Journal of Cell Science 113, 169-177 (2000) 169 Printed in Great Britain © The Company of Biologists Limited 2000 JCS0310

Lymphoid adhesion promotes human thymic epithelial cell survival via NF-κB activation

Maria T. Scupoli1,*, Emma Fiorini1, Pier C. Marchisio2,3, Ornella Poffe1, Elda Tagliabue4, Monica Brentegani5, Giuseppe Tridente1,5 and Dunia Ramarli1,5 1Department of Pathology, Section of Immunology, University of Verona, 37134 Verona, Italy 2DIBIT, Department of Biological and Technological Research, San Raffaele Scientific Institute, 20132 Milano, Italy 3Department of Biomedical Sciences and Human Oncology, University of Torino, 10126 Torino, Italy 4Oncologia Sperimentale E-Istituto Nazionale Tumori, Milano, Italy 5Clinical Immunology, Azienda Ospedaliera Verona, 37134 Verona, Italy *Author for correspondence (e-mail: [email protected])

Accepted 27 October; published on WWW 9 December 1999

SUMMARY

Inside the , thymic epithelial cells and analysis. The activation of NF-κB was necessary to promote show an interdependent relationship for their functional survival, since its inhibition by acetyl salicylic acid differentiation and development. As regards possible prevented the promoting effect. The mAb-mediated interdependency for their mutual survival, it is clear that crosslinking of α3β1 was considered as a potential inducer lympho-epithelial adhesion can control the survival of of TEC survival, since we have previously demonstrated developing thymocytes whereas the effects of lymphoid that the engagement of this integrin was able to induce NF- adhesion on epithelial cell survival have never been κB activation in TEC. The crosslinking of α3β1, which described. To address this issue, we performed co-cultures clustered at the lympho-epithelial contact sites, partially between normal human thymic epithelial cells (TEC) and reproduced the promoting activity of cell adhesion. These a mature lymphoid line (H9) or unfractionated results highlight that lympho-epithelial adhesion can thymocytes. TEC were induced to by growth control the survival of thymic epithelial cells through an factor deprivation and the level of cell death was measured intracellular pathway which requires the activation of NF- by flow cytometry. TEC stimulated by cell adhesion showed κB and is triggered by integrins of the β1 family. a significant reduced apoptosis when compared to the control and this phenomenon was associated with increased Key words: Thymus, Apoptosis, Cell interaction, Adhesion binding activity of NF-κB, as measured by gel shift molecule,

INTRODUCTION proteins and increased IL-6 mRNA transcription (Couture et al., 1992; Meilin et al., 1995; Cohen-Kaminski et al., 1993). The thymic epithelium is composed of various subsets of Moreover, adhesion with mature lymphoid T cells results in epithelial cells characterized by tonofilaments, interconnected activation of NF-κB (Ramarli et al., 1996). Previously our by desmosomal junctions, and surrounded by extracellular results have also provided evidence that integrin recruitment at matrix, to form an intralobular meshwork filled with the TEC surface during their adhesion to thymocytes initiates developing T lymphocytes. It is known that thymic epithelial an intracellular signaling leading to increased IL-6 production cells (TEC) play a pivotal role in the homing, intrathymic via activation of the IL-6 gene transcription factors, NF-κB and migration, and differentiation of developing T lymphocytes NF-IL6 (Ramarli et al., 1998). through the release of , the secretion of extracellular Adhesion to extracellular matrix (ECM) components as well matrix components, and the establishment of adhesive as cell-to-cell contact are known to regulate cell survival. In interactions (Crisa et al., 1996; Schluns et al., 1997). On the this processes the major deliverers of the adhesion-derived other hand, it is now apparent that these adhesive interactions signals are members of the integrin family. In fact, in epithelial are also required by TEC to modulate their own functions and and endothelial cells the disruption of integrin-mediated development. In mouse models, the adhesion of thymocytes is interactions with the ECM induces a form of apoptosis termed necessary for the epithelial component of the thymus to anoikis (Frish and Francis, 1994; Montgomery et al., 1994; undergo compartmental differentiation and to properly develop Brooks et al., 1994; Boudreau et al., 1995). Intercellular during embryogenesis (for a review, see Boyd et al., 1993). In adhesion has also been shown to protect from apoptosis and addition, in vitro experiments indicate that adhesion there is evidence to suggest that these processes are also to TEC results in tyrosine phosphorylation of cytoplasmic mediated by integrins. Homotypic cell adhesion mediated by 170 M. T. Scupoli and others

