human small-intestinal epithelium contains functional natural killer lymphocytes

Human Small-Intestinal Epithelium Contains Functional NaturalKiller Lymphocytes

FRANCISCO LEON,* ERNESTO ROLDAN,* LAURA SANCHEZ,* CRISTINA CAMARERO,‡

ALFREDO BOOTELLO,* and GARBINE ROY*Departments of *Immunology and ‡Paediatrics, Hospital Ramon y Cajal, Madrid, Spain

Background & Aims: CD3� non-T lymphocytes constitutethe second most abundant lymphoid subset in the hu-man small-bowel epithelium, and these CD3� intraepi-thelial lymphocytes are virtually absent in active celiacdisease. Phenotypically, they resemble natural killercells and have been termed natural killer–like intraepi-thelial lymphocytes. Because of the limited availabilityof appropriate human samples, functional studies havenot yet been reported, and it is not yet clear whetherthese are true natural killer cells. Methods: We usedmagnetic bead–based purification and flow cytometryto study several aspects of normal human small-bowelnatural killer–like intraepithelial lymphocytes: intracellu-lar cytokine content (basally and after activation); abilityto lyse natural killer–sensitive K562 target cells; andexpression of perforins, Fas ligand, and other functionalmarkers. Results: CD3� intraepithelial lymphocytes cul-tured in interleukin-2 showed a higher lymphokine-acti-vated killer activity than CD3� intraepithelial lympho-cytes (48%–83% lysis exerted by CD3� intraepitheliallymphocytes at an effector–target cell ratio of 2:1 vs.8%–18% by CD3� intraepithelial lymphocytes). Perforincontent correlated with this lytic potential (75% � 4% inCD3� vs. 5% � 4% in CD3� intraepithelial lymphocytes).Both CD3� and CD3� cells displayed a type I cytokineprofile (interferon-� > tumor necrosis factor-� > inter-leukin-2; undetectable interleukin-4 and interleukin-10).In addition to their activated phenotype, subsets of nat-ural killer–like intraepithelial lymphocytes expressedCD8�� and intracellular CD3� chain, showing the exis-tence of heterogeneity within this cell lineage.Conclusions: This is the first demonstration of functionalnatural killer cells within the human gut epithelium.These cells might play an important role in innate mu-cosal immunity (host defense and tumor surveillance)and tolerance.

Human intraepithelial lymphocytes (IEL) constitutea heterogeneous and partly unexplored lymphoid

compartment whose components, of controversial ontog-eny (thymic and extrathymic),1–4 display unique pheno-typic and molecular properties5 and ill-defined functions.

Besides the classically described T-cell receptor(TCR)-�� and -�� IEL components,2,6 a thirdCD3� lymphoid subset has been recently characterizedamong human IEL that presents an activated naturalkiller (NK)–like phenotype.7,8 First noticed by Spenceret al.9 in 1989, these CD3� IEL have been almostignored except for 2 subsequent studies by Jarry et al.10

in 1990 and Lundqvist et al.2 in 1995. We have previ-ously reported that the CD3�CD7� IEL subset consti-tutes a numerically relevant fraction within the IELcompartment, accounting for 42% � 20% of the totalIEL in healthy children.11 These cells display a homoge-neously activated phenotype and a unique pattern ofadhesion molecules (lacking CD18)8 and express typicalNK markers, such as CD122 (� chain of interleukin[IL]-2R) and CD161 in their totality and CD2, CD94,CD56, and CD16 in variable proportions.8 Presently, nodata are available about their potential role in mucosalimmune responses, except for the observation that theyare drastically diminished in inflamed celiac disease mu-cosa.9,11

Functional studies with intestinal IEL in humans arescarce, mainly because of difficulties in the collection ofnormal tissue and in the purification of these cells.12,13

There is evidence of the cytolytic capabilities of CD3�

IEL in redirected lysis assays with anti-CD3 antibod-ies14–16 and of their spontaneous lysis against T-celllines16 and epithelial tumor cells.17 Freshly isolatedCD3� IEL do not exert spontaneous killing of the tu-moral NK-sensitive target K562,18 and they do so onlymoderately after a 72-hour culture with IL-2, IL-7, or

Abbreviations used in this paper: APC, allophycocyanin; CM, com-plete medium; DiO, 3,3�-dioctadecyloxacarbocyanine; FCM, flow cyto-metric analysis; FITC, fluorescein isothiocyanate; IEL, intraepitheliallymphocytes; IFN, interferon; i-NK, intraepithelial intestinal NK cells;LAK, lymphokine-activated killer; mAb, monoclonal antibody; NK, nat-ural killer; PE, phycoerythrin; PerCP, peridinin chlorophyll protein;PMA, phorbol myristate acetate; PI, propidium iodide; TCR, T-cellreceptor; TNF, tumor necrosis factor.

© 2003 by the American Gastroenterological Association0016-5085/03/$30.00

doi:10.1016/S0016-5085(03)00886-2

GASTROENTEROLOGY 2003;125:345–356

IL-15 (lymphokine-activated killer [LAK ] assays).19,20

IEL cytokine production has been also investigated, witha prominent type I profile (interferon [IFN]-� and tumornecrosis factor [TNF]-�)14,21–23 predominant over nearlyundetectable IL-422 and IL-10 production.14 The evi-dence thus supports a potential role of IEL in mucosalsurveillance against infections24,25 and tumors,17,26 aswell as in epithelial growth27 and in immunomodulationand maintenance of oral tolerance.28,29 Such diverse func-tions are likely exerted by different IEL subsets. So far,there is no evidence of a direct implication of any IELsubset in mucosal enteropathy.

In this article, we discuss some aspects of phenotypicheterogeneity in the NK-like IEL subset and functionallycharacterize their cytotoxic potential and cytokine pro-duction. Our results are in agreement with the functionalproperties of similar NK IEL subsets that have beenrecently characterized in the mouse,30 chicken,31 andrat32 and show the existence of a functional NK IELsubset in human intestinal mucosa (i-NK).

Materials and MethodsMucosal Specimens

Histologically normal biopsy samples (1–2 mm2) wereobtained from children and adults undergoing endoscopicexamination for the investigation of small-bowel disease inwhich the final diagnoses excluded the involvement of theintestine. Duodenal specimens (4–6 cm2) were obtained fromhealthy adult individuals undergoing Scopinaro operations formorbid obesity. Jejunal and proximal ileal mucosa were alsoobtained from the resected intestine of a child with microvil-lous inclusion disease who was undergoing intestinal trans-plantation. In total, samples from 47 individuals—20 childrenand 27 adults—were used. All the specimens were obtainedafter informed, written consent was given. This study wasapproved by the local Ethics Committee of Ramon y CajalHospital.

Isolation and Purification of IntestinalIntraepithelial Lymphocyte Subsets

Mechanical dispersion of the mucosa. Single cellsuspensions were prepared from the epithelial layer of small-intestinal biopsy samples or fragmented surgical specimens(cut into fragments of approximately 1–5 mm2) by using apreviously described protocol33 with minor modifications.Briefly, IEL and epithelial cells were liberated from the mu-cosal specimens by incubation for 1 hour under intense stirringwith 1 mmol/L dithiothreitol and 1 mmol/L EDTA in RPMI1640 medium (Gibco BRL Life Technologies, Vienna, Austria)supplemented with 10% fetal calf serum and antibiotics (com-plete medium [CM]). The suspension of released cells was thenwashed and further processed according to their further use: (1)labeled with the appropriate conjugated monoclonal antibod-

ies (mAb) for 3- or 4-color flow cytometric analysis (FCM), (2)kept in culture, or (3) subjected to further purification of IELby Percoll density gradients.

