CD8 naive T cell counts decrease progressively in HIV infected adults

(1)

CD8 naive T cell counts decrease progressively

in HIV-infected adults.

M Roederer, … , L A Herzenberg, L A Herzenberg

J Clin Invest.

1995;

95(5)

:2061-2066.

https://doi.org/10.1172/JCI117892

.

We show here that CD8 naive T cells are depleted during the asymptomatic stage of HIV

infection. Although overall CD8 T cell numbers are increased during this stage, the naive

CD8 T cells are progressively lost and fall in parallel with overall CD4 T cell counts. In

addition, we show that naive CD4 T cells are preferentially lost as total CD4 cell counts fall.

These findings, presented here for adults, and in the accompanying study for children,

represent the first demonstration that HIV disease involves the loss of both CD4 T cells and

CD8 T cells. Furthermore, they provide a new insight into the mechanisms underlying the

immunodeficiency of HIV-infected individuals, since naive T cells are required for all new T

cell-mediated immune responses. Studies presented here also show that the well-known

increase in total CD8 counts in most HIV-infected individuals is primarily due to an

expansion of memory cells. Thus, memory CD8 T cells comprise over 80% of the T cells in

PBMC from individuals with < 200 CD4/microliter, whereas they comprise roughly 15% in

uninfected individuals. Since the naive and memory subsets have very different functional

activities, this altered naive/memory T cell representation has significant consequences for

the interpretation of data from in vitro functional studies.

Research Article

Find the latest version:

(2)

CD8 Naive T Cell Counts Decrease

Progressively

in

HIV-infected

Adults

Mario Roederer, J.Gregson Dubs, Michael T. Anderson, Paul A. Raju, Leonore A. Herzenberg, and Leonard A. Herzenberg

DepartmentofGenetics, StanfordUniversity SchoolofMedicine, Stanford, California94305

Abstract

We show here that CD8 naiveT cells aredepleted

during

the asymptomatic stage ofHIV infection. Although overall CD8 T cell numbers are increased during this stage, the

naiveCD8 T cells areprogressively lost and fall in parallel withoverall CD4 Tcell counts. In addition, we show that

naive CD4 Tcells are preferentially lost as total CD4 cell counts fall. These findings, presented here for adults, and in the accompanying study for children, represent thefirst demonstration that HIV disease involves the loss of both CD4T cells and CD8 Tcells.Furthermore, they providea

newinsight intothemechanisms underlyingthe immunode-ficiency of HIV-infected individuals, since naive T cellsare

requiredfor all newTcell-mediated immune responses. Studies presented here also show that the well-known increase in totalCD8countsinmostHIV-infected

individu-alsisprimarily dueto anexpansion ofmemorycells.Thus, memory CD8 T cells compriseover 80% of theT cells in PBMC fromindividuals with <200CD4/Vl, whereas they comprise roughly 15% in uninfected individuals. Since the naive and memory subsets have very different functional activities, this altered naive/memory T cell representation

has significant consequences for the interpretation of data from in vitro functional studies. (J. Clin. Invest. 1995.

95:2061-2066.)

Key words: memory T cells * CD45RA A

CD62L *absolute Tcell counts *

immunodeficiency

Introduction

Recent flow cytometric studies have identified several pheno-typically andfunctionally distinct subsets ofCD4 and CD8 T cells (1-5). Functionalstudies have assigned specificroles to

some of these subsets. In particular, the terms memory and

naivedistinguish subsetswhicheither containor donot contain (respectively) long-livedcellscapableofmountingan

immedi-ate response to a specific antigen. In vitro studies have also demonstrated thatthesesubsets have distinct functional

capacit-ies: in general, the memory subsets do notproliferate as well

asnaivesubsets inresponse togeneric mitogenic stimuli; how-ever, the memory subsets producea widervariety andgreater amounts of many cytokines (1, 6, 7).

Address correspondence to LeonardA.Herzenberg, Department of Ge-netics, Stanford University, Beckman B007, Stanford, CA 94305. Phone: 415-723-5054; FAX: 415-725-8564.

Received for publication 16 June 1994 and in revised form 14 November 1994.

