ver the past several years, radiolabeled antibodieshave been used clinically for the external imaging oftumors, myocardial necrosis, pulmonary emboli, andinfection(1—12).Currently,the methodsfor screeningpotential antibodies for immunoscintigraphic applications include: in vitro cell binding, biodistribution studies, and external imaging in small animals. In order tooptimize the screening process, a better understandingof the in vivo localization of radiolabeled antibodieswithin tissues at the light microscopic level is critical.

While immunoscintigraphy with conventional singlephoton agents is useful for localizing focal sites ofpathology, absolute tissue concentration of radiolabeledprotein cannot be determined. In addition, externalimaging does not provide any information about thedistribution of radioactivity in different tissue spaces orcell types. For example, in myocardial infarction im

ReceivedJan. 27, 1989;revision accepted June 7, 1989.For reprints contact: Ronald 0. Tompkins, MD, ScD, ACC

3A Suite 364, Massachusetts General Hospital, Boston, MA02114.

aging, it is impossible to determine the distribution ofanti-myosin between necrotic myocytes, normal myocytes, inflammatory cells, and interstitial space.

Recently, a method of quantitative autoradiographywas developed to measure concentrations of radioiodinated proteins in microscopic volumes of tissue and todetermine the degree of association of radiolabel withdifferent microscopic tissue components (13). Thismethod was recently used to determine iodine-l25-(1251) labeled low-density lipoprotein (LDL) tissue con

centrations in vascular (14) and extravascular tissues(15).

Since indium-i 11 (“In)is the most commonly usedradionuclide in immunoscintigraphy, this autoradiographic method was adapted to ‘‘‘In-labeledproteins.The method was used to measure tissue concentrationsof these ‘‘‘In-labeledproteins in an effort to betterunderstand mechanisms that localize radiolabeled proteins in immunoscintigrams. Indium-i 11 decays byelectron capture and thus does not have direct particulate emission, however, significant quantities of verylow energy internal conversion and Auger electrons are

The Journal of Nuclear Medicine1538 Morrel,Tompkins,Fischmanetal

Autoradiographic Method for Quantitationof Radiolabeled Proteins in Tissues UsingIndium- 111Eric M. Morrel, Ronald G. Tompkins, Alan J. Fischman, Robert A. Wilkinson,John F. Burke, Robert H. Rubin, H. William Strauss, and Martin L. Yarmush

The Surgical Service ofthe Massachusetts General Hospital, Department ofSurgery, and TheNuclear Medicine Division ofthe Massachusetts General Hospital, Department of Radiology,Harvard Medical School, Boston, Massachusetts

A quantitative autoradiographic method was developed to measure 111ln-labeledproteins inextravascular tissues with a spatial resolution sufficient to associate these proteins withtissue morphology. A linear relationship between measured grain density and isotopeconcentration was demonstrated with uniformly-labeled standard sources of epoxy-embeddedgelatin containing [111lnjalbumin;half-distance of spatial resolution was 0.6 @m.The techniquewas illustrated by measuring 24-hr accumulation of diethylenetriaminepentaacetic acidcoupled 111ln-labeledhuman polyclonallgG and human serum albumin(HSA)in a thighinfectionmodel in the rat. Gamma camera images localized the infectionand showed targetto-background ratios of 2.5 ±0.3 for lgG and 1.4 ±0.02 for human serum albumin(mean ±s.d., n = 3). Usingquantitative autoradiography, significantlyhigher average tissueconcentrations were found in the infected thighs at 4 to 4.5% of the initialplasmaconcentrations as compared to 0.2 to 0.3% of initialplasma concentrations in the noninfectedthigh (p < 0.05); these radiolabeled proteins were not inflammatorycell associated andlocalized primarilywithinthe edematous interstitialspaces of the infection.

J Nucl Med 30:1538—1545,1989

produced during decay. The low energy of these emissions make it possible to obtain autoradiograms ofsufficientresolutionto localizeradioactivityto specificcell types and to provide quantitation of local proteinconcentrations.