β1 integrins is required by mammary epithelial cells for F12 medium (3:1 mixture), 10% FCS (fetal calf serum), 5 µg/ml survival on the basement membrane (Pullan et al., 1996) and insulin, 5 µg/ml transferrin, 0.18 µM adenine, 0.4 µg/ml survival of bronchial epithelial cells is enhanced by homotypic hydrocortisone, 0.1 nM cholera toxin, 2 nM triiodothyronine, 10 cell contact mediated at least partially by αv integrins (Aoshiba ng/ml EGF (epidermal growth factor), 4 mM glutamine, and 50 IU/ml et al., 1997). As regards heterotypic cell adhesion, it has been penicillin-streptomycin. Confluent TEC primary cultures were shown that engagement of α β and/or α β integrins on detached by trypsin-EDTA treatment, plated on a 3T3-J2 feeder layer L 2 4 1 and expanded to confluent secondary cultures in growth medium. TEC germinal center B cells by their ligand on follicular dendritic destined for co-culture experiments were derived from secondary cells can protect B cells from apoptosis (Koopman et al., 1994). cultures, already devoid of 3T3-J2 cells, and grown to confluence in Moreover, rescue from apoptosis was described in early medium containing one third of the concentration of insulin, hematopoietic cells by α4β1 integrin-dependent adhesion to transferrin, adenine, hydrocortisone, cholera toxin, triiodothyronine, bone marrow stromal cells (Wang et al., 1998). So far, the and EGF, as we have previously observed that this culture condition effects of heterotypic cell adhesion in modulating the life/death minimizes the NF-κB basal activation in unstimulated cells, thus balance of epithelial cells have never been described. giving the optimal NF-κB gel-shift signal following cell adhesion Recently, it has been demonstrated that integrin-mediated (Ramarli et al., 1998). This medium composition is referred to survival requires the activation of the transcription factor NF- throughout the text as ‘medium’. κB, which has been indicated to have a key role in preventing Media were purchased from Seromed (Berlin, FRG), EGF was from Austral Biological (San Ramon, CA), and supplements were apoptosis in many cell types (Beg and Baltimore, 1996; Wang from Sigma-Aldrich (Milan, Italy). et al., 1996; Van Antwerp et al., 1996; Scatena et al., 1998). The H9 human T cell line (kindly provided by Dr A. De Rossi, NF-κB is a pleiotropic regulator of many genes involved in Istituto di Oncologia, Università di Padova, Italy) was maintained in immune and inflammatory responses. The NF-κB family of culture in RPMI 1640 medium (Seromed) supplemented with 10% proteins comprises homo- or heterodimers of the Rel family, FCS, 4 mM glutamine, and antibiotics. including p50 and p65 (Rel A) subunits. In unstimulated cells, Human thymocytes were prepared by mechanical disruption of NF-κB is localized in the cytosol bound to inhibitors of the thymus specimens. At least 95% viable cells were isolated from the IκBs family which mask its nuclear localization sequence (for cell suspension by Ficoll-Hypaque gradient centrifugation, washed, a review, see Baldwin, 1996). Various extracellular stimuli and used immediately after preparation. are able to induce the serine hyperphosphorylation and Adhesion assay and mAb treatment proteosomal degradation of inhibitor (Palombella et al., 1994), Co-culture experiments were performed by adding H9 cells or or the tyrosine phosphorylation and subsequent dissociation of thymocytes to TEC monolayer at 1:1 H9/TEC or 5:1 thymocyte/TEC inhibitor (Imbert et al., 1996), thus allowing NF-κB to move ratios, respectively. Co-cultures were carried on for 24 hours in into the nucleus, bind to promoter regions, and transactivate medium devoid of insulin, transferrin, adenine, hydrocortisone, genes. cholera toxin, triiodothyronine, and EGF (minimal medium). When In the present study, we explored whether the increased NF- indicated, H9 cells or thymocytes were replaced with supernatatants κB activity we have previously observed in TEC upon adhesion derived from the same cells cultured for 24 hours in minimal with H9 cells or thymocytes (Ramarli et al., 1996, 1998) could medium. At the end of co-culture, H9 cells as well as thymocytes also promote TEC survival. We observed that cell adhesion were removed by vigorous washings. Untreated and supernatant- with H9 cells or thymocytes partially inhibited TEC from treated TEC were equally subjected to vigorous washing, to prevent artifacts eventually created by washing. TEC were then cultured for apoptosis induced by growth factor deprivation and this κ 1 to 4 days in minimal medium. Blocking experiments were phenomenon was associated with increased NF- B binding performed by preincubating TEC monolayers with mAbs anti-β1 activity. The protective effect mediated by lymphoid cell (MAR4, kindly provided by Dr S. Ménard, Istituto Nazionale adhesion was abolished when NF-κB activation was inhibited Tumori, Milano, Italy), anti-ICAM-1 (84H10 clone, Immunotech, by acetyl salicylic acid. Moreover, a reduction of apoptosis was Coulter, Hialeah, Fl), or with nonspecific IgG (Sigma-Aldrich), used also observed in TEC upon cross-linking of α3β1 integrin, purified at 5 µg/ml, at 4¡C for 30 minutes. Then, H9 cells or which was found to cluster at the lympho-epithelial contact thymocytes were added to TEC monolayers and co-cultures were sites. We propose that heterotypic cell adhesion might carried out at 37¡C, as above described. For co-culture experiments modulate TEC survival by initiating an intracellular signaling in the presence of acetylsalicylic acid or indomethacin (Sigma- which requires NF-κB activation. Moreover, we suggest that Aldrich), these drugs were added to TEC monolayers 30 minutes α β before H9 were dispensed on. 3 1 integrin heterodimers might initiate this intracellular Treatment with monoclonal antibodies (mAbs) were performed by pathway. incubating TEC monolayers with mAbs anti-β1 (MAR4), anti-α3 (K20 clone), anti-CD44 (J.173 clone), or anti-ICAM-1 (84H10 clone), all from Immunotech (Coulter) for 1 hour, at 37¡C. mAbs were used MATERIALS AND METHODS purified at 5 µg/ml. The incubation with primary antibody was followed by F(ab)2 goat anti-mouse IgG (Pierce, Oud Beijerland, the Cell cultures Netherlands), at 10 µg/ml, for 