Selective modulation of human natural killer cells in vivo after prolonged infusion of low dose recombinant interleukin 2

Selective modulation of human natural killer

cells in vivo after prolonged infusion of low

dose recombinant interleukin 2.

M A Caligiuri, … , T R Klumpp, R J Soiffer

J Clin Invest.

1993;

91(1)

:123-132.

https://doi.org/10.1172/JCI116161

.

The immunologic consequences of prolonged infusions of rIL-2 in doses that produce

physiologic serum concentrations of this cytokine were investigated. rIL-2 in doses of

0.5-6.0 x 10(6) U/m2 per d (3.3-40 micrograms/m2 per d) was administered by continuous

intravenous infusion for 90 consecutive days to patients with advanced cancer. IL-2

concentrations (25 +/- 25 and 77 +/- 64 pM, respectively) that selectively saturate

high-affinity IL-2 receptors (IL-2R) were achieved in the serum of patients receiving rIL-2

infusions of 10 micrograms/m2 per d and 30 micrograms/m2 per d. A gradual, progressive

expansion of natural killer (NK) cells was seen in the peripheral blood of these patients with

no evidence of a plateau effect during the 3 mo of therapy. A preferential expansion of

CD56bright NK cells was consistently evident. NK cytotoxicity against tumor targets was

only slightly enhanced at these dose levels. However, brief incubation of these expanded

NK cells with IL-2 in vitro induced potent lysis of NK-sensitive, NK-resistant, and

antibody-coated targets. Infusions of rIL-2 at 40 micrograms/m2 per d produced serum IL-2 levels

(345 +/- 381 pM) sufficient to engage intermediate affinity IL-2R p75, which is constitutively

expressed by human NK cells. This did not result in greater NK cell expansion compared to

the lower dose levels, but did produce in vivo activation of NK cytotoxicity, as evidenced by

[…]

Research Article

Find the latest version:

Selective Modulation of Human Natural Killer Cells In Vivo

After Prolonged

Infusion of Low Dose Recombinant

Interleukin 2

MichaelA.

Caligiuri,I

ChristineMurray,*Michael J.Robertson,* EuniceWang,*Keith Cochran, * ChristineCameron, *

PeterSchow,* MaryE.Ross,' Thomas R.Klumpp,*Robert J. Soiffer,*Kendall A.Smith,t and Jerome Ritz*

*Division of Tumor Immunology, Dana-Farber Cancer Institute, Department of Medicine,Harvard MedicalSchool, Boston, Massachusetts 02115; tDepartment of Medicine, Dartmouth MedicalSchool, Hanover, NewHampshire 03756;and

IDepartments

of Medicine and Molecular MedicineandImmunology, Roswell Park CancerInstitute, Buffalo,New York 14263

Abstract

Theimmunologicconsequencesofprolongedinfusions of rIL-2 in doses thatproducephysiologicserumconcentrations of this cytokine wereinvestigated. rIL-2 in doses of0.54.0 x 106

U/im2per d(3.3-40

_,gg/m2

perd)wasadministered by continu-ous intravenous infusion for 90 consecutivedays to patients

with advanced cancer.IL-2 concentrations (25±25 and 77±64

pM, respectively) thatselectively saturate high-affinity IL-2 receptors(IL-2R) wereachieved in theserumofpatients

re-ceivingrIL-2infusions of 10

Agg/m2

per d and30

gg/m2

per d. Agradual, progressive expansion of naturalkiller(NK)cells was seen in theperipheralbloodofthesepatientswithno evi-dence ofaplateau effect

during

the 3mooftherapy.A preferen-tialexpansion of

CD56"'b"t

NK cellswasconsistentlyevident. NK cytotoxicity against tumor targets was only slightly en-hancedatthesedoselevels.However, brief incubation of these expanded NK cells withIL-2 invitro induced potentlysis of

NK-sensitive,NK-resistant,and

antibody-coated

targets.

Infu-sions ofrIL-2at40

,g/m2

per d

produced

serumIL-2levels

(345±381 pM) sufficienttoengageintermediate

affinity

IL-2R p75, which isconstitutively expressed by human NK cells.This did not result in greater NK cell

expansion compared

tothe lower dose levels, but did produce in vivo activation ofNK

cytotoxicity, as evidenced by lysis of NK-resistant targets. There was no consistent change in the numbers of CD56

-CD3+Tcells,CD56+

CD3+

MHC-unrestricted T

cells,

orB cellsduring infusions ofrIL-2atanyofthedosagesused.This

study demonstrates thatprolonged infusions ofrIL-2 in doses that saturateonly

high

affinity

IL-2Rcan selectively expand

human NK cellsforanextendedperiod of time withonly mini-mal toxicity. Further activation ofNK cytolytic activity can also be achieved invivo,butit requiresconcentrationsofIL-2 that bind intermediate

affinity

IL-2R p75. Clinical trials are

underway

attempting

to

exploit

the

differing

effects of various concentrations ofIL-2on human NK cells in vivo. (J. Clin.

Invest. 1993. 91:123-132.) Key words: IL-2 * natural killer cells-immunotherapy-IL-2 receptor *cytotoxicity

Introduction

IL-2 plays a central role in thevertebrate immune response

(

1

).

Produced

by

T

lymphocytes

activated

through

exposureto Addresscorrespondence to JeromeRitz, M.D., Dana-Farber Cancer

Institute,44BinneySt., Boston, MA 02115.

Receivedfor publication 13 May 1992.

cognate antigen, IL-2 promotes proliferation and

differentia-tion ofhelper T cells, cytotoxic T cells, and B cells ( 1, 2).

Consequently, IL-2 participatesinboth cell-mediated

immu-nity and humoral immunity, and is probably necessary for bothprimary and secondary adaptive immune responses. IL-2 mayalso augmentinnateor"natural"immunity by

stimulat-ing natural killer (NK)' cells(3, 4) and monocytes (5, 6).

Furthermore, IL-2 can induce theclassic signsofdelayed-type

hypersensitivity even in the absence of specific antigen (7). Commensurate with thecrucial involvement ofIL-2 in im-muneresponses,congenital absenceof IL-2 production in

hu-mans causesaseverecombined immune deficiency with life-threatening infections ( 8, 9). Replacement therapy with

recom-binant IL-2 (rIL-2) can be effective treatment for this

dis-order(8).

The diverse effects ofIL-2 are mediatedthrough specific

cell surfacereceptors on lymphocytes, monocytes, and other

celltypes( 10).The IL-2Riscomprised ofatleasttwosubunits (10). Theintermediate affinity IL-2R is a70-75-kD protein thatbindsIL-2withanequilibriumKdof -1 nM; IL-2R p75 caninitiateintracellular signal transduction afterIL-2binding

(11).The lowaffinityIL-2R(p55orCD25)isa 55-kDprotein

thatbindsIL-2withaKdof - 10 nM.IL-2Rp55 has not been

showntomediateinternalizationofIL-2 orsignal transduction by itself. Expressed togetheron thecell surface, the p75 and

p55 chainscanassociate noncovalentlytoformaheterodimer thatbindsIL-2with high

affinity

(Kd, - 10pM).Mostresting

Tcellsand B cells express neither IL-2R p75 nor p55 and do notrespondtoexogenous IL-2 ( 10). However,after activation by

antigen

ormitogens,Tcells and B cells expresshigh affinity

IL-2Rheterodimersandproliferate inanIL-2-dependent fash-ion (10, 11). By comparison, almost all resting NK cells in humanperipheral blood constitutivelyexpressisolated IL-2R

p75 chains (12-14).Thus, NK cells can be activated by

nano-molarconcentrations ofIL-2inthe absenceofadditional

stim-uli. SinceNKcells

comprise

- 10% ofPBL, as many as 109

circulating lymphocytes are potentially immediately

respon-sivetoIL-2.

Thepredominant therapeuticuseofIL-2 to date has been as animmunostimulant in patients withcancer( 15,16). Based on the

paradigms

ofcytotoxic chemotherapy, IL-2has been administered inprogressively higherdoses todefine the maxi-mally tolerated dose ( 17, 18), and bolus intravenous infusions

of15-150 x

106

U(1-10mg)ofIL-2 have been usedinmost

subsequent clinical

trials

( 15, 16). Approximately 15-20% of

patientswith advanced renal cell carcinoma or melanoma

dem-1.Abbreviations usedinthispaper:ADCC,antibody-dependent cellu-larcytotoxicity; E/T, effector to target (ratio); NK, natural killer (cells); PE, phycoerythrin; rIL-2,recombinant interleukin 2. J.Clin.Invest.