The present report illustrates the method by using‘I ‘In-labeled polyclonal human IgG as a model anti

body to evaluate the potential of ‘‘‘Inquantitativeautoradiography. This antibody preparation has beendemonstrated to be an effective agent for external imaging of focal sites of infection in both animal modelsof infection and in human subjects by a mechanismthat is not known (11). Possible mechanisms for localization include an enhanced specific binding of antibody to tissue components such as Fc receptors on thesurface of leukocytes or simply increased vascularpermeability that naturally occurs in infected tissues.By quantitatively comparing the localization of “InIgG with that of ‘‘‘In-labeledhuman serum albumin (avascular permeability marker) at focal sites of infection,these two possibilities may be distinguished.

METhODS

AntibodiesHuman nonspecific polyclonal IgO (Sandoz, Inc., East

Hanover,NJ) containingIgG antibodiesthat normallyoccurin a broadly-baseddonor populationwasused.This preparation (Sandoglobulin) is 96% IgO with a distribution of subclasses corresponding to normal serum.

Sandoglobulin and human serum albumin (HSA) (CutterBiologicals, Berkeley, CA) were radiolabeled with ‘‘‘Invia thediethylenetriamine pentaacetic acid (DTPA) antibody chelatemethod (16). Briefly, isobutylchloroformate(0.25 mmol) wereadded to a cooledsolutionof the triethylaminesalt of DTPA(0.2 mmol in 2 ml of acetonitrile) and stirred for 30 mm. Analiquot of the resulting solution of the mixed carboxy anhydride of DTPA was added to a cooled solution of antibodydissolved in 0.1 M NaHCO3, pH 7 to 8. After overnightstorageat 4°C,the mixturewasdialyzedagainst0.lM acetatebuffer pH 5.0 and applied to a Sephadex 0-25 column. Theprotein fractions were dialyzed against normal saline. A 0.5-ml aliquot ofthe resulting antibody solution, containing ‘@‘0.5mg of protein, was diluted with 0.5 ml l.OM citrate buffer(pH 5.0),combinedwith 1—2mCi ofindium chloride(Amersham, ArlingtonHeights,IL), incubated for 30 mm at roomtemperature, and applied to a Sephadex 0-25 column toisolate protein fractions.

Animal ModelA clinical isolate ofE. Co/i was storedat —70°Cin a freezing

media containing 20% glycerol and 80% dextrose phosphateuntil use. Aliquots of bacteria were defrosted and colonycounts performed on serial dilutions grown overnight on BBLBrucella agarplates. Based on the colony count, freshly thawedbacterial suspensions were washed and diluted with saline toa final concentration of 8 x l0@0organisms/ml. Aliquots (0.1ml) ofbacterial suspensions (8 x i0@organisms) were injectedinto the right thigh muscle of 150 g male Sprague-Dawley rats

(Charles River Breeding Laboratories, Burlington, MD).Twenty-four hours later, gross swelling in the right thighreadily appeared representing -@-5%of the total-body weightofthe animal. Twenty-four hours after bacterial inoculations,500 @gofantibody or human serumalbumin (specificactivity.e#3 @Ci/@igfor either tracer) was injected via the tail vein. Tail

vein blood samples were collected at 10 mm, 4 hr, 20 hr, andprior to killing at 25 hr. At 24 hr after injection of theradionuclide, the animals were anesthestized with ketamineand scintigrams recorded using a standard field-of-view scmtillation camera (Technicare 420, Solon, OH) equipped witha parallel hole, medium energy collimator. Images were recorded for a preset time of2 mm per view. Following imaging,the animals were killed and pressure-perfusion fixed for 20mm with 2.5% glutaraldehyde in 0.1M sodium cacodylate,pH 7.4.Portionsofthe infectedand noninfectedthigh musclewere excised and stored overnight in fresh fixative. The following day, whole tissue ‘‘‘Incounts of the excised thigh musclesamples and total plasma radioactivities (Ce, cpm/ml) weremeasured in a gamma counter (Compugamma, LKB Pharmacia; counting efficiency, 80%).