2 hours, at 37¡C. When indicated, Thymic epithelial primary cultures were derived from normal primary antibody was replaced by nonspecific mouse IgG (Sigma- thymuses of children (<5 years of age) undergoing cardiac surgery Aldrich), used purified at 5 µg/ml. (Green et al., 1979). Briefly, thymus specimens were minced and treated with a 0.05% trypsin-0.01% EDTA solution at 37¡C for 3 Apoptosis detection hours. Cells were collected every 30 minutes, pooled, plated onto For light microscopy analysis, TEC at 2nd culture passage were lethally irradiated 3T3-J2 cells (a gift from Dr H. Green, Harvard seeded onto round glass coverslips fitting the wells of a 24-well plate Medical School, Boston, MA) at 2.5×104/cm2, and cultured in a at 1.2×104/cm2, grown to confluence, and subjected to the various humidified atmosphere of 5% CO2 in growth medium composed as treatments. The coverslips were fixed in methanol/acetic acid (3:1) for follows: DMEM (Dulbecco’s modified Eagle’s medium) and Ham’s 20 minutes, stained with a Giemsa solution 1:10 in water for 15 NF-κB activation in adhesion-mediated TEC survival 171 minutes, and mounted on slides. The slides were photographed with apparatus (Coulter). Results were expressed as H9 cells: TEC ratio ± a Zeiss Axioscope compound at ×40 magnification. s.e.m. Apoptotic nuclei were identified by their diminished DNA stainability (hypodiploid peak) which has been shown to rely on DNA degradation and leakage from the cells (Darzynkiewicz et al., 1992). RESULTS For the measure of DNA nuclear content, TEC monolayers were µ treated with hypotonic fluorochrome solution (50 g/ml propidium Lymphoid adhesion protects thymic epithelial cells iodide in 0.1%, w/v, sodium citrate and 0.1%, v/v, Triton X-100), in wells of a 24-well culture plate. The cells were gently resuspended from apoptosis and recovered in polypropylene tubes, then placed at 4¡C in the dark The ability of cell-cell contacts in regulating TEC survival was overnight before analysis (Nicoletti et al., 1991). The PI fluorescence investigated by co-culturing TEC monolayers induced to intensity was measured by a XL flow cytometry (Coulter). PI was apoptosis and mature lymphoid T cells (H9) or unfractionated excited with a 488 nm argon laser and fluorescence was detected with thymocytes. TEC used in the co-culture experiments were at a 600 nm dichroic mirror and a 620 nm band pass filter (band width the third passage of culture and expressed high amount of β1 615-625 nm). Fluorescence and scatter signals were recorded on a and β4 integrins, CD44, MHC class I molecules, but low levels logarithmic scale. Cell debris were excluded from analysis by raising of ICAM-1 (Ramarli et al., 1998). TEC cultures also retained the fluorescence threshold. The flow rate was at about 100 nuclei × seconds−1 and 104 nuclei of each sample were analyzed. Each the ability to form polarized, continuous monolayers, produce experiment was carried on in triplicate and results were expressed as ECM components, and secrete IL-6 both in a constitutive and the mean of the percentage of hypodiploid peaks ± s.e.m. inducible manner (Ramarli et al., 1998). Apoptosis was efficiently induced in TEC grown to organized monolayers by EMSA (electrophoretic mobility shift assay) deprivation of growth factors. In fact, after one to two days Nuclear extracts were prepared from TEC as previously described following growth factor withdrawal, the cells began to display (Schreiber et al., 1989), with minor modifications. Protein concentration was determined by Coomassie protein assay reagent (Pierce). Binding activity of nuclear proteins was determined by using the following [γ-32P]ATP end-labeled double-strand oligonucleotides as probes: 5′-CAGAGGGGACTTTCCGAG-3′, corresponding to the κΒ enhancer element from the Igκ gene; or 5′-ATTCGATC- GGGGCGGGGCGAGC-3′, containing the DNA binding site for Sp1. Nuclear extracts (6 µg) were incubated for 30 minutes at room temperature with the probe (1×105 cpm) in 20 µl of binding buffer (20 mM Tris-HCl (pH 7.5), 0.1 M NaCl, 1 mM dithiothreitol (DTT), 1 mM EDTA, 1 mg/ml bovine serum albumin (BSA), 0.1% Nonidet P-40 (NP-40), 4% glycerol) containing 1 µg of poly(dI-dC) (Pharmacia, Uppsala, Sweden). Sample were then electrophoresed onto a 5% (30:1.2) polyacrylamide gel in 0.05 M Tris-borate/0.001 M EDTA at 150 V. When indicated, polyclonal antibodies directed against the NF-κB p65 or p50 subunits (Santa Cruz Biotecnology, Santa Cruz, CA) were incubated with nuclear extracts for 30 minutes on ice before the addition of the probe. Gels were dried and the retarded bands were revealed by exposure to Hyperfilms films (Amersham, Little Chalfont, UK). Film densitometry was performed with an Ultroscan Densitometer and the built-in software (LKB, Bromma, Sweden). Flow cytometry Phenotype analysis of TEC after co-culture and washes from H9 cells or thymocytes, was performed by indirect immunofluorescence and flow cytometry on a XL apparatus (Coulter). Anti-CD3 (Leu-4) and anti-CD2 (S5.2) mouse monoclonal antibodies (mAbs) were from Becton-Dickinson (Mountain View, CA); FITC-conjugated F(ab)2 goat anti-mouse Ig was from Immunotech (Coulter). Binding assay H9 cells were labeled with PKH 26 GL (Sigma-Aldrich) according to the manufacturer’s recommendations. Optimal cell labeling was achieved by incubating 2×107 cells/ml with 20 µM PKH 26 GL for 2 minutes, at room temperature. The reaction was stopped by adding an equal volume of FCS. H9 cells were then washed and dispensed on TEC confluent monolayer at a 1:1 H9/TEC ratio, in minimal medium plus 3% BSA. Co-cultures were incubated for 1 hour at 37¡C, in a humidified atmosphere of 5% CO2. At the end of the co-culture, the non-adherent lymphoid cells were removed by three subsequent Fig. 1. Morphological features of cell death induced in TEC by gentle washes. TEC monolayer and bound H9 cells were then growth factor deprivation. Photomicrographs of Giemsa-stained TEC detached by tripsin-EDTA treatment, washed, vigorously resuspended monolayers maintained for 4 days in medium (A), or in minimal to disrupt aggregates, and analyzed by flow cytometry on a XL medium showing the nuclear condensation and fragmentation (B). 172 M. T. Scupoli and others