Studies characterizing changes in T cell representation in AIDS have generally focused on the decrease in total CD4

countsthatoccurs asthedisease progresses.Indeed, forover a

decade, thespecific lossof these cells has beenthe most

com-monly used surrogate marker for disease progression. Unlike

CD4 T cell counts, absolute CD8 counts have been shown to

rise earlyafter HIV infection and maintain arelatively steady

level until very late in thedisease (8) when CD4 counts drop

below

_IOO/yIl

(9).

Studies here,whichfocusonthe functionally distinctsubsets within both CD4 and CD8 T celllineages,challenge this

para-digm. We demonstrate aprofound loss of naive T cells, both

CD4 and CD8, in HIV-infected adults. Similarly, in studies presented in anaccompanying manuscript(10),wedemonstrate that CD8 naive T cells are also lost in HIV-infected children. Inaddition,weshow that theearlyincrease in total CD8 counts isdueprimarilyto the expansion of the memory CD8 subsets. Previous attempts to quantitate the relative levels of naive and memory subsets showed little or no preferential loss of

either subset during the progression of AIDS (8, 11-15);

how-ever, the combination of reagents used in these previous studies is insufficient to resolve the naive T cells from other subsets. In contrast, the specific three-color immunofluorescence FACSa methodology we use here, i.e., the combination of CD45RA and CD62L together with CD4 orCD8, uniquelyidentifies and enumerates the naive and memory T cell subsets. Thus, we were able to determine the representation of these subsets in over250 HIV-infected individuals andtherebydemonstrate the selective loss of naive T cells during the progression of HIV disease.

Methods

Human samples. We recruited 266 HIV-infected adults from the San Francisco area. Since this study was part of a clinical trial in which an entry criterion was <500 CD4 T cells per microliter, our cohort is weighted towards infected individuals with fewer CD4 T cells than thegeneral infected population. Also excluded from participation were

patientswho hadconcurrentopportunistic infectionsor weretaking very largeamountsof antioxidants, vitamins,orminerals. Theinfection status of each individual wasconfirmed by anti-p24 ELISA. Inaddition, we recruited 44 HIV-uninfected adults, in good health,ascontrolsubjects. Allclinical trial subjects signedaninformedconsentform.

From each HIV-infected patient, blood was drawn by venipuncture for FACS®analysis; for 242 of the 266 patients, bloodwasdrawn for

acompletebloodcountandanabsolute CD4 and CD8count(performed byanaccredited commercial laboratory). From control subjects, only bloodfor FACS®analysiswasdrawn. All sampleswereprepared within 8 h ofthe draw.

Reagents. Allfluorochrome-conjugated monoclonals wereobtained from PharMingen (San Diego, CA). Ficol-Paque was obtained from PharmaciaAB(Uppsala, Sweden). Biotin, flavin-deficient RPMI-1640 (hereafter, RPMI) was obtained from Irvine Scientific (Santa Ana,

CA). Other chemicals were obtained from Sigma Chemical Co. (St.

Louis, MO).

FACS® analysis. PBMCwere prepared from 6 ml of heparinized J.Clin. Invest.

©TheAmerican Societyfor ClinicalInvestigation,Inc. 0021-9738/95/05/2061/06 $2.00

(3)

100-

10-

1-CD4

T

cells

dem

10 100

I r l

1

CD62L

CD8

T

cells

1 10 100

CD62L

Figure1. Tcell subsetphenotypes.

Progressively gated PBMC sample from a healthy adult. The lympho-cytes are selected by scatter gates (top left). Within lymphocytes, CD4 and CD8 T cells can be discriminated (top right). Also evident are NK cells (dim CD8 staining) and B cells (neg-ative for both CD4 and CD8). Within CD4 or CD8 cells (boxes) are several subsets (bottom panels). The naive subset(darkbox) of both CD4 and CD8 lineages are defined as

CD45RA'andCD62L+; the re-maining cells (gray box) comprise the memorysubsets.Within CD8 T cells

(bottomright) are at least three mem-orysubsets: cells expressing only one ofCD45RA or CD62L, and those

ex-pressingneither. Within CD4 T cells

(bottom left)aretwomajormemory subsets: both are CD45RA-, but one is CD62L + and the other is CD62L -. Inhealthyadults,there are veryfew CD4 T cells with the CD45RA+

CD62L-phenotype (averaging 5% of

totalCD4); however, inHIV-infected

adults, thispopulationbecomes much moreprevalent(averaging15% of all CD4and as much as40-50%in some

patients;seeFig.2 B foranexample).