AutoradiographyThe autoradiography protocol has been presented in detail

elsewhere (13, 14). Briefly, glutaraldehyde-fixed samples werecut into small cubes (‘@-3mm on edge), processed by routinehistological techniques ofpostfixation with osmium tetroxide,dehydration in graded alcohols, and embedded in epoxy resin(EMbed812, ElectronMicroscopySciences,Ft. Washington,PA).Theseplastic-embeddedsamplesweretrimmed to a smalltrapezoid, (‘@-0.5mm on a side), 1-sm sections were cut withglass knives on an ultramicrotome (Sorvall MT-2, DuPont,Newton, CT) and 10 to 15 sections were placed in a centrallocation on a microscope slide.

Within 3 days of the in vivo experiment (approximatelyone half-lifeof ‘‘‘In,2.83 days), the slides were dipped inmolten (40°C)nuclear emulsion (Kodak NTB-2; EastmanKodak, Rochester, NY) in a darkroom and dried vertically at27—28°and 80% relative humidity for 1 hr. After packagingin light-tight boxes containing desiccant (anhydrous calciumsulfate), the slides were stored for 14—21days at 4°C.Slideswere developed, stained with 0.5% toluidine blue, and allowedto dry.

Grain density measurements were made visually using aZeissPhotomicroscopeIII (CarlZeiss,NewYork, NY) and acalibrated eyepiece reticle grid containing rectangular boxes(5.6 x 28 @mat 880X). Transmitted light brightfield ordarkfleld microscopy was used for grain density measurements. At least ten tissue sections from each embedded tissuesample were counted and at least two tissue samples fromeach thighwerequantitativelyanalyzed.

Nonradioactive tissue sections of similar structure to thatbeing analyzed were included with each group on a separateslide as a control for positive or negative chemography. Chemography is the subsequent development of autoradiographicgrains as a result of chemical reduction of the silver halidewithin the nuclear emulsion rather than as a result of aradioactive decay event. Such artifactual grains can greatlyincrease background grain densities and thereby reduce detec

tion sensitivity of radioactive decay.To provide a relationship between measured grain density

and ‘‘‘Inconcentration, uniformly-labeled standards were

Volume30 •Number9 •September1989 1539

N@ndi@dMeanGroitDensttyI

\

. E*@ Unibm Souvce

prepared as previously reported (13) using [‘‘‘In]HSA.Briefly,l-ml aliquots ofknown concentrations of[' ‘‘In]HSAin 0.l5MNaCI were added to identical volumes of 14% (w/v) gelatinin water at 37°C.The mixtures were transferred to individualplastic vials and allowed to gel at 4°Cfor 8 hr after which 10

ml of 2.5% (w/v) glutaraldehydein 0.lM cacodylatebufferwas added to each vial and stored at 4°Covernight. The fixedgels were removed from the vials, cut into small cubes (“-3mm on edge), and processed for embedding as outlined abovefor the tissue samples. Samples at each concentration were cutout of the epoxy resin blocks, weighed, and analyzed forradioactivity in a gamma counter (CompuGamma) creating aseries ofgelatin standards.

These standard calibrationsampleswere dipped, exposed,and developed at the same time with each group of slidescontaining radioactive tissues. The calibration data was usedto convert the measuredsilvergrain densityover 1 @imtissuesections into absolute tissue isotope concentration CT (cpm/ml). Calculated tissue concentrations were normalized to the10-mm plasma concentration at the time of the experiment,C,@(cpm/ml), to yield the relative tissue concentration CT/C,5,.

As previouslydescribed for 1251(13), resolution of “Inautoradiography was assessed using high concentration “Ingelatin standards to produce a step change from a uniformlylabeled high concentration standard to an immediately adjacent nonradiolabeled plastic epoxy resin thus, creating a resolution target. Sections, consisting of roughly half embeddedradioactive gelatin and half pure nonradioactive epoxy resin,were autoradiographed. The calibrated eyepiece reticle (usedat a magnification of 220X corresponding to grid boxes 2.2 x11 aim) was used to measure the grain density over the gelatinas a function of distance. From these measurements,a halfdistance (distance from the edge ofthe ‘‘‘In-labeledsource atwhich the grain density falls to halfofits value over the source)was obtained.