©TheAmericanSociety for ClinicalInvestigation, Inc.

0021-9738/93/01/0123/10 $2.00

onstrateobjectivetumorresponses after

"high

dose" IL-2 ther-apy. Unfortunately, such treatment is associated with life-threatening toxicities, including severe hypotension,

pulmo-naryedema, renalfailure,cardiacarrhythmias,andneurologic dysfunction ( 19, 20).It is unclear whether beneficial responses canbeinduced by lower, more

physiologic

doses of IL-2 that would notleadtosuchseveresystemic toxicity.

Werecentlypublishedthepreliminary resultsofaphaseI

trialofrIL-2 givenin low doses(0.5-6.0 X 105

U/M2

per

d;

3.3-40,ug/m2perd) bycontinuous intravenous infusion for 3 mo topatients with advancedcancer

(21

).

Toxicity

was mod-estandgenerallydidnot

impair

the normal

daily

activities of these ambulatory patients. Almost all patients

receiving

be-tween 1.5 X 105 U/M2perd(10

,g/m2

per

d)

and 6.0 X

105

U/m2

per d(40

gg/m2

perd) demonstrated

significant

periph-eralbloodlymphocytosiscausedbytheselective

expansion

of

circulating NKcells. In this report, we describe in detail the

immunologic

consequencesof

prolonged

low-dose rIL-2

infu-sion. Our resultsdefine a range of IL-2 doses that

produces

serum IL-2 concentrations sufficient to saturate

high affinity

IL-2 receptors and induces a marked selective

expansion

of

circulating

NKcells.

Methods

Descriptionofthe clinicalstudy.21 patientswith advancedcancer

re-ceived prolonged continuous intravenous infusions of rIL-2atdoses thatrangedfrom 0.5X 105U/rM2perd(3.3

Ag/m2

perd)to6.0X105 U/M2perd(40

,4g/m2

perd)(specificactivity 1.5x 107U/mg

pro-tein).Atleast threepatients received2 10wkof rIL-2infusionateach

of thefour dose levels tested. Recombinant human IL-2wassupplied byHoffmann-LaRoche(Nutley, NJ). This clinical trialwascarriedout

with the approval of the Human Subjects Protection Committee, Dana-Farber Cancer Institute (Boston, MA). All patients gave in-formedconsentforparticipationin this study. Patient characteristics andtoxicities have been described elsewhere (21). Blood specimens

werecollected atweekly orbiweekly intervals in heparinized60-cm3 syringes,and PBMCwere immediatelyseparatedbyFicoll-Hypaque densitygradientcentrifugation.Cellswerewashed threetimes and

re-suspended in RPMI 1640 medium (Gibco Laboratories, Grand Island, NY) supplementedwith 5% human antibodyserum for phenotypic and functionalassays. Aliquotsof PBMCwerealso cryopreserved in liquid nitrogen for lateruse inproliferation andcytotoxicityassays.

Bloodspecimenswerealsoobtained periodically for determinations of complete bloodcounts, serumchemistries,andserumIL-2 levels.

Analysisofcellsurfaceantigens. Single- and two-color

immunofluo-rescenceanalysiswasperformed usingFITC-andphycoerythrin

(PE)-conjugated murine mAb as previously described (22). CD2, CD3,

CD4,CD5, CD8, CD20,CD25, CD56, and anti-IL-2R p75 antibodies weresupplied by Coulter Immunology (Hialeah, FL).

FHTC-conju-gatedanti-IL-2Rp75 mAb TU-27waskindlyprovided by Dr. Kazuo

Sugamura,Tohoku School of Medicine (Sendai, Japan).CD16mAb

3G8waskindly providedby Dr. Jay Unkeless, Mount Sinai School of

Medicine (NewYork). Flow cytometry was performed on an Epics V orEpics Elite (CoulterElectronics, Hialeah, FL) and results were

dis-playedassinglecolorhistogramsor as two-color orthographic

projec-tions plottinglog greenfluorescenceversus log red fluorescence.

Back-groundfluorescencewasdetermined by incubating cells with

fluoro-chrome-conjugated murinemAbof irrelevant specificity.

Cytotoxicityassays.NKcytolyticactivitywasmeasuredusing

stan-dard4-h5'Cr-releaseassaysaspreviouslydescribed(23). 5,000

5'Cr-la-beled target cellswereaddedtowellsof V-bottom 96-well microtiter plates,andeffector cellswereplatedintriplicatetoyield various

effec-tor totarget(E/T) ratios. PatientPBMCwereassayed forcytotoxicity immediately after separation fromperipheralbloodorafterovernight (18-h) incubation at 37°C in media alone or in media containing

variousconcentrations ( 10 pM-10 nM) ofrIL-2.Target cells included K562 (anNK-sensitive human myeloid leukemia cellline),COLO 205 (an NK-resistant human colon adenocarcinoma cell line), and P815(an NK-resistant murine mastocytoma cellline). Antibody-de-pendent cellularcytotoxicity (ADCC) assayswereperformed against P815 cellspreincubated with either medium aloneorwith a 1:100 dilutionof polyclonal rabbit anti-mouse lymphocyteserum(Accurate Chemical & Scientific Corp., Westbury, NY) as previously de-scribed (23).

Proliferation assays. Cryopreservedpatient PBMC werethawed, stainedwithPE-conjugatedCD56mAb,and sorted into CD56+ and CD56- subsetsby standard methods using flow cytometry. PBMC fromhealthyvolunteer donorswereisolatedbyFicoll-Hypaque den-sitygradient centrifugation, enrichedforNKcells by negative selection using immunomagneticbeads, and sorted into CD56dim and

CD56bright

subsetsby flow cytometry aspreviouslydescribed(23). Lymphocyte subsetswereplatedat30,000 cells per well in 96-well U-bottom micro-titer plates and culturedat37°C;1

ACi

of tritiated thymidinewasadded to each well after4 d. Samples were collected 18 h later using a cell harvester, andthymidine incorporationwasdetermined usingaliquid scintillation counter aspreviously described (23).

Measurement of serum IL-2 concentrations. Serum sampleswere

assayed for IL-2 using an IL-2 enzymeimmunoassay kit from Ad-vanced Magnetics, Inc. (Cambridge, MA). Each sample was assayed in duplicate(undiluted ordiluted 1:2, 1:4, and 1:8) and comparedto a

standard curve using homogeneous rIL-2 (Takeda Chemical Indus-tries, Ltd.,Osaka, Japan).Thelowerlimit of detection in this assay is

10 pM.

Results

Changesinlymphocytesubsets

during

continuousinfusion

rIL-2. Phenotypic analyses of T cell, B cell, and NK cell popula-tions were performed every 7-14 d while patients were receiv-ing rIL-2. Quantitative changes in circulating NK (CD56+)

cellsaresummarized inFig. 1A. Atthe lowest dose level, (0.5 x I05U/iM2 per d or 3.3

_Ag/m2

per d), there was no apprecia-blechange in NK cell number. However, at dose levels of 1.5, 4.5, and 6.0 x 01 _{U/m2 per d ( 10, 30, and 40 Ag/m2 per d,}

respectively), there were consistent increases in the NK cell population during the course of IL-2 infusion. In the peripheral blood of three patients who completed 2 10 wk of therapy at 1.5 X 105 U/iM2 per d, the absolute number of total NK cells increasedfrom a mean of 134/,ul at week 0 to 410/,ul at week 12. Inpatients who completed 2 10 wk of therapy at 4.5 X

I0'

U /m2 per d, the absolute number of total NK cells increased

from a mean of

165/AI

at week 0 to

2,425/,l

at week 12. Similarly, at a dose of 6.0 X 105 U/M2 per d, the absolute numberof NK cells increased from 247

/l

atweek 0 to 1,337/

Iul

atweek 12. Inthese

_patients,

theincrease in

_circulating

NK cells occurred gradually throughout the entire period of IL-2 infusion, and there was no evidence to suggest decreased re-sponsiveness over time. It is also noteworthy thata dose-depen-dentincrease in NK cells was observed in patients who received 1.5 x105 U/M2 per d or 4.5x 105 U/M2 perd,whileafurther increase in dose to 6.0x105 U/iM2per ddidnotresult in any additional expansion in the NK cell population. Infact, onlya fiveto sixfold increase in NK cells was noted at 6.0x I0 5 U/ m2 per d compared to an average 15-fold increase in patients treatedwith 4.5 x 105 U/iM2 per d.

The absolute number of CD3 + T cells remainedessentially

the same for all patients at each of the final three dose levels, with a mean value of 1,170/,l at week 0 and a mean value of

per-25o A NKCells

_Figure

1.