A B

60 C Fig. 2. Effect of lymphoid adhesion on TEC apoptosis upon growth Control factor withdrawal. (A) Flow cytofluorimetric analysis of TEC 50 Minimal medium preparations after co-culture with H9 cells and subsequent washes. Abscissa: CD3-specific fluorescence as detected by indirect +H9 immunofluorescence with mouse mAb and FITC-conjugated F(ab)2 40 goat anti-mouse Ig. Primary mAb was replaced by isotype matched mouse IgG as control. The values are expressed in mean fluorescence 30 (a.u.) on a logarithmic scale (B) DNA content frequency histograms of TEC cultured for 4 days in medium (control), in minimal medium (minimal medium), in minimal medium plus H9 cells (+ H9), or in 20 the presence of H9 supernatant (H9 sup). The percentage of hypodiploid peaks is indicated. Histograms are representative of Hypodiploid nuclei (%) Hypodiploid nuclei 10 seven independent experiments. Each experiment was carried on in triplicate. (C) Time course analysis of the percentage of hypodiploid peaks in TEC cultured in medium (control), in minimal medium 0 (minimal medium), or in minimal medium plus H9 cells (+H9). Data ± 012345 represent the average of triplicates s.e.m. of a representative experiment of at least three experiments performed independently by Time (days) using TEC derived from different donors. morphological features consistent with substantial apoptotic nuclei for fragmentation and analyzing the hypodiploid DNA death such as detachment from culture substratum, membrane content. The results of Table 1 and Fig. 2B, respectively, blebbing, and nuclear condensation and fragmentation (Fig. 1). obtained by performing the two criteria of analysis, show that Moreover, TEC maintained in minimal medium displayed a growth factor withdrawal-induced apoptosis significantly noteworthy increase in the percentage of hypodiploid peak diminished when TEC were co-cultured with H9 cells whereas when compared to the control, as assessed by measuring the the treatment with H9 supernatant was ineffective. The time DNA content of propidium iodide-stained nuclei by flow course analysis shown in Fig. 2C indicates that the hypodiploid cytometry (12±3 versus 6±2, 36±2 versus 6±2, 52±1 versus peak in TEC cultured in minimal medium increased in a time- 7±1 percentage of apoptosis ± s.e.m. at days 2, 3, and 4 from dependent manner and that the difference in the degree of growth factor withdrawal, respectively) (Fig. 2C). apoptosis was significantly maintained throughout the period Co-culture experiments were firstly performed with TEC of observation (38±14%, 41±7%, and 30±4% reduction ± induced to apoptosis and H9 cells. At the end of co-cultures, s.e.m. at days 2, 3, and 4, respectively, P<0.05). H9 cells were removed by extensive washes, as described in To extend these findings to a more physiological setting, we the Materials and Methods. After washes, the H9 cell next co-cultured TEC with unfractionated thymocytes for 24 contamination of TEC was less than 2%, as assessed by hours in minimal medium. After thymocytes were washed off, immunofluorescence with anti-CD3 mAb and flow cytometry the thymocyte contamination of TEC was ∼2%, as assessed by (Fig. 2A). Then, TEC apoptosis was assessed 4 days after immunofluorescence with anti-CD2 mAb and flow cytometry growth factor deprivation both by scoring Giemsa-stained (not shown). After 4 days from growth factor withdrawal, TEC NF-κB activation in adhesion-mediated TEC survival 173

Table 1. Effects of lympho-epithelial cell adhesion on TEC the apoptosis induced by growth factor withdrawal, and that apoptosis this protection is dependent upon cell-cell adhesion rather than on soluble factors secreted by H9 cells or thymocytes. % Apoptosis ± s.e.m. Treatment n=3 Lymphoid cell adhesion increases NF-κB binding Medium 6±2 activity in TEC cultured in minimal medium Minimal medium 47±8 We have previously shown that adhesion with mature lymphoid Minimal medium 28±7 κ + H9 cells cells or thymocytes induced increased NF- B binding activity Minimal medium 45±5 in TEC cultured in the presence of exogenous growth factors + H9 supernatant (Ramarli et al., 1996, 1998). To investigate whether TEC Minimal medium 33±2 adhesion-dependent protection was accompanied by NF-κB + thymocytes activation, we analyzed the NF-κB binding activity of H9 Minimal medium 44±8 adhesion-stimulated and unstimulated TEC cultured in + thymocyte supernatant minimal medium. Nuclear extracts from H9-stimulated and unstimulated TEC were therefore subjected to electrophoretic TEC grown onto round glass coverslips were co-cultured with H9 cells or κ thymocytes for 24 hours in medium devoid of growth factors (minimal mobility-shift assay (EMSA) of a B sequence as probe. As medium), as described in Materials and Methods. After 4 days from growth shown in Fig. 4, the constitutive NF-κB binding activity factor withdrawal, the coverslips were stained with a Giemsa solution and detected in unstimulated TEC was increased more than TEC nuclei were scored for fragmentation. Percentage mean values ± s.e.m. threefold upon lymphoid adhesion, as assessed by gel were calculated from three independent experiments. densitometry (not shown). In contrast, H9 supernatant failed to induce any increase in NF-κB binding activity above the nuclei were both scored for fragmentation and analyzed for constitutive level, thus excluding the possibility that NF-κB hypodiploid DNA content. The results summarized in Table 1 activation could depend on soluble factors secreted by H9 cells and Fig. 3 show that TEC co-cultured with thymocytes (Fig. 4). The translocated complex contained p65/p50 NF-κB displayed a significative reduction of apoptosis when heterodimers, since when nuclear extracts were preincubated compared to the control (30±10% and 27±1% reduction ± with polyclonal antibodies anti-p50 or p65 the primary band s.e.m., for fragmented nuclei scoring and hypodiploid DNA was supershifted (Fig. 4). These results indicate that the partial content analysis, respectively). The treatment with supernatant inhibition from apoptosis observed in TEC upon lymphoid derived from thymocytes cultured for 24 hours in minimal adhesion is associated with increased binding activity of their medium had no effect on TEC survival. endogenous NF-κB. These results indicate that lympho-epithelial interactions with mature T cells or thymocytes partially inhibit TEC from Inhibition of NF-κB prevents protection from apoptosis induced by cell adhesion To explore whether the increase of NF-kB binding activity induced by H9 adhesion was functionally related to the enhanced survival, we decided to inhibit NF-κB activation by acetyl salicylic acid (ASA). To this aim, we evaluated in control

minimal medium

+ thy

+ thy sup

0 20406080100 κ Hypodiploid nuclei (%) Fig. 4. Induction of NF- B binding activity in TEC by H9 cell adhesion. NF-κB binding activity of nuclear extracts from TEC Fig. 3. Effect of thymocyte adhesion on TEC apoptosis upon growth cultured for 24 hours in minimal medium (lane 1), in minimal factor withdrawal. TEC were cultured for 24 hours in medium medium plus H9 cells (lane 2), or plus H9 supernatant (lane 3). (control), in minimal medium (minimal medium), in minimal EMSA was performed with 6 µg of nuclear extract probed with 32P- medium plus thymocytes (+thy), in minimal medium plus thymocyte labeled κB oligonucleotide at saturating amounts. Nuclear complexes supernatants (+thy sup). After treatments, TEC were further cultured contain p50 and p65 heterodimers, as assessed by the band in minimal medium for 3 days and the percentage of hypodiploid supershifting obtained by preincubating the same extract of lane 2 peaks was assessed. Data from TEC cultured in minimal medium are with anti-p50 (lane 5) or p65 (lane 6) antisera. A section of gel is referred as 100% of apoptosis and data from other cultures are shown. The results are representative of at least three experiments expressed as percent of this maximun. The average of triplicates ± performed independently by using TEC derived from different s.e.m. of a representative experiment out of three are shown. donors. 174 M. T. Scupoli and others