Statistical comparison of populations was performed byStudents's unpaired t-tests (two-way). A probability of a type1 error of 5% or less was considered significant.

RESULTS

The resolution of ‘‘‘Inautoradiography was determined by measurement ofgrain densities from the edgeof a uniformly-labeled ‘‘‘Ingel to background graindensity levels. An autoradiogram ofa uniformly labeledI 1‘In gel along with a plot of the normalized grain

densities is shown in Figure 1. Grain densities decreasedto half the average level over the uniformly-labeledgelatin source yielding a half-distance of 0.6 tim. Theaverage grain density over the ‘‘‘In-labeledgel was 6.6grains/24.2 @m2and over the nonradioactive adjacentepoxy (representing background grain density) was 0.75grains/24.2 @m2.A standard curve, relating autoradiographic grain density (grains/l000 @m2)to “Inconcentrations(cpm/ml), is shown in Figure 2. As is readilyapparent, the data was described satisfactorily by alinear model with an R2 of 0.99.

Plasma clearance of ‘‘‘In-labeled HSA was more

rapid than [‘‘‘In]IgG(Fig. 3). After 25 hr of tracercirculation, plasma ‘‘‘In-IgGand [‘‘‘In]HSAassociatedradioactivity averaged 26% and 13% of initial levels.[‘‘‘InJHSAplasma concentration were also less thanthose of [‘‘‘In]IgGat 2 and 20 hr.

Representative gamma camera images after injectionofeither [“In]IgGor [“InJHSAare shown in Figure4. These images demonstrate that both radiolabeledproteins preferentially localize to the infected thighs.Corresponding average counts per pixel were 582 ±58in the infected thigh versus 233 ±10 in the noninfectedthigh for [‘‘‘In@IgG,and 519 ±68 in the infected thighversus 371 ±31 in the noninfected thigh for theF'‘‘InJHSA(Mean±s.d.,n=3).Asignificantdifference(p < 0.05) between the counts per pixel in the infectedand noninfected thighs for both [“In]IgG and[“In]HSAwas demonstrated. The target-to-back

BI'- 12

@ 10

@ 08

c:@06

‘3@ 04

@ 02

-. ..‘ .@. ..;.. . , . , ,.@. r

..- . @.‘ .,..@.4. ‘..

.‘,t :‘- •.@

. .e. ‘ •..

‘; ,t /‘ @u

•:@•:..@ :

...ft.*.@lu..,.

.;.:@ ‘:.@i@'r@. .@. t•. b . . ..@-—@e

:@ •@@Ak,

. , I I-20 -45 -10 -5 0 5 40 $5 20 25

-02 .@ .j@@ -5 0 5 10 15@

DISTANCE PR@@$1£VQEc@c@twiro@u SOURCE, #4W

FIGURE1Half-distance measurement of the ‘“Inautoradiographymethod. A: Autoradiogram of epoxy-embedded uniformlylabeled source consisting of [111ln]HSAin gelatin on theleft half and nonradioactive epoxy on right. B: Grain densities, measured by visual grain counting at 2200X, overthe uniformly-labeledsource and immediately adjacentepoxy (Mean ±s.d., n = 4). Densities normalized withrespect to the mean grain density over the 111ln-labeledsource. Half-distance at which the gram density falls to50% of the mean grain density over the source (0.6 @m)indicated by the dashed lines.

1540 Morrel,Tompkins,Fischmanet al TheJournalof NuclearMediane

.@

—7(cprn/m/ x 10 1

FIGURE 2Representative standard calibration curve. Visuallycounted grain densities over a series of epoxy-embeddeduniformly-labeled gelatin standards plotted versus radioactivityconcentration in the embedded standard sourcesof a [111ln]lgGor [1111ln]HSAexperiment; at least 1000grain/gel concentration were counted. (Mean ±s.d.)

ground ratio (ratio ofthe counts per pixel in the infectedand noninfected thighs) was determined for each animal. The average target to background ratio for[“In]IgG was 2.5 ±0.3 and for [“In]HSA as 1.4 ±0.2 (Mean ±s.d., n = 3). These ratios were significantlygreater than unity (p < 0.05) indicating preferentiallocalization of radiolabel at the infection sites.

