Abstract
TNF is a proinflammatory cytokine with opposing death/no-death effects in vivo and in vitro. Our studies showed that TNF regulates mouse thymocyte production, inducing both apoptosis and proliferation of the most immature CD3−CD4−CD8− triple negative (TN) subset within a broad range of dosages (101–105 pg/ml) in the presence of IL-7. TNF apoptosis affected only the TN3 (CD44−CD25+) and TN4 (CD44−CD25−) subsets that expressed both TNFR-p55 and -p75. Although each TNFR alone could mediate TNF apoptosis, maximal apoptosis was seen in C57BL/6J wild type, which expressed both TNFRs. TNF also induced proliferation of TN3 cells at higher doses (104–105 pg/ml) mediated only by TNFR-p75. Both anti-TNFR-p55 and -TNFR-p75 mAb inhibited apoptosis but only anti-p75 inhibited proliferation. TNF also regulated TN proliferation to IL-7 because TNFR knockout (KO), TNF KO, and TNF/lymphotoxin α and β triple KO mice showed 2- to 3-fold increased responses not seen in C57BL/6J wild type. In vivo, TNFR KO mice showed thymic hypertrophy with a 60% increase in total thymocytes, with no effect on the CD4/CD8 subsets. We conclude that TNF maintains homeostatic control of total thymocyte production by negative selection of TN3 and TN4 prothymocytes and down-regulation of their proliferation to endogenous IL-7.
Tumor necrosis factor was originally defined by its ability to produce necrosis of tumors in vivo and kill some tumor cells in vitro without affecting normal cells (1). In the past decade, it became clear that TNF mediates a broad range of cytotoxic and noncytotoxic cellular responses in vivo and in vitro and plays a key role in host reactions to infection and inflammation (1, 2, 3, 4, 5, 6, 7). However, the physiological functions of TNF under steady-state conditions are still undefined (2, 3, 4, 5, 6, 7). We decided to test the possible physiological role of TNF in thymocyte development for five reasons: 1) the only site of constitutive TNF production is the normal thymus, shown in transgenic mice with a chloramphenicol acetyltransferase-TNF reporter (8); 2) cells with TNF message by in situ hybridization are found mainly in the subcortical regions and medulla (our unpublished work with A. Stopacciaro and E. Lattime), regions where thymocyte selection and apoptosis take place (9); 3) TNF is a modest costimulator of thymocyte proliferation in vitro and augments the effects of other stimulatory cytokines (10, 11, 12, 13, 14); 4) we find that the interactions of TNF with other cytokines in thymocyte proliferation are complex and show bidirectional stimulatory-inhibitory effects (seen with IL-1, IL-2, IL-4, and IL-6) depending on cytokine and costimulatory mitogen concentration (15); and 5) we have previously shown findings that TNF can induce apoptosis of a fraction of normal mouse thymocytes in vitro (15).
This study shows that TNF may indeed play a role in the control of thymocyte production in vivo. TNF regulates thymocyte production by acting primarily on some of the early T lymphocyte precursors that are negative or very dull for the expression of CD3, CD4, and CD8, CD3−CD4−CD8− (triple negative (TN)3 thymocytes). Because the 18- and 24-h assays needed to test apoptosis and proliferation of TN thymocytes showed high spontaneous cell death background, most of the studies described were performed in the presence of IL-7, which substantially reduced spontaneous TN thymocyte death (16). TNF showed a dual effect, inducing both apoptosis and proliferation of TN thymocytes. Apoptosis was seen in the two more mature TN thymocyte subsets defined by CD44/CD25 expression, CD44−CD25+ (TN3) and CD44−CD25− (TN4, Ref. 17); proliferation was only seen in the TN3 subset and required higher TNF doses. We found that TN thymocytes coexpressed both TNFRs. Based on the effects of anti-TNFR mAbs and TNF on TN thymocytes from TNFR-deficient (knockout, KO) mice, apoptosis is mediated by both TNFR-1/p55 and TNFR-2/p75, whereas proliferation is mediated solely by TNFR-p75. For apoptosis, the most effective TNF dose responses were seen in p55+/p75+ > p55+/p75− > p55−/p75+ TN cells, suggesting that the optimal apoptosis is obtained by the engagement of both TNFRs. An unexpected finding, showing the complex regulatory functions of TNF in the thymus, was that proliferation of TN cells to IL-7 alone was 2- to 3-fold higher in the TNFR KO and TNF KO than in the normal wild-type (WT) mice.
However, the finding that gives strongest support to a physiologic role of locally produced TNF in the thymus is the thymic hypertrophy seen in the TNFR KO mice. Thymocyte numbers were significantly increased in TNFR-p55, TNFR-p75, and TNFR-p55/p75 KO mice above those in sex- and age-matched WT controls, without major changes in the thymocyte subsets as defined by CD4 and/or CD8 expression. Our hypothesis is that absence of the TNF apoptosis-negative control on the TN3 and TN4 subsets, combined with the enhanced IL-7-mediated proliferation of TN3 (which is normally down-regulated by TNF), causes an increase in the daily input of TN4 prothymocytes that move into the CD4+CD8+ double positive (DP) pool and are further differentiated and selected into the single positive (SP) CD4-SP and CD8−SP, generating the overall thymic hypertrophy seen in TNFR KO mice.