The density of IELs per 100 epithelial cells in control biopsysamples was 9.7% � 4.3% (mean � SD). The proportion ofNK-like IEL was 44% � 16% in children and 30% � 12%in adults (mean � SD). We did not observe phenotypic orfunctional differences with respect to age. On average, themechanical dispersion procedure yielded 50,000–150,000 IELper biopsy. Cell clumps were disaggregated by passagethrough a 25-gauge needle.

Cell cultures. IEL were cultured in CM. When re-quired, phorbol myristate acetate (PMA, 5 ng/mL; Sigma, St.Louis, MO) and calcium ionophore (ionomycin, 100 ng/mL;Sigma) or recombinant human IL-2 (R&D Systems, Minneap-olis, MN) was added as indicated.

Purification of intraepithelial lymphocyte sub-sets. First, Percoll density gradients containing 20-40-55%Percoll solution (Amersham Pharmacia Biotec, Uppsala, Swe-den) in CM were used to enrich IEL after mechanical disrup-tion of the mucosa. The cell fraction between 40% and 55%Percoll was collected, and IEL were further purified by immu-nomagnetic methods. We took advantage of the high expres-sion of phosphatidyl serine at the cell surface of enterocytes atthis step of the purification protocol (�95% annexin V�

enterocytes with 9000 means fluorescence channel vs. 15%annexin V� IEL with 200 means fluorescence channel) andused annexin V coated to magnetic microbeads (Dead CellRemoval kit; Miltenyi Biotec, Bergich Gladbach, Germany) toretain dead cells in MiniMACS magnetic columns (MiltenyiBiotec). Eluted IEL (�95% pure and viable) were stained withanti-CD3/fluorescein isothiocyanate and separated into theCD3� and CD3� fractions by using anti–fluorescein isothio-cyanate–coated magnetic beads (Miltenyi Biotec). The result-ing fractions were �90% pure after 2–3 rounds of separation.Limitations in the size of biopsy specimens and the low yieldsobtained with this lengthy approach did not allow us to obtainenough freshly isolated CD3� cells to perform the cytotoxicityassays.

Spontaneous intraepithelial lymphocyte releasefrom mucosal specimens and their expansion with inter-leukin-2. Only mucosal fragments with more than 30%CD3�CD103� IEL were cultured ex vivo without processing,and the cells that were detached after 24–48 hours of culturein CM were harvested and tested for CD103, CD3, and CD19expression to assess their intraepithelial origin and the level ofany eventual lamina propria or peripheral blood contamina-tion. Only samples with less than 10% non-IEL contaminationwere further processed. An initial IEL phenotype after dithio-threitol/EDTA de-epithelization in a parallel sample was per-formed as a reference of the mucosa in the study. IEL werecleared of cellular debris and expanded through successivepasses in CM supplemented with 10 ng/mL recombinanthuman IL-2 (R&D Systems). Expanded CD3� and CD3�

subsets were positively and negatively selected by immuno-magnetic procedures by using goat anti-mouse immunoglob-

346 LEON ET AL. GASTROENTEROLOGY Vol. 125, No. 2

ulin (Ig)G-coated magnetic microspheres (Dynabead M-450;Dynal, Oslo, Norway) previously incubated with anti-CD3mAb (Beckman Coulter, Brea, CA), by following the commer-cial instructions. This procedure of IEL expansion rendered lowcell recoveries (10%) of a high purity (�93%) on bothselected subsets (as controlled by FCM analysis), but there wereenough cells to perform cytotoxic assays with CD3� and CD3�

IEL. Peripheral blood NK cells were purified by immunomag-netic procedures from healthy donor blood: peripheral bloodmononuclear cells were plated on plastic for 1.5 hours at 37°C,and the nonadherent cells were depleted of T and B cells byincubation with anti-CD3 and anti-CD19 antibodies (BDPharmingen, San Diego, CA) followed by goat anti-mouseIgG-coated magnetic microspheres (Dynal), by following themanufacturers’ protocols. The purity of the NK cells wasassessed as �95% by CD45 and CD56 staining, and the cellswere used as a positive control for the lysis assays after stim-ulation with IL-2 (50 ng/mL) for 7 days.

Flow Cytometric Analysis

The expression of surface markers on IEL was analyzedby 3- or 4-color FCM by using a FACScan or a FACSCalibur(BD Biosciences, San Diego, CA) after staining with fluoro-chrome-conjugated mAbs (see below), and data were processedwith Lysis II and Cell Quest software (BD Biosciences). Ac-quisition of multiparameter data was performed with an ap-propriate forward-scatter threshold to exclude erythroid cellsand debris. IEL cells were selected from the entire populationof epithelial cells after gating on the basis of their low 90°light scattering, and their intraepithelial localization was con-firmed by the expression of CD103 and by the absence ofCD19� B cells in the suspension. Perforin intracellular stain-ing was performed after cell fixation (2% paraformaldehyde inphosphate-buffered saline) and permeabilization (0.1% sapo-nin; Sigma). The rest of the intracytoplasmic stainings (cyto-kines and CD3ε) were performed with the commercial kitCytofix/Cytoperm (BD Pharmingen). Fluorochrome-conju-gated isotype control Igs were used for all analyses. Thespecificity of the anti-CD95L/FasL staining was investigatedby blockade with recombinant human CD95L: 1 g of recom-binant human CD95L (rhsFasL; Alexis, CA) per 1 g ofanti-CD95L.

Monoclonal antibodies. For FCM, fluorochrome- orbiotin-conjugated mAbs directed against CD3� (clone Leu4),CD3� (UCHT1), CD8 (SK1), CD10 (HI10a), CD11a (G-25.2), CD16 (NKP15), CD18 (L130), CD28 (L293), CD29(MAR4), CD31 (L133.1), CD44 (L178), CD45 (2D1), CD45(HI30), CD69 (L78), CD80 (L307.4), CD86 (IT2.2), CD95(Dx2), CD103 (BerACT8), CD117 (104D2), CD152/CTLA-4(BNI3), TCR-�� (WT31), TCR-�� (11F2), IL-2 (MQ1-17H),IL-4 (MP4-25D2), IL-10 (JES3-19F1), IFN-� (4S.B3), TNF-�(Mab11), perforins (�G9), and isotype IgG controls were ob-tained from BD Pharmingen; mAbs against CD4 (T4), CD8�(T8), CD8� (2ST8.5H), and CD56 (NKH1) were obtainedfrom Beckman Coulter; mAbs against CD3 (OKT3) and CD54(6.5B5) were obtained from DAKO Diagnostics (Glostrup,

Denmark); and mAbs against CD95 ligand (NOK-1), CD120a(2H10), and CD120b (4D1B10) were obtained from CALTAG(Burlingame, CA). PE-conjugated streptavidin was obtainedfrom BD Pharmingen.