Representative autoradiograms of tissue samplesfrom infected and uninfected thighs are shown in Figure5. The noninfected muscle tissue, shown in Panels Aand B, exhibits normal skeletal muscle morphology;muscle fibers within each fascicle are tightly packedwith little space between fibers. No morphological features of infection, such as cellular infiltration by neutrophils and/or edema were visible in the sections fromuninfected high muscle. By contrast, infected tissue(Panels C and D) displayed significant cellular infiltration that consisted primarily of neutrophils withinloose, edematous interstitial connective tissue.

Qualitatively, the autoradiographic silver grain densities were distinctly different over the infected (PanelB) versus uninfected muscle (Panel D). The distributionofthese grains was nearly identical for IgG and albumin.Silver grain densities were low, often at or near background levels, over the noninfected soft tissues. How

RAD/O4C7JWTY CONCENTRATION

TIME (Hours)FIGURE 3Plasma decay curves. [111ln]lgGand [111ln]HSA(albumin)concentrationin plasma(Cr)as a functionof time.Concentration are expressed as a percentage of the labeledlgG or HSA concentration measured 10 mm after tracerinjection(C,@).(Mean ±s.e.m., n = 3)

ever, grain densities in autoradiograms ofinfected tissuewere uniformly greater than background, particularlywithin the connective tissues surrounding muscle bundies. No particular tendency for an association betweengrains and any tissue component such as the leukocyteswas seen. Grain densities over the muscle bundles varied from background levels to levels that were indistinguishable from the high grain densities ofthe surrounding interstitial space.

Using the standard curve (Fig. 2), tissue concentrations (CT) were estimated from grain densities measuredover the tissues. Tissue concentrations were normalizedand expressed as relative tissue concentrations (CT/C,@).Averages of relative tissue concentrations representing six regions for infected and noninfected tissueswithin an animal (animal averages) were combined forthree animals to yield grand animal averages (Fig. 6).The animal average tissue concentrations of both IgGand albumin within infected thighs ranged from 4.0 to4.5% of Cpo. This represents an actual tissue proteinconcentration (with a Cp@of 1.2 x 108dpm/ml and aspecific activity of 3 @Ci/@ig)of 0.73 to 0.82 @ig/ml.Animal average tissue concentrations in noninfectedtissues were significantly less than those in the infected

Volume 30 •Number 9 •September1989 1541

‘@..FIGURE 4Scintigrams demonstrating localization of unilateralE. coil thigh infections imaged 24 hr after tracer injectionof either 111ln-labelednonspecificpolyclonallgG or HSA.

thighs, averaging 0.2 to 0.3% ofthe Cp@(p < 0.05). Theaverage target-to-background ratios based upon the individually calculated ratios of infected to noninfectedtissue concentration for each animal were 9.9 ±8.7 for[“In]IgG and 7.8 ±1.1 for [“In]HSA (Mean ±s.d.,n = 3).

One major advantage of autoradiography over othermethods of tissue concentration measurement is theability both to localize radiolabel to specific tissue structures and to determine concentrations of radiolabeledproteins that are associated with those tissues. An example of this feature of autoradiography is shown inFigure 7. For both radiolabel tracers, direct quantitativecomparisons of the average normalized concentrationsmeasured over the muscle fibers in infected tissues tothose concentrations in the region of edematous connective tissue surrounding the fibers may be made. Thisanalysis indicates that concentrations outside of themuscle cells were higher than those over the fibers,

FIGURE 5Representative toluidineblue stainedautoradiogramsof [111ln]lgGin noninfected (A, B) and infected (C, D)thigh muscle tissue following 24-hrcirculationin the rat. A:Noninfectedtissue showing minimaledema andsilver grain density. Phase contrastphotomicrograph; bar represents 50

@m.B: Darkfleldphotomicrograph ofPanel A. C: Representative infectedtissue showing interstitial edema andincreased grain density. Phase contrast. D: Darkfield photomicrographof Panel C showing higher grain densities in the infected tissue; grainsprimarily localize within the interstitialtissues surrounding muscle fibers.

yielding ratios of these concentrations of 1.6 and 1.3for [“In]IgGand [“In]HSA,respectively.