bloodby Ficol-Paque density centrifugation. For analyses in thisreport,

weused the following antibody combinations: (a) fluorescein-conju-gated (FITC) CD62L, phycoerythrin (PE)' CD45RA, and CyChromeIN (Cy5-PE) CD8; (b)FITCCD62L, PE CD45RA, and CyS-PE CD4; (c) FITC CD1la, PE CD45RA, and Cy5-PE CD4; (d) FITC CD1la, PE CD45RA, and CyS-PE CD8; (e)FITC CD3, PE CD8, and CyS-PE CD4;and(f )FITCCD14,PECD16,andCyS-PECD45.Concentrated antibodypreparations weretitered in combinations withunconjugated antibodytoobtainsaturatingreagentsthatwereon-scale influorescence.

Enoughreagent had beenpremixed tostudy the entire cohort thatwe

reporton here, ensuring comparable fluorescence distributions across

allexperiments.

Afterstaining,cellswerewashed three times inbiotin, flavin-defi-cient RPMI-1640 and thenresuspended in0.5% paraformaldehyde in RPMI.Flowcytometric analysiswasperformedon adual laser(argon 360nm,argon488nm)FACStar PlusO_(Becton_Dickinson

Immunocy-tometrySystems, SanJose, CA)interfacedtoaVAX6300 computer

(Digital Computer, Maynard, MA). For each cell, datafor forward

scatter, sidescatter,andthe fluorescences of fluorescein(515-545-nm bandpass filter),phycoerythrin(570-600 nm),andCyS-PE (650-690 nm)werecollected. For eachstain,data from50,000cellswerecollected andanalyzed byFACS-Deskl& software (16).Forsubset frequencies,

a lymphocyte gate was used. In addition, gates uniquely identifying

CD4orCD8 T cells and their subsets(asshown inFig.1)wereapplied.

Absolute numbersofsubsets ofCD4or CD8 T cellswerefoundby multiplyingtheirrepresentation bytheabsolute subsetcountsobtained

atthe clinical laboratory. Forinstance, thenumber of naive CD4per

microliter of whole blood is calculatedby multiplying thefraction of CD4 T cells which are naive(from FACSO gating) by theabsolute CD4countpermicroliter ofblood. We didnotobtain CD4 and CD8

1. Abbreviations used in thispaper:CTL,cytotoxicTlymphocyte; PE, phycoerythrin.

countsfor the control (HIV-uninfected) population. ValuesinTableI

werederived from thesameclinical laboratory that performed the

abso-lutecountsfor HIV-infected adultsinourstudy.

Statistical analyses. For analysis and display of statistical

compari-sons, weused JMP for the Apple Macintosh (SAS Institute, Cary, NC).

Comparisonsofdistributionswereperformed bythenonparametric two-sample Wilcoxon ranktest.

Results

Both CD4 andCD8Tcellsfrom bloodcanbefurther subdivided

into subsets basedonsurfacephenotypes. Thecombination of

CD62L (L-selectin)andCD45RA definesatleast four pheno-typically distinct subsets in each lineage (Fig. 1). Functional studies have shown that naive T cells belong to the subsets which stainbrightly with both of these antibodies, anddimly

forCD1la(LFA-1a) (1, 4, 5). Threeother subsets ofcells,

consisting of the other combinations of CD62L and CD45RA expression, comprisethememorycompartment(4, 5).

Naive CD4 T cells differ from naive CD8 T cells in that the CD45RA expression on CD8 T cells is uniformly higher

thanonCD4 T cells.Indeed,eventheCD45RAnegativeCD8 cells still displaylow levels of CD45RA. Since someofthese cellsexpressasmuchCD45RAasCD4 naivecells,the

frequen-ciesof naive cells of eachlineagemustbe determined indepen-dently (i.e., usingCD3to identifyTcells, it is notpossibleto use agatebasedonCD45RA and CD62Lexpressionthat will

include all CD8 and CD4 naiveT cells and will notinclude

memory T cells). Thus, enumerating CD8 and CD4 naive T

cells in PBMCsamples requiressimultaneous FACS@

measure-mentof three cell surface antigens intwoseparate tubes(Fig. 1), i.e.,CD8 (or CD4), togetherwithCD45RA and CD62L.