Materials and Methods
Mice
C3H/HeJ (C3H) and C57BL/6J (B6) female mice (6- to 8-wk old) were purchased from The Jackson Laboratory (Bar Harbor, ME). C57BL/6bcl-2 transgenic (bcl-2 tg) mice (6- to 10-wk old) were a gift from Dr. H. T. Petrie (Memorial Sloan-Kettering Cancer Center, New York, NY). The bcl-2 tg mice were derived by mating heterozygous transgenics with (B6 × SJL/Wehi)F1 (18) and backcrossing into B6 (19). RAG-1 KO mice in a B6 background (20) were obtained from The Jackson Laboratory and were 4–5 wk old. All mice were maintained in our Core Animal Facilities as described (15). The p55 and p75 TNFR KO mice in a B6 background (21) were a gift from Dr. J. Peschon (Immunex, Seattle, WA), and breeding was initiated in our animal facilities. Young adult TNF (22), and TNFIII (TNF-α,lymphotoxin α and β (LT)) KO mice were a gift from Dr. M. W. Marino (Ludwig Institute at Memorial Sloan-Kettering Cancer Center) and Drs. S. Nedospasov and D. Kuprash (National Cancer Institute-Frederick Cancer Research and Development Center, Frederick, MD), respectively. The double TNFR-p55−/−/p75−/− KO and RAG-1/TNFR-p75−/− KO mice were produced in our laboratory by intercrossing p55 and p75 KO mice and RAG-1 and p75 KO mice, respectively. DNA from tail fragments of the potential founders was screened for TNFR-p55 and -p75 sequences using primer sequences for PCR provided by Dr. J. Peschon. The deficiency for RAG-1 was tested by phenotyping peripheral lymphocytes from the orbital sinus blood.
Cytokines, Abs, and reagents
Murine (Mu) and human (Hu) rTNF-α (10 μg/ml) and IL-7 (2 μg/ml) were purchased from PharMingen (San Diego, CA); stem cell factor (SCF; 10 μg/ml) was obtained from Genzyme (Boston, MA). The following mAbs were used: anti-mouse TNF (clone 2E2) produced in our laboratory (1523). Anti-mouse TNF (XT-22) was purchased from Endogen (Cambridge, MA). Unless otherwise stated, all other reagents were obtained from Sigma (St. Louis, MO).
TNF quantitation
TNF in supernatants was measured by ELISA. Supernatants were incubated with an anti-TNF capture mAb (G281-2626) coated on a 96-well plate overnight and detected by a different biotinylated anti-TNF (MP6-XT3) and streptavidin alkaline phosphatase.
Thymocyte cultures
Tissue culture medium (MED) optimized for thymocytes (15) was high glucose DMEM from our Media Preparation Facility supplemented with 10% FCS (HyClone, Logan, UT), 1% l-glutamine, 2% vitamins, 1% sodium pyruvate, 1% nonessential amino acids, 1% penicillin/streptomycin and ethidium bromide (Life Technologies, Grand Island, NY) and 5 × 105 M 2-ME (Fisher, Pittsburgh, PA). Thymi were removed aseptically and teased on wire mesh to prepare cell suspensions (15). Thymocytes and subsets were cultured with TNF (10 pg to 100 ng/ml) and/or IL-7 (5 μg/ml) at 4 × 105 cells/well in 96-well flat-bottom plates in a total volume of 200 μl for 18 h at 37°C and 5% CO2. IL-7 was used at 0.5–50 ng/ml, however, once the optimal dose was established, 0.5 ng/ml was used in most experiments. SCF (100 ng/ml) was also used alone or with TNF and IL-7. To measure proliferation as DNA replication, 5 × 105 TN cells from C3H, B6, p55 KO, P75 KO, TNF KO, and TNFIII KO mice were cultured in MED as above in 96-well plates for 24 h with TNF and/or IL-7 and with 0.8 μCi/well [3H]TdR ([methyl-3H]thymidine, 2 Ci/nmol; New England Nuclear, Boston, MA) during the last 6 h, the cells were harvested onto glass-fiber 96-well filters, and incorporation was determined by beta-counter (15). Anti-TNF mAbs were added at the beginning of the cultures as described (15). The anti-TNFR mAbs were tested by culturing the TN cells with the mAbs for 1 h at 37°C before adding TNF and/or IL-7.
Thymocyte subset separation, immunofluorescence staining, and sorting
Subset isolation was performed by incubating 1.5 × 109 thymocytes with rat anti-CD4 (H129.19) and anti-CD8 (53-6.7) for 30 min on ice, followed by a 30-min incubation with goat anti-rat IgG-coated magnetic beads (Perceptive Biosystems, Framingham, MA) at a bead-cell ratio of 10:1. The bead-coated thymocytes were removed with a Dynal Magnetic Particle Concentrator (Dynal, Lake Success, NY). The remaining cells were treated in a second incubation cycle with anti-CD4 (GK1.5), anti-CD8 (53-5.8), anti-CD45R/B22a, anti-CD11b/Mac-1, and anti-CD3, and then incubated with sheep anti-rat IgG-coated magnetic beads (Dynal) at 5:1 ratios (to obtain CD4−-CD8−-DN). These cells, negatively selected, were stained for sorting with PE anti-CD24, FITC anti-CD3, CyC anti-CD4, and CyC anti-CD8 to get the TN thymocyte subset or with PE anti-CD24, FITC anti-CD25, and CyC anti-CD44 to obtain the four TN thymocyte subsets CD44+CD25− (TN1), CD44+CD25+ (TN2), CD44−CD25+ (TN3), and CD44−CD25− (TN4). Thymocytes were also stained with FITC anti-CD4 and PE anti-CD8 to obtain DP, CD4+SP, CD8+SP, and CD4−CD8−. After sorting, the cells showed ∼100% purity for each subtype. The gates were set on the fluorescence obtained with the isotype-matched irrelevant Ig controls. For all FACS analysis, data was from 2 to 5 × 104 viable cells defined by forward and 90° side light scatter gating that excludes dead cells. A FACScan flow cytometer (Becton Dickinson, Mountain View, CA) and a FACStarPlus instrument (Becton Dickinson) equipped with argon ion lasers were used for analysis and sorting. Data analysis was performed with LYSIS II software (Becton Dickinson).