DISCUSSION

This study demonstrates the first measurement of‘‘‘In-labeled protein concentrations within microscopic

volumes of tissue using quantitative autoradiographyand therefore demonstrates a useful method to obtainhigh tissue resolution of clinically relevant ‘‘‘In-labeledproteins. Indium-i 11 has properties that are particularly well suited for high resolution autoradiography,since its decay is associated with a high yield of verylow energy conversion and Auger electrons. Gammaemissions are not effective in producing discrete grainsin autoradiographs and do not produce significant background degradation. Indium-i 11 was used in this studynot only because ofits potential as a suitable radioactive

The Journal of Nuclear Medicine1542 Morrel,Tompkins,Fischmanetal

1111n-IgG “tn—HSA

860 -

40

20

0-

6

80 10

Tissue@.: Non-infected

@ Infected

4

\K

2AlbuminFIGURE 6Grand average relative tissue concentrations (CT/C@)of‘11ln-labeledlgG and HSA within noninfected and infectedtissues. CT is the concentration within the tissue and C@is the radiolabelconcentration withinthe plasma measuredat 10 mm after tracer injection.(Mean ±s.e.m., n = 3) C@for the radiolabel proteins in infected tissue was greatlyincreased as compared to the noninfected thigh; C@foralbumin was only slightly higher than C1 for lgG in boththe infected and noninfected thighs.

source for autoradiography but also because it is currently the most commonly used radionuclide for invivo antibody imaging. Indium-i 11 is a popular imaging radionuclide for several reasons including: (a) protein labeling via chelating group derivitized intermediates such as DTPA is a relatively simple and reproducible procedure (16); (b) ‘‘‘In-labelingis known to bestable intracellularly (17—19);(c) half-life ‘‘‘In(2.83days) is relatively well matched to the long circulationtimes of antibodies; and (d) gamma emissions of 171and 247 keY are readily imaged with conventionalgamma cameras equipped with medium-energy collimation.

The validity of quantitative autoradiography with‘‘‘In-labeled proteins is demonstrated by the very sat

isfactory statistical fit of a linear model to the standardcurve relating measured autoradiographic grain densityto radioactivity concentration. The resolution of ‘‘‘Inautoradiography, whose half-distance was 0.6 @m,wasbetter than that of 125Iautoradiography, whose halfdistance was 1.7 @mas determined by the same method(13). Indium-i 11 resolution was comparable to a halfdistance of 0.4 @mfor hydrogen-3 autoradiography as

0Albumin

FIGURE 7Average relative tissue concentrations of 111ln-labeledlgG(n = 6) and albumin(n = 3) withinmuscular and nonmuscular regions of infected thighs; nonmuscular regions arethe interstitialspaces exduding vascular regions; muscularregions are areas occupied by muscles. (Mean ±s.e.m.).

reported by Rogers although the exact method for Roger's half-distance determination is not known (20).

The primary limitation of ‘‘‘Inautoradiography isthe short physical half-life of 2.83 days for ‘‘‘In.Thisrapid decay requires that large amounts of radiolabeledprotein are injected and that tissue processing (embedding in plastic and sectioning) is rapid in order toproduce autoradiographs with grain densities that aresignificantly greater than background. In general, thistime from the in vivo experiment to placement of theemulsion over the autoradiographic epoxy-embeddedtissue sections was approximately one half-life. However, when these conditions are met, accurate tissueconcentration measurements and high resolution autoradiograms may be achieved.

The ‘‘‘Inquantitative autoradiography method wasused to further investigate gamma camera studies localizing radiolabeled IgG and HSA to sites of infectionin animals and humans (11, 21). Results of gammacamera imaging of focal E. co/i infections in this studywith ‘‘‘In-labeledIgG and HSA reproduce those previ

Volume30 •Number 9 •September1989 1543

@i::.:.:JNon-Muscle

@Muscle

ously reported observations (11, 21). Both [‘‘‘In]IgGand [‘‘‘In]HSAaccumulated to an extent sufficient toproduce clear images of the infection site, with targetto-background ratios of 2.5 ±0.3 and 1.4 ±0.02,respectively. Quantitative autoradiography of tissuesamples from the animals used in these experimentssupport these imaging results. Tissue radionuclide concentrations, measured relative to initial plasma radioactivity for each animal, were an order of magnitudehigher in the infected thigh muscles compared to thenoninfected contralateral thighs.