_(A)

Absolute number of NK cells in

" 2000 / peripheral bloodof

pa-tientsreceiving

continu-U, 15O / ousinfusionof rIL-2at

each of four dose levels.

low.

1O

/Values

wereobtained

3£ 2

by

multiplying

the

ab-A ' * solute

lymphocyte

0 count

_by

thepercent

o 4 8 12 _{CD56+ cells within the}

B TCells

lymphocyte

population

identified by flow cyto-metricanalysis. Each point represents the meanvalueforatleast

threepatients

complet-aE 1000 a Cf 2

_aA

* _{A ing}2

₁₀

_wk

oftreat-. ° ment at aspecificdose

0 ._._._. level. The dose levels

Weeks on rlL-2 12 used in this study were

Weeks on rlL-2 _X1,UM , 0.5 x 105 U/i2 per d; &, 1.5x 105U/M2perd;m, 4.5 x 105U/M2perd;o, 6.0x 105 U/iM2perd.(B) Absolute number of T cells inperipheralbloodof patientsreceivingcontinuous infusion of rIL-2ateach of four dose levels. Values were obtained bymultiplyingthe absolutelymphocyte countby the percent CD5 + cells identifiedbyflowcytometric analy-sis. Eachpointrepresents themeanvalueforatleast threepatients completing> 10 wkoftreatment at aspecificdose level. Thedose levels used in this study were o, 0.5x 105U/M2perd;A, 1.5x 105

U/m2perd; *, 4.5 x 105U/m2perd;o, 6.0x 105U/M2perd.

formed to assess whether T cell activation occurred in vivo

during infusion oflow dose rIL-2. Proliferation studies

per-formedin vitrofailedtodemonstrateanyspontaneous prolifer-ation ofpatient T cells,yet showed normal responsesafterin vitro activation with either PHAoranti-CD2antibodies(data

not shown). Although no patient exhibited a persistent

in-creaseincirculating Tcells,transient increases inTcellswere

noted in severalpatients duringrIL-2infusion. Forexample,at

the finaltwodoses,atransient increase in absoluteTcell

num-ber(arbitrarily defined byatwofold increase frompretherapy value)wasnoted infourpatients, followed byareturntonear

baseline values within 14 d. Each episode ofT cell increase occurredwithin thefirst5wkofinitiatingtheinfusion,with the exceptionofonepatientwhodemonstratedasecondincrease duringweeks 9 and 10.Phenotypic analyses performed during theseincreases did notreveal anyselectivechanges ineither CD4or CD8T lymphocyte subsets. Absolute numbers of B

lymphocytes (CD20+ cells) remained relatively unchanged duringthe entireperiodof IL-2 infusion. Themeanvalue for

absolute numbersofBcellsatthe threehighestdose levelswas

901,ul

atweek 0and

_12O/h1

atweek 12. Inaddition,therewere

nosignificant changes inserum IgG, IgM,orIgA levelseither

duringorafter thecontinuousin vivo infusion of rIL-2(data

notshown).

IL-2 induced expansion of

CD56bri'gh

NK cells. The

CD56b"Ot

subsetofNK cells representsonly- 10%of human

NKcells and < 1%of all restinghumanPBL,yetis theonly lymphocyte population that constitutivelyexpressesthehigh affinityIL-2Randexhibitsaprofoundproliferativeresponseto

low concentrationsof rIL-2invitro(24-26). As showninFig. 2 forarepresentative patient, phenotypic examinationof

pe-mu

z

-i

.1

mU C.)

Week 3

NKH1+ 21%

1

_-Week 6 NKH1+ 57%

Week 8 (off tx)

NKH1+ 35%

CD56 Fluorescence Intensity

Figure2. Expression of

the CD56 antigenby PBLobtainedfroman

individualpatientat

in-dicatedtimepoints

be-fore,during,andafter

the continuousinfusion

of4.5x 105U/m2per

dof rIL-2.Freshly

iso-latednonadherent PBL

werestained with

PE-conjugated mAb NKH I(CD56)and

an-alyzed by flow cytome-try.PercentCD56+cells

is indicatedintheupper

right.Note the promi-nentexpansionof

CD56bfl60t

cells during therIL-2infusion, and

subsequent regression

oftheCD56brihlt

popu-lationafter

discontinua-tionof IL-2. Thelast

analysiswasperformed aftera2-wk

interrup-tionin IL-2 therapy.

ripheral blood NK cells during IL-2 infusionatdoses of 1.5-6.0

X 105 U/M2perd revealedagradual increase in the CD56b lgt

population compared to the CD56dim cells. While the

CD56b1Ot

cells accounted for< 10% ofCD56+ cellsinall

pa-tients before the infusion ofrIL-2, thispopulation accounted for the majority of CD56+ cells after the 12-wk continuous infusion of rIL-2atthetwohighest doses. Insomecases,there was a2 100-fold increase in the absolute number of CD56

blOlt

cells. Thediscontinuation ofrIL-2infusionwasinvariably

fol-lowed byadecrease in the number ofCD56bflsht cells in periph-eral blood withindaystoweeks(Fig. 2, A-D).

CD16 (FcyRIII) is expressed by the majority ofhumanNK cellsand is the receptorthattriggersNKcellADCC(27, 28). In normal peripheral blood, the

CD56bight

population ex-presseslittleornoCD16 and mediatesADCC relatively poorly

(25, 29). In patients receivingcontinuousinfusion IL-2, the expanded CD56b

'lt

cells demonstratedlittlechangeinCD16 antigen density, and thevastmajoritycoexpressesCD16 (Fig.

3). Nevertheless,asignificant fraction of the CD56+ cells in all

patientswerealso CD16negative. Additional two-color immu-nofluorescence analysis demonstrated that the CD56+ cells consistently lacked CD3, CD4, and CD5, while> 95% of these

lymphocytesexpressed CD2 and - 40% expressed CD8 (data notshown).

ExpressionofIL-2receptorsonexpandedNKcells. During

the courseofrIL-2 infusion wealso examinedexpression of

bothp55 and p75 subunits oftheIL-2receptorbyNKcells and Tcellsusing direct two-color immunofluorescence analysis. As shownforarepresentative patientin Fig. 4, fewNKcells

ex-pressed IL-2R p55 by flow cytometric analysis either beforeor

duringrIL-2infusion.In contrast,expression of IL-2R p75was

evidentonalmostallCD56+ cells beforetreatment,and there appearedtobeanincreaseinantigen density during the period

of IL-2 infusion.Simultaneousanalysis ofT cells failedto dem-Pre-treatment

WEEK 12 IL-2

L

A

4.4% 17.0% - 8.6%Y

CD56-PE CD56-PE CD56-PE

Figure 3.Coexpression of CD56 and CD 16(Fcy RIII) byperipheralbloodNKcells inarepresentative patient receivingcontinuous IL-2 infusion of 4.5x105U/M2perd.Nonadherent PBL obtainedattheindicated timepointswerestained withPE-conjugated anti-CD56 mAb and FITC-conjugated anti-CD 16 mAb and analyzed by flow cytometry. Note thepredominanceof CD16+CD56dimNKcellspriortotherapyand the

ex-pansion of bothCDl6+6 CD56bh' andCD16-CD56b'Otcellsduringthe rIL-2 infusion.

onstratesignificant expression of either IL-2R p55 or IL-2R p75 on CD3+ cells as a result of IL-2 infusion (data not

shown). Furthermore, Northern blot analysis performed on

purified

populations

ofTcellsfrom selectedpatients receiving low dose rIL-2 infusion showed only minimal expression of

IL-2R p75 mRNA when compared with NK cells isolated

underidenticalconditions,orcompared withTcellsafter CD3 activation(data notshown).

Effects of

rIL-2

infusion

onnon-MHC-restrictedcytotoxic Tlymphocytes. Asmall subsetof resting CD3+ cells expresses theCD56antigenatlowdensity;theseuniqueTcells comprise

- 2% of normal PBL(30). CD56+ CD3+ cells exhibit the

morphology of large granular lymphocytes, mediate spontane-ous non-MHC-restricted cytotoxicity when

freshly

isolated,

and demonstrateaugmented lysis of NK-resistanttumorcell lines after short-term incubation in nanomolar concentrations of rIL-2 (30). While detected in

virtually

all

patients

invarying

percentagesbefore theinitiation oftherapy, this population did

notexpand during continuous low doserIL-2 infusion. This

canbe bestappreciated inone

patient

withanunusuallyhigh

percentageofCD56+ CD3+ non-MHC-restrictedTcells be-foretreatment

(Fig.