Apoptosis and cell cycle analysis
Spontaneous and TNF-induced thymocyte apoptosis were measured with hypotonic propidium iodide (PI) nuclear staining and FACS analysis (24), with some modifications (15). This method is based on the decreased DNA PI staining of apoptotic cells due to chromatinolysis (24) and has been validated with correlative DNA fragmentation by gel electrophoresis (25, 26). Apoptotic cells are hypodiploid, whereas nonapoptotic cells are diploid or hyperdiploid (24, 25, 26). Cell debris, doublets, and aggregates were removed with the Doublet Discriminating Module (Becton Dickinson) and gated before each analysis to ensure singlet evaluation (15, 25). For apoptosis, 2 × 104 events were recorded, data analyzed with the LYSYS II program, and expressed as percent apoptotic cells (15). Cells were harvested after 18-h cultures, washed once in PBS (PBS was obtained from our Core Media Preparation Facility), resuspended in 0.2 ml PI hypotonic solution (0.1% sodium citrate, 0.1% Triton X-100, 50 μg/ml PI), and placed at 4°C in the dark overnight before FACS analysis (15). Fresh cells and cells cultured with MED, TNF, and/or IL-7 for 18 h were studied. Specificity of the effect was tested with affinity-purified anti-MuTNF mAb (2E2) that totally inhibits TNF-induced thymocyte apoptosis (23 μg/ml; Ref. 15).
To test earlier apoptotic DNA events, the exogenous TdT method, which labels in situ DNA strand breaks with biotin (b)-dUTP (27), was used in a method that combines PI cell cycle analysis with TdT/b-dUTP apoptosis (26). Unseparated (UNS) and TN thymocytes were cultured in MED or TNF at 37°C for 3, 6, 12, and 18 h, washed in PBS, fixed in cold 70% ethanol (in PBS), rehydrated in PBS, resuspended in 50 μl cacodylate buffer (0.2 M sodium cacodylate; 2.5 mM Tris-HCl, pH 6.6; 2.5 mM CoCl2; 0.25 mg/ml BSA (Calbiochem-Novabiochem, La Jolla, CA)) with 5 U TdT and 0.5 nM b-dUTP, and incubated at 37°C for 30 min. Cells were resuspended in 100 μl of 0.52 M NaCl, 8 mM sodium citrate, 0.1% Triton X-100, 5% nonfat dry milk, and 2.5 mg/ml avidin-FITC; incubated at room temperature for 30 min in the dark; washed in PBS; resuspended in PBS-0.1% RNase. PI (5 μg/ml) was added to each sample 1 min before analysis of the respective controls were without TdT (26). Singlets were analyzed by FACS as described above. TdT, b-UTP, and a-FITC were obtained from Boehringer Mannheim (Indianapolis, IN).
Statistical analysis
Data were expressed as means ± SD of at least three independent experiments. Comparison between groups was made using Student’s t test. Differences with a probability (p) value <0.05 were considered significant.
Results
TNF induces apoptosis of CD4−CD8−CD3− TN thymocytes
We previously showed that TNF induces apoptosis of a fraction of UNS murine thymocytes in 18-h cultures (15). Table I⇓ shows the effect of TNF on UNS thymocytes and the four subsets defined by expression of CD3/CD4/CD8 (DP, CD4+SP, CD8+SP, and TN) from C3H, B6, and bcl-2 tg mice. Significant TNF-induced apoptosis was seen in the TN subset in the three strains at the TNF doses tested. Modest but significant TNF apoptosis of UNS thymocytes was seen in C3H and B6 (15) but not bcl-2 tg. TNF apoptosis was not detected in DP, CD4+SP, and CD8+SP from any of the mouse strains (corroborated by DNA electrophoresis performed as in Ref. 25 ; data not shown). Conversely, Table I⇓ shows that anti-Fas (28) induced significant apoptosis of UNS and DP thymocytes but had no effect on the TN subset in C3H and B6 mice (SP were not tested). The anti-TNF neutralizing mAb 2E2 (15) completely inhibited apoptosis of UNS and TN thymocytes in C3H and B6 mice (Table I⇓). Similar inhibition was seen with a rat anti-mouse TNF mAb XT-22 (data not shown). Because the TN thymocyte population is only a 1–2% minority (29) and the SP are not affected by TNF (Table I⇓), it is likely that TNF also induces apoptosis of some DP, which are the ∼85% majority (29). This would explain the apoptosis of UNS thymocytes seen in our previous study (15). However, we could not verify this because, after sorting, the DP showed high spontaneous apoptosis, and the TNF-induced apoptosis increase was nonsignificant (Table I⇓). TNF apoptosis of UNS thymocytes was regulated by bcl-2, whereas TNF apoptosis of TN thymocytes was independent of this proapoptotic factor (Table I⇓). Table I⇓ also shows that TNF induced significant apoptosis of thymocytes from RAG-1 KO mice, which are arrested at the CD44−CD25+TN stage (20), supporting the results with TN thymocytes from normal mice and the data with TN subsets to be discussed below.