Cytotoxicity assessment. The lysis of the NK-sen-sitive K562 cell line exerted by IEL subsets was measured byusing a previously described protocol34 with minor modifica-tions. Briefly, purified subpopulations of IEL, either incubatedwith recombinant human IL-2 (10 ng/mL) for 12 hours iffreshly isolated from biopsy samples or directly added to theassay if obtained from in vitro expanded cultures, were mixedat the indicated effector–target ratios, centrifuged at low speed(800 rpm) to favor cell contact, and incubated for 4 hours.After washing, cells were stained with propidium iodide (PI;Molecular Probes, Eugene, OR). K562 target cells, a humanerythroleukemic cell line (American Type Culture Collection,Rockville, MD), were cultured to log phase growth. For assaysperformed with freshly isolated IEL (which showed more de-bris than cultured IEL), K562 cells were previously labeledwith the lipophilic membrane dye 3,3�-dioctadecyloxacarbo-cyanine (DiO; Molecular Probes) at a concentration of 30mol/L for the 18 hours before the assay. DiO emits greenfluorescence and allows for the identification of K562 in thefluorometric biparametric analysis of DiO/PI staining. Themortality index was calculated as the percentage of compro-mised K562 PI� in the assay.

Intracellular cytokine detection. This analysis wasperformed on IEL stimulated (or not) with PMA (5 ng/mL;Sigma) and ionomycin (100 ng/mL; Sigma) for 12 hours in CMin the presence of monensin (GolgiStop; BD Pharmingen).After stimulation, cells were washed and labeled with conju-gated mAb (PerCP� and APC�) against surface antigens.They were then fixed and permeabilized (Cytofix/Cytoperm)for 20 minutes, washed with WashPerm (BD), and labeledwith cytokine-specific FITC- and PE-conjugated mAbs andisotype-matched Ig controls. IEL were gated by their sidescatter and CD45� profiles.

ResultsPhenotypic Heterogeneity Within theNatural Killer–Like IntraepithelialLymphocytes (CD3�CD7� IntraepithelialLymphocytes)

Signs of cellular immaturity. The characteristicsof the NK-like IEL8 strongly suggested that they wereindeed NK cells, but we sought to investigate the po-tential coexistence of CD7�CD3� NK/T precursor cellsamong them. We first analyzed the intracellular expres-sion of the CD3� chain by FCM and found it in 10%–25% of the NK-like IEL (n 10), which are by defini-tion negative for surface membrane staining of CD3(Figure 1B). None of the characteristic NK markers—CD2, CD16 and CD56, or CD8�—showed a selectivedistribution between the intracellular CD3ε� or intra-

August 2003 INTESTINAL INTRAEPITHELIAL NK CELLS 347

cellular CD3�� fractions (Figure 1C–F). The immature-cell markers CD1a, CD117, and CD10 were expressed inCD3� IEL at very low proportions (5% � 4% for CD1a,2% � 2% for CD117/c-kit, and 3% � 3% for CD10).terminal deoxynucleotidyl transferase and CD34 werenegative. No immature-cell markers were found in theCD3� subset (Figure 1G–J).

Extrathymic maturation markers. The expressionof the CD8�� homodimer is attributed to an extrathy-mic ontogeny of IEL.10–38 Only a minor fraction ofNK-like IEL expressed CD8 (10% � 6%), but thesewere predominantly CD8�� (83% � 5% of the CD8�

NK-like IEL were CD8��; Figure 1K and L). By con-trast, the CD8�� homodimer represented less than halfof the CD8� TcR-�� IEL (45% � 6%) and a minor

percentage of the CD8� TcR-�� IEL (15% � 7%;Table 1).

Activation antigens and accessory molecules. Inagreement with previous reports,39,40 we noticed a majorexpression of CD69 on CD3� IEL, most of which were

Figure 1. Markers associatedwith ontogeny and immaturityin the IEL compartment. IELwere first identified and gatedin region R on the basis of theirlow side-scatter (SSC) andCD45 expression (A). Shownare representative dot plots ofCD3� intracytoplasmic staining(icCD3� fluorescein isothiocya-nate [FITC]) on surface CD3�

and CD3� IEL subsets (smCD3PerCP) (B). Second, smCD3�

were gated in and analyzed forthe expression of NK markers(C, D, and E) and CD8� (F ).Panels G–J are representativeof the expression of precursor-cell markers on CD3� andCD3� IEL subsets, and panelsK and L show the expression ofCD8� and CD8� on these sub-sets, evidencing the predomi-nant expression of the CD8��homodimer in the NK-like sub-set.

Table 1. Percentage of CD8� and CD8� Expression on IELSubsets (mean � SD; n 10)

Variable TcR-�� IEL NK-like IEL TcR-�� IEL

CD8a 73 � 20 10 � 6 48 � 5CD8��b 15 � 7 83 � 5 45 � 6CD8��b 85 � 7 17 � 5 55 � 6

aPercentage of CD8� cells for each of the 3 major IEL subsets.bPercentage of CD8�� and CD8�� cells within the CD8� fraction ofeach major IEL subset.


CD25� (not shown), and this marker was also present inNK-like IEL (97% � 2% were CD69�CD25�). Only2% � 2% of the CD3� IEL expressed major histocom-patibility complex class II, whereas 39% � 25% of theCD3� IEL did, as reported by others.41 The transferrinreceptor, CD71, was negative on all IEL (not shown).

IEL are more completely activated by ligation of theirco-receptors than by stimulation of the T-cell recep-tor.40,42–44 CD28 and CTLA-4 were undetectable onNK-like IEL, whereas these co-stimulatory moleculescould be readily detected on CD3� IEL (25% � 19%CD28�, n 20; 5% � 4% CTLA-4�, n 10; Figure 2).

The pairs B7/CD80 and B7.2/CD86 are ligands forCD28 and CTLA-4. CD80 was not detected on CD3� (asreported previously39,44) or on NK-like IEL (Figure 2D),and CD86 was scarcely expressed on both CD3� (4% �4%) and NK-like IEL (10% � 9%; n 10; Figure 2C).

CD44 and lymphocyte function–associated antigen-1,two adhesion molecules involved in co-stimulation, weremainly expressed in the CD3�IEL subset (95% � 4%and 45% � 10%, respectively, while undetected on theNK-like IEL), whereas ICAM-1 was predominantly lo-calized in the CD3� IEL fraction (86% � 6% vs. 34% �16% in the CD3� IEL; not shown). These differencestended to disappear when IEL were cultured with IL-2:CD44 and lymphocyte function–associated antigen-1were induced up to 80% in the CD3� fraction, whereasintercellular adhesion molecule 1 was found all throughthe IEL compartment (not shown). CD31/PECAM-1, amolecule that has been implicated in NK cell adhesion toendothelium,45 was present in a small fraction of CD3�

IEL (57% � 9%; n 4), but not in NK-like IEL (notshown).

In summary, we found differences in the expression ofmarkers associated with ontogeny, homing, and activa-tion requirements among IEL subsets, suggesting dis-tinct functional roles for T and NK-like IEL.

Lytic Potential of the Natural Killer–LikeIntraepithelial Lymphocytes

Natural killer cytotoxic assays (K562 as target):cytolytic capacity of freshly isolated CD3� intraepitheliallymphocytes. We performed 3 independent NK lyticassays with freshly isolated IEL (from duodenal biopsies)after 12 hours of preincubation with IL-2. We achievedhigh IEL purity (�90%) but low yields (10%), and wecould not obtain enough freshly isolated CD3� IEL. Thelytic capacity of CD3� IEL was modest, in the range of3% to 8% (percentage of PI� K562) at an effector–targetratio of 1:1 and 4%–10% at a ratio of 5:1 (Figure 3).

Cytolytic capacity of CD3� and CD3� intraepithe-lial lymphocytes cultured with interleukin-2. We per-formed 4 independent NK lytic assays with IEL har-vested from cultured mucosal specimens and expanded inculture for 2–3 weeks with IL-2 (as described in Mate-rials and Methods). After fractionation into CD3� andCD3� subsets, the purity of the fractions was �93%.