PBMC

100-w 0

;o

I , 1 10 100

Side

Scatter

(4)

Table I. Representationof CD4 and CD8 T Cells in Whole Blood*

Vf+

(CD4/pI)t

Cells 0-200 200-500 > 500

HIV-Fraction of T cells that are naive (%)

CD8 12(7-18) 16 (11-24) 23(16-33) 52(41-59)

CD4 27(18-37) 39(30-49) 45 (30-57) 50(41-58)

Absolutenumberofcellslsl ofblood

Total CD8 520(310-780) 790(580-1110) 750(590-1100) 620(330-910)'

Naive CD8 58(34-94) 120(93-160) 175(140-230)

32011

Naive CD4 18(7-37) 120(85-178) 250(160-230) 4901

*_Numbers_presented_are_the_{median values}_for_{each group}_(either_percent_{of each lineage}_{that is naive in the upper}_portion_of_the_table,_or_cells/

kilfor each subset in thelowerportion).Theinterquartilerangeisgiveninparentheses. *_{Number of individuals in each}

group:HIV-=44; HIV+,

0-200= 109;200-500= 105;> 500= 28. 'Thisisthe value for all HIV-adultsseen atthesameclinic whichperformedourabsolute counts,

including subjectsnotinourcohort. 11This value is basedonthetotalCD8count(620) multiplied bythe average percentofCD8 T cells which

arenaive in uninfected adults fromourstudy (52%). 1_{The average value for total CD4}count(980),for all HIV- adultsseen attheclinic, multiplied

bythe fraction of CD4 T cells whicharenaive (50%).

FACS®D

plots

in

Fig.

2 show therelative

frequencies

of T

significant

fraction of naive T cells of both CD4 and CD8

cell subsets in PBMC from

representative

HIV-infected and

lineages.

This loss is visibleas adecrease in the representation uninfected adults. Inagreement with results fromour

pediatric

of the CD45RA+

CD62L'

cells. Absolute counts

(cells

per

study (10),

these

plots

show that HIV-infected adults lose a microliter of

blood)

of these

subsets,

calculated by combining

A

eCD Figure 2. Examples of T cell subsets

A.CD8 _in_healthy_{and HIV-infected adults.}

(A) CD8Tcells. Fourexamples each

from the

HIV-negative

cohort

(bot-_00

tom) and theHIV-positivecohort

YUYWT+100 X

|(top)

n wereselectedtoshow the range

HIV

_Ii

of naive T cellrepresentationin these

groups, and are arranged in order of

inrain av _CD8_T_cell

represen-tation

(left

to

_ight).

OOnlyCD8 T

If) v~ _cellsareshown(basedongating

simi-lartothat shown in

Fig.

1).Theloss

U

100- ofthenaivesubset in the infected

in-dividualsis seenbythe decrease of

10

_~

_{~ ~ ~}

JtheCD45RA' CD62L'subset,anda

corresponding increase in memory.

1 _Note_{that an}_increase_{of memory}_cells

.,__________________________,.inoneHIV+adultmayreflect

expan-1 10 100 1 10 100 1 10 100 1 10 100 sion ofadifferent subset than in

an-CD62L other HIV+adult (e.g., compare the

third and fourthtoppanels). While there isamarked variation in the

rep-Bo

_B.

CD

_CD4

4 HIV+

HIV

resentation of the various subsets

withinindividuals(bothHIV-infected anduninfected), thephenotypesof

these subsets areconstant. (B) CD4 Tcells. Acomparisonofan unin-dO

10-.

Eggfected

individual(right)andan

HIV-:

//li~l

/infected

individual(left)isshown,

U 7 with the boxdenotingthepositionof

the naive cells.(Theexactphenotype

1 10 100 1 10 100 ofnaive CD4 T cells is invariant in

CD62L

alluninfectedadults and in those

in-fected adults who still retain some

na-ivecells.)ThisHIV-infectedadult has alarge proportionofCD45RA+ CD4 T cells. However,theseare notnaive Tcells, sincetheyhave at least several characteristics distinguishing themasmemory:asshownin thefigure, theyareconsiderably brighterforCD45RAthan naive CD4 cells and they are low for CD62L expression.