TNF apoptosis of thymocyte subsets defined by CD4/CD8 expressiona
Dose response, time kinetics, time of exposure, and cell cycle effects on TNF apoptosis of TN thymocytes
Little is known about the physiological concentrations required for the in vivo biological effects of TNF. Dose-response curves for TNF apoptosis of UNS (15) and TN thymocytes were asymptotic. Apoptosis reached saturation levels at 10 ng and remained unchanged within a broad range of TNF concentrations up to 500 ng/ml. Unless stated, 10 pg to 10 ng/ml of TNF were used. Time kinetics of TNF apoptosis was comparable for C3H or B6 UNS and for B6 TN thymocytes. Significant TNF apoptosis (at 25 ng/ml) was seen both by PI staining or gel electrophoresis at 12 h, which reached plateau levels at 18 h but was not detected at 3, 6, and 8 h (data not shown). Apoptosis time kinetics was not changed by increasing the TNF dose (data not shown). As was seen with tumor cells (25), short pulses with TNF are sufficient to induce apoptosis. C3H TN thymocytes that were pulsed with TNF (25 ng/ml) for different time periods (30 min, 1, 2, 6, and 12 h), washed to remove TNF, and put back in culture for the remaining time (up to 18 h) showed that 1-h exposure to TNF was sufficient to induce apoptosis. In a representative experiment of three performed, apoptosis was 40.1% in MED controls, 70.0% in cells exposed to TNF for the whole 18 h, and 71.7% in cells pulsed with TNF for 1 h, washed, and cultured to complete the remaining 18 h. Similar results were obtained when the anti-TNF mAb 2E2 was added to the culture after the 1-h exposure to TNF. The TdT/dUTP method combined with PI staining was used to detect early apoptotic events and cell-cycle stage. Table II⇓ shows that TNF apoptosis was detected at 12 h but not earlier (supporting our time-course results) and that most of the apoptosis (both spontaneous or TNF induced) affects TN cells in the G1 phase of the cell cycle, as we described with growth-arrested tumor cells (25). We interpret the paradoxical decrease of apoptosis of cells in S/G2/M at 18 h (Table II⇓) as due to the increase of nonapoptotic cells in S phase produced by spontaneous or TNF-induced proliferation. In our studies with growth-arrested and synchronized tumor cells, we showed that cells in S phase are totally resistant to TNF apoptosis (25).
TNF induces apoptosis on TN thymocytes mainly in G1 of the cell cyclea
TNF induces apoptosis of CD44−CD25+ and CD44−CD25− TN subsets in the presence of IL-7
Because IL-7 is the cytokine of choice to maintain TN cells in vitro (16, 17, 30), we tested its effects on spontaneous and TNF-induced apoptosis. IL-7, at doses ranging from 0.5 to 50 ng/ml, significantly reduced spontaneous TN thymocyte apoptosis (Table III⇓). The addition of SCF alone or with IL-7 did not increase the inhibition of spontaneous (16, 30) or TNF-induced apoptosis (data not shown). Therefore, all of the subsequent studies with TNF on TN subsets were performed in the presence of a constant dose of IL-7 (0.5 ng/ml). Table III⇓ shows that, although IL-7 reduced spontaneous apoptosis, TNF still produced significant apoptosis of TN B6 thymocytes (51.1 vs 25.3% in the IL-7 controls). Both the time kinetics and TNF pulsing studies mentioned above showed comparable results when performed in the presence of IL-7 (data not shown).
TNF induced apoptosis of TN thymocyte subsets defined by CD44 and/or CD25 expression in the presence of IL-7a
To further analyze the TN thymocyte subset, TN cells were separated based on CD44 (Pgp-1) and CD25 (IL-2Rα) expression (17, 19, 30, 31, 32); the results are shown in Table III⇑. By three-color labeling and sorting, four distinct TN thymocyte subsets were obtained (17): TN1, TN2, TN3, and TN4. A maturational sequence from TN1 (less mature CD4low lymphoid progenitor) → TN2 (prothymocyte before TCRβ rearrangement) → TN3 (prothymocyte that starts to rearrange TCRβ) → TN4 (more mature subset that gives rise to the CD3+DP) is generally accepted (17, 30, 31). The CD44− subsets (TN3 and TN4) account for most of the TN cells, ∼56 and ∼30%, respectively (17, 19, 32). Table III⇑ shows that spontaneous apoptosis was different for each subset with TN3 and TN4 the highest (71 and 77%, respectively) and TN2 the lowest (27%). Table III⇑ also shows that IL-7 had no effect on TN1 and reduced spontaneous apoptosis of TN2, TN3, and TN4 subsets (6, 22, and 30%, respectively). For TNF-induced apoptosis, no effects above background were seen with TN1, but IL-7 allowed significant TNF apoptosis of the TN3 (49%) and TN4 subsets (55%) above the IL-7 controls. The effect of IL-7 on TNF apoptosis of the TN2 subset (which is ∼14% of the TN cells) was somewhat different; TNF increased apoptosis from 6 to 14% in the presence of IL-7, although TNF alone produced 74.5% vs a spontaneous apoptosis of 27.7% in the absence of IL-7. This could be due to the strong proliferative effect of IL-7 on this TN thymocyte subset (17). In summary, TNF produces apoptosis even in the presence of IL-7 but mainly of the more mature TN3 and TN4 subsets.