Figure 4 shows a representative 2-parameter cytogramanalysis of the cell populations involved in an NK assay,performed with the CD3� (Figure 4F and G) and CD3�

(Figure 4H and I) IEL fractions expanded from ilealmucosa. The percentage of compromised PI� K562 wassuperior when CD3� (NK-like) IEL and purified IL-2–cultured peripheral blood NK cells were used as effec-tors. Increasing the effector–target ratio up to 50 CD3�

IEL per target cell implied a dose–response cytotoxicitythat reached a plateau in the range of 10%–25% IP�

K562 (n 2; not shown). Table 2 shows the range of thelytic capacity displayed by each IEL fraction and com-

Figure 2. Expression of co-stimulatory markers on CD3� and CD3�

IEL subsets. CD4� IEL expressed CD28 (B), unlike CD3� IEL (A). (C)Expression of CD86 on a small fraction of both CD3� and CD3� IEL.By contrast, CD80 was not found on IEL (D). (F ) Intracellular detectionof CTLA-4 (icCTLA4), expressed in only a small fraction of CD3� IEL(intracellular isotype matched IgG control in panel E; icIgG-PE). Ana-lyzed cells were selected on gate R, as in Figure 1.


pared with that of peripheral blood NK cells. As can beseen, NK-like IEL possess a cytotoxic LAK activity thatis clearly superior to that of CD3� IEL.

Potential mechanisms of intraepithelial lympho-cyte–mediated cytotoxicity: perforin content. Freshlyisolated NK-like IEL showed a greater content of intra-cellular perforins (75% � 4%; n 7) than their CD3�

counterparts (5% � 4%, similar to what has been re-cently reported16; Figure 5A). When IEL were expandedin culture with IL-2, the perforin content remained highin the CD3� subset, particularly in the CD56�CD3�

IEL (Figure 5C), and increased up to 25% in the CD3�

IEL subset (Figure 5D).CD95L (FasL) expression. CD95L expression on

freshly isolated IEL was low (6% � 3% in CD3� and 5%� 3% in NK-like IEL; n 10; Figure 5E) but specific,as suggested by its blockade with recombinant humanCD95L protein (Figure 5, histograms). This contrastswith the widespread expression on IEL of its receptor,CD95 (Fas)8,46 (Figure 5F).

The expression of other members of the TNF receptorsuperfamily implicated in apoptosis and cell death wasalso evaluated. CD120a (TNF receptor 1)47 was hardlydetected in the IEL compartment (Figure 5G), whereas

CD120b (TNF receptor 2)48 was highly expressed inCD3� (61% � 2%) and more variably in NK-like IEL(31% � 27%; Figure 5H). Other members of the family,such as the activation antigens CD30 and CD70, werenegative.

In conclusion, CD3� IEL were capable of strong lyticactivity after in vitro expansion with IL-2, comparable toIL-2–stimulated peripheral blood NK cells. In contrast,CD3� IEL were weaker effectors against K562.

Figure 3. Cytolytic capacity of freshly isolated CD3� IEL; (A) granu-larity and size (forward scatter [FSC]/side scatter [SSC]) of coculturedCD3� IEL and K562 cells in a 1:1 ratio. (B) Fluorescence of DiO-labeled K562 vs. the PI staining of the IEL (DiO�) and K562 coculturein an ungated acquisition (upper right quadrant, DiO�PI� compro-mised K562 cells; lower right quadrant, DiO�PI� intact K562 cells).(C) FL2/PI histograms analyzed on gated DiO� K562 cells. The re-sults of this representative experiment are shown as percentage oflysis at 0:1, 1:1, and 5:1 effector–target cell ratios. FL, fluorescence.

Figure 4. Cytotoxic capacity of CD3� and CD3� IEL long-term culturedwith IL-2. (A) Granularity and CD3 expression of purified IEL culturedwith IL-2 before magnetic separation of the CD3� and CD3� fractions.(B) Morphology of cocultured K562 (target) and IEL (effector) in a 1:1ratio. (C) Autofluorescence in the FL1 channel (FL1-AF) of mixedeffector and target cells, showing how the difference between thesepopulations allows for their discrimination during FCM analysis. Thedot plots on the right represent the analysis of PI incorporation intoK562 cells, previously selected on the basis of their light scatterparameters (region R in panel B). The effector cells were peripheralblood (PB) NK cells (D, E), ileal expanded CD3� IEL (F, G), or CD3� IEL(H, I ). The upper right quadrants show the percentage of com-promised K562 PI� at the effector–target ratios of 2:1 and 1:1in this representative experiment. SSC, side scatter; FSC, forwardscatter. FL, fluorescence.

Table 2. NK Lytic Activity of IL-2–Expanded IEL SubsetsFrom Human Small Intestine Versus IL-2–CulturedPeripheral Blood NK Cells

Effector—targetcell ratio

Peripheralblood NK(n 2)

NK-like (CD3�) IELsubset (n 4)

CD3� IELsubset (n 4)

1:1 16, 49 19, 27, 44, 78 3, 5, 13, 172:1 29, 60 48, 50, 58, 83 8, 10, 19, 19

NOTE. Data represent the proportion of PI� K562 cells in each sam-ple.


Interferon-� and Tumor Necrosis Factor-�Are the Main Cytokines Produced byNatural Killer–Like IntraepithelialLymphocytes

The intracellular content of type I (IFN-�,TNF-�, and IL-2) and type II (IL-10, IL-4, and IL-5)cytokines was determined by FCM on freshly isolatedIEL. Basal cytokine content in IEL, without in vitrostimulation, was difficult to evaluate because of the lowfrequencies of cytokine-producing cells, which were inthe limit of the sensitivity threshold of the FCM methodof analysis. In general, IEL had a negligible spontaneousproduction of IFN-� and TNF-� (fewer than 2% cyto-kine-producing cells), whereas IL-2, IL-4, IL-10, andIL-5 were undetectable (Figure 6).

The production of cytokines by CD3� (TcR-�� and-��) and NK-like IEL subsets was then investigatedafter activation with a phorbol ester (PMA) and ionomy-cin. Type I cytokines were readily detectable in all IELsubsets, but not type II ones (Figure 6). The frequency ofIFN-�–, TNF-�–, and IL-2–producing cells was higherin the CD3� TcR-�� IEL subset than in the TcR-��

or NK-like IEL. The results are summarized in Table 3and indicate that (1) IFN-� was mainly synthesized by

Figure 6. Cytokine production by freshly isolated IEL. IEL isolatedfrom duodenal biopsies were cultured for 12 hours with or withoutPMA and ionomycin in the presence of monensin, surface-labeled fordifferent antigens (CD3 and CD45), and fixed and permeabilizedbefore intracellular cytokine staining. Representative dot plots areshown.

Figure 5. Flow cytometric analysis of some potential cytolytic mech-anisms of IEL. (A) Representative analysis of intracellular perforincontent of freshly isolated IEL from a duodenal biopsy, with thecorresponding isotype-matched IgG-PE control (B). Intracellular per-forin content on CD3� (C) and CD3� (D) IEL subsets after in vitro IL-2expansion for 2–3 weeks. Representative dot plots of CD95L (FasL)(E), CD95 (Fas) (F ), CD120a (G), and CD120b (H ) expression onCD3� and CD3� IEL subsets. The histograms on the right suggest thespecificity of the fluorometric CD95L-PE signal by its partial blockagewith recombinant human CD95L (rh-CD95L) protein. PE, phyco-erythrin.