For both IgG and HSA, ratios of radiolabel concentration between the infected and noninfected musclemeasured autoradiographically were significantlygreater than that measured on the gamma camera images. This difference is most likely the result of inclusionof residual intravascular activity in the imaging-basedmeasurements.Thus, one clear advantageof the autoradiographic method in animal studies is the ability todetermine tissue concentrations ofradioactivity withoutthe contaminating intravascular radioactivity.

A principal feature of autoradiography is its abilityto associate radiolabel concentrations with specific tissue spaces and cell types. In this study, in the infectedmuscle, accumulation of “In-labeled IgG and HSAwas observed primarily over the edematous interstitialspace in a diffuse pattern that was not inflammatorycell associated. However, in areas with extensive involvement in the infectious process, silver grains werealso seen over muscle fibers suggesting that radiolabeledproteins may also have entered some of the musclespaces possibly as a result of damage to the muscle cellmembranes.

One hypothesis to explain the marked accumulationsof nonspecific polycional IgO within infection is thebinding of IgG to Fc receptors on the surface of infiltrating leukocytes (21). As a result ofthe current study,this hypothesis is not likely for two reasons: (a) if IgGbinding to Fc receptors on leukocytes were important,then autoradiographic grains should have been leukocyte-associated; however, the silver grains were notfound to be associated with leukocytes and (b) if Fcbinding is important for IgG to accumulate withininfections then, the total accumulation and tissue distribution of this accumulation for IgG should be verydifferent from albumin which does not have specific Fcbinding properties; however, the total accumulationand tissue distribution for albumin and IgG were identical in this study. It has been reported that Fc receptorsmay be shed from leukocytes within an inflammatoryreaction (22), however, the fact that the total accumulation oflgG and albumin were nearly identical stronglyargues against any additional specific Fc receptor binding of IgG to shed Fc receptors. Therefore, the Fcreceptor binding hypothesis is not only unnecessary toexplain IgG accumulation but an unlikely mechanism

for IgG accumulation. An enhanced vascular permeability would lead to accumulation of these macromolecules and would be a more reasonable explanation.

In conclusion, the results of this study establish thefeasibility of quantitative autoradiography of tissuesamples after injection of ‘‘‘In-labeledproteins. Thisapproach can be applied to a variety of problems inwhich accurate measurements of tissue concentrationsof ‘‘‘In-labeledradiopharmaceuticals are required. Furthermore, the high resolution of this technique mightbe of general use in protein autoradiography in instances where the resolution of 1251autoradiography isnot sufficient and tritium labeling is not feasible.

ACKNOWLEDGMENTS

This work was in part supported by the National Institutesof Health Grant in General Medical Sciences GM 2 1700 andGM T32-07035,ShrinersHospitalsforCrippledChildren,theGeorge Link Jr. Foundation, New York, NY, the EdwardMallinkrodtJr. Foundation, St. Louis, MO, and Ortho-Biotech Imaging Products, Washington's Crossing, NJ.

REFERENCES

1. Kennan AM. Radiolabeled monoclonal antibodies:current status and future outlook. In: Nuclear Medicine Annua/. New York: Raven Press, 1988:171—207.

2. Carrasquilo JA, Krohn KA, Beaumier P, et al. Diagnosis and therapy for solid tumors with radiolabeledantibodies and immune fragments. Cancer Treat Rep1984;68:317—328.

3. Larson SM. Radiolabeled monoclonal anti-tumor antibodies in diagnosis and therapy. J Nuc/ Med 1985;26:583—545.

4. LarsonSM. Lymphoma,melanoma,coloncancer@diagnosisand treatment with radiolabeled monoclonalantibodies. The 1986 Eugene P. Pendergrass NewHorizons Lecture. Radiology 1987; 165:297—304.