5).The percentageof CD56+Tcells

pro-gressively diminishesasthe absolute numberofCD56+ CD3-NK cellsselectively expandsduring continuous infusion low

dose rIL-2therapy. Similar resultswereobtained with all

pa-tientstreated onthis study (not shown).ThusCD56+ CD3+

lymphocytes, like conventional CD56-CD3+Tcells, do not appear to expresshigh affinityIL-2R or toproliferateinvivoin the presenceoflowconcentrationsofIL-2.

Effects

of

IL-2

infusion

oncytolyticactivity

ofNK

cells.The

cytolytic

activity ofPBMCagainstbothNKsensitive(K562) andNKresistant (COLO 205)target cells wasexamined

peri-odically during

thecourseofrIL-2infusion.Agradual increase

incytotoxicity against K562 target cells was detected during

IL-2 therapy, generally

correlating

withthe degreeofexpansion of

circulating

NKcells(21).Patientstreated at thehighestdose level (6.0 x 105

U/iM2

per d) exhibited low but significant killing ofCOLO 205target cellsduringrIL-2infusion.NKcells incubatedinvitro with nanomolar concentrations ofIL-2 de-velop non-MHC-restrictedcytotoxicity againsttumorcell

tar-getsgenerally resistanttolysis by restingNKcells lymphokine-activated killer (LAKactivity) (31, 32).Furtherstudieswere

undertakentodeterminewhether thecytolytic activity ofNK

cells expanded in vivo by low dose rIL-2 infusion could be

further enhanced by incubation withrIL-2invitro. Fig. 6

dem-onstratesresults obtained using PBMC froma representative patient after 8 wk of continuous rIL-2 infusion at6.0 x I05

U/M2

perd. PBMC incubated overnight in mediumalone dem-onstrated only minor cytotoxicity against NK-resistanttarget

cells, and thepresenceof low concentrations ofIL-2(10pM)

didnotenhancecytotoxicity. Cytotoxicitywasenhancedafter incubation ofPBMCwith 100 pM IL-2, butthiswasevident onlyat the

higher effector/target

cell ratios. Incubation with higher concentrations ofIL-2( 1-10 nM)producedmarkedly enhanced

killing

oftheNK-resistanttarget, even at low

effec-tor/targetratios.

ADCCby PBMCwasalsotestedduring thecourseof rIL-2 infusion. PBMC isolatedbefore lowdoserIL-2infusionfailed

todemonstrateany

significant lysis

of the P815 murine masto-cytomacelllinein the absence

(Fig.

7A)orpresence(Fig.7B) ofthe polyclonal rabbit anti-mouse lymphocyteserum;further incubation with varying concentrations ofexogenous rIL-2 alsofailedtoinduce

killing

of P815target cells. Afterinvivo expansion of CD56+cellsby low dose rIL-2infusionfor6-8 wk, nosignificant cytotoxicitywasseenin the absence of

poly-clonal rabbit anti-mouselymphocyte serum.

Despite

further stimulation withrIL-2 invitro,minimal

cytotoxicity

occurred evenatthe

highest

concentration ofIL-2

(10

nM)

(Fig.

7

C).

After addition of polyclonal rabbit anti-mouse lymphocyte

serumand exogenous IL-2

(Fig.

7D), however,the

expanded

CD56+ cells exhibited a

high

degree ofADCC. Aswith NK

activityand

lymphokine-activated

killer

activity, high

concen-trations ofIL-2 (

1-10 nM)

were

required

todemonstrate en-hancedADCC.

Proliferative

responses

of

NKcells

expanded

invivo

by

rIL-2

infusion.