CD3� IEL T cells (45% � 16% of TcR-�� and 17%� 3% of TcR-��), but also by NK-like cells (11% �8%); (2) CD3� IEL synthesized TNF-� (27% � 13% ofTcR-�� and 4% � 1% of TcR-��), as did 11% �7% of the NK-like cells; (3) IL-2 was produced at lowerpercentages by both CD3� (6% � 2%) and NK-like IEL(4% � 6%), whereas IL-4 and IL-10 were almost unde-tectable (less than 2%) in all subsets. Dual cytokine cellcontent for IFN-� and TNF-� was also evaluated by4-color FCM and was observed in approximately 30% ofthe positive cells (Figure 6, lower right dot plot).

Cytokine synthesis was also evaluated in IEL culturedwith IL-2. IEL from 1 jejunal specimen and 1 ilealmucosal specimen, expanded in vitro with IL-2 (10 ng/mL) for 2–3 weeks, showed almost undetectable cytokineproduction without mitogenic stimuli (Figure 7A andC). After a 4–6-hour stimulation period with PMA/ionomycin, the cytokine pattern in CD3� IEL was sim-ilar to that observed in freshly isolated ones, but withhigher frequencies of secreting cells (51%–80% pro-duced IFN-�, and 30%–70% produced IL-2; Figure 7Band D). Expanded NK-like cells produced IFN-� (range,10%–38%) and less IL-2 (5%; Figure 7B and D). IL-4and IL-10 were almost undetectable in both IL-2 cul-tured samples.

In conclusion, a significant fraction of the NK-likeIEL displays a proinflammatory cytokine profile, which iscompatible with a true NK nature.

DiscussionSince the initial observation made by Spencer et

al.9 in 1989 of the presence of CD3�CD7� IEL cells inthe intraepithelial compartment of human small intes-tine, few studies have referred to the existence of thesecells,2,10 with a complete lack of information about theirfunctional relevance. In previous works, we found thatCD3�CD7� cells represented a numerically relevantfraction of small-intestinal IEL,11 characterized theirphenotype as NK-like,7,8 and pointed out that these cellswere drastically diminished in active celiac disease.9,11

We now extend these observations by showing that atleast a subset of the NK-like IEL seem to constitute theintestinal NK cell lineage that has been recently de-

scribed in other species,30–32 on the basis of their phe-notype, their ability to lyse NK-specific targets, andtheir ability to secrete proinflammatory cytokines.

The vast majority of CD3� IEL share a commonphenotype,7,8 but the heterogeneity we report now sug-gests that precursor cells2,3,49,50 may be present withinthe CD7�CD3� IEL subset. We have observed intracel-lular expression of the CD3� chain (found in NK and Tprogenitor cells51) in 10%–25% of the NK-like IEL, butthese cells did not differ from the rest in terms of othersurface markers. These intracellular CD3� IEL may con-stitute the benign counterparts of the aberrant IEL foundin intestinal T-cell lymphoma (enteropathy associated ornot), which are usually surface-membrane CD3�/intra-cellular CD3��/CD2�/CD56�, and in refractorysprue.52,53 CD3� has also been detected in fetal hepatic

Table 3. Frequency of Cytokine-Producing Cells on IEL Subsets Isolated From Human Duodenal Biopsies, as Determined byFlow Cytometry (mean � SD)

Variable IFN-� (n 18) TNF-� (n 12) IL-2 (n 9) IL-4 (n 13) IL-10 (n 13)

TcR-�� IEL 45 � 16 27 � 13 6.5 � 2 1 � 0.5 1.4 � 1TcR-�� IEL 17 � 3 4 � 1 0.1 � 0.2 0.1 � 0.2 0.1 � 0.1NK-like IEL 11 � 8 11 � 7 4 � 6 0.1 � 0.2 0.2 � 0.1

NOTE. IEL were stimulated for 12 hours with PMA and ionomycin before cytokine detection.

Figure 7. Cytokine production by IL-2–expanded IEL. IEL were ob-tained by spontaneous detachment from small-intestinal mucosa andcultured with IL-2 (10 ng/mL) for 2–3 weeks. Before cytokine detec-tion, IEL cultures were stimulated with PMA and ionomycin for 5 hoursin the presence of monensin. Representative dot plots show IFN-� andIL-2 production with (B, D) or without (A, C) PMA with Io stimulation.IO, ionomycin.


NK cells54 and uterine NK cells,55 but not in peripheralblood NK cells unless induced by activation.56 There-fore, the presence of intracellular CD3ε could alterna-tively reflect an ontogenic or functional relationshipamong intestinal, uterine, and fetal hepatic NK cells. Asa third possibility, it could be a sign of cellular hyper-activation. A second aspect of the heterogeneity of NK-like IEL is their partial expression of CD8��, a proposedmarker of extrathymic intestinal ontogeny,57–59 as hasbeen shown for CD8�� IEL in mice.30,37,60 We did notobserve any significant expression of immature lymphoidcell markers (CD34, Tdt, or c-kit61) on NK-like IEL, butmost are CD16low/�, which resembles uterine NK cellsand a minor subset of peripheral blood NK that havebeen considered NK precursors.62–64 In addition, func-tionally immature human NK cells also lack CD16,65

reinforcing the hypothesis of an in situ pathway of NKdifferentiation in the intestinal epithelium.35 A finalphenotypic aspect to note is that the unique pattern ofadhesion molecules the NK-like IEL display is essentiallylost after in vitro culture, indicating that it might re-spond to functional requirements that influence theirmigratory capacity,66 function, or both.67,68

Regarding the functionality of NK-like IEL, cytokineproduction by normal human IEL has infrequently beenreported, and most studies have focused on CD3� IEL,which show a predominant type I cytokine profile.14,69 Inour study, the frequency of IEL that spontaneously pro-duced cytokines was near the sensitivity threshold ofFCM but also showed a proinflammatory T-helper type 1bias, similar in magnitude to that in previous re-ports.14,21 Mitogenic stimulation of freshly isolated IELclearly showed a type I cytokine pattern on both CD3�

and NK-like IEL subsets, with the former as the mainsource of IFN-� and TNF-�. We did not find IL-4 orIL-10 production by normal IEL, even after mitogenicstimulation.

To our knowledge, this is the first time that cytokineproduction by human intestinal intraepithelial NK-likecells has been reported. When stimulated with PMA andionomycin, a significant fraction of NK-like cells pro-duced IFN-� (11% � 8%), TNF-� (11% � 7%), andsome IL-2 (4% � 6%), a type I pattern that was main-tained after in vitro culture. We did not observe type IIcytokine production by human NK-like IEL, unlike thereported IL-4 secretion detected by ELISPOT (enzyme-linked immunospot assay) in rat intestinal NK IEL,32

although the latter cells show notorious differences com-pared with human NK-like IEL (they are CD25�).32 Ourdata suggest an innate cytolytic effector function rather

than an immunoregulatory or tolerogenic one70 for NK-like IEL, although we cannot exclude it.