(5)

_Figure3. Representation of naive T cells

401- 20- within theCD4andCD8 lineages. Inboth

HIVE _ lineages,thereisasubstantialdecrease in the

(n=162) 20- 10-

_{representation}

of naiveTcells anda

corre-_ _ sponding increase in memory. This is much

0-

X-

1 moredramatic in CD8 T cells

(left)

than CD4

-

10

10] T cells (right): 90% of HIV-infected adults

_*(top)have a lower fraction of naive CD8s

HIV-

₅₅

1 thanalmost all

uninfected

adults (bottom).

(n=34) 5

_

_ __The

decrease in the CD4 naive T cells may I01X beX X P-~1-evengreater,

_|

than these datashow,because 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 of theoverlap in expression of CD45RAon

CD8:

_Naive/Total

CD4: Naive/Total memory CD4 T cells and naive CD4 T cells

(e.g., see Fig. 2 B). This overlap results in anoverestimateofnaive CD4representation forinfected adults. Finally, these data were calculated from the frequencies ofCD62L', CD45RA+

Tcells; however, virtually identical results were obtained for each subject bydetermining the frequency of theCDl la-dim, CD45RA+cells of bothCD4andCD8Tsubsets.

theseFACS® frequency data with complete blood count data, showasimilardecline (see below).

Previous researchers have found little consistent change in memory and naive representation among T cells between in-fectedand uninfected adults (8, 11-15). However, these studies used only CD62L or only CD45RA to identify subsets. Thus, thecontribution of a memory subset to this single phenotype confoundedquantitation of naive cells. This is borne out in the examples shown in Fig.2.Allof the HIV-infected individuals shown have very few naive cells. Even so, the third individual has aconsiderable fraction of memory CD62L+ cells and the fourth has a

considerable

fraction of memory CD45RA+ cells. Measurement of only one or the other marker would therefore not have revealed the naive cell loss in these individuals.

The decreased frequency of naive T cells is particularly

dramatic

in the CD8

lineage

(Fig.

3). 50% of CD8 T cells in healthy adults typically belong to the naive subset; however, in

mostHIV-infectedadults, < 15% ofCD8 cellsarenaive. This change in naive/memory

representation

does not necessarily reflectadecrease in the

absolute

numbers ofthe naive cells.In

fact, among the HIV-infected individuals with the highest CD4 counts, absolute naive CD8 countsaresometimes in the normal range, but significant elevation of the memorycountdrives the ratio of naive/memory well below the normal ratio ofonehalf

(Table

IandFig. 4).

Ingeneral,astotal CD4 counts fall, naive CD8Tcell counts also fall

(Fig.

4 A). Since the absolute number of CD4 Tcells per microliter of blood

provides

areasonable indication (surro-gate

marker)

forthe

progression

of

AIDS,

we

interpret

this data

asindicating that the naive CD8Tcellsareselectively lostas

HIV diseaseprogresses. While there isaloose correlation

be-tweenthesetwocounts, it is also apparent than many individuals haveeither

considerably

more, and others fewer, naive CD8T

cells than would be

predicted

by their total CD4count. There-fore, thenaiveCD8Tcellcountmay

provide

auseful marker

todistinguish individuals who arevery different

immunologi-cally

and yet would be stratified

together by

the use of CD4

counts.

Naive CD4Tcells arealsopreferentially lost duringHIV

disease

progression (Fig.

4C). While morethan half of CD4 T cells in

healthy

adults are in the naive

subset,

about

one-fourtharenaive in HIV-infected adults with CD4countsunder

200/1lI

(Table I).

Since the

preferential

loss of naive CD4 T

cellsparallels the loss of naive CD8 T cells, a similar mecha-nism might account for the decrease in both naive subsets.

Discussion

We have used three-color immunophenotyping to address the impact of HIV infection on the

representation

of defined T

cell subsets in adults (here) and children (10). By specifically distinguishing the naiveTcells from those in various memory subsets, we found apreferential loss of the naive T cells that correlates with progression ofHIVdisease (as measured by the decline of total CD4 T cells). Significantly, the loss of the naive cells begins relatively early after infection, when these individuals areotherwise

asymptomatic

and still have substan-tial numbers of CD4Tcells.

Thesefindings reveal apreviously

unsuspected

complexity in the

dynamics

ofTcell

representation

inHIV-infected individ-uals. Despite the

progressive

decrease of naive CD8 T cells, the overall CD8 Tcell number

persists

atelevated levels until very late in the disease. The increased numbers of CD8Tcells must therefore reflect a selective increase in memory CD8 T

cells. Since naive T cells are memory precursors, their loss, whichprecedes the eventual loss of memory cells, may

contrib-utesubstantiallytotheeventual loss of the total CD8

population.