SinceNKcellnumbers progressively increasedin

patients receiving

low dose continuous infusion

rIL-2,

it seemed

likely

thatthese

expanded

NKcellswould

proliferate

vigorouslyto exogenous IL-2. Surprisingly,

CD56+

cells iso-latedbyflowcytometriccellsorting fromthe bloodof

patients

PRE

IL-2

v . ... .... I . . . * ... II Ill I I I IIII**

.X la la ees90

CD56

-

PE

WEEK

5

IL-2

*-~~~~~~

~~~~~~~~~~l

ol N I r- _Z.gf 11 _..

.1

CD56

-

PE

Figure4. Expressionofp55 andp75 IL-2Rby peripheralbloodCD56+NK cells fromarepresentativepatientbefore andduring (week 5)

continuous rIL-2 infusion of 4.5 x I05U/rm2perd. NonadherentPBLwerestainedwith

_{PE-conjugated}

anti-CD56 mAbandeither FITC-con-jugatedanti-IL-2Rp55(AandB)orFITC-conjugatedanti-IL-2Rp75 (CandD)andanalyzed byflow cytometry. Actual values forpercent

double-positivecells is indicatedintheupperright quadrantof eachhistogram.

on low dose IL-2 therapy showed very little proliferation to

exogenousIL-2 in vitro (Table I). These expanded NK cells

demonstratedalmostnoproliferationtoIL-2 in concentrations of 10pM (Table I),and anti-IL-2Rp55 mAb failedto

abro-gate theirproliferation tohigher concentrations ofIL-2 (not shown). Although some proliferation was consistently seen

when expanded NK cellswerecultured with 1 nMor 10 nM IL-2, thedegree of proliferationwasmuchless than thatseen

withCD56b

i"'t

NK cells freshly isolated from normaldonors (Fig. 8). Indeed, the proliferative response ofNK cells

ex-panded in vivo closely resembles that ofnormalCD56dlm NK cells.

SerumconcentrationsofIL-2duringrIL-2infusion.Serial

measurementsofserumIL-2 concentrationswerecarriedout

in 11 patientswhiletheywerereceiving continuous infusions

of IL-2. As shown in Fig. 9, IL-2wasundetectable

(concentra-A

2.4%

p55+

CD56+

-_m~~c

W-

S.-

qm.-B

4.4% p55+ CD56+

I IH--I- r_-~

A

4-

:-LO~

Le)

I Ln

IL

UL

JL

10

a.

Al[LO

I I

-Pre IL-2

U-IL

cv,

a

WEEK 2

CD56-PE

CD56-PE

WEEK 11

CD56-PE

Figure 5. Relative decrease in peripheralbloodCD56+ CD3+cytotoxic T cells inasingle patient receiving continuous rIL-2 infusionat4.5x 105

U/M2perd. Nonadherent PBLwereobtainedattheindicated time points duringrIL-2 therapy, stained with PE-conjugated anti-CD56 mAb andFITC-conjugated anti-CD3 mAb, and analyzed by flowcytometry. DespiteanunusuallyhighpercentageofCD56+ CD3+cellsin the

pe-ripheral blood of this patient before IL-2 therapy, thislymphocyte populationfailedtodemonstrateanyexpansion during prolonged rIL-2

infu-sion, whileCD56+CD3-NK cellsweremarkedlyincreased (lower right quadrant ofeach histogram).

tion< 10pM)intheserumofpatients treatedatthe first dose levelof0.5x 105 U/m2perd. Atadoseof 1.5x 105 U/m2per

d, the mean serum IL-2 concentration during infusion was

25±25 pM, butoneof three patients didnothaveconsistently detectable levels. Atadose of 4.5x 105 U/M2perd, themean

serumIL-2concentration during infusionwas77±64 pM.At

thehighest dose used, 6.0x I05 U/iM2perd, therewasgreater variation in serumIL-2 concentration seenbetweenpatients.

Onepatient hada meanconcentration of 39±4pM,asecond

patient had a mean concentration of 383±216 pM, and the

thirdpatient treatedatthis dose hada meanconcentration of

776±336pM. Of these three patients, theonewith the lowest

serum IL-2concentration had themostprofound increasein NKcell number during rIL-2 infusion, while theonewith the

highestserumIL-2concentration showed the least increase in

NKcells.

During the courseof thistrial, one patient failedto show

any increase in NK cell number as seen in the majority of

patients treatedatsimilar doses. Thepatientwastreatedat1.5

x 105 U/M2 per d, yet had undetectable (< 10 pM) serum

concentrations ofrIL-2throughoutthe infusion. This is in

con-trast tothetwootherpatientstreatedatthisdose,who showed

100- Figure 6. Lysisof

NK-> 80_- resistant_targetcells_by

lymphocytes expanded

60 invivobyrIL-2

infu-o sion. Nonadherent PBL

Q 40 - /t (64%CD56+ cells)

wereisolated from the

20.- ...

,

4blood

ofa

representa-0 SE tivepatient after 8wk

20 30 40 _{of rIL-2 infusionat6.0}

E:T ratio X105U/m2perd,

cul-tured for 18h ineither media aloneormediacontaining 10 pM, 100 pM, 1 nM,or 10 nM rIL-2asindicated, and tested for lysis of5'Cr-labeledCOLO 205

tar-get cells atvariousE/T ratios. Resultsrepresentthemeanpercent lysis for triplicate samples. COLOtargetcells:--- ---,media; 0

-, 10 pM; -, 100 pM; -o -, 1 nM; , 10 nM.

measurableserumlevels of rIL-2 and hadmorethana

seven-fold increase inthe absolute NK cell number during the 3mo

of rIL-2therapy. A second patient demonstrated increases in NKcell number during the infusion, butataratewhichwas

significantly slower than the majority of patients. This patient

wastreatedatthehighest dose level (6.0x 105U/M2perd),

and had serumrIL-2 concentrations thatwere thehighest in

thestudy, ranging between 419 and 1,230 pM. In addition, this patient uniquely exhibitedaprofound increase in both

neutro-phil and eosinoneutro-philcounts,aswellas asignificant decrease in

plateletcount,suggesting that additional activation via the IL-2Rp75 mayhave resulted inamultitude of stimulatory and

inhibitory hematologic effects, possibly through the release of additionalcytokines (33-35).

The relationship between the IL-2 dose administered,

serum IL-2 levels attained, and the calculated percentage of IL-2receptoroccupancy areshown in Table II. Since the Kd for

the three classes of IL-2binding sites differ by three orders of magnitude, there isamarked difference in the proportion of

receptorsoccupiedateach IL-2dose level. Thus, thehigh affin-ity IL-2receptor(Kd, 10pM) is> 90% saturated when the IL-2

concentration is 100pM, while< 10%of the intermediate

af-finity receptorsareoccupied atthis concentration. However, 26% of the intermediate affinity receptors are occupied at a

serumIL-2 levelof 345 pM.

Discussion

We haveinvestigated the immunologicconsequencesof

main-taining near-physiologic serumconcentrations of IL-2 for

ex-tendedperiods by prolongedintravenous infusions of the cyto-kine. The doses ofrIL-2 administeredcould produce serum

IL-2concentrations(10-100pM)thatselectivelysaturatehigh affinity IL-2R heterodimers, as well as concentrations ( 1-nM) thatcanengageisolated intermediateaffinityIL-2Rp75. Suchtherapyinduceda profoundand selectiveexpansion of NK cellswithoutanysubstantialchangeinthe absolute

num-ber ofT cells, B cells, or monocytes. A clear dose-response effectwas seenbetween 1.5 x 105 U/M2perd and 4.5 x

105

2

-4

* . x Ils

I

PRE IL-2

Targets + Media

so- A

60

-40

-20

-0-.

9

-0--O

-0-5

media 10pM

100PM

1nM 10 nM

10

B Targets + Antibody

60

-40

-20

-5

E:T ratio

10

6-8 WEEKS IL-2

C Targets + Media

80 D Targets + Antibody

0 5

E:T ratio

Figure7.ADCC mediated bylymphocytes expanded in vivo by rIL-2 infusion. PBL obtained from three -0 patients before(A andB)orafter (CandD)IL-2

therapywerecultured for 18 h in media aloneor

me-diacontainingIL-2atthe indicated concentrations

A4 and then tested forcytotoxicity againstP815 cells preincubatedwithmediaalone(AandC)or polyclo-nalrabbit anti-mouselymphocyteserumreactivewith P815

(B

and

_D).

PBL before

therapy

contained14%,

13%, or 26% CD56+ cells,respectively.PBLobtained after 6-8 wk of rIL-2 infusionat6.0x I0 U/im2per dcontained 50%,77%,or79%CD56+cells. Each

10 _{point represents themeanvalueofpercentlysisby}

PBLfrom the threepatients.

U/m2perd, which resulted inserum IL-2 concentrations of

- _25pMand 77pM,respectively. Increasingthe IL-2 dose

from 4.5x 10 5 U/rm2perdtoaslightly higher dose of6.0xI05 U/M2perd didnotresult in further NK cell expansion. Serum IL-2levelsatthis highest dosewerevariable, whichmayinpart

reflect poorrenal excretion of IL-2 as a consequence of

pre-vious chemotherapy-induced nephrotoxicity. Intwo of three patients, this dose producedserumIL-2levelssufficientto en-gageintermediate affinity IL-2R (Kd, 1 nM). Commensurate

with this, the circulating NK cells from these patients demon-strated higher levels of NK cytotoxicity compared with NK cells frompatients receiving the lower doses of IL-2. The obser-vation that absolute peripheral blood NK cell numberswere