TNF also induces proliferation of B6 TN thymocytes in the presence of IL-7
Table IV⇓ shows that, in 24-h cultures, IL-7 (0.5 ng/ml) induced modest but significant thymidine uptake, which was further increased by TNF (10 ng/ml) in the whole B6 TN and the TN3 subset, but had no effect on TN4. Lower doses of MuTNF (100 pg/ml) and HuTNF (10 ng/ml) had no effect above the IL-7 controls. It is worth noting that, in the same experiments (in 18-h assays), apoptosis was detected at both the 10 and 100 pg/ml MuTNF dosages. Thus, MuTNF can induce both proliferation and apoptosis in TN3 thymocytes in the presence of IL-7, although proliferation requires higher TNF doses.
Proliferation of TN subsets in response to IL-7 and TNFa
TN thymocytes express both TNFR-1 (p55) and TNFR-2 (p75), and the effects of TNF on TN thymocytes are inhibited by anti-TNFR mAb
Fig. 1⇓ shows that both p55 and p75 TNFR are expressed on B6 TN thymocytes, based on staining with specific mAb and FACS analysis (23). TNFRs were not detected in UNS thymocytes from B6 mice. For the purpose of this paper, it is important to note that TNFR-p55 and -p75 are easily detected in the TN thymocyte population, which is the target of the TNF effects.
Expression of TNFRs on TN thymocytes. TN thymocytes were sorted as described in Materials and Methods and immunostained with hamster anti-mouse p55 or p75R and FITC-conjugated anti-hamster IgG. The negative control (CTR) was a matching isotype, hamster IgG. The cells were incubated with the primary Ab for 30 min in ice and for an additional 30 min with the secondary Ab-FITC. The staining was analyzed by FACS, cell counts vs p55R or p75R expression.
Fig. 2⇓ shows the effects of mAbs against p55 and p75 TNFR (23) on TNF-induced apoptosis and proliferation of B6 TN thymocytes in the presence of IL-7. Anti-p55 significantly inhibited TNF-induced apoptosis, although anti-p75 also produced a modest but significant reduction. Only anti-p75 significantly reduced TNF-induced proliferation. This last finding supports the lack of effect of HuTNF on TN thymocyte proliferation (Table IV⇑) because HuTNF does not bind to murine p75 (33).
Inhibition of WT TN thymocyte apoptosis and proliferation by anti-p55 or p75 mAb. A, TN thymocytes were tested for apoptosis by culturing them with 5 μg of anti-p55 or -p75 mAb for 1 h at 37°C before adding IL-7 (0.5 ng/ml) or IL7/TNF (10 pg/ml), and then cultured for an additional 18 h. In B, the cells were cultured in the same condition as above except that 10 ng/ml of TNF was added to induce proliferation as described in Table IV⇑. Values (% and cpm) are mean ± SD of five experiments.
TNF effects on TN thymocytes from TNFR KO mice
TNFR KO mice in a B6 background (21) were used to further study the role of the TNFRs on the negative (apoptosis) and positive (proliferation) effects of MuTNF on TN thymocytes. TN thymocytes from p55, p75, p55/p75 KO, and WT B6 controls were tested for TNF dose responses measured in the presence of a constant dose of IL-7 in 18-h apoptosis and 24-h proliferation assays (Fig. 3⇓). Fig. 3⇓A shows that apoptosis by MuTNF was seen in TN thymocytes from WT and both p55 and p75 KO mice above IL-7 controls. However, some differences were seen, depending on TNF dose. Apoptosis was demonstrated at all doses tested, from 10–105 pg/ml in B6 WT TN, at 103–105 pg/ml for p75 KO (which express p55), and only at the higher 104–105 doses for p55 KO (which express p75). TNF had no apoptotic effect on TN cells from p55/p75 KO at any of the above doses. In summary, “optimal” TN apoptosis by the lowest dose of MuTNF was seen in WT mice that express both TNFR, followed by p75 KO and p55 KO. The anti-TNFR mAbs used in Fig. 2⇑ were tested on the effects of MuTNF on TN cells from TNFR KO. In five experiments, apoptosis by MuTNF (10 ng/ml) was reduced from 37.2 ± 1.7 to 22.7 ± 1.6% by anti-p55 mAb in p75 KO (21.8 ± 1.0% in IL-7 controls) and from 29.1 ± 0.8 to 22.0 ± 0.8% by anti-p75 mAb (22.3 ± 1.1% of the IL-7 controls) in p55 KO (mAb against the missing TNFR had no inhibitory effect; data not shown). HuTNF was studied to further test the role of p55 in apoptosis of TN thymocytes. Fig. 3⇓B shows that HuTNF induced apoptosis of TN cells from WT and p75 KO but had no effect on p55 and p55/p75 KO cells. At the doses tested, HuTNF, which binds only to p55 (33), was less effective than MuTNF as an apoptosis inducer (compare Fig. 3⇓, A and B).
TNF induced apoptosis and proliferation of p55−/−, p75−/−, and p55−/−p75−/− (KO) TN thymocytes. TN thymocytes from different TNFR KO mice were cultured with IL-7 (0.5 ng/ml) and with the indicated doses of mouse TNF (A and C) or HuTNF (B and D) for 18 h to study apoptosis (A and B) and proliferation (C and D). Apoptosis and proliferation were analyzed as in Tables I⇑ and IV⇑, respectively. The data were calculated as mean ± SD of 10 experiments. A, ∗, Significant differences between IL-7- and IL-7/TNF-treated cells in p55−/− (p < 0.05) and in p75−/− (p < 0.01) and ‡, Significant in relation to IL-7/TNF-induced apoptosis in WT (p < 0.001, p < 0.05 for p75−/− and p55−/−, respectively). B, ∗, Significant differences in comparison to WT (p < 0.01 for p55−/− and p < 0.05 for p75−/−. C, ∗, p < 0.001 in all three KO when compared with WT and also ‡, Significant difference between IL-7- and IL-7/TNF-induced proliferation in WT (p < 0.01) and in p55−/− (p < 0.05).
