (CoA), magnesium chloride, and glucose as inbuffer alone. RBC that had been stored in albu-min and buffer at 5° overnight before incubationincorporated more from the media containing theadded nutrients. There was no detectable uptakeinto triglycerides by the stored cells unless thesenutrients were added (Table I). Although theinitial rate of 14C-labeled FFA incorporation intophospholipids was the same, after 1 hour the in-corporation in the absence of nutrients was less.

Effect of concentration of free fatty acids on the

TABLE II

Effect of ratio of palmitic acid to albumin on theincorporation of palmitic acid into RBC

TG and PL

Palmitic acidincorporated into:

Incubation FFA (16:0)/mixture 55 mgalbumin TG PL TG/PL

pumoles mjsmoles mpmoles1 0.2 0.8 0.7 1.12 0.3 1.5 0.9 1.63 0.7 3.8 1.8 2.14 0.9 5.2 1.8 2.75 1.2 7.4 2.4 3.06 2.2 11.0 4.4 2.5

incorporation into phospholipids and triglycerides.When the quantity of albumin in the incubationmixture was held constant, the micromoles of fattyacid incorporated into phospholipids and triglyc-erides increased with increase in the concentra-tion of FFA on the albumin over the range from0.2 to 2.2 jLmoles per 55 mg albumin. With in-crease in the molar ratio of FFA to albumin, in-corporation into triglycerides was even more

markedly increased than into phospholipids (Ta-ble II).

Uptake of different fatty acids into RBCphos-pholipids and triglycerides. At the end of 3 hoursof incubation, in a typical experiment in which theratio of albumin to RBCwas low, one-fifth of thetotal radioactivity was associated with the cellsand four-fifths was washed out with the albumin.Approximately 20% of the RBCradioactivity was

in the triglyceride fraction, 12% in the phospho-

TABLE III

Fatty acid incorporation into RBCfrom albumin after 3 hours of incubation*

Distribution of 14C

cpm X 10-6 %of totalTotal incubated after 3 hours 5.25 12:0 14:0 16:0 18:0 18:1 18:2

Albumin (10.5 mg) 4.20 :1 0.15 5.2 8.9 19.4 16.7 30.3 19.7RBCtotal (2 ml) 1.05 :1 0.15RBCdistribution

FFA 0.77 i 0.04 3.0 8.2 22.1 45.8 17.4 3.1(73 ±44%) ±0.3 41.1 41.1 ±1.5 ±1.6 ±0.3

Triglyceride 0.19 -t 0.04 2.8 10.0 22.5 7.7 36.0 21.4(18 ± 4%) 40.3 ±1.1 ±3.0 +0.5 ±t2.0 ±2.2

Phospholipid 0.095 i 0.007 0 5.5 27.1 9.7 38.6 19.7(9.1% ±1 0.7 %) ±0.8 ±1.5 ±1.7 41.4 ±1.8

* Each incubation mixture contained 2.0 ml packed RBCand 3.0 ml buffer containing 10.5 mg albumin. Thealbumin contained 0.061Amole FFA and 5.25 X 106 cpm '4C (approximately 3 lsc). The analyses are averages ± standarddeviations of results obtained by incubating RBC from three healthy subjects, each in duplicate. The numbers inparentheses below the mean and standard deviation are the percentages of the total counts in the RBClipids.

1020

FATTY ACID INCORPORATIONINTO HUMANERYTHROCYTES

lipid, and the remainder, or almost 70%o, was freefatty acid (Table III). The distribution of 14Camong the fatty acids was different in the tri-glycerides, phospholipids, and FFA fractions ofthe red cells, and each, in turn, was different fromthe distribution in the FFA on the albumin. Thetotal radioactivity and the distribution of radio-activity in each lipid were similar in each of theblood samples tested. This similarity is exempli-fied in the analyses shown in Figures 2 and 3.Comparison of the distribution of radioactivityamong the FFA of the albumin and the RBCshowed that larger fractions of free stearic andpalmitic acids were associated with the RBCcom-

pared to oleic and linoleic. The distribution inthe phospholipids and triglycerides was similarto that in the albumin-FFA except for less 14C-labeled stearic acid in phospholipids and triglycer-ides and the lack of uptake of 14C-labeled lauricacid into phospholipids (Table IV).