actually lower after continuous infusion of 6.0x I05 U/rm2per

d ofrIL-2 thanafterinfusion of4.5x l0 U/M2perdis

intrigu-ing. Activation ofNK cells through IL-2R p75causesthe

upreg-TableI. Proliferation ofNKCellsExpandedIn VivobyrIL-2

Infusion

Proliferation (cpm)in the presenceofIL-2 concentrationof:

Week

Patient onIL-2 None 10 pM 100 pM 1nM 10 nM

1 6 231 149 448 4,351 N.D.

2 12 238 230 403 8,391 13,102

3 11 284 467 485 2,834 2,948

4 12 580 295 3,528 10,680 4,664

TotalCD56+NKcellsfrom theperipheral blood of patientsreceiving rIL-2infusionsof 6x l0 U/M2per d for 6-12 wk were cultured in vitrowith medium alone or medium containing various concentra-tions of IL-2 asindicated. Tritiated thymidine incorporation was determinedasdescribed in Methods. Results are mean cpm from triplicatewells. N.D., not done.

ulation of certain NK cell adhesion molecules (22),andthus the higher serum IL-2 concentrations produced by the 6.0

x 105 U/rm2perd dose might have induced the extravasation

ofexpandedNKcellsinto tissues.Alternatively, secondary

cy-tokines may have been produced byNK cellsor monocytes stimulatedthroughIL-2Rp75, resultinginbothsystemic toxic-ity andaninhibitory effectonNKcellexpansion. Ofnote,the number of circulating cytotoxic effectors induced by chronic lowdose IL-2therapy insomeinstancesequaled the number of

effectors that can be generatedex vivousing high dose IL-2 ( 15 ). Furthermore, this expanded NK populationwasevident

formonths and showednosign of diminution throughout the

courseofIL-2infusion. Despite the dramatic increaseinNK

cell numbers,NKcytotoxicity againsttumortargetswasonly slightly enhancedatdose levels of 1.5 x 105 U/rm2perdand 4.5x 105 U/rm2perd. Thus, prolonged low-dose IL-2 infusion

50X00

_Figure

_{8. IL-2-induced}

f 40000 proliferation of normal

CD56""' NKcells,

o 30 _{_ normal}

_CD56"m

_NK

cells,and NK cells

ex-pandedin vivoduring

o10000- _ rIL-2infusions. Sorted

0 _ _ o _

CD5614"g"

(A)and

A B C CD56dIm(B)NKcell

subsets fromhealthy

donorsorsorted totalCD56'NKcells(C)frompatients receiving

rIL-2 infusionsat6.0x iOs U/M2perd for 6-12 wkwereisolated

byflow cytometry and cultured for5dinmediumcontaining 1nM

IL-2.Tritiatedthymidine incorporationwasdeterminedasdescribed inMethods. Datarepresentmean±SDofcpmfor NK cells derived

fromfour normal donors(AandB)orfour IL-2patients (C).

Back-groundcpminthepresenceof medium alone(278±35forA,

322±119 forB,and245±27 forC)have beensubtracted.

0

._U

0

0-0

w

0

0

4-80

-60

40

20

-i

os

c

z

J.

-i

I06

80

20X

v v v -v- v

0. 5 0. 51.5 1.5 1.5 4.5 4. 5 4. 5 6. 0 6. 0 6. 0

IL-2 Dose Level

Figure9. SerumrIL-2concentrationsin 11patients receiving

contin-uousintravenous infusions ofrIL-2for 90 d.Patients 1 and2 re-ceived 0.5x10'U/rm2perd; patients 3, 4, and 5 received 1.5x 10'

U/m'perd; patients 6, 7, and 8 received 4.5x 105U/M2perd; and

patients 9, 10, and 11received 6.0x 105U/M2perdofrIL-2.Each

pointrepresentsaseparate measurement.Serum levelsweredrawn

randomly throughoutthe infusions. IL-2wasnotdetectable in the

serumofanypatientbefore the initiation of rIL-2therapy.

canproduce prominentNKcellexpansion without activation

of substantial NK cytolytic activity. However, brief in vitro incubation ofexpandedNKcells with concentrations ofIL-2 sufficientto bind IL-2R p75 resultedin marked cytotoxicity against NK-sensitive, NK-resistant, and antibody-coatedtarget cells. This is consistent with theexpression ofisolated interme-diateaffinity p75 detectedon expandedNK cellsby flow

cy-tometry.Theabilitytoexpand theNKpopulationatlow,

non-toxic doses ofrIL-2 and thento activatehighly specific anti-body-directed target cell killing using a higherdose of rIL-2

(Fig. 7)suggestsfuturestrategiesforimmunotherapy. Despite inducingup to 15-fold increases in the absolute number ofcirculatingNKcells, lowdosecontinuous infusion IL-2 caused no significant increase in B cells, CD56- CD3+ conventional T cells, or CD56+ CD3+ MHC-unrestricted T

cellsin mostpatients.The selective NK cellexpansion reported herewasexpected basedontheknown distribution of IL-2Ron

normalresting human lymphocytes (24).Thevastmajorityof Tcells and B cellsexpressneitherintermediatenorhigh affinity IL-2Rbinding sites until activated by antigen,and these lym-phocytesarethereforelargely unresponsivetoexogenousrIL-2

Table II. Relationship betweenIL-2 Dose, Serum IL-2 Concentration, and IL-2R Occupancy

Percent receptoroccupancy

Serum

IL-2dose (IL-2) High Intermediate Low

Ag/m2perd pM

3.3 <10 -

-10 25 71 2 <1

30 77 88 7 1

40 345 97 26 3

CalculationbasedontheKdof each class ofreceptor:high affinity, Kd

= 10pM; intermediate affinity, Kd= 1nM; low affinity, Kd = 10 nM(1 1).

(IL-2)

Percent

RC.

= x100, where

Ro.

isthe_receptoroccupancy.

( 10, 24).Incontrast, - 90% ofNKcells_expressIL-2Rp75in

the absenceof IL-2Rp55anddemonstrateenhancedcytolytic activityinresponsetoconcentrations(- 1-10nM)ofrIL-2

thathalf-saturate intermediateaffinity IL-2R. Anovelsubset

ofNKcells,identifiedby high density expressionofthe CD56 antigen, appearstoexpress constitutively high affinityIL-2R

p55 /75heterodimers,aswellas anexcessofisolated

interme-diateaffinityIL-2Rp75.These CD56" t' NKcellscan

prolifer-ate to concentrations(- 10-100 pM)of rIL-2 that saturate

only high affinityIL-2R. Incontrast,althoughbriefincreasesin

Tcell number occurred in severalpatients duringthe infusion ofrIL-2,thenumber ofTcellsgenerallyreturnedtobaseline levels within 2-3wk; acumulative increase in T cell number

wasnotobserved. Transientproliferation ofTcellsduring rIL-2infusion isnotunexpected,since T cellsarepresumably acti-vatedintermittently invivo afterencountering cognate

anti-gen.Afterantigen stimulation,Tcellsexpresshigh affinity

IL-2R p55/p75 heterodimers andproliferate in the presenceof picomolarconcentrations of IL-2.However, high affinity

IL-2RaredownregulatedonTcells 10-14 d after activation( 10), and thus Tcell expansion would notbe expected to persist despitethe continuedpresenceofexogenousrIL-2. In contrast toactivatedTcells,activatedNKcells donotappearto

pro-ducesignificantamountsofIL-2 (27, 36);thusNKcellsare

probably incapableof autocrine stimulationthroughtheIL-2

pathway. However,theexpandedNKcells inourpatients dem-onstratestableor even enhanced surfacestainingwith IL-2R

p75 mAb, and their cytolytic activity can be augmented by nanomolar concentrations ofIL-2 invitro.Thus,wehave no

evidence to suggestthat these NK cells downregulate IL-2R

p75afterprolongedexposuretoexogenousIL-2.

Mostpreviouscancerimmunotherapytrials have involved

very high doses of rIL-2 (- 15-150 X 106 U or 1-10 mg)

administeredintravenouslyoverseveraldaysorintermittently

over several weeks ( 15, 16). The immunologic and clinical effects ofhighdose IL-2therapycontrastmarkedlywith those

seen inourtrial. Highdosetherapy commonlycauses a pro-foundlymphopeniafollowedbyreboundlymphocytosis

com-prisedof bothTcells andNKcells( 17, 37, 38).Most human

NK cells (12-14) and monocytes (6) constitutively express isolated IL-2Rp75 chains,and nanomolar concentrations of IL-2caninduce thesecretionof severalcytokines bythese cells

(6, 28, 36, 39). The T cell activation detected inpatients

re-ceiving highdoseIL-2islikelycausedby secondary cytokines

derived from activated NKcells andmonocytes, rather than

directly caused by rIL-2 itself. Indeed, PBMC from patients receiving highdose IL-2infusions demonstrate enhanced

ex-pressionof mRNA for severalcytokines, includingIL-1, IL-5, IL-6, INF-'yand tumor necrosisfactor,and elevatedblood lev-els of these cytokines canbe detected in somepatients after IL-2therapy (33-35). Furthermore,suchcytokinesmay

con-tribute to theseveretoxicitiesofhighdose IL-2therapy,

includ-ing hypotension (40)and the vascular leaksyndrome (41 ).In thisregard,the selectiveNKcell modulation and mildtoxicity

thatwedescribeafter infusion of low dose IL-2maydelineate the direct effects ofIL-2as_opposedto theconfoundingeffects ofsecondary cytokines.

AlthoughselectiveNKcellexpansionwas_expectedinthis

trial,thephysiologicbasisfor themarkedlyincreasedNKcell