5. Khaw BA, Gold HK, Yasuda T, et al. Scintigraphicquantification of myocardial necrosis in patients afterintravenous injection of myosin-specific antibody.Circulation 1986; 74:501—508.

6. Khaw BA, Yasuda T, Gold HK, et al. Acute myocardial infarct imaging with indium-i 1i-labeled monoclonal antimyosin Fab. J Nucl Med 1987; 28:1671—1678.

7. Rosebrough SF, Grossman ZD, McAfee JG, et al.Thrombus imaging with indium-i 11 and iodine-i3i-labeled fibrin-specific monoclonal antibody and itsF(ab')2and Fab fragments.J Nud Med 1988;29:1212—1222.

8. Knight LC, Maurer AH, Ammar IA, Shealy DJ, MattisJA. Evaluation of indium-i 1i-labeled anti-fibrin antibody for imaging vascular thrombi. J Nud Med1988; 29:494—502.

9. Oster ZH, Srivastava SC, Som P, et al. Thrombusradioimmunoscintigraphy: an approach using monoclonal antipiatelet antibody. Proc NatlAcad Sd USA1985;82:3465—3468.

10. Locher JT, Seybold K, Andres RY, Schubiger PA,

1544 Morrel,Tompkins,Fischmanet al TheJournalof NudearMedicine

Mach JP, BucheggerF. Imaging of inflammatory andinfectious lesions after injection of radioiodinatedmonoclonal anti-granulocyte antibodies. Nucl MedCommun 1986; 7:659—670.

11. Fischman AJ, Rubin RH, Khaw BA, et al. Detectionof acute inflammation with ‘‘‘In-labelednonspecificpolycionalIgG.SeminNuc/Med 1988;18:335—344.

12. Fairweather DS, Bradweli AR, Dykes PW, VaughanAT, Watson-James SF, Chandler S. Improved tumourlocalization using indium-i 11 labeled antibodies. BrMedJ 1983;287:167—170.

13. Schnitzer JJ, Morrel EM, Colton CK, Smith KA,StemermanMB.Absolutequantitativeautoradiography of low concentrations of ‘25I-labeledproteins inarterial tissue. J Histochem Cytochem 1987;35:1439—1450.

14. Tompkins RG, Schnitzer JJ, Yannush ML. Macromolecular transport within heart valves. Circ Res1989;64:1213—1223.

15. Tompkins RG, Schnitzer JJ, Yarmush ML, ColtonCK, Smith KA. Measurement of ‘251-lowdensity lipoprotein uptake in selectedtissuesofthe squirrel monkey by quantitative autoradiography. Am J Pathol1988;132:526—542.

16. Krejcarek GE, Tucker KL. Covalent attachment of

chelatinggroups to macromolecules. BiochemBiophysRes Commun 1977; 77:58 1—585.

17. Thakur ML, Segal AW, Louis L, Welch MJ, HopkinsJ, Peters TJ. Indium-i 11-labeled cellular blood components: mechanisms of labeling and intracellular location in human neutrophils. J Nucl Med 1977;18:1022—1026.

18. Meares CF, Goodwin DA, Leung CS, et al. Covalentattachment of metal chelates to proteins: The stabilityin vivo and in vitro of the conjugate of albumin witha chelate of ‘‘‘indium.Proc NatlAcad Sci USA 1976;73:3803—3806.

19. Goodwin DA, Meares CF, McCall MJ, et al. Chelateconjugates of monoclonal antibodies for imaginglymphoid structures in the mouse. J Nucl Med 1985;26:493—502.

20. Rogers AW. Techniques of autoradiography, 3rd edition. New York: Elsevier, 1979:74—75.

21. Rubin RH, Young IS, Hansen WP, et al. Specific andnonspecific imaging of localized Fisher Immunotype1Pseudomonas aeruginosa infection with radiolabeledmonocional antibody. J Nuc/ Med 1988; 29:651—656.

22. Huizinga TW, Van der Schoot CE, Jost C, et al. TheP1-linked receptor FcRIII is released on stimulation ofneutrophils. Nature 1988; 333:667—669.

Volume 30 •Number 9 •September1989 1545