Fig. 5. Analysis of TEC-H9 the adhesion-mediated NF-κB activation in a dose-dependent binding in the presence of manner, reaching the maximum inhibitory effect when used at ASA. Binding assay was 5 mM. In contrast, the non-steroidal anti-inflammatory drug performed between TEC minimal indomethacin used as control had no effect on the activation of medium monolayers grown to tight NF-κB. The retarded complexes were identified by using anti- confluence and PK H26-GL- p50 or p65 specific antibodies in supershift analysis: the upper labelled H9 cells in minimal minimal medium or in minimal band was composed by p50/p65 heterodimers, previously medium demonstrated to be transcriptionally active (Baldwin, 1996), medium containing 5 mM + ASA ASA (minimal medium + whilst the lower band by p50 homodimers (Fig. 6A). The ASA). Results are expressed difference in size of the lower band with respect to the EMSA ± as mean s.e.m. of H9: TEC 0 1 analysis of Fig. 4 reflects differences in the balance between ratio triplicates. Shown are hetero- and homodimers we can observe among different TEC data from a representative H9/TEC ratio preparations. In the same extracts, a constitutive transcription experiment out of two. factor, Sp1, was unaffected by the presence of ASA (Fig. 6B), thus excluding.the possibility that ASA could interfere with preliminary assays possible effects of ASA on TEC which transcription in a general manner. might hamper our study. Firstly, a possible dose-dependent We next evaluated whether the blockage of NF-κB activation cytotoxic effect of ASA was analyzed by trypan-blue staining. could prevent the protective effect mediated by lymphoid The results indicated that treatment of TEC with 1 to 5 mM adhesion. H9/TEC co-cultures were carried on in minimal ASA for 24 hours didn’t affect TEC viability (not shown). medium in the presence of 5 mM ASA and TEC apoptosis was Secondly, we investigated whether ASA could affect the analyzed by flow cytometry. Results obtained 24 hours after expression of some of the adhesion molecules known to be H9 removal are summarized in Fig. 7. When NF-κB was expressed on the TEC surface (Ramarli et al., 1998). Flow inhibited by ASA, lymphoid adhesion failed to have protective cytometry analysis showed that treatment of TEC with 5 mM effects on TEC. Conversely, H9 adhesion performed in the ASA for 24 hours did not affect the expression of β1, β4, α3, and α6 integrin subunits, ICAM-1, and CD44 (not shown). Lastly, we evaluated the possibility that ASA could interfere with lympho-epithelial binding by analyzing the H9/TEC binding at a single cell level by flow cytometry. H9/TEC co-culture experiments were carried out with PKH 26-labeled H9 cells, in the presence or absence of 5 mM ASA. The ratio of bound H9/TEC was similar either in the absence or in the presence of ASA (Fig. 5), thus ruling out that ASA could affect the extent of lympho-epithelial binding. Lympho-epithelial co-culture experiments were therefore performed in the absence or presence of 1,3, or 5 mM ASA. As shown in Fig. 6A, ASA inhibited

Fig. 6. Effect of acetyl salycilic acid (ASA) on DNA binding activity of NF-κB and Sp1 transcription factors. (A) Dose-response inhibition of ASA on NF-κB activity induced in TEC by lymphoid adhesion. NF-κB binding activity from TEC cultured for 24 hours in minimal medium (lane 1), in minimal medium plus H9 in the presence or absence of the indicated concentration of acetyl salicylic acid (ASA) (lanes 2-5). In lane 6 ASA was replaced by 20 µM indomethacin. EMSA was performed with 6 µg of nuclear extract probed with 32P-labeled κB oligonucleotide at saturating amounts. The retarded complexes were identified as p50/p65 heterodimers (upper band) and p50/p50 homodimers (lower band) by the band supershifting obtained by preincubating the same extract of lane 2 with anti-p50 (lane 8) or p65 (lane 9) antisera. A section of gel is shown. The results are representative of at least three experiments performed independently by using TEC derived from different donors. (B) Effect of ASA on Sp1 activity in TEC stimulated by lymphoid adhesion. The same extract of A were analyzed for Sp1 binding activity. TEC were cultured for 24 hours in minimal medium (lane 1), in minimal medium plus H9 in the presence or absence of the indicated concentration of acetyl salicylic acid (ASA) (lanes 2-5). In lane 6 ASA was replaced by 20 µM indomethacin. EMSA was performed with 6 µg of nuclear extract probed with 32P-labeled Sp1 oligonucleotide at saturating amounts. A section of gel is shown. The results are representative of at least two experiments performed independently by using TEC derived from different donors. NF-κB activation in adhesion-mediated TEC survival 175

A control control minimal medium minimal medium mouse IgG minimal medium +ASA anti-CD44 α +H9 anti- 3 anti-β1 +H9 +ASA anti-ICAM-1 +H9 +indometh. anti- ICAM1 +-α3

020406080100 020406080100

HHypodiploidypodiploid nuclei (%) Hypodiploid nuclei (%)