Regarding the cytotoxic potential of IEL, most previ-ous reports have focused on CD3� IEL in redirected lysisassays14,16–19,71 or after cytokine stimulation.19,71 Thelimitations in the availability of healthy human small-bowel specimens to purify CD3� IEL prompted us toexpand these cells with low concentrations of IL-2 toobtain enough to assess their function in vitro. By doingso, we observed that NK-like IEL lyse K562 cells farmore efficiently than CD3� IEL. This LAK activityparallels the intense cytotoxicity against K562 exertedby CD3�CD56�CD16� large granular lymphocytes ob-tained from unfractionated duodenal mucosa,72 as well asby uterine large granular lymphocytes73 and by circulat-ing CD56�CD16� peripheral blood NK.66 Neither ofthese NK subsets displayed a significant baseline killingwithout IL-2 stimulation. Because K562 does not ex-press Fas and because NK-like IEL possessed a higherperforin content, it seems plausible these molecules mayhave been responsible for the lysis, as has been shown inmurine IEL.74 However, the role of other lytic co-stim-ulators, such as NKG2D15 or CD94,75 cannot be disre-garded. Although our data do not directly support aregulatory or tolerogenic role of this NK-like subset inmucosal immune responses, this possibility is attractivebecause of their absence in celiac disease, a situation ofgluten intolerance,9–11 and because of their phenotypicand functional similarities to uterine NK cells,73,76–78

which have been associated with the development ofmaternal/fetal tolerance.79–81

In conclusion, this is the first functional description ofhuman intestinal intraepithelial NK cells. These NKcells produce type I cytokines, mainly IFN-� andTNF-�, and have a cytolytic LAK activity against NKtargets. Further work is required to investigate the roleof these cells in intestinal immune responses, possibly asinnate lytic effectors and/or as regulators of homeostaticinflammation and oral tolerance, and to determine thereason for their absence in celiac disease.

References1. Gek-Kee S. Intraepithelial lymphocytes and the immune system.

Adv Immunol 1995;58:297–343.2. Lundqvist C, Baranov V, Hammarstrom S, Athlin L, Hammarstrom

ML. Intraepithelial lymphocytes. Evidence for regional specializa-tion and extrathymic T cell maturation in the human gut epithe-lium. Int Immunol 1995;24:1703–1705.

3. Lynch S, Kelleher D, McManus R, O’Farrelly C. RAG1 and RAG2expression in human intestinal epithelium: evidence for extrathy-mic T cell differentiation. Eur J Immunol 1995;25:1143–1147.

4. Rocha B, Guy-Grand D, Vassalli P. Extrathymic T cell differentia-tion. Curr Opin Immunol 1995;7:235–242.

5. Guy-Grand D, Rocha B, Mintz P, Malassis-Seris M, Selz F, Malis-


sen B, Vassalli PJ. Different use of T cell receptor transducingmodules in two populations of gut intraepithelial lymphocytes isrelated to distinct pathways of T cell differentiation. J Exp Med1994;180:673–679.

6. Deusch K, Luling F, Reich K, Classen M, Wagner H, Pfeffer K. Amajor fraction of human intraepithelial lymphocytes simulta-neously expresses the �/� T cell receptor, the CD8 accessorymolecule and preferentially uses the V�1 gene segment. EurJ Immunol 1991;21:1053–1059.

7. Eiras P, Roldan E, Camarero C, Olivares F, Bootello A, Roy G. Flowcytometry description of a novel CD3�/CD7� intraepithelial lym-phocyte subset in human duodenal biopsies: potential diagnosticvalue in coeliac disease. Cytometry 1998;34:95–102.

8. Eiras P, Leon F, Camarero C, Lombardia M, Roldan E, Bootello A,Roy G. Intestinal intraepithelial lymphocytes contain aCD3�CD7� subset expressing natural killer markers and a sin-gular pattern of adhesion molecules. Scand J Immunol 2000;52:1–6.

9. Spencer J, McDonald TT, Diss TC, Walker-Smith JA, Ciclitira PJ,Isaacson PG. Changes in intraepithelial lymphocyte subpopula-tions in coeliac disease and enteropathy associated T cell lym-phoma (malignant histiocytosis of the intestine). Gut 1989;30:339–346.

10. Jarry A, Cerf-Bensussan N, Brousse N, Selz F, Guy-Grand G.Subsets of CD3� (T cell receptor �� and ��) and CD3� lympho-cytes isolated from normal human gut epithelium display pheno-typic features different from their counterparts in peripheralblood. Eur J Immunol 1990;20:1097–2103.

11. Camarero C, Eiras P, Asensio A, Leon F, Olivares F, Escobar H,Roy G. Intraepithelial lymphocytes and coeliac disease: perma-nent changes in CD3�/CD7� and T cell receptor �� subsetsstudied by flow cytometry. Acta Paediatr 2000;89:285–290.

12. Todd D, Singh AJ, Greiner DL, Mordes JP, Rossini AA, Bortell R. Anew isolation method for rat intraepithelial lymphocytes. J Immu-nol Methods 1999;224:111–127.

13. Yamamoto M, Fujihashi K, Kawabata K, McGhee JR, Kiyono H. Amucosal intranet: intestinal epithelial cells down-regulate intra-epithelial, but not peripheral, T lymphocytes. J Immunol 1998;160:2188–2196.

14. Lundqvist C, Melgar S, Mo-Wai M, Hammarstrom S, Ham-marstrom ML. Intraepithelial lymphocytes in human gut have lyticpotential and a cytokine profile that suggest T helper 1 andcytotoxic function. J Immunol 1996;157:1926–1934.

15. Roberts AI, Lee L, Schwarz E, Groh V, Spies T, Ebert EC, Jabri B.NKG2D receptors induced by IL-15 costimulate CD28-negativeeffector CTL in the tissue microenvironment. J Immunol 2001;167:5527–5530.

16. Melgar S, Bas A, Hammarstrom S, Hammarstrom ML. Humansmall intestinal mucosa harbors a small population of cytolyti-cally active CD8� �� T lymphocytes. Immunology 2002;106:476–485.

17. Taunk J, Roberts AI, Ebert EC. Spontaneous cytotoxicity of humanintraepithelial lymphocytes against epithelial cell tumors. Gastro-enterology 1992;102:69–75.

18. Cerf-Bensussan N, Guy-Grand D, Griscelli C. Intraepithelial lym-phocytes of human gut: isolation, characterisation and study ofnatural killer activity. Gut 1985;26:81–88.

19. Ebert EC, Roberts AI. Lymphokine-activated killing by humanintestinal lymphocytes. Cell Immunol 1992;146:107–116.

20. Ebert EC. Interleukin 15 is a potent stimulant of intraepitheliallymphocytes. Gastroenterology 1998;115:1439–1445.

21. Ebert E. Intra-epithelial lymphocytes: interferon-gamma produc-tion and suppressor/cytotoxic activities. Clin Exp Immunol 1990;82:81–85.

22. Carol M, Lambrechts A, Van Gossum A, Libin M, Goldman M,Mascart-Lemone F. Spontaneous secretion of interferon-� and

interleukin-4 by human intraepithelial and lamina propria gutlymphocytes. Gut 1998;42:643–649.

23. van Damme N, de Keyser F, Demetter P, Baeten D, Mielants H,Verbruggen G, Cuvelier C, Veys EM, De Vos M. The proportion ofTh1 cells, which prevail in gut mucosa, is decreased in inflam-matory bowel syndrome. Clin Exp Immunol 2001;125:383–390.

24. Buzoni-Gadel D, Lepage AC, Dimier-Poisson IH, Bout DT, KasperLH. Adoptive transfer of gut intraepithelial lymphocytes protectsagain murine infection with Toxoplasma gondii. J Immunol 1997;158:5883–5889.

25. Muller S, Buhler-Jungo M, Mueller C. Intestinal intraepitheliallymphocytes exert potent protective cytotoxic activity during anacute virus infection. J Immunol 2000;164:1986–1994.