Theearly

disappearance

of the naive CD8 cells is

particu-larly

surprising,

since mostdiscussions ofT cell loss in HIV infection have centered around the specific infectability of CD4-bearing cells. However, since

Bonyhadi

and

colleagues

(17) have shown that the

thymocytes expressing

both CD8 and CD4

are

infectable

by

HIV and will die when

infected,

theloss of naive CD8 Tcells could well be dueto the

depletion

of these double

positive

precursors.

Alternatively,

loss of

thymopoietic

capability

could bedueto a

general

destruction of the

thymus

or a

progressive

lossof the bonemarrow's

capacity

to

produce

thymocyte

progenitors. Hypotheses

basedon

general

failure of

thymopoiesis

are attractive since

they

could also

explain

the

preferential

loss of naive CD4T cells

(Fig. 4).

In any case, the loss of the naive T cells has

important

consequences for the

development

of immune responses in

HIV-infected individuals.Asthe naive subsets

disappear,

there will be a

progressive inability

to mount responses to novel

(6)

B

0 200 400 600

HIV-Total

CD4/4l

C

400

L

300

-U

200-Z

100-0 0 200 400 600 HIV"

Total

CD4/ul

*

7-..

,e.

-

o.-.*2.Fe...

do.

0 200 400 600

Total

CD4/4l

Figure 4. Correlation ofnaive Tcell representationwith absoluteCD4count:naive cells in both CD8 and CD4lineagesarelostpreferentially during the progressionofHIV disease. (A)Theabsolutenumber of naiveCD8 T cells isplotted againstthe absolutetotalCD4count. Eachdot representsoneindividual. The lineconnectsthe median values for individuals grouped by CD4counts(eachgroupis basedon arangeof100CD4

cells/pl).

The shadedareaisthe interquartilerangefor thegroups.Therangefor uninfected adults is from TableI.(B)The total CD8countsare

plottedasin A. Most infected adults havemoretotalCD8cells than uninfected adults. Since the number of naive T cellsisdeclining,theincrease

isdue solelytothe expansionofmemoryCD8Tcells.(C)Theabsolutenumber of naive CD4 T cells isplotted.The shadedareais theinterquartile

range;the line connecting medians isnotdrawn. IfnaiveandmemoryCD4 T cellswerelost from HIV-infected individualsatthesamerate, then

the shaded region wouldbe centeredonthedashed line. (The slopeof the dashedline, 0.5,is the fraction of CD4 T cells whicharenaivein

uninfected adults;seeTable I.) Since the distribution actuallycurveswellbelowthisline,naive CD4 T cellsarelostpreferentially tomemoryCD4

T cells inHIV-infectedindividuals.

ability to deal with the constantly mutating virus itself: novel strains, whichareimmunogenically unique, willencounterless resistance from T cell immunity than early in the progression of disease.

The impairment of immune function duetotheloss of naive Tcells from HIV-infected adults and children also has important

consequences for therapeutic strategies for AIDS. Since re-sponses to novel antigens have to arise (by definition) from the naivecompartment, the magnitudeoreffectiveness of such

responses is necessarily dependenton the availability of naive

T cells. Therefore,anHIV-infected individual withnonaive T cells is likelyto fail to respond to anyprimary immunization

involvingTcells, be it therapeutic vaccinationorotherwise. As apracticalmatter,thismeansthattherapeutic vaccination trials

canbe severely compromised by failuretocontrol for the vari-ability in the number of naive T cells in study subjects.

Similar considerations applyto therapies involving immu-nomodulators and ex vivo expansion of cytotoxic T

lympho-cytes(CTL) for therapeutic reinfusion. Since the representation ofCTLprecursorsandCTL effector cells changes dramatically during the progression of AIDS, the effectiveness of these thera-pies (or the ability to expand cells capable of carrying out

immunity) may varysignificantly between patients, dueto the variable representation of CD8 subsets. Thus,ourfindings

sug-gestthat therapeutic trials for vaccines andimmunomodulating

methodologies should be stratified withrespect tonaive T cell

countinadditiontototal CD4count.Efficacy trials for therapeu-tic drugs should similarly benefit from thistypeof stratification, which will eliminate the confounding variable of

nonrespon-siveness duetothe lack of naive T cells.