Fig. 7. Inhibition of the adhesion-dependent protective effect by B acetyl salicylic acid (ASA). TEC were cultured for 24 hours in +H9 medium (control), in minimal medium (minimal medium), in minimal medium plus 5 mM acetyl salicylic acid (minimal medium mouse IgG + ASA), in minimal medium plus H9 cells (+H9), in minimal medium plus H9 cells in the presence of 5 mM acetyl salicylic acid (+H9 +ASA), or 20 µM indomethacin (+H9 +Indometh.). After treatments, TEC were further cultured in minimal medium for 24 anti-β1 hours and the percentage of hypodiploid peaks was assessed. Data from TEC cultured in minimal medium are referred as 100% of apoptosis and data from other cultures are expressed as percent of this maximum. The average of triplicates ± s.e.m. of a representative anti-ICAM experiment out of three are shown. presence of indomethacin, which was ineffective on NF-κB activity, did not prevent the adhesion-mediated protection. 0 20406080 These data indicate that activation of NF-κB is required for the Inhibition of protection (%) lymphoid adhesion-mediated protection observed in TEC upon growth factor withdrawal. Fig. 8. Involvement of α3β1 integrin in adhesion-dependent protection from apoptosis. (A) Protective effects of α3β1 crosslinking Cross-linking of α3 and β1 integrins partially inhibits on TEC apoptosis upon growth factor withdrawal. TEC cultured in thymic epithelial cells from apoptosis minimal medium were untreated (minimal medium) or treated with κ the indicated mAbs (or a mouse nonspecific IgG) at 5 µg/ml We have recently shown that NF- B binding activity in TEC µ α β followed by F(ab)2 goat anti-mouse at 10 g/ml. mAbs were was efficiently induced by the clustering of 3 1 integrin replaced by a nonspecific goat anti-mouse IgG as control. After the reproduced by mAb-mediated crosslinking. We have also removal of stimuli, TEC were further cultured for 24 hours in demonstrated that this integrin re-polarized and clustered at the minimal medium and the percentages of hypodiploid peaks were intercellular boundaries between TEC and thymocytes assessed. Data from TEC cultured in minimal medium are referred as (Ramarli et al., 1998) and between TEC and H9 cells 100% of apoptosis and data from other cultures are expressed as (unpublished observation) during adhesion. These previous percent of this maximum. The average of triplicates ± s.e.m. of a findings, together with the present observation on the representative experiment out of three are shown. (B) Inhibition of requirement of NF-κB binding activity for the adhesion- H9-mediated TEC protection by mAb anti-β1 integrin. TEC were β mediated survival to occur, prompted us to explore whether precoating with a nonspecific mouse IgG, anti- 1 mAb, or anti- α β ICAM-1 mAb (used at 5 µg/ml, at 4¡C, for 30 minutes), before the 3 1 could be functionally involved in promoting TEC survival co-culture with H9 cells, at 37¡C. After H9 were washed off, TEC during the lympho-epithelial adhesion. With the purpose of α β were further cultured for 24 hours in minimal medium and the reproducing the clustering of 3 1, we crosslinked the percentages of hypodiploid peaks were assessed. Data are expressed heterodimer on TEC monolayers induced to apoptosis with as percentage inhibition of the protection achieved in the presence of mAbs recognizing the extracellular domain of α3 or β1 H9 alone. The average of triplicates ± s.e.m. of a representative subunits. Then, we measured the level of TEC apoptosis 2 days experiment out of three are shown. after the removal of growth factors. The results of the flow cytometry analysis of hypodiploid peaks are summarized in ± s.e.m. for α3 and β1, respectively). Conversely, TEC survival Fig. 8A. The crosslinking of either α3 or β1 integrin chain was unaffected by the crosslinking of the CD44 adhesion reduced TEC apoptosis of more than twenty percent compared molecule, whilst the crosslinking of ICAM-1 induced only a to the mouse IgG used as control (22±1% and 25±2% reduction slight and not statistically significative decrease of TEC 176 M. T. Scupoli and others apoptosis. Moreover, the simultaneous crosslinking ICAM-1 defective mice die during embryogenesis, apparently for and α3 integrin did not alter the extent of the protection extensive death of liver cells (Beg et al., 1995). In addition, mediated by the α3 crosslinking alone. To confirm the cells functionally defective of NF-κB p65 are sensitive to the functional involvement of β1 integrins in the cell adhesion- pro-apoptotic effects of TNF-α (Beg and Baltimore, 1996; dependent protection, inhibition experiments were performed. Wang et al., 1996; Van Antwerp et al., 1996; Scatena et al., As shown in Fig. 8B, the extent of the protection was 1998). To determine a role for NF-κB in protecting human considerably reduced when TEC were incubated with anti-β1 epithelial cells from apoptosis, we inhibited NF-κB activation mAb before the binding with H9 cells (65±5% reduction ± by means of acetyl salicylic acid at a concentration (5 mM) s.e.m.). A lower but notable reduction was also exerted by close to those generally attained in the plasma during clinical mAbs anti-ICAM-1 (41±2% s.e.m.) whereas a 18±2% s.e.m. treatments (1-3 mM). Acetyl salicylic acid has been recently reduction was observed in the presence of nonspecific mIgG. shown to specifically inhibit the activity of IκB kinase-β, which Similar results were obtained by incubating TEC with anti-β1 phosphorilates IκB, leading to its degradation and translocation mAb before the binding with thymocytes (59±4% reduction ± of NF-κB to the nucleus (Yin et al., 1998). We could therefore s.e.m.) (not shown). These results were reproducible on take advantage of this pharmacological property in our independent cultures derived from at least three different experimental system, which consists of normal cells and, donors and, taken together, indicate that α3β1 integrin because of this, is not easily amenable to complex crosslinking is able to specifically modulate cell survival in manipulations such as the delivery of NF-κB inhibitor- TEC upon growth factor deprivation. expressing constructs. Our findings indicate that NF-κB activation is necessary for the protection observed in TEC upon lymphoid adhesion, since the pharmacological blockage of NF- DISCUSSION κB nuclear translocation abolishes this protection. Thus, they are