Fig. 3⇑, C and D, shows the proliferation results in 24-h assays with MuTNF and HuTNF in the presence of IL-7. Unexpectedly, IL-7 alone produced a 2- to 3-fold increase in thymidine uptake in TN cells from TNFR KO mice when compared with WT. The actual thymidine uptake values produced by 0.5 ng/ml of IL-7 in Fig. 3⇑C were (×103 cpm): 6.3 ± 1.2 in WT, 19.9 ± 1.2 in p75 KO, 29.2 ± 3.2 in p55 KO, and 30.3 ± 2.0 in p55/p75 KO mice. However, no differences between WT and TNFR KO mice were observable when the TN cells were stained for the expression of IL-7 receptor CD127 or bcl-2 (data not shown; five experiments). We also tested TNF production by TN thymocytes from WT B6 and TNFR KO, cultured for 18 h with IL-7, by using ELISA. We found that TNF was released in the supernatants, between 10 and 20 pg/ml from the WT mice, and undetectable amounts from the TNFR KO (five experiments). MuTNF increased proliferation above IL-7 alone in WT (also shown in Table IV⇑) and p55 KO was 11.2 × 103 and 40 ± 5.1 × 103, respectively, but only at the highest concentration tested (104 pg/ml). Interestingly, TNF at the highest dose actually decreased the IL-7 proliferative response in p75 KO, probably because the deficiency of this receptor abolishes the TNF proliferative effect but not apoptosis. Fig. 3⇑D shows that, as expected, HuTNF had no proliferative effects above IL-7 controls. These findings combined indicate that p75 plays a key role in proliferation induced by MuTNF in the presence of IL-7, not shared by p55, supporting our findings in WT cells with the anti-TNFR mAbs (Fig. 2⇑).
TNF effects on TN thymocytes from TNF and TNFIII KO mice
Fig. 4⇓ shows studies on MuTNF apoptosis and proliferation of TN thymocytes from TNF and TNFIII KO mice. Fig. 4⇓A shows the apoptosis data compared with WT. At the same TNF dose range as in Fig. 3⇑, apoptosis was seen at 102–105 pg/ml in TNF KO and at 103–105 in TNFIII KO. Although significantly different from the IL-7 controls, TNF apoptosis in TNF KO and TNFIII KO was significantly lower than in WT. This could suggest a role for endogenous TNF production in the WT cultures (shown above) that supplements the exogenous TNF and/or a role for endogenous TNF in the regulation of TNFR expression lacking in the TNF KO mice. However, the expression of TNFR on TN cells was similar between the WT and the TNF KO (data not shown from five experiments).
TNF induces apoptosis but not proliferation of TNF−/− and TNFIII−/− TN thymocytes. The cells were tested for apoptosis (A) and proliferation (B) with mouse TNF as in Tables I⇑ and IV⇑, respectively. Data are expressed as the average (mean ± SD) of three experiments. A, ∗, Apoptosis was significantly lower in TNF−/− and TNFIII−/− TN thymocytes when compared with WT TN thymocytes, p < 0.05 and p < 0.01, respectively. B, TNF does not induce proliferation in TNF−/− or TNFIII−/− TN. ∗, Significant difference in relation to IL-7 WT (p < 0.001 and p < 0.01 for TNF−/− and TNFIII−/−, respectively).
TN thymocyte proliferation induced by TNF is shown in Fig. 4⇑B. Both TNF KO and TNFIII KO mice show the same increased proliferation to IL-7 alone seen in the TNFR KO (Fig. 3⇑C). Although WT had 6.3 ± 1.1 × 103 cpm, TNF KO cells had 16.9 ± 1.1 × 103 cpm and TNFIII KO had 11.9 ± 1.9 × 103 cpm, which, although lower than in TNFR KO, are significantly higher than normal WT. However, the TN cells from neither TNF KO nor TNFIII KO showed significant MuTNF-induced proliferation above the IL-7 controls. Because TNF-induced proliferation is a “higher” dose response (Table IV⇑ and Fig. 3⇑C) it is possible that endogenous TNF, absent in these KO mice, may play a role in triggering TN thymocyte proliferation. Similarly, because membrane-associated TNF is a strong stimulator of both TNFRs (34), it is possible that the lack of endogenous membrane-associated TNF in these mice may also explain the lack of a proliferative response to TNF above IL-7 controls. IL-7 receptors on TN thymocytes were expressed in TNF KO at the same level as in WT (data not shown; five experiments).