26. Schilbach KE, Geiselhart A, Wessels JT, Niethammer D, Hand-gretinger R. Human �/� T lymphocytes exert natural and IL-2-induced cytotoxicity to neuroblastoma cells. J Immunother 2000;23:536–548.

27. McDonald TT. Effector and regulatory lymphoid cells and cyto-kines in mucosal sites. Curr Top Microbiol Immunol 1999;263:113–135.

28. Weiner HL, Friedman A, Miller A, Khoury SJ, al-Sabbagh A, SantosL, Sayegh M, Nussenblatt RB, Trentham DE, Hafler DA. Oraltolerance. Annu Rev Immunol 1994;12:809–837.

29. Sachdev GK, Dalton HR, Hoang P, DiPaolo MC, Crooty B, JewellDP. Human colonic intraepithelial lymphocytes suppress in vitroimmunoglobulin synthesis by autologous peripheral blood lym-phocytes and lamina propria lymphocytes. Gut 1993;34:257–265.

30. Guy-Grand D, Cuenod-Jabri B, Malassis-Seris M, Vassalli P. Com-plexity of the mouse gut T cell immune system: identification oftwo distinct natural killer T cell intraepithelial lineages. Eur J Im-munol 1996;26:2248–2256.

31. Gobel TWF, Kaspers B, Stangassinger M. NK and T cells consti-tute two major, functionally distinct intestinal epithelial lympho-cyte subsets in the chicken. Int Immunol 2001;13:757–762.

32. Todd DJ, Greiner DL, Rossini AA, Mordes JP, Bortell R. An atypicalpopulation of NK cells that spontaneously secrete IFN-� and IL-4is present in the intraepithelial lymphoid compartment of the rat.J Immunol 2001;167:3600–3609.

33. Madrigal L, Lynch S, Feighery C, Weir D, Kelleher D, O’Farrelly CJ.Flow cytometry analysis of surface major histocompatibility com-plex class II expression on human epithelial cells prepared fromsmall intestinal biopsies. J Immunol Methods 1993;158:207–214.

34. Chang L, Gusewitch GA, Chritton DBW, Folz JC, Lebeck LK,Nehlsen-Cannarella SL. Rapid flow cytometric analysis for theassessment of natural killer cell activity. J Immunol Methods1993;166:45–54.

35. Spits H, Blom B, Jaleco A-C, Weijer K, Verschuren MCM, vanDongen JJM, Heemskerk MHM, Res PCM. Early stages in thedevelopment of human T, natural killer and thymic dendritic cells.Immunol Rev 1998;165:75–86.

36. Lin T, Matsuzaki G, Yoshida H, Kobayashi N, Kenai H, Omoto K,Nomoto K. CD3�CD8� intestinal intraepithelial lymphocytes(IEL) and the extrathymic development of IEL. Eur J Immunol1994;24:1080–1087.

37. Page ST, Bogatzki LY, Hamerman JA, Sweenie CH, Hogarth PJ,Malissen M, Perlmutter RM, Pullen AM. Intestinal intraepitheliallymphocytes include precursors committed to the T cell receptor�� lineage. Proc Natl Acad Sci U S A 1998;95:9459–9464.

38. Guy-Grand G, Cerf-Bensussan N, Malissen B, Malassis-Seris M,Briottet C, Vassalli P. Two gut intraepithelial CD8� lymphocytepopulations with different T cell receptors: a role for the gutepithelium in T cell differentiation. J Exp Med 1991;173:471–481.

39. Sillett HK, Southgate J, Howdle PD, Trejdosiewicz LK. Expression


of activation and costimulatory elements by human intestinalintraepithelial lymphocytes. Scand J Immunol 1999;50:52–60.

40. Ebert EC. Proliferative responses of human intestinal intraepithe-lial lymphocytes to various T-cell stimuli. Gastroenterology 1989;98:1372–1381.

41. Abuzakouk M, Kelleher D, Heighery C, O’Farrelly C. IncreasedHLA-DR and decreased CD3 on human intestinal intraepitheliallymphocytes: evidence of activation? Gut 1996;39:396–400.

42. Cerf-Bensussan N, Begue B, Gagnon J, Meo T. The human intra-epithelial lymphocyte marker HML-1 is an integrin consisting of abeta 7 subunit associated with a distinctive alpha chain. EurJ Immunol 1992;22:885.

43. Sarnacki S, Begue B, Buc H, Le Deist F, Cerf-Bensussan N.Enhancement of CD3-induced activation of human intestinal in-traepithelial lymphocytes by stimulation of the beta 7-containingintegrin defined by HML-1 monoclonal antibody. Eur J Immunol1992;22:2887–2892.

44. Ebert EC, Roberts AI. Costimulation of the CD3 pathway by CD28ligation in human intestinal lymphocytes. Cell Immunol 1996;171:211–216.

45. Schneider MKJ, Forte P, Seebach JD. Adhesive interactions be-tween human NK cells and porcine endothelial cells. ScandJ Immunol 2001;54:70–75.

46. Hongo T, Morimoto Y, Iwagaki H, Kobashi K, Yoshii M, UrushiharaN, Hizuta A, Tanaka N. Functional expression of Fas and Fasligand on human colonic intraepithelial T lymphocytes. J Int MedRes 2000;28:132–142.

47. Galon J, Aksentijevich I, McDermott MF, O’Shea JJ, Kastner DL.TNFRSF1A mutations and autoinflammatory syndromes. CurrOpin Immunol 2000;12:479–486.

48. Chan FK-M, Chun HJ, Zheng L, Siegel RM, Bui KL, Lenardo MJ. Adomain in TNF receptors that mediates ligand-independent recep-tor assembly and signaling. Science 2000;288:2351–2354.

49. Howie D, Spencer J, DeLord D, Pitzalis C, Wathen NC, Dogan A,Akbar A, MacDonald TT. Extrathymic T cell differentiation in thehuman intestine early in life. J Immunol 1998;161:5862–5872.

50. Kumberova A, Howie D, Spencer J, McDonald TT. Extrathymic Tcell development in the human intestine. Scand J Immunol 2001;54(Suppl 1):14 (abstr).

51. Wang B, Wang N, Whitehurst CE, She J, Chen J, Terhost C. Tlymphocyte development in the absence of CD3 epsilon or CD3gamma delta epsilon zeta. J Immunol 1999;162:88–94.

52. Chott A, Haedicke W, Mosberger I, Fodinger M, Winkler K,Mannhalter C, Muller-Hermelink HK. Most CD56� intestinal lym-phomas are CD8�CD5�T-cell lymphomas of monomorphic smallto medium size histology. Am J Pathol 1998;153:1483–1490.

53. Cellier C, Patey N, Mauvieux L, Jabri B, Delabesse E, Cervoni JP,Burtin ML, Guy-Grand D, Bouhnik Y, Modigliani R, Barbier JP,Macintyre E, Brousse N, Cerf-Bensussan N. Abnormal intestinalintraepithelial lymphocytes in refractory sprue. Gastroenterology1998;114:471–481.

54. Phillips JH, Hori T, Nagler A, Bhat N, Spits H, Lanier LL. Ontogenyof human natural killer cells: fetal NK cells mediate cytolyticfunction and express cytoplasmic CD3 epsilon, delta proteins. JExp Med 1992;175:1055–1066.