consistent with other reports indicating a key role for NF- In this study we have explored the possibility that heterotypic κB in promoting cell survival. cell adhesion could regulate the survival of thymic epithelium. Among the possible inducers of the NF-κB-mediated TEC Our experimental model consisted of epithelial cells derived protection during the lympho-epithelial adhesion, we from normal human thymus co-cultured with mature lymphoid considered integrins and among them we focused on the α3β1 T cells or normal thymocytes, and was aimed at partially heterodimer based on the following considerations: (i) it has mimicking in vitro the lympho-epithelial interactions occurring been lately become evident that integrin engagement by ECM inside the thymus. Our results indicated that adhesion to H9 components or cell ligands may initiate intracellular signaling cells or thymocytes significantly reduced the level of apoptosis cascades which promote cell survival (Frish and Francis, 1994; observed in TEC monolayers upon growth factor withdrawal. Koopman et al., 1994; Boudreau et al., 1995; Pullan et al., To investigate the downstream mechanisms of this protection, 1996; Wang et al., 1998); (ii) a recent report has further we analyzed the binding activity of NF-κB, which has been demonstrated a functional link between the integrin-mediated recently described to play a key role in apoptosis inhibition cell survival and the activation of NF-κB in endothelial cells (Beg and Baltimore, 1996; Wang et al., 1996; Van Antwerp et (Scatena et al., 1998); (iii) a previous study from our laboratory al., 1996; Scatena et al., 1998). We found that lympho- demonstrated that α3β1 integrin efficiently induces increased epithelial cell adhesion induced the activation of NF-κB, and NF-κB binding activity in TEC when cross-linked, repolarizes that the activity of this transcription factor was necessary for at the TEC-thymocyte contact sites during adhesion, and is promoting the adhesion-dependent TEC survival. In addition actively involved in their binding (Ramarli et al., 1998). we observed that α3β1 integrin, which clustered at the lympho- Moreover, we observed that α3β1 integrin re-distributes also at epithelial contact sites, could also modulate TEC survival when the TEC-H9 boundaries without any evidence of interposing cross-linked. known ECM ligands (unpublished observation). Herein, we Different lines of evidence have suggested an report that the crosslinking of this integrin can initiate a interdependence between thymic epithelial cells and signaling cascade leading to increase of TEC survival, thus developing T lymphocytes. This communication involves the reproducing the promoting activity of lymphoid or thymocyte activity of adhesion molecules, cytokines, and adhesion. This property appears to be specific, as we observed receptors (for a review, see Boyd et al., 1993). While the effects that the crosslinking of CD44 or ICAM-1 adhesion receptors of epithelial cell adhesion on thymocyte proliferation and failed to protect TEC from the growth factor-induced maturation are well documented, there is only scanty evidence apoptosis. In addition, the functional blocking of β1 integrins about the effects of thymocyte adhesion on thymic epithelial on TEC surface notably reduced the protective effect mediated cells. Results obtained from either naturally occurring or by T cell adhesion. However, a lower but considerable experimental models mainly pertain to effects of thymocyte reduction of protection was also observed by inhibiting ICAM- adhesion on TEC development or functions (Couture et al., 1 at the TEC surface. With regard to this, it could be supposed 1992; Cohen-Kaminski et al., 1993; Meilin et al., 1995; that the reduced adhesion potential induced by ICAM-1 Ramarli et al., 1996, 1998). So far, the ability of thymocyte blocking during TEC/thymocyte contact (Ramarli et al., 1998), contact to modulate the balance between TEC survival and might hamper the β1 recruitment, therefore impairing the death has never been described. Our results first highlighted a intracellular signal leading to TEC protection. role for T cell-adhesion in controlling the survival of thymic The cell ligand(s) recognized on H9 cells or thymocytes by epithelial cells. the α3β1 integrin expressed on the TEC surface is (are) still Recently, different groups have indicated a key role for NF- unknown. The finding that similar clustering is occurring when κB in protecting cells from apoptosis. Transgenic p65- TEC are interfaced with both developing thymocytes and NF-κB activation in adhesion-mediated TEC survival 177 mature lymphoid T cells may suggest that putative ligand(s) is Frish, S. M. and Francis, H. (1994). Disruption of epithelial cell-matrix (are) not restricted to the early state of lymphoid development. interactions induces apoptosis. J. Cell Biol. 4, 619-626. It has been described a α β homotypic interaction in Green, H., Kehinde, O. and Thomas, J. (1979). Growth of cultured human 3 1 epidermal cells into multiple epithelia suitable for grafting. Proc. Nat. Acad. keratinocytes (Symington et al., 1993). A similar interaction Sci. USA 76, 5665-5668. seems unlikely to occur between TEC and thymocytes in our Imbert, V., Rupec, R. A., Livolsi, A., Pahl, H. L., Traenckner, B. M., experimental setting, as thymocyte preparations we used Muelle-Dieckmann, C., Farahifar, D., Rossi, B., Auberger, P., Bauerle, P. and Peyron, J. F. (1996). Tyrosine phoshorylation of Iκ-B-α activates apparently did not expressed α3β1 integrin (not shown). α β NF-κB without proteolytic degradation of Iκ-Bα. Cell 86, 787-798. However, since thymocytes may express 3 1 heterodimers Koopman, G., Keehnen, R. M. J., Lindhout, E., Newman, W., Shimizu, Y., (Salomon et al., 1994), we cannot exclude that a homotypic van, Seventer, G. A., de, Groot, C. and Pals, S. T. (1994). Adhesion interaction could occur within the thymus between TEC and through the LFA-1 (CD11a/CD18)-ICAM-1 (CD54) and the VLA-4 α3β1 integrin-expressing thymocytes. (CD49d)-VCAM-1 (CD 106) pathways prevents apoptosis of germinal Although it is well established that survival of epithelial cells center B cells. J. Immunol. 152, 3760-3767. Meilin, A., Shoham, J., Schreiber, L. and Sharabi, Y. (1995). The role of mainly depends upon adhesion to ECM ligands, taken together thymocytes in regulating thymic epithelial cell growth and function. Scand. our results support the hypothesis that not only homotypic, but J. Immunol. 