SUBJECT K.". A A kL2 FFA

SUBJECT ItD.

FFA

PL

P1

A ~~~~~TG

18:2 18:1 18:0 16:0 14:0 12:0

FIG. 2. DISTRIBUTION BY GAS-LIQUID CHROMATOGRAPHY

(GLC) OF 14C AMONGALBUMIN FFA, RBC-FFA, RBCPHOSPHOLIPIDS, AND RBC TRIGLYCERIDES AFTER 3 HOURS

OF INCUBATION OF RBC OF A NORMALHUMANSUBJECTWITH ALBUMIN.

18:2 18:1 18:0 16:0 14:0 12:0

FIG. 3. DISTRIBUTION BY GLC OF '4C AMONGALBUMIN

FFA, RBC-FFA, RBC PHOSPHOLIPIDS, AND RBC TRI-

GLYCERIDES AFTER 3 HOURSOF INCUBATION OF RBC OF A

SECONDNORMALHUMANSUBJECT WITH ALBUMIN.

Samples of incubation mixtures processed al-most immediately after completion contained asmuch FFA associated with the RBC as thosetaken at the end of 6 hours of incubation.

TABLE IV

Relative fractional incorporation of fatty acids fromalbumin into RBClipids (relative to

palmitate = 1.00)*

Fatty acid

12:0 14:0 16:0 18:0 18:1 18:2

RBCFFA 0.51 0.81 1.00 2.40 0.50 0.14

RBCTG 0.46 0.96 1.00 0.40 1.02 0.94

RBCPL 0 0.44 1.00 0.41 0.90 0.71

* Calculated from data presented in Table III.

1021

RICHARD K. DONABEDIANANDARTHURKARMEN

TABLE V

Independence of relative fractional uptakesof fatty acid composition*

Fatty acid

12:0 14:0 16:0 18:0 18:1 18:2

Albumin I 12C % 2.4 3.1 45.2 7.5 30.1 11.6Albumin I 14C % 2.5 1.9 25 16 35 19

Phospholipid I 14C % 0 1.1 30 8.8 41 18

Relative fractional uptake 0 0.48 1.0 0.46 1.0 0.79from albumin I tophospholipid I

Albumin II 12C % 2.2 3.5 31 5.7 38 20Albumin II 14C % 2.2 3.3 23 15 35 21

Phospholipid II 14C % 0 1.9 26 10 39 23

Relative fractional uptake 0 0.51 1.00 0.59 0.98 0.97from albumin II tophospholipid II

* Each tube contained albumin, 17 mg, '4C-labeled FFA, 3.5 X 106cpm, and buffer, 3.0 ml, and was incubated for 3 hours.

Independence of relative fractional uptake ofthe composition of FFA on albumin. When thecomposition of the 12C-labeled FFA on the albu-min (and on the RBC-FFA) was intentionallychanged, the relationship between the distributionof 14C in the RBC lipids and on the albumin wasnot significantly affected (Table V). We con-cluded that the relative fractional uptakes wereindependent of the relative concentrations of FFApresent.

Transfer of FFA from albumin to RBC. Whensolutions of albumin-FFA-'14C and of RBCweremixed together, the 14C-labeled FFA were dis-tributed between the two in proportion to therelative quantities present. The ratio of 14C permilligram albumin to 14C per milliliter cells variedonly from 1.7 to 0.9 as the albumin with its associ-ated FFA was increased from 3 to 78 mg per mlcells (Table VI). Approximately the same quan-tity of fatty acid was therefore associated with thecells. When the ratio of fatty acid to albumin wasincreased, correspondingly more fatty acid wasassociated with the cells. The RBC relative frac-tional uptakes of FFA from the albumin were notchanged when the composition of the FFA waschanged. Increasing the palmitic acid on thealbumin from 20 to 70%7o of the total FFA did notchange the distribution of radioactivity in theRBC-FFA fraction.