numberduringlow doserIL-2infusionis notentirelyclear.A

preferential expansionof CD56'"t NKcellswasseen in this

study,as wellasother IL-2 trials_{(42, 43).}Sincefreshlyisolated

10 0

0

0o

9

10 0

o 0

5o 0 0

oS2

CD56b"ht

NK cells from normal donorsexpresshigh affinity IL-2R and proliferate vigorously to low concentrations of IL-2 invitro, we initially speculated that

CD56b'60t

circulating NK cells also selectivelyproliferatedin vivoduring IL-2 administra-tion. However, several observations argue against this

hypothe-sis.

CD56bfl"t

NKcells in the peripheralbloodofhealthy per-sons express little or no surface CD16, while the expanded

CD56b'r*t

NK cells from our IL-2 patients are largely

CD16-positive. Furthermore,CD56bflht NK cells from normal persons express functional

high affinity

IL-2R, whereas most

expanded

CD56bfght

NK cells appear to lackhigh affinity IL-2R, bothby flow cytometricanalysis and byinvitrofunctional

studies using various concentrations of IL-2. Finally,

CD5b6lrhtNKcells from normalpersonsproliferatevigorously

to low concentrations ofexogenous IL-2 in vitro, while NK

cells expanded in vivo demonstrate very little proliferation

even inresponse tonanomolar concentrations ofIL-2. Itis possiblethat boththeCD56b

l't

andCD56dimNKcells that accumulate in the peripheral blood ofpatients receiving

prolongedinfusionsoflow dose IL-2arisefrom immatureNK cell progenitors that canproliferate to low concentrations of IL-2. Theregulation ofhuman NKcelldifferentiation remains poorlyunderstood.TheearliestNKcellprogenitorsarealmost

certainly generated in the bone marrow, but it is uncertain whether theseprogenitors differentiate into fullymature

effec-torcells in the marrow, orwhether NK precursors leave the

marrowandundergo differentiation elsewhere (27, 28). How-ever, human andanimal data indicatethat immature NK cell

progenitorscanbe stimulated withexogenous IL-2 toproduce

mature NK cells (44-47). Thus, the expanded NK cells in

patients receiving prolonged infusions oflow dose IL-2could reflectproliferation of bothmature

peripheral

blood

CD56b'rt

NK cells and immature NK cell precursors

possessing high

affinity

IL-2R.Ofnote, NK cells in these

patients

progressively

expanded

during

continuousrIL-2 infusionand

subsequently

decreased in number after discontinuation ofIL-2treatment.

Furthermore, reinstitution of IL-2 therapy resulted in recur-rent NKcell

expansion.

Thus, theNKcell subset of

lympho-cytes appearstorespondto IL-2invivo ina mannersimilarto the response of certain

myeloid

cell subsetsto

hematopoietic

growth

factors,

suchas

erythropoeitin

orgranulocyte

colony-stimulating

factor (48).

Extending

this

speculation further,

NKcell

production

under

physiologic

conditions

might

also be regulatedin afashionmoreanalogousto

myeloid

cellsthan to TandBlymphocytes. Monocytes, granulocytes, and

erythro-cytes areterminally differentiated cells,so aneedfor increased numbers ofmaturemyeloid cells ismetby increasedmarrow

production ofthese cells. In contrast, adaptive immune re-sponses

require

theclonal

expansion

of

specific

Tcells and B

cells, and these lymphocytes

proliferate extensively

after acti-vation via their

antigen-specific,

clonotypicreceptors.There is currently little evidencetosuggest that NK cells haveclonally distributed, highlyvariable receptors (27, 28),sothat there is no obviousadvantage toexpanding individualclones of ma-ture NK cells.Therefore,the demandfor increasednumbersof

NKcells could be metbyincreasingmarrowproductionfrom immatureprogenitors. Thismightalsoexplain the apparently

limitedproliferative potential of the vast majorityof mature

peripheral

blood NK cells(29, 36).

Our resultsindicatethat smallsubpopulations of lympho-cytescanbeselectivelyandprofoundlyexpanded byprolonged

infusion oflow dose rIL-2. This

study

stronglysuggeststhatT

cellsexpress functional

high affinity

IL-2Rfor

only

alimited time after antigen activation, while NK cells can respondto

IL-2 without known previous antigen activation and for an

apparently unlimited period of time. Attemptsmustnow be

made toidentify patientswho mightbenefit from the

expan-sion ofthese cytotoxic lymphocytes. For example,

patients

withcertain hematologic malignancieshave ahigh risk of recur-rent disease despite myeloablative chemoradiotherapy and bone marrowtransplantation.Recentstudies have shown that

prolonged low doserIL-2 therapy can bewell tolerated after either autologous or Tcell-depleted allogeneicbone marrow

transplantation

(49). Intermittent infusions of higher doses of IL-2might alsobeadministeredtoactivate expandedNKcells

via IL-2R p75, in hopesof improvingtheantitumor activity of

these cells. Furthermore, this approach could be combined

with serotherapy using mAbreactive withtumorcells or

vi-ruses,sincemostoftheexpanded NK cellsexpressthe

FcyRIII

(CD16) and can mediate ADCC. Clinical trials to evaluate these approaches are currently underway. Thedesign of our IL-2 regimenwasbased on thepharmacokinetics of IL-2, the binding affinities oftheIL-2R, and theknowndistribution of

IL-2R on lymphocyte subsets. Further rational clinical trials withIL-2 and other cytokines could be undertakenusing fun-damental

principles

ofimmunologyandendocrinology rather

than the paradigms of cytotoxic chemotherapy.

Acknowledgments

This researchwassupported by National Institutes of Healthgrants

CA41619,CA01572,andAI17643. M. A.Caligiuriwassupported by

anawardfrom the Dr. Louis SklarowMemorial Fund,aninstitutional researchgrantfrom theAmerican CancerSociety,and the Coleman Leukemia Research Fund. M. J. Robertson is supported by the Claudia Adams BarrProgram in Cancer Research.

References

1.Smith, K. A. 1988. Interleukin-2: inception, impact, and implications. Science(Wash.DC). 240:1169-1175.

2. Mingari,M.C., F.Gerosa, G. Carra, R. S. Accolla,A.Moretta, R. H. Zubler, T.A.Waldmann, and L. Moretta. 1984. Humaninterleukin-2promotes proliferationof activatedBcells viasurface receptors similartothoseof activated

Tcells. Nature(Lond.).312:641-643.

3. Henney, C. S.,K.Kuribayashi,D. E.Kern,andS.Gillis. 1981.

Interleukin-2augments natural killeractivity. Nature (Lond.). 291:335-338.

4.Trinchieri, G.,M.Matsumoto-Kobayashi, S. C. Clark,J.Seehra,L.

Lon-don, and B. Perussia. 1984. Response of resting human peripheral blood natural killer cellstointerleukin 2. J. Exp.Med. 160:1147-1-169.

5.Malkovsky, M.,B.Loveland,M.North, G.L.Asherson,L.Gao,P.Ward, andW.Fiers. 1987. Recombinant interleukin-2 directlyaugmentsthe cytotoxic-ity of human monocytes. Nature(Lond.). 325:262-265.

6.Espinoza-Delgado,I.,J. R.Ortaldo,R.Winkler-Pickett,K.Sugamura, L. Varesio, and D.L. Longo.1990. Expressionand roleofp75interleukin2receptor

onhumanmonocytes.J.Exp.Med. 171:1821-1826.

7.Kaplan, G.,R.Kiessling, S. Teklemariam, G. Hancock, G. Sheftel, C.K.

Job,P.Converse,T.H.M.Ottenhoff,M.Becx-Bleumink,M.Dietz, and Z.A.

Cohn. 1989. The reconstitution of cell-mediated immunity in thecutaneous

lesions of lepromatousleprosy by recombinant interleukin 2. J. Exp. Med.

169:893-907.

8. Pahwa, R.,T.Chatila, S. Pahwa, C. Paradise,N. K.Day,R.Geha, S.A.

Schwartz,H.Slade,N.Oyaizu,andR. A.Good. 1989. Recombinant interleukin

2therapy in severe combined immunodeficiency disease. Proc. Natl.Acad.Sci. USA.86:5069-5073.

9.Weinberg,K.,andR.Parkman.1990. Severecombinedimmunodeficiency due to aspecificdefect in theproduction of interleukin-2.N. Engi.J.Med. 322:1718-1723.

10.Smith,K. A.1989.Theinterleukin2receptor.Ann.Rev.Cell Biol.

5:397-425.

11.Wang, H.-M.,and K. A.Smith. 1987. Theinterleukin2receptor:

12.Siegel, J. P., M. Sharon,P. L.Smith, andW.J. Leonard.1987. The IL-2 receptorflchain(p70): role in mediating signals for LAK, NK, andproliferative activities. Science(Wash. DC). 238:75-78.

13.Kehrl, J. H.,M.Dukovich,G. Whalen, P. Katz,A.S.Fauci,and W. C. Greene. 1988. Novelinterleukin2(IL-2)receptor appearstomediate IL-2-in-duced activation of natural killer cells.J.Clin. Invest. 81:200-205.