Increased numbers of thymocytes in the thymus of TNFR KO mice
An unexpected finding was the thymic hypertrophy seen in the TNFR KO mice. Table V⇓ shows that the total number of thymocytes was ∼60% increased in the thymi of 6- to 7-wk-old female p55-KO, p75 KO, and p55/p75 KO mice when compared with sex- and age- matched WT B6 controls. However, the percentage of subsets defined by CD4/CD8 expression was comparable to WT (data not shown). This finding strongly points to the biological significance of the TNF effects observed in vitro on TN thymocytes as a key control mechanism of thymic homeostasis. Our studies, mainly with p75 KO female mice (which are better breeders and of which we had a larger stock) showed some decrease of thymocyte numbers with age. However, in every case, the values were higher than for the age- and sex- matched WT controls. For example, at 6 mo, p75 KO had 143.5 ± 14.0 × 106 thymocytes (with 22.4 ± 0.9 body weights), whereas WT B6 had 93.6 ± 20 × 106 thymocytes (with body weights of 22.5 ± 2.6). In an in vivo study, we obtained RAG-1/p75 double KO by breeding both KOs to test whether the deficiency of the p75R would affect the number of immature thymocytes. Interestingly, we found that the RAG-1/TNFR-p75 double KO mice had a 7-fold increase in TN thymocytes per thymus, from 4.0 × 106 in RAG-1 to 29.5 × 106 in RAG/p75 KO.
TNF-R KO mice have increased number of total thymocytesa
A more limited study of the thymi of TNF KO and TNFIII KO showed the same trend. However, the data were not included in Table V⇑ because 1) the number of mice available at given age points was limited; and 2) ∼25% of both KO mice have an atrophic thymus that could be secondary to the problem of these mice in dealing with some infections (22). The fact that the thymus of the TNFIII KO is not substantially different from that of the TN KO suggests that LTα, which uses the same p55 and p75 TNFR as TNF (35), is not replacing the absence of TNF in these KO mice.
Discussion
These studies expand our early finding that TNF induced apoptosis of a fraction of normal thymocytes in mice (15). Here we show that the main target for TNF-induced apoptosis in the presence of IL-7 is the TN thymocyte subset in WT (C3H, B6), B6 bcl-2 tg, B6 RAG-1 KO, B6 TNFR-p55 KO, B6 TNFR-p75 KO, B6 TNF KO, and TNFIII KO mice. Within the TN thymocytes the more mature TN3 and TN4 subsets are the main TNF targets. We also found that in addition to apoptosis, TNF at higher doses can induce proliferation of the TN3 subset but not of TN4 in the presence of IL-7. IL-7 was used to maintain viable immature thymocytes in culture (16), and its interaction with TNF might be of importance because IL-7 is a nonredundant cytokine for T cells in the thymus (36). IL-7 also appears to have a role in periphery (37) and in thymic and extrathymic development of TCRγδ T cells (38). TNF is a strong stimulator of peripheral γδ T cells via p75 (39).
Both TNF-induced apoptosis and proliferation were inhibited by anti-TNF mAbs and were not inhibited in cells from TNFR-p55/p75 double KO mice, showing that the effects are TNF specific and direct where apoptosis was inhibited by both anti-p55 and -p75 mAb, whereas only anti-p75 inhibited proliferation (Fig. 2⇑). Therefore, the ambiguous behavior of TNF, providing both death and no-death signals which have been studied mostly with established tumor cell lines in vitro (1, 2, 3, 4, 5, 6, 7, 40), is also seen with fresh normal mouse thymocytes in short-term cultures (18 and 24 h). Mutated HuTNF molecules that bind to either TNFR-p55 or -p75 (41, 42, 43) and TNFR KO mice have been extensively used to determine the predominant role of each or both receptors in a variety of in vivo and in vitro models (21, 44, 45, 46, 47, 48, 49, 50, 51, 52). One of the interpretations of the TNF death/no-death paradox is that the two TNFRs can mediate different functions (33, 53). However, both p55 and p75 can mediate apoptosis of cell lines, whereas only p55 has a cytoplasmic death domain (54). The models that test apoptosis by p75 alone usually require induced overexpression of this receptor on cell lines (41, 55). Our studies show that although there are clear and significant differences in magnitude of the response, both p55 and p75 alone can mediate TNF apoptosis of TN thymocytes (Fig. 3⇑A). However, optimal (or maximal) apoptosis at low TNF dosages is seen when both TNFRs are expressed, as in normal B6 WT mice. This last point supports the in vitro evidence (studied mostly with cell lines and TNFR overexpression) that mutual interactions between the two TNFRs are important for optimal TNF responses. Such interactions include “ligand passing” by p75 to p55 (56), cytoplasmic domain interactions of the TNFR-associated factors involved in the signaling cascade (34, 42, 57, 58, 59, 60), and/or induction of endogenous production of TNF by p75 (61). Fig. 3⇑, A and C, shows that whereas apoptosis is a “low dose” TNF response, proliferation is a “higher dose” TNF response. The unusually high proliferative responses to IL-7 alone seen in the TNFR KO (and TNF KO and TNFIII KO) also show that both receptors or the lack of TNF may be involved in the proliferative response to IL-7 (Figs. 3⇑C and 4B), which is not due to an increase in IL-7 receptors or bcl-2 expression. It has been shown that the IL-7 trophic effect on TN subsets correlates with high levels of bcl-2 (62). Interestingly, the incubation of WT TN thymocytes for 18 h with IL-7 induces release of picogram levels of TNF in the supernatants (with nondetectable levels in the TNFR KO), suggesting a potential intrathymic autocrine regulatory circuit, with TNF and IL-7 as the interacting partners. This effect brings to mind the bidirectional regulation that TNF exerts on the costimulation of UNS mouse thymocytes by other cytokines (15). A similar bidirectional modulatory role of TNF was seen with murine hemopoietic stem cells in vitro, where TNF counteracted the effects of SCF, IL-6, and IL-11 and acted synergistically with IL-1 (63) and a human erythroleukemia cell line in vitro (64). In contrast, p75 seems necessary and sufficient for TNF-induced proliferation of TN3 thymocytes in the presence of IL-7 (Fig. 3⇑C). A similar dominant role of p75 was shown in the modest costimulation of UNS thymocytes with TNF and mitogens (33, 53, 65). Our apoptosis data uphold the view that TNFR interactions are critical even under physiological conditions and are not the consequence of transfected overexpression of TNFRs, as may be the case in cell lines (33, 42, 43, 57, 58, 59, 60). Therefore, normal TN thymocytes could be an ideal tool to study the biochemical events that participate in TNFR activation by its ligand and the consequent cross-talk between the two TNFRs.