The rate of transfer of FFA from albumin toRBCwas rapid. There was no more 14C-labeled

FFA associated with cells after 4 hours of incu-bation with albumin than after a brief exposure(about a minute) followed by dilution in 10 volof saline and centrifugation. Exactly how rapidlythis transfer occurred could not be determined be-cause of the time required to separate the albu-min completely from the cells (about 10 minutes).

The FFA could also be removed from the RBCrapidly by suspending the FFA in albumin. Twoml of washed RBC was mixed with 1 ml cold(40 C) saline containing 12 mg albumin to whicha mixture of six 14C-labeled fatty acids was bound.The albumin was then removed by three times re-peated centrifugation and resuspension in freshcold saline. The RBC were then resuspendedbriefly in a new albumin solution containing itsnormal unlabeled FFA, but no 14C, after whichthis albumin was in turn removed. This entireprocess of resuspending the cells in albumin andthen removing the albumin was repeated fourtimes. Measurement of the radioactivity associ-ated with the cells after each centrifugation re-vealed that most of the 14C was removed in thefirst albumin solution. After four resuspensionsin albumin, less than 1%of the added radioactivityremained associated with the cells.

The role of free fatty acid associated with RBCin the incorporation of fatty acid into phospholipidsand triglycerides. The uptake of labeled fatty acidinto phospholipids and triglycerides was similar incells exposed only momentarily to albumin-FFA-14C, and then incubated in buffer, and in cells in-cubated in albumin-FFA-14C for the entire incu-bation period. Six ml of RBCwas suspended in

TABLE VI

Effect of albumin concentration on immediate uptakeof FFA (six FFA, 12:0-18:2) by RBC*

ALB/RBC 14C/RBC 14C/ALB ALB/RBC

mg/ml cpm/ml cPm/mg (cPm/mg)/(thousands) (thousands) (cpm/ml)

3 41 70 1.718 19 31 1.63

13 13.7 19 1.3818 11.5 14 1.2228 7.7 9 1.1753 5.0 5 1.078 3.6 3 0.9

* Each incubation mixture consisted of 1.0 ml RBCand buffer containing albumin in the quantities shown ina total volume of 3.0 ml. The albumin in each instancecontained approximately 0.006 Mumoles FFA per mg.

1022

FATTY ACID INCORPORATIONINTO HUMANERYTHROCYTES

TABLE VII

Incorporation of 14C-labeled fatty acid into RBCphospholipids and triglycerides*

ALB present ALB removed

cpm after incubation (thousands)/2 ml RBC

Experiment 1RBCFFA 612 480RBCPL 80 76RBCTG 101 39

Experiment 2RBCFFA 717 754RBCPL 88 78RBCTG 132 125

* Comparison of uptake when albumin was removedafter 2 minutes and uptake when albumin was presentthroughout the 3-hour incubation period. In each ex-periment, the incubation mixture was prepared by suspend-ing 6 ml packed cells in 6 ml buffer containing 35 mgalbumin and 22 X 106 cpm 14C-labeled fatty acids. Theresults above are analyses of 4-ml samples containing2 ml RBC.

buffered albumin to which a mixture of six la-beled fatty acids was bound. After about 1 min-ute, three equal samples were removed and placedin separate tubes. The first was incubated asusual. The second and third were centrifuged, re-suspended in saline, and recentrifuged to removethe albumin. Chloroform-methanol was thenadded to the second sample to extract the lipids.The third was incubated along with the first. Atthe end of 3 hours of incubation, similar quanti-ties of 14C were incorporated into phospholipidsin the first and third tubes. Moderately less was

PL TG

ALB INC

BUF INC

. 6w0 l0 Wl *2 120U 10 30 W *2

FIG. 4. DISTRIBUTION OF "C AMONGFATTY ACIDS IN

RBC TRIGLYCERIDES AND PHOSPHOLIPIDS WHEN CELLSWEREEXPOSEDTO ALBUMIN BRIEFLY THEN INCUBATED IN

BUFFER (BUF-INC) COMPAREDTO THE DISTRIBUTIONWHENTHE SAMECELLS WEREINCUBATED IN ALBUMIN FORTHE ENTIRE PERIOD (ALB-INC).