14.Phillips, J. H., T.Takeshita,K.Sugamura,andL. L.Lanier. 1989. Activa-tion of natural killer cells via thep75 interleukin 2 receptor. J. Exp. Med. 170:291-296.

15.Rosenberg, S. A.,M.T.Lotze, L. M.Muul,S.Leitman,A.E.Chang,S. E. Ettinghausen,Y. L.Matory, J. M.Skibber,E.Shiloni,J. T.Vetto,etal. 1985. Observationsonthesystemic administration of autologous

lymphokine-acti-vated killer cells and recombinant interleukin-2 topatients with metastatic cancer.N.Engl. J.Med.313:1485-1492.

16.Rosenberg, S. A.,M. T.Lotze, J. C.Yang,P. M.Aebersold,W.M. Line-han, C. A.Seipp, andD.E.White. 1989.Experiencewith theuseof high-dose interleukin-2 in the treatment of 652cancerpatients.Ann.Surg.210:474-485.

17. Lotze, M.T., Y. L.Matory, S. E. Ettinghausen, A. A.Rayner,S. 0. Sharrow, C.A. Y. Seipp,M.C. Custer,and S.A. Rosenberg. 1985.Invivo administration ofpurified human interleukin 2:II.Half-life,immunologic ef-fects, andexpansion ofperipherallymphoidcellsin vivo withrecombinant IL 2. J.Immunol. 135:2865-2875.

18.Atkins, M. B., J.A.Gould, M.Allegretta,J.J.Li,R.A.Dempsey,R.A.

Rudders, D.R.Parkinson,S.Reichlin,and J. W. Mier. 1986. Phase I evaluation of recombinantinterleukin-2 in patientswith advancedmalignantdisease. J. Clin.Oncol.4:1380-1391.

19. Margolin, K. A., A. A. Rayner, M.J. Hawkins, M. B. Atkins,J.P. Dutcher,R.I.Fisher,G. R.Weiss,J. H.Doroshow,H.S.Jaffe,M.Roper,etal. 1989. Interleukin-2 andlymphokine-activatedkiller celltherapy of solidtumors:

analysis oftoxicity and managementguidelines. J.Clin. Oncol.7:486-498. 20.Siegel, J. P., andR. K.Puri. 1991.Interleukin-2toxicity.J. Clin.Oncol. 9:694-704.

21.Caligiuri, M. A., C.Murray,R. J.Soiffer, T. R. Klumpp, M. Seiden, K. Cochran, C. Cameron, C. Ish, L. Buchanan, D. Perillo, et al. 1991. Extended continuousinfusionlow-doserecombinant interleukin-2 in advanced cancer: prolonged immunomodulation without significant toxicity. J. Clin. Oncol. 9:2110-2119.

22. Robertson, M. J., M. A. Caligiuri, T. J. Manley, H. Levine, and J. Ritz. 1990. Human natural killer cell adhesion molecules: differential expression after activation and participation in cytolysis.J.Immunol. 145:3194-3201.

23. Robertson, M. J., R. J. Soiffer, S. F. Wolf, T. J. Manley, C. Donahue, D. Young, S.H. Herrmann, and J. Ritz. 1992. Response of human natural killer (NK) cellsto NKcellstimulatoryfactor(NKSF):cytolytic activity and prolifera-tion ofNKcellsisdifferentiallyregulated by NKSF. J. Exp. Med. 175:779-788. 24.Caligiuri,M. A., A.Zmuidzinas, T. J. Manley, H. Levine, K. A. Smith, and J.Ritz. 1990. Functional consequencesofinterleukin 2 receptor expression

onrestinghumanlymphocytes: identification ofanovel naturalkiller cell subset with highaffinity receptors.J.Exp.Med. 171:1509-1526.

25. Nagler, A., L. L. Lanier, S.Cwirla,and J. H.Phillips. 1989. Comparative studies of human FcR III-positive andnegativenaturalkiller cells.J.Immunol.

143:3183-3191.

26. Nagler, A., L. L. Lanier, and J. H. Phillips. 1990. Constitutive expression ofhighaffinity interleukin2 receptorsonhumanCD16- natural killer cells in vivo.J.Exp. Med. 171:1527-1533.

27.Trinchieri, G. 1989. Biology ofnatural killer cells. Adv. Immunol. 47:187-337.

28. Robertson, M. J., and J. Ritz. 1990. Biology and clinical relevance of human naturalkillercells.Blood.76:2421-2438.

29. Baume, D. M., M. J. Robertson, H.Levine,T. J. Manley, P. W. Schow, and J.Ritz. 1992.Differentialresponses tointerleukin-2define functionally dis-tinct subsets of human natural killer cells. Eur. J.Immunol.22:1-6.

30.Schmidt,R.E.,C.Murray,J.F.Daley, S. F.Schlossman, and J. Ritz. 1986.Asubsetof natural killer cells in peripheral blood displays a mature T cell phenotype.J.Exp.Med. 164:351-356.

31. Schmidt, R. E., J.M. Michon, J. Woronicz, S. F. Schlossman, E. L. Reinherz, and J. Ritz. 1987. Enhancement of natural killer function through activation of theT I1Erosette receptor.J.Clin. Invest. 79:305-308.

32.Phillips,J.H., and L. L. Lanier. 1986. Dissection of the

lymphokine-acti-vated killer phenomenon: relative contribution of peripheral blood natural killer cells and T lymphocytes to cytolysis. J. Exp. Med. 164:814-825.

33.Kasid, A., E. P. Director, and S. A. Rosenberg. 1989. Induction ofendoge-nouscytokine mRNA incirculatingperipheral blood mononuclear cellsbyIL-2 administration to cancer patients. J. Immunol. 143:736-739.

34. Heslop, H. E., D. J. Gottlieb, A. C. M. Bianchi, A. Meager, H. G. Prentice, A. B. Mehta, A. V.Hoffbrand, and M. K. Brenner. 1989. In vivo induction of gammainterferon andtumornecrosis factorbyinterleukin-2 infusionfollowing intensive chemotherapy or autologous bone marrow transplantation. Blood. 74:1374-1380.

35.Schaafsma,M.R., J. H. F.Falkenburg, J. E.Landegent,N.Duinkerken,S. Osanto, P.Ralph, K.Kaushansky,G.Wagemaker,J. V.Damme,R.Willemze, and W.E.Fibbe. 1991. In vivo production ofinterleukin-2, granulocyte-macro-phagecolony-stimulatingfactor, macrophagecolony-stimulatingfactor,and IL-6 duringintravenous administration ofhigh-doseinterleukin-2 incancerpatients. Blood. 78:1981-1987.

36. Robertson, M. J., andJ.Ritz. 1992. RoleofIL-2receptors inNKcell activation andproliferation. InNKCellMediatedCytotoxicity: Receptors, Sig-nalling,and Mechanisms. E. Lotzova, andR. B.Herberman, editors. CRC Press, BocaRaton, FL. 183-206.

37.Phillips, J. H.,B. T.Gemlo,W. W.Myers,A. A.Rayner,and L. L.Lanier. 1987. Invivo and invitroactivation of natural killer cells in advancedcancer

patientsundergoing combined recombinantinterleukin-2andLAKcelltherapy. J. Clin.Oncol.5:1933-1941.

38.Boldt,D.H.,B.J.Mills,B. T.Gemlo,H.Holden,J.Mier,E.Paietta,J. D. McMannis, L. V. Escobedo, I.Sniecinski,A. A.Rayner,etal.1988.Laboratory correlatesof adoptiveimmunotherapywith recombinant interleukin-2 and lym-phokine-activated killer cells in humans. Cancer Res. 48:4409-4416.

39. Musso,T., I.Espinoza-Delgado,K.Pulkki,G. L.Gusella,D. L.Longo,

and L.Varesio. 1992. IL-2 induces IL-6productionin human monocytes. J Immunol. 148:795-800.

40. Mier, J. W., G. Vachino, J.W. M.VanDerMeer, R. P. Numerof,S. Adams, J. G.Cannon,M. A.Bernheim,M. B.Atkins,D. R.Pinkerson,and C.A.

Dinarello. 1988. Inductionof circulatingtumornecrosisfactor(TNF a)asthe mechanismofthefebrileresponsetoIL2 incancerpatients.J.Clin.Immunol. 8:426-436.

41.Thijs, L. G., C.E.Hack,R.J.M.Strack,vanSchijndel,J. H.Nuijens,G. J. Wolbink, A. J. M.Eerenberg-Belmer, H.vanderVall,andJ.Wagstaff. 1990. Activationof the complement systemduringimmunotherapywith recombinant IL-2:relation to thedevelopmentof sideeffects. J. Immunol. 144:2419-2424.

42.Ellis, T. M., S. P.Creekmore,J.D.McMannis,D.P.Braun,J.A.Harris,

and R. I. Fisher. 1988.Appearance andphenotypiccharacterization of circulat-ing Leu 19+ cells incancerpatientsreceivingrecombinant interleukin 2. Cancer Res.48:6597-6602.

43.Weil-Hillman,G., P.Fisch,A.F.Prieve,J.A.Sosman,J.A.Hank,and P. M. Sondel. 1989.Lymphokine-activated killeractivityinducedbyin vivo interleukin 2therapy:predominantrolefor lymphocyteswith increased expres-sion of CD2 and Leu 19 butnegativeexpressionofCD16antigens.Cancer Res. 49:3680-3688.

44.Yoda, Y.,Z.Kawakami,A.Shibuya,and T. Abe. 1988. Characterization ofnatural killer cells cultured from human bonemarrowcells.Exp.Hematol.

16:7 12-717.

45.Keever, C. A.,K.Pekle, M. V.Gazzola,N. H.Collins,J.H.Bourhis,and A.Gillio. 1989.Naturalkiller andlymphokine-activatedkiller cell activities from human marrow precursors: II. The effects of IL-3 and IL-4. J. Immunol.

143:3241-3249.

46.Migliorati, G.,L.Cannarile,R. B.Herberman,A.Bartocci,E.R.Stanley,

andC. Riccardi. 1987. Role of interleukin 2(IL-2)andhemopoietin-1(H-I)in thegenerationofmousenaturalkiller(NK)cells fromprimitivebone marrow precursors. J. Immunol. 138:3618-3625.

47.Van denBrink,M.R.M., S. S.Boggs,R. B.Herberman,and J.C. Hiser-odt. 1990. Thegenerationof natural killer(NK)cellsfromNKprecursor cells in ratlong-term bonemarrow cultures.J.Exp.Med. 172:303-313.

48.Clark, S.C.,and R.Kamen.1987. The humanhematopoietic

colony-stim-ulating factors. Science(Wash. DC).236:1229-1237.

49.Soiffer,R.J., C.Murray,K.Cochran,C.Cameron,E.Wang,P.W.Schow,

J.F. Daley, and J. Ritz. 1992. Clinicalandimmunologiceffectsofprolonged

infusion of low-dose recombinant interleukin-2followingautologousand