Finally, the loss of naive T cells coupled with the relative overrepresentation ofmemoryCD8 T cells introduces the

neces-sity for reexamination of all conclusions drawn fromfunctional

studies of PBMC from HIV-infected individuals. Memory

sub-setsare the primary producers of certain cytokines suchas

IL-4, IL-10, and y-IFN; naive subsets tend to produce mainly

IL-2. Therefore, simply changing the representation of these

functionallydistinct subsets willchangethe cytokine profile of

aculture.

For example, cultures of PBMC fromatypicalHIV-infected

individual with< 100CD4/1alwillhaveveryfewnaive Tcells,

andan overrepresentation ofmemoryCD8 cells. Thus,evenif the functions of naive andmemorycellsareequivalent in HIV-infected and unHIV-infected individuals, PBMC cultures from the HIV-infected individuals will produce less IL-2 andmoreIL-4

than in comparable cultures from the uninfected individuals. Similarly, since naive T cells proliferate better in responseto

mitogen than domemorycells, the PBMC cultures from HIV-infected individuals with few naive T cells will proliferate less in response to, e.g., phytohemagglutinin. Unfortunately, this meansthat previous demonstrations of thesetypesoffunctional differences between PBMC from individualsatdifferentstages

ofdisease (or between PBMC from infectedversusuninfected donors) could merely reflect differences in the underlying

repre-sentation of naive andmemory cells.

Theloss of T cellsand theimpairmentof Tcellfunctionality

arehallmarks of HIV disease. However, such deficiencies are

notnecessarily restricted to this disease. For example, chronic alcohol consumptionfrequentlyresults inaloss of T cell func-tion andadecreased representation of naive T cell subsets ( 18).

Thus,the mechanisms underlying the naive T cell loss in HIV

diseasemaybeuniquetoHIVdiseaseorreflect general immu-noregulatory processesthat HIV infection setsinmotion.

Insummary,regardless of how naive T cellsarelostorthe

specificity of the naive TcelllosstoHIV disease, this loss has important consequences for thedisease itself and the ways in which we study it. At afunctional level, the progressive loss

of naive cellscontributestotheoverallimmunodeficiencyinthe diseaseby decreasingtheabilitytorespondtonovelantigens. At

a methodological level, theselective loss of the naive subset,

A

(7)

and the corresponding overrepresentation ofmemory subsets, impacts the interpretation ofdata fromin vitro studies. Thus,

thedemonstration that both CD4 andCD8naive T cell subsets

are depleted relatively early in HIV infection of adults and

children introduces significantnewperspectivesfor paradigms

of immune function in HIV disease.

Acknowledgments

We thank Dr. Stan Deresinski for helping establish andruntheclinical

trial; Dr. Hajime Nakamura, Dr. Mitra Dependra, Dr. Steve DeRosa, EricWunderlich,IwanTjioe, and Kai-Uwe Belge for help in the

prepa-ration and analysis of samples; and William Clawson, David Hook, Laura Goldbaum, and Sharon O'Leary forexpertnursingsupport.We especially thank Dr. Thomas Merigan for making the Stanford Center for AIDS Research available for these studies, and Dr. David Parks and the other members of the Stanford Shared FACSt facility staff for flow

cytometrysupport.

Mario Roederer is aSenior Fellow of the Leukemia Society of America. This workwas supported in part by National Institutes of Health grants LM-04836, AI-31770, and CA-42509 together with a

special supplementtoCA-42509 for the clinical trial.

References

1.Sanders, M. E., M. W.Makgoba,andS. Shaw. 1988. Human naive and

memoryTcells. ImmunoL Today.9:195-199.

2.Sohen,S., D.Rothstein,T.Tallman,D.Gaudette,S.Schlossman,and C. Morimoto. 1990. The functionalheterogeneity of CD8+ cells defined by anti-CD45RA (2H4) and anti-CD29(4B4)antibodies.Cell. ImmunoL 128:314-328. 3.De Jong,R.,M.Brouwer,F.Miedema,and R.A.vanLier. 1991. Human CD8'Tlymphocytescanbe divided intoCD45RA'andCD45RO'cells with different requirements for activation and differentiation. J. ImmunoL 146:2088-2094.

4.Picker,L.J.,J. R.Treer,D. B.Ferguson,P. A.Collins,D.Buck,and L. W.