FIG. 5. DISTRIBUTION BY GLC OF 14C AMONGFFA OFALBUMIN (ALB-FFA), FFA ON RBC AFTER EXPOSURETO ALBUMIN FOR 1 MINUTE (1-FFA), AND 1-FFA INCU-BATED FOR 4 HOURSIN BUFFER (2-FFA). The fatty acidsare, right to left, 12: 0, 14: 0, 16: 0, 18: 0, 18: 1, and18:2.

incorporated into triglycerides in the mixture fromwhich the albumin had been removed (TableVII). Distribution of 14C among fatty acids ofthe phospholipids was also similar in both tubes,and there were only minor differences in the in-corporation into triglycerides (Figure 4). Thetotal 14C and the distribution of 14C in the RBC-FFA in all three tubes were also similar (Figure5). These results indicated that the 14C-labeledfatty acids incorporated into RBC phospholipidsand triglycerides during incubation could have

FFA

2

FFA

1023

A

RICHARD K. DONABEDIANANDARTHURKARMEN

TABLE VIII

Fractional incorporation of fatty acids from RBC-bound FFA into RBClipid esters(relative to palmitate = 1.00)*

Fatty acid

12:0 14:0 16:0 18:0 18:1 18:2

RBCFFA (from data in Table III) 0 0.54 1.00 0.16 1.80 5.07

From total of 11 different blood samples 0 0.48 1.00 0.15 2.06 4.104:10.15 :10.02 :4:0.75 ±41.27

RBCFFA (from data in Table III) .90 1.18 1.00 0.16 2.04 6.71

From total of 11 different blood samples 1.00 1.81 1.00 0.16 1.70 5.03±0.79 :i 1.01 :10.05 40.69 ±2.60

* The relative fractional uptakes and standard deviations thereof given here were means calculated from 11 differentblood samples each analyzed in duplicate. The relative fractional uptakes calculated from the three samples in theexperiment summarized in Table III, included in the 11, are presented for comparison.

come from (or through) the RBC-FFA fraction.On this basis, the relative fractional uptakes ofthe different fatty acids into triglycerides andphospholipids from the RBC-FFA were calcu-lated. Fractional uptakes of linoleic and oleicacids from RBC-FFA were from two to seven

times greater than those of palmitic, whereas thoseof stearic were only 0.15 as high. The fractionaluptakes of lauric and myristic acids into triglycer-ides were approximately equal to that of palmitic.A smaller fraction of the myristic acid and none

of the lauric were incorporated into phospholipid(Table VIII).

Comparative binding of various fatty acids toalbumin. To estimate the relative affinities (8)of albumin for different fatty acids, a mixture ofsix fatty acids dissolved in 2 ml heptane was care-

fully layered over 20 ml phosphate-saline buffer,pH 7.35, containing 70 mg albumin. This was

shaken gently in a Dubnoff shaker at room tem-perature for 36 hours, when most of the albuminsolution was carefully removed from below the

heptane. Most of the heptane (more than 90%o)was then removed and equilibrated with 40 mlphosphate-saline buffer for 48 hours. The distri-bution of 14C among the fatty acids in the albumin,saline, and heptane after its exposure to the al-bumin was then analyzed by GLC.

At this level of concentration of FFA on albu-min, lauric and myristic acids were bound to al-bumin less strongly than palmitic, stearic, oleic,and linoleic acids. Oleic acid was most stronglybound (Table IX).

