IMMUNOREGULATORY ROLE OF CD8α IN THE VETO EFFECT: Transforming Growth Factor-b1 Activation: 1 : Transplantation

Journal Logo

Experimental Transplantation


Transforming Growth Factor-b1 Activation1

Asiedu, Clement2; Meng, Yuru2; Wang, Weila2; Huang, Zhi2; Contreras, Juan L.2; George, James F.3; Thomas, Judith M.2,4

Author Information
  • Free


*Abbreviations: allo-BMC, allogeneic bone marrow cell; BMC, bone marrow cell; CML, cell-mediated lymphocytotoxicity; CTL, cytotoxic T lymphocyte; CTLp, cytotoxic T lymphocyte precursor; DMSO, dimethyl sulfoxide; ELISA, enzyme-linked immunosorbent assay; FasL, Fas ligand; FBS, fetal bovine serum; FITC, fluorescein isothiocyanate; IL, interleukin; JNK, Jun N-terminal kinase; mAb, monoclonal antibody; MHC, major histocompatibility complex; MLR, mixed lymphocyte reaction; pAb, polyclonal antibody; PBMC, peripheral blood mononuclear cell; PCR, polymerase chain reaction; PHA, phytohemagglutinin; PMA, phorbol myristate acetate; RT, reverse transcription; rTGF-β1, human recombinant transforming growth factor β1; SAPK, stress-activated protein kinase; TGF-β1, transforming growth factor β1; TUNEL, TdT-mediated end-labeling; XL, cross-linking or cross-linked.

Infusion of allogeneic donor bone marrow cells (allo-BMC*) induces immunologic tolerance to major histocompatibility complex (MHC)-incompatible renal allografts in rhesus macaques (1-3). The tolerogenic effect is ascribed to a veto mechanism (2,3) that involves functional deletion of recipient allo-specific cytotoxic T lymphocyte precursor (CTLp) and T helper cells by the specific donor's allo-BMC (1,4). The immunobiology of the veto phenomenon has been extensively studied in murine models, although details of the molecular and cellular mechanisms remain uncertain (5-7). In the primate model, the veto effect is dependent on a minor allo-BMC subset, putatively natural killer or dendritic cell precursors, with a CD2+CD3-CD8+CD16+DR- phenotype (2). Studies of the functional characteristics of these cells suggest that the CD8 molecule may play an important role in the veto effect (8-12). Blocking the interaction of CD8 and the MHC class I α3 domain with monoclonal antibodies (mAbs) resulted in the loss of CTLp suppression by the allo-BMC (8). CD8 expression on the donor antigen-bearing BMC may lead to responder cell inactivation by two nonexclusive pathways. The interaction of CD8α with the MHC class Iα3 domain on responder CTLp may transmit a signal via the class I molecule that results in functional deletion of the CTLp (10,13). Alternatively, the CD8α/MHC class Iα3 domain interaction may initiate signaling via the CD8 molecule on the allo-BMC, inducing a factor that facilitates deletion of the CTLp. Available evidence supports the latter possibility and suggests transforming growth factor-β1 (TGF-β1) as a candidate for such a factor. Active TGF-β1 was detected in supernatants of conventional mixed lymphocyte reaction (MLR) cultures supplemented with allo-BMC, and the addition of TGF-β1 neutralizing antibodies abrogated the veto effect of BMC (8). Most relevant to the present studies, is the suggestion of a functional linkage between CD8 and TGF-β1 by experiments showing that the perturbation of the CD8 surface molecule of BMC by cross-linking (XL) elicited secretion of active TGF-β1 (8).

These observations led us to examine the hypotheses that (i) purified CD8+ allo-BMC mediate veto activity, (ii) CD8 XL induces accumulation of TGF-β1 mRNA, and (iii) active TGF-β1 facilitates deletion of activated T cells through apoptosis. We show that perturbation of the CD8 surface glycoprotein on normal rhesus BMC and peripheral blood mononuclear cells (PBMC) induces increased steady state levels of TGF-β1 mRNA and protein; bulk CD8+ BMC, as well as CD3-CD16+ and CD3+CD16- subsets of CD8+ BMC, upregulate TGF-β1 mRNA in response to CD8 XL and exhibit veto activity in vitro; and finally, TGF-β1 induces apoptosis of phytohemagglutinin (PHA)-activated rhesus macaque T cells in vitro. These data suggest that TGF-β1 activation via CD8 is a key player in the veto effect, contributing to the deletion of responding T cells via apoptosis.


mAbs and other reagents. Anti-CD8 mAbs were (IgG2a) OKT8F, a generous gift of Dr. Robert Knowles (Ortho Pharmaceutical Research Institute, Raritan, NJ). IgG1 anti-human CD8 mAb was purchased from PharMingen (San Diego, CA). IgG1 anti-CD16 mAbs and rat anti-mouse IgG2- and IgG1-conjugated magnetic beads were obtained from Immunotech (Westbrook, ME) and Miltenyi Biotech (Sunnyvale, CA), respectively. Each of the anti-human mAbs used was strongly cross-reactive with rhesus macaque cells. Human recombinant TGF-β1 and the TGF-β1 Emax™ immunoassay system were purchased from Promega (Madison, WI). Rabbit anti-human TGF-β1 polyclonal antibody (pAb) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Goat anti-rabbit IgG-PE, actinomycin D, and dimethyl sulfoxide (DMSO) were obtained from Sigma Chemical (St Louis, MO). Ficoll-Paque was from Pharmacia (Piscataway, NJ).