Furthermore, the intrathymic regulatory role of TNF is strongly supported by the findings described in Table V⇑, showing that the lack of TNF signaling in vivo, as in the TNFR KO mice, is accompanied by a significant thymic hypertrophy expressed as an overall increase in total thymocytes per thymus. The fact that this hypertrophy shows a normal distribution of thymocyte subsets defined by CD4 and/or CD8 expression may explain why it was overlooked in original descriptions of the p55 KO (46) and the p75 KO mice (21, 47), which assumed a normal thymus because the CD4/CD8 subsets were normal. The removal of a fraction of the TN3 and TN4 subsets by TNF-mediated apoptosis acts as a homeostatic control of overall thymocyte production by keeping steady-state numbers of TN prothymocytes into the thymocyte differentiation pathway. Because the thymocyte hyperproduction in the absence of one of the two TNFRs or both, is across the board rather than selective, it is probable that the negative selection of TN3 and TN4 cells is random and not related to actual repertoire selection when CD3, CD4, CD8, and TCR are expressed (66, 67). Because TNF also induces proliferation of TN3 at higher dosages, it is possible that local concentration gradients of TNF could guide the TN thymocyte maturational development in a dual manner. At lower doses, TNF removes by apoptosis a fraction of TN3 and TN4 cells and, at higher doses, rescues by proliferation a fraction of TN3 for further maturation. This is supported by our findings with tumor cell lines showing that cells become totally refractory to TNF apoptosis when they enter the S phase (Ref 25 . Table II⇑). In the absence of TNF-mediated dual control on apoptosis and proliferation, higher numbers of TN3/TN4 prothymocytes are produced; this results in an overall increase of thymocytes. The study in vivo of the p75−/−/RAG-1 double KO showing a 7-fold increase of TN thymocytes indicates that TNF certainly controls immature thymocytes, which accumulate but cannot differentiate any further because they cannot rearrange the TCR β-chain. This is in accordance with the finding that progression in the cell cycle of TN thymocyte is independent of the TCR-β gene recombination (68) and supports a role of TNF in homeostatic control of thymocyte production.
Whether TNF can affect other thymocytes than the early TN cells is still an open matter. The studies of negative-repertoire selection of mature thymocytes in TNFR KO mice have shown a TNF effect in some models (such as anti-CD3 in in vivo treatment) but not in other in vivo models of Ag negative-selection (69). Similarly, the effects of TNF on peripheral T cells show variable responses. Fresh UNS T cells from spleen or lymph nodes are not killed by TNF (15); however, TNF kills T lymphoblasts induced in vitro (70). TNF selectively kills resting CD8 T cells via p75 engagement in normal lymph nodes (71), and in Fas-deficient lpr mice the lymphadenopathy is greatly accelerated in p55 KO mice (72).
In summary, we propose that constitutively produced TNF in the thymus plays a key role in maintaining homeostasis of thymocyte production by acting on some of the TN prothymocyte subsets before the actual CD3/TCR/CD4/CD8 selection takes place. Such a role of TNF involves both negative and positive signals on TN cells and regulatory interactions with IL-7 responses.
Acknowledgments
We thank Dr. Jacques Peschon (Immunex) for providing the initial breeding pairs of TNFR KO, Dr. Michael Marino (The Ludwig Institute) for providing the TNF KO and for his help in phenotyping the TNFR double KO, Dr. S. Nedospasov (National Cancer Institute-Frederick Cancer Research and Development Center) for the TNFIII KO, Sylvaine Frances and Katherine Jagiello for their technical support, and Diane Domingo for help with the flow cytometry analysis.
Footnotes
-
↵1 This work was supported in part by grant R01 CA55068 (to O.S.), Cancer Center Support Grant P30 CA08748, and Immunology Training Grant T32 CA09149 from the National Institutes of Health.
-
↵2 Address correspondence and reprint requests to Dr. Juana Gonzalez, Department of Pathology (F-726), Albert Einstein College of Medicine, Yeshiva University, 1300 Morris Park Avenue, Bronx, NY, 10461. E-mail address: jmgonzal{at}aecom.yu.edu
-
↵3 Abbreviations used in this paper: TN, triple negative; KO, knockout; WT, wild type; DP, double positive; SP, single positive; C3H, C3H/HeJ; B6, C57BL/6J; bcl-2 tg, C57BL/6bcl-2 transgenic; LT, lymphotoxin α and β; SCF, stem cell factor; PI, propidium iodide; UNS, unseparated; TNFIII, TNF-α,lymphotoxin α and β; MED, tissue culture medium; MuTNF, murine TNF; HuTNF, human TNF; CyC, Cy-Chrome.
- Received March 13, 2000.
- Accepted August 28, 2000.
- Copyright © 2000 by The American Association of Immunologists