Studies of abnormal human RBC. FFA trans-fer from albumin to RBC and incorporation offatty acids into phospholipids and triglycerideswere surveyed in blood samples from a few pa-

tients with a variety of diseases, including one eachwith primary hypercholesterolemia, congenital he-molytic anemia, abetalipoproteinemia, and Fabry'sdisease. RBC drawn from a normal individualwere incubated at the same time with the same

mixtures as the cells from each patient.The transfer of FFA to RBCand their uptakes

TABLE IX

Comparative binding of fatty acids to albumin: distribution of FFA radioactivity in bufferin equilibrium with albumin-bound FFA

14C distribution

4C total 12:0 14:0 16:0 18:0 18:1 18:2

cPm %Albumin 38 X 106 11 10 19 16 27 17Buffer 42.7 X 103 35 29 10 8 10 9

Albumin 12:0/buffer 12:0 1.0 1.0 5.8 7.0 9.9 6.4

1024

FATTY ACID INCORPORATIONINTO. HUMANERYTHROCYTES

and relative fractional uptakes into triglyceridesand phospholipids were similar to those in the con-trol normal cells. The relative fractional uptakeof linoleic acid into phospholipids and triglyceridesin the RBC from the patient with abetalipopro-teinemia was normal, even though the 12C-labeledlinoleic acid was less than 17o of the fatty acidspresent.

Discussion

When human erythrocytes were shaken withserum albumin to which labeled FFA were bound,a fraction of the FFA was transferred to the cells.Although the precise rate of this transfer was toorapid to be measured by the technique employedhere, there was no more 14C in the RBC-FFA inthe samples incubated for 4 hours than in thesamples exposed only briefly to the albumin. Theassociation of the FFA with the cells was reversi-ble; the labeled acids could be removed completelyby washing the cells several times with solutionsof albumin. Increasing the concentration of agiven acid on the albumin increased the quantityof that acid transferred to the cells proportionately.Increasing the ratio of albumin to cells increasedthe fraction of the total labeled acid that was boundto the albumin. It thus appeared that the FFAwere distributed between the albumin and theRBCjust as they might have been distributed be-tween two immiscible solvents that were shakentogether. The distribution coefficient, calculatedas FFA-14C per milligram albumin: FFA-14C permilliliter RBC, decreased by only one-half whenthe albumin was increased 25-fold. Each of thefatty acids was distributed between the albuminand the cells differently; the fraction of thestearate and palmitate associated with the cellsat equilibrium was larger than the fraction ofoleate and linoleate.

Goodman has reported that RBC bind FFAwith affinities similar to those of the "secondclass" of binding sites of albumin (14). The re-sults we obtained are not inconsistent with Good-man's model.

When the albumin and RBC were incubatedtogether, the labeled fatty acids were slowly in-corporated into cellular triglycerides and phos-pholipids. This reaction requires energy (2).Since addition of glucose, CoA, ATP, and mag-nesium ions had no effect on incorporation in

freshly drawn erythrocytes, an adequate quantityof these nutrients was probably present. The re-quirement for these nutrients was more ap-parent in cells stored overnight and then incu-bated in buffer alone before the labeled acidswere introduced.

The FFA esterified by the RBCwere apparentlyfirst bound to the cells as FFA. The pattern andquantity of incorporation of 14C when the cellswere exposed to the albumin momentarily, re-moved, and then incubated in buffer were verysimilar to those when the cells were incubated inthe presence of albumin.

Of three groups that measured uptakes of 14C-_labeled FFA into RBCphospholipids (2, 3, 15),only Michaels, Wood, Mullin, and Kinsell alsoobserved uptake into triglycerides. In our ex-periments, although there was very little triglycer-ide in the RBC samples, the identification of thelipid into which the 14C was taken up as triglycer-ide is fairly sure. The retardation factor of the4C on TLC was the same as that of standard tri-

glycerides. It was clearly separated from the FFAand from any of the other usual fatty acid-con-taining lipids. Its 14C fatty acid composition wasmarkedly different from that of the FFA fraction.The possibility was considered that it was a sub-fraction of the FFA isolated as an artifact of TLC.Against this was the fact that it was not present insamples taken from the mixture early in the in-cubation and that it increased slowly. Of all theparameters we measured, the quantity of 14C in-corporated into triglycerides was most sensitiveto changes in the concentration of FFA, the lengthof time the cells were stored before incubation,the presence of nutrients, and the method of han-dling the cells. This sensitivity probably explainsthe different results obtained in different labora-tories.