Terstappen. 1993. Control oflymphocyterecirculation in man.I. Differential regulation of the peripherallymphnodehomingreceptorL-selectinonTcells

duringthe virgintomemorycell transition. J. ImmunoL 150:1105-1121. 5.Okumura, M.,Y.Fujii,Y.Takeuchi,K.Inada,K.Nakahara,and H.

Ma-tsuda.1993.Age-relatedaccumulation of LFA-lIghcells in aCD8+CD45RA"hb

T cellpopulation.Eur. J.Immunol.23:1057-1063.

6.Swain,S. L. 1993. Effector functions of helperTcells. J.Immunother. 14:150-154.

7.Ernst, D.N.,W. 0. Weigle,D. J. Noonan, D. N. McQuitty, and M. V. Hobbs. 1993.Theage-associated increase inIFN-gamma synthesis by mouse

CD8+ T cells correlates withshiftsinthefrequenciesof cellsubsets definedby membrane CD44, CD45RB, 3G11, and MEL-14 expression. J. Immunol.

151:575-587.

8.Giorgi,J.V.,andR. Detels.1989. T-cell subsetalterationsinHIV-infected homosexualmen:NIAIDmulticenter AIDS cohortstudy.Clin.Immunol. Immuno-pathol.52:10-18.

9.McGrath, M. S. 1990. T cellAbnormalities.InAIDSKnowledge Base. Textbook onHIV from theUniversityofCalifornia, San Franciscoand San Francisco GeneralHospital.P.T. Cohen,M. A.Sande,andP.A.Volberding,

editors. Medical Publishing Group, Waltham. 3.2.2-1-3.

10.Rabin,R.L.,M.Roederer,Y.Maldonado,A.Petru,L. A.Herzenberg,

and L. A.Herzenberg. 1995.Alteredrepresentationofnaiveand memory CD8

TcellsubsetsinHIV-infected children.J. Clin. Invest.95:2054-2060.

11.Froebel,K.S.,K. V.Doherty,J. A.Whitelaw,R. A.Hague,J.Y.Mok, and A.G.Bird. 1991.Increased expressionof theCD45RO(memory)antigen

onTcellsinHIV-infected children. AIDS(Phila).5:97-99.

12.Borkowsky, W.,T.Moore,K.Krasinski,K.0. Ajuang-Simbiri,and R.

Holzman.1992.Evolutionofphenotypicmemory T cells in HIV-1 infected infants andchildren. Clin. Immunol. Immunopathol. 63:280-284.

13.Reddy, M. M.,and M.H. Grieco.1991.Quantitative changes inThelper inducer(CD4+ CD45RA-), T suppressor inducer (CD4+ CD45RA +), T suppres-sor(CD8'CD1lb'),and Tcytotoxic(CD8'CD1 lb-)subsets inhuman immu-nodeficiency virus infection. J.Clin. Lab.Anal. 5:96-100.

14.Prince, H. E.,S. Kleinman,C.Czaplicki,J.John,and A. E.Williams.

1990.Interrelationshipsbetweenserologicmarkers of immune activation andT lymphocyte subsetsinHIVinfection. J. Acquired ImmuneDefic.Syndr. 3:525-530.

15.Teitel,J.M.,J.J. Freedman,M. B.Garvey,and M. Kardish. 1989.

Two-yearevaluationofclinicalandlaboratoryvariables of immune function in 117 hemophiliacs seropositiveorseronegativeforHIV-1. Am.J. Hematol. 32:262-272.

16.Moore,W.A.,and R. A.Kautz. 1986.Data analysis inflowcytometry.

InHandbookofExperimental Immunology.D. M.Weir,L. A.Herzenberg,and

L.A.Herzenberg, editors.Blackwell ScientificPublications, Edinburgh. 30.1-30.11.

17.Bonyhadi,M.L.,L.Rabin, S. Salimi, D.Brown,J.Kosek,J.McCune,

and H.Kaneshima. 1993.HIVinduces thymus depletioninvivo. Nature (Lond.). 363:728-732.

18.Cook,R.T.,T.J. Waldschmidt,Z.K.Ballas,B. L.Cook,B. M.Booth,

B. C. Steward,and M. J. Garvey. 1994. Fine T-cell subsets in alcoholics as determined by the expression ofL-selectin, leukocytecommonantigen,and