55. King A. Functions of human decidual NK cells. Am J ReprodImmunol 1996;35:258–262.

56. Lanier LL, Chang C, Spits H, Phillips JH. Expression of cytoplas-mic CD3� proteins in activated human adult natural killer (NK)cells and CD3�� complexes in fetal NK cells. J Immunol 1992;149:1876–1880.

57. Rocha B, Vassalli P, Guy-Grand D. Thymic and extrathymic originsof gut intraepithelial lymphocyte population in mice. J Exp Med1994;180:681–686.

58. Rocha B, Vassalli P, Guy-Grand D. The V� repertoire of mouse guthomodimeric � CD8� intraepithelial T cell receptor �� lympho-

cytes reveals a major extrathymic pathway of T cell differentia-tion. J Exp Med 1991;173:483–486.

59. Cruz D, Sydora BC, Hetzel K, Yakoub G, Kronenberg M, CheroutreH. An opposite pattern of selection of a single T cell antigenreceptor in the thymus and among intraepithelial lymphocytes. JExp Med 1998;188:255–265.

60. Reimann J, Rudolphi A. Co-expression of CD8 alpha in the CD4�T cell receptor alpha�beta� T cells migrating into the murineintestinal epithelial layer. Eur J Immunol 1995;25:1580–1588.

61. Puddington L, Olson S, Lefrancois L. Interactions between stemcell factor c-kit are required for intestinal immune system ho-meostasis. Immunity 1994;1:733–739.

62. Lanier LL, Le AM, Civin CI, Loken MR, Phillips JH. The relationshipof CD16 (Leu-11) and Leu-19 (NKH-1) antigen expression onhuman peripheral blood NK cells and cytotoxic T lymphocytes.J Immunol 1986;136:4480–4486.

63. Nagler A, Lanier LL, Cwirla S, Phillips JH. Comparative studies ofhuman FcRIII-positive and negative natural killer cells. J Immunol1989;143:3183–3191.

64. Cooper MA, Fehniger TA, Caligiuri MA. The biology of humannatural killer cells. Immunol Today 2001;22:633–640.

65. Bennett IM, Zatsepina O, Zamai L, Azzoni L, Mikheeva T, Perus-sia B. Definition of a natural killer NKR�P1A�/CD56�CD16�functionally immature human NK cell subset that differentiates invitro in the presence of interleukin 12. J Exp Med 1996;184:1845–1856.

66. DeGrendele HC, Estess P, Picker LJ, Siegelman MH. CD44 andits ligand hyaluronate mediate rolling under physiologic flow: anovel lymphocyte-endothelial cell primary adhesion pathway. JExp Med 1996;183:1119–1130.

67. Sconocchia G, Titus JA, Segal DM. Signaling pathways regulatingCD44-dependent cytolysis in natural killer cells. Blood 1997;90:716–725.

68. Marquez C, Trigueros C, Fernandez E, Toribio ML. The develop-ment of T and non-T cell lineages from CD34� human thymicprecursors can be traced by the differential expression of CD44.J Exp Med 1995;181:475–483.

69. Nilsen EM, Jahnsen FL, Lundin KEA, Johansen F-E, Fausa O,Sollid LM, Jahnsen J, Scott H, Brandtzaeg P. Gluten induces anintestinal cytokine response strongly dominated by interferongamma in patients with celiac disease. Gastroenterology 1998;115:551–563.

70. Strober W, Kellsal B, Fuss I, Marth T, Ludviksson B, Ehrhardt R,Neurath M. Reciprocal IFN-� and TGF� responses regulate theoccurrence of mucosal inflammation. Immunol Today 1997;18:61–64.

71. Roberst AI, O’Commell SM, Biancone L, Brolin RE, Ebert EC.Spontaneous cytotoxicity of intestinal intraepithelial lympho-cytes: clues to the mechanism. Clin Exp Immunol 1993;94:527–532.

72. Pang G, Buret A, Batey RT, Chen QY, Couch L, Cripps A, Clancy R.Morphological, phenotypic and functional characteristics of apure population of CD56� CD16� CD3� large granular lympho-cytes generated from human duodenal mucosa. Immunology1993;79:498–505.

73. Ferry BL, Starkey PM, Sargent JL, Watt GMO, Jackson M, RedmanCWG. Cell populations in the human early pregnancy decidua:natural killer activity and response to interleukin-2 of CD56-positive large granular lymphocytes. Immunology 1990;70:446–452.

74. Corazza N, Muller S, Brunner T, Kagi D, Mueller C. Differentialcontribution of Fas- and perforin-mediated mechanisms to thecell-mediated cytotoxic activity of naive and in vivo-primed intes-tinal intraepithelial lymphocytes. J Immunol 2000;164:398–403.

75. Jabri B, de Serre NP, Cellier C, Evans K, Gache C, Carvalho C,Mougenot JF, Allez M, Jian R, Desreumaux P, Colombel JF,Matuchansky C, Cugnenc H, Lopez-Botet M, Vivier E, Moretta A,


Roberts AI, Ebert EC, Guy-Grand D, Brousse N, Schmitz J, Cerf-Bensussan N. Selective expansion of intraepithelial lymphocytesexpressing the HLA-E-specific natural killer receptor CD94 inceliac disease. Gastroenterology 2000;118:867–879.

76. Nishikawa K, Saito S, Morii T, Hamada K, Ako H, Narita N, IchijoM, Kurahayashi M, Sugamura K. Accumulation of CD16�CD56�natural killer cells with high affinity interleukin 2 receptors inhuman early pregnancy decidua. Int Immunol 1991;3:743–750.

77. Schallhammer L, Walcher W, Wintersteiger R, Dohr G, SedlmayrP. Phenotypic comparison of natural killer cells from peripheralblood and from early pregnancy decidua. Early Pregnancy 1997;3:15–22.

78. Verma S, King A, Loke YW. Expression of killer cell inhibitoryreceptors on human uterine natural killer cells. Eur J Immunol1997;27:979–983.

79. Clark DA, Vince G, Flanders KC, Hirte H, Starkey P. CD56�lymphoid cells in human first trimester pregnancy decidua as asource of novel transforming growth factor-beta 2-related immu-nosuppressive factors. Hum Reprod 1994;9:2270–2277.

80. Ponte M, Cantoni C, Biassoni R, Tradori-Cappai A, Bentiroglio G,

Vitale C, Bertone S, Moretta A, Moretta L. Inhibitory receptorssensing HLA-G1 molecules in pregnancy. Decidua-associatednatural killer cells express LIR-1 and CD94/NKG2A and acquirep49 and HLA-G1 specific receptors. Proc Natl Acad Sci U S A1999;96:5674–5679.

81. Yamamoto T, Takahashi Y, Kase N, Mori H. Role of decidual NKcells in patients with missed abortion: differences betweencases with normal and abnormal chromosome. Clin Exp Immunol1999;116:449–452.

Received November 30, 2002. Accepted April 17, 2003.Address requests for reprints to: Garbine Roy, M.D., Ph.D., Servicio

de Inmunologıa, Hospital Ramon y Cajal, Ctra. de Colmenar Viejo, Km9,1. 28034 Madrid, Spain. e-mail: [email protected]; fax: (34)91-336-88-09.

This work was supported by Spanish Fondo de Investigacion Sani-taria Grants 00/0196 and 01/9417 (to F.L.).

We thank Dr. Eiras for sharing his expertise on intraepithelial lym-phocytes and Dr. Quijano and Dr. Lobo for their collaboration in thecollection of the surgical specimens.


human small-intestinal epithelium contains functional natural killer lymphocytes

Documents