Wealso considered the possibility that the FFAwere incorporated into triglycerides by the leuko-cytes remaining in the mixture rather 'than theRBC. However, Elsbach, in his studies of uptakeof FFA by leukocytes (16), found that they in-corporated fatty acids from albumin into phos-pholipids and triglycerides at almost the same rate.If we attribute the uptake of 14C into triglyceridesto leukocytes, we should also attribute at' leastsome of the uptake into phospholipids to them aswell.' Unfortunately, our attempts to remove the

1025

RICHARD K. DONABEDIANAND ARTHURKARMEN

leukocytes completely were not successful. Therewas, however, no correlation between the num-ber remaining and the quantity of FFA incorpo-rated into triglycerides. There was abundant up-take in mixtures containing as little as 300 to 400leukocytes per mm3 RBC. On this basis, wetentatively concluded that the RBCwere respon-sible for the triglyceride incorporation.

The relative fractional uptakes of the differentFFA into RBC lipids did not change appreciablywhen the composition of the FFA pool waschanged. It was therefore not necessary to usea precursor pool of equimolar or any other fixedcomposition to measure them. The albumin wastherefore used with its complement of FFA assupplied or as altered by the addition of differentquantities of labeled or unlabeled fatty acids.

The uptake of fatty acids into RBC triglycer-ides and phospholipids approximated a first-orderreaction in the range of FFA concentrations usedhere (close to those the RBCand albumin wouldbe exposed to in the circulating blood). In afirst-order reaction, the relative fractional uptakeis directly related to the rate constant. Relativefractional uptakes can also be measured by in-cubating individual albumin-bound 14C-labeledfatty acids with the RBCand comparing the up-takes of radioactivity. The advantages of measur-ing them by the method described here are thatthe radiochemical purity of the individual acidsis somewhat less important and that each labeledacid serves in each separation step as an internalstandard for the recovery of the rest.

Since the relative fractional uptakes are char-acteristic of the RBC rather than of a particularFFA pool, they are well suited for describing theselectivity of the RBC for different acids. Giventhe composition of the precursor FFA pool, therelative rates of uptake of different fatty acids canbe predicted from the relative fractional uptakevalue. They thus may help in predicting how achange in the composition of the precursor poolmay affect the composition of the RBC lipid.

Oliveira and Vaughan observed that the uptakesof oleic and linoleic acids into phospholipids ofRBCghosts were greater than those of the satu-rated acids and that all these acids were incorpo-rated into the beta position of the lecithin. Itwas therefore tempting to try to explain the rela-tively high concentrations of unsaturated acids in

the RBC phospholipids, particularly in the betaposition, by this selectively higher incorporation.However, it is probable that the RBC in vtivo areexposed to FFA bound to albumin, not FFA asfree ions. As shown here, the fractional incor-poration of the various fatty acids from albuminto RBC lipid esters does not vary greatly fromone acid to another except for stearic acid. (Thisis in contradistinction to the fractional incorpora-tion of RBC-FFA into the same lipids, which wasdifferent from one fatty acid to another.) Onemight therefore expect the beta position of thelecithin to resemble the FFA on the albumin.More experimental data are needed to clarify thispoint.

The uptake of fatty acids by the RBC offereda rather unique opportunity to dissect the processof FFA transport because of the rapid rate oftransfer of FFA from albumin to cells and therelatively slow rate of incorporation of these acidsinto cellular lipids. In the studies of fatty acidincorporation into leukocytes described by Elsbach(16), the acids found in the cells were almostcompletely esterified into triglyceride. In thestudies of fatty acid incorporation into ascites tu-mor cells reported by Spector and Steinberg (17),the quantity of fatty acid remaining as FFA wasalso small compared to that esterified into tri-glyceride, even early in the incubation period. Webelieve, however, that there are no marked differ-ences in the pictures of fatty acid transfer andmetabolism given by the studies of these differentsystems.