42, 185-190. also heterotypic cell adhesion can modulate the survival of Montgomery, A. M. P., Reisfeld, R. A. and Cheresh, D. (1994). Integrin α β epithelial cells. Within the thymus, T cell adhesion may v 3 rescues melanoma cells from apoptosis in three-dimensional dermal collagen. Proc. Nat. Acad. Sci. USA 91, 8856-8860. represent one of the mechanisms which finely tune the survival/ Nicoletti, I., Migliorati, G., Pagliacci, M. C., Grignani, F. and Riccardi, C. death balance of thymic epithelial cells. (1991). A rapid and simple method for measuring thymocyte apoptosis by propidium iodide stainig and flow cytometry. J. Immunol. Meth. 139, 271- We gratefully aknowledge the Cardiosurgery Division at Azenda 279. Ospedaliera Verona-University of Verona for providing thymic Palombella, V. J., Rando, O. J., Goldberg, A. L. and Maniatis, T. (1994). specimens. The financial support of Telethon-Italy (grant no. 1178) is The ubiquitin-proteasome pathway is required for processing the NF-κB gratefully acknowledged. This work was also supported by: Azienda precursor protein and the activation of NF-κB. Cell 78, 773-785. Ospedaliera Verona, research fund 1997; Regione Veneto, Ricerca Pullan, S., Wilson, J., Metcalfe, A., Edwards, G. M., Goberdhan, N., Tilly, J., Hickman, J. A., Dive, C. and Streuli, C. H. (1996). Requirementof Sanitaria Finalizzata no. 711/06/96; MURST 60% 1998. basement membrane for the suppression of programmed cell death in mammary epithelium. J. Cell Sci. 109, 631-642. Ramarli, D., Reina, S., Merola, M., Scupoli, M. T., Poffe, O., Riviera, A. P., Brentegani, M., Fiorini, E., Vella, A., Varnier, O. and Tridente, G. REFERENCES (1996). HTLV type IIIB infection of human thymic epithelial cells: viral expression correlates with the induction of NF-κB binding activity in cells Aoshiba, K., Rennard, S. I. and Spurzem, J. R. (1997). Cell-matrix and cell- activated by cell adhesion. AIDS Res. Hum. Retroviruses 12, 1217-1225. cell interactions can modulate apoptosis of bronchial epithelial cells. Am. J. Ramarli, D., Scupoli, M. T., Fiorini, E., Poffe, O., Brentegani, M., Villa, Physiol. 272, L28-L37. A., Cecchini, G., Tridente, G. and Marchisio, P. C. (1998). Thymocyte Baldwin, A. S. (1996). The NF-κB and IκB proteins: new discoveries and contact or mAb-mediated clustering of α3β1 or α6β4 integrins activate IL-6 insights. Annu. Rev. Immunol. 14, 649-681. transcription factors (NF-kB and NF-IL6) and IL-6 production in human Beg, A. A., Sha, W. C., Bronson, R. T., Ghosh, S. and Baltimore, D. (1995). thymic epithelial cells. Blood 92, 3745-3755. Embryonic lethality and liver degeneration in mice lacking the Rel A Salomon, D. R., Mojcik, C. F., Cang, W. C., Wadsworth, S., Adams, D. H., component of NF-kappa B. Nature 376, 167-170. Coligan, J. E. and Shevach, E. M. (1994). Constitutive activation of α4β1 Beg, A. A. and Baltimore, D. (1996). An essential role for NF-κB in defines a unique stage of human thymocytes development. J. Exp. Med. 179, preventing TNF-α-induced cell death. Science 274, 782-784. 1573-1584. Boudreau, N., Sympson, C. J., Werb, Z. and Bissel, M. J. (1995). Scatena, M., Almeida, M., Chaisson, M. L., Fausto, N., Nicosia, R. F. and Suppression of ICE and apoptosis in mammary epithelial cells by Giachelli, C. M. (1998). NF-κB mediates αvβ3 integrin-induced endothelial extracellular matrix. Science 267, 891-893. cell survival. J. Cell Biol. 141, 1083-1093. Boyd, R. L., Tucek, C. L., Godfrey, D. I., Izon, D. J., Wilson, T. J., Schluns, K. S., Cook, J. E. and Le, P. T. (1997). TGF-beta differentially Davidson, N. J., Bean, A. G. D., Ladyman, H. M., Ritter, M. A. and modulates epidermal growth factor-mediated increases in leukemia- Hugo, P. (1993). The thymic microenvironment. Immunol. Today 14, 445- inhibitory factor, IL-6, IL-1 alpha and IL-1 beta in human thymic epithelial 449. cells. J. Immunol. 158, 2704-2712. Brooks, P. C., Montgomery, A. M. P., Rosenfeld, M., Reisfeld, R. A., Hu, Schreiber, E., Matthias, P., Müller, M. M. and Shaffner, W. (1989). Rapid T., Klier, G. and Cheresh, D. A. (1994). Integrin αvβ3 antagonists promote detection of octamer binding proteins with ‘mini-extracts’, prepared from a tumor regression by inducing apoptosis of angiogenic blood vessels. Cell small number of cells. Nucl. Acids Res. 17, 6419. 79, 1157-1164. Symington, B. E., Takada, Y. and Carter, W. C. (1993). Potential role in Cohen-Kaminski, S., Devergne, O., Delattre, R. M., Klingel-Schmitt, I., keratinocyte intercellular adhesion. J. Cell Biol. 120, 523-535. Emilie, D., Galanaud, P. and Berrih-Aknin, S. (1993). Interleukin 6 Van Antwerp, D. J., Martin, S. J., Kafri, T., Green, D. R. and Verma, I. overproduction by cultured thymic epithelial cells from patients with M. (1996). Suppression of TNF-α-induced apoptosis by NF-κB. Science myastenia Gravis is potentially involved in thymic iperplasia. Eur. Cytokine 274, 787-789. Net. 4, 121-132. Wang, C. Y., Mayo, M. W. and Baldwin, A. S. (1996). TNF- and cancer Couture, C., Amarante-Mendes, G. and Potworowski, E. F. (1992). therapy-induced apoptosis: potentiation by inhibition of NF-κB. Science Tyrosine kinase activation in thymic epithelial cells: necessity of thymocyte 274, 784-787. contact through the gp23/45/90 adhesion complex. Eur. J. Immunol. 22, Wang, M. W., Consoli, U., Lane, C. M., Durett, A., Lauppe, M. J., 2579-2585. Champlin, R., Andreeff, M. and Deisseroth, A. B. (1998). Rescue from Crisa, L., Cirulli, V., Ellisman, M. H., Ishii, J. K., Elices, M. J. and apoptosis in early (CD34-selected) versus late (non-CD34-selected) human Salomon, D. R. (1996). Cell adhesion and migration are regulate at distinct hematopoietic cells by very late antigen 4- and vascular cell adhesion stages of thymic T cell development: the roles of fibronectin, VLA 4 and molecule (VCAM) 1-dependent adhesion to bone marrow stromal cells. Cell VLA 5. J. Exp. Med. 184, 215-228. Growth Differ. 9, 105-112. Darzynkiewicz, Z., Bruno, S., Del, Bino, G., Gorczyca, W., Hotz, M. A., Yin, M.-J., Yamamoto, Y. and Gaynor, R. B. (1998). The anti-inflammatory Lassota, P. and Traganos, F. (1992). Features of apoptotic cells measured agents aspirin and salicylate inhibit the activity of Iκ-B kinase-β. Nature by flow cytometry. Cytometry 13, 795-808. 396, 77-80.