Our experiments indicated that the differentfatty acids were all bound to albumin with aboutthe same affinity, whereas Goodman (8) reportedthat linoleate was bound less strongly than someof the others. The relative affinities of the RBCvaried much more. If other cells have the samerelative affinities as the RBC, the greater affinityfor free stearic and palmitic acids could provide amechanism by which unsaturated fatty acids,which seem to be incorporated into cellular lipidsat a greater rate than the saturates once theyreach the cells, are conserved in the bloodstream.

References

1. Farquhar, J. W., and E. H. Ahrens, Jr. Effects ofdietary fats on human erythrocyte fatty acid pat-terns. J. clin. Invest. 1963, 42, 675.

1026

FATTY ACID INCORPORATIONINTO HUMANERYTHROCYTES

2. Oliveira, M. M., and M. Vaughan. Incorporation offatty acids into phospholipids of erythrocyte mem-

.ane s. J. Lipid Res. 1964, 5, 156.3. Mu der, E., J. De Gier, and L. L. M. Van Deenen.

Selective incorporation of fatty acids into phospho-lipids of mature red cells. Biochim. biophys. Acta(Amst.) 1963, 70, 94.

4. Mulder, E., and L. L. M. Van Deenen. Metabolismof red-cell lipids. Incorporation in vitro of fattyacids into phospholipids from mature erythrocytes.Biochim. biophys. Acta (Amst.) 1965, 106, 106.

5. Gordon, R. S., Jr., and A. Cherkes. Unesterifiedfatty acid in human blood plasma. J. clin. Invest.1956, 35, 206.

6. Borgstr6m, B. Investigation on lipid separationmethods. Separation of cholesterol esters, glycer-ides and free fatty acids. Acta physiol. scand. 1952,25, 111.

7. Hornstein, I., J. A. Alford, L. E. Elliott, and P. F.Crowe. Determination of free fatty acids in fat.Analyt. Chem. 1960, 32, 540.

8. Goodman, D. S. The interaction of human serumalbumin with long-chain fatty acid anions. J.Amer. chem. Soc. 1958, 80, 3892.

9. Ways, P., and D. J. Hanahan. Characterization andquantification of red cell lipids in normal man. J.Lipid Res. 1964, 5, 318.

10. Whyte, M., A. Karmen, and D. S. Goodman. Fattyacid esterification and chylomicron formation dur-ing fat absorption. II. Phospholipids. J. LipidRes. 1963, 4, 322.

11. James, A. T., and E. A. Piper. Automatic recordingof the radioactivity of zones eluted from the gas-liquid chromatogram. J. chromatogr. 1961, 5, 265.

12. Karmen, A. Analysis of radioactive compounds bygas chromatography. J. Ass. off. agricul. Chem.1964, 47, 15.

13. Karmen, A., I. McCaffrey, and R. L. Bowman. Aflow-through method for scintillation counting ofcarbon-14 and tritium in gas-liquid chromato-graphic effluents. J. Lipid Res. 1962, 3, 372.

14. Goodman, D. S. The interaction of human erythro-cytes with sodium palmitate. J. clin. Invest. 1958,37, 1729.

15. Michaels, G., P. Wood, M. Mullin, and L. W. Kinsell.Lipid metabolism in human red blood cells. Fed.Proc. 1965, 24, 295.

16. Elsbach, P. Comparison of uptake of palmitic,stearic, oleic and linoleic acid by polymorphonu-clear leukocytes. Biochim. biophys. Acta (Amst.)1964, 84, 8.

17. Spector, A. A., and D. Steinberg. The utilization ofunesterified palmitate by Ehrlich ascites tumorcells. J. biol. Chem. 1965, 240, 3747.

1027