Preparation of BMC and culture conditions. Normal rhesus BMC were isolated from ribs as previously described (8) and by aseptic aspiration from the humerus. The aspirate was diluted 3-fold, and BMC were purified by Ficoll-Paque density gradient centrifugation. CD8+ cells were purified by immunoaffinity using biotinylated anti-CD8 mAb (PharMingen) and streptavidin Cellpro columns (Cellpro Incorp, Bothell, WA) according to the manufacturer's instructions. CD8+ cells were routinely >85% pure as determined by flow cytometry using a FACSCalibur (Becton Dickinson, Mountain View, CA). Where indicated, purified CD8+ cells were labeled with anti-CD3 or anti-CD16 mAb and sorted into CD3-CD16+ and CD3+CD16- subpopulations using a Becton Dickinson FACStarPLUS sorter.

XL CD8 and CD16 on BMC. BMC (∼3×105 to 4×10 (6) cells) were initially incubated for 20 min on ice with saturating amounts of anti-CD8 and/or anti-CD16 mAb, or with an isotype-matched-negative control mAb. After three washes, XL of the surface molecules was achieved by adding the appropriate rat anti-mouse IgG isotype-conjugated beads. The cells were then cultured in a 24-well plate in 1 ml of RPMI 1640 containing 10% fetal bovine serum (FBS) or autologous serum for the indicated times in a humidified CO2 incubator. Where indicated, actinomycin D was dissolved in DMSO at 10 mg/ml and diluted in medium to a final concentration of 10 µg/ml.

Semiquantitative reverse transcription-polymerase chain reaction (RT-PCR). Total RNA was isolated from cells (1×105 to 2×106 using Trizol reagent (GIBCO, Gaithersburg, MD) according to the manufacturer's instructions. Total RNA (0.3-1 µg) was reverse transcribed using MLV-RT (Promega) or Superscript II (GIBCO) according to the manufacturer's instructions. Then 2 µl of the cDNA was amplified by PCR in a total volume of 100 µl, containing 1× PCR buffer, 0.2 mM dNTP, 0.4 µM each of sense and antisense primers, and 2.5 U of Taq polymerase (GIBCO). The sequences (5′ to 3′) of the β-actin and TGF-β1 specific primers used are as follows: β-actin sense: 5′-GTGGGGCGCCCCAGGCACCA-3′; β-actin antisense: 5′-CTCCTTAATGTCACGCACGATTTC-3′ (550-base pair product); TGF-β1 sense: 5′-GCCCTGGACACCAACTATTGCT-3′; TGF-β1 antisense: 5′-AGGCTCCAAATGTAGGGGCAGG-3′ (150-base pair product).

Amplifications were performed in the linear range (25-32 cycles) in a DNA thermal cycler (Perkin-Elmer, Norwalk, CT) using the following conditions: β-actin: 94°C, 3 min, 94°C, 45 sec, 57°C, 1 min, 72°C, 1.15 min, 72°C, 10 min; TGF-β1: 95°C, 3 min, 95°C, 1 min, 55°C, 1 min, 72°C, 1 min, 72°C, 10 min. Equivalent aliqouts (20 µl) of PCR reactions were analyzed in ethidium bromide-stained agarose gels. The gels were scanned by densitometry and quantitated by using NIH image analysis software. The relative TGF-β1 mRNA level was estimated as arbitrary units obtained by normalizing differences in cDNA content with the corrected actin band intensity (RT actin band intensity minus the no RT actin band intensity). The arbitrary units cannot be compared from one experiment to another because of differences in cell numbers used for total RNA isolation in different experiments.

TGF-β1 enzyme-linked immunosorbent assay (ELISA). CD8-enriched cells were XL as above and incubated in complete medium for 48 hr. Supernatants were harvested for ELISA and stored at -20°C before use. Coating of 96-well plates with the TGF-β1 mAb and all subsequent steps of the assay were done according to the manufacturer's instructions. Samples were diluted and latent TGF-β1 was heat-activated for 10 min at 80°C (14). Plates were read at 450 nm wavelength in a microplate reader (BioRad, Richmond, CA).

Intracellular staining for TGF-β1. Normal rhesus BMC were isolated by aseptic aspiration from the humerus followed by density gradient centrifugation on Ficoll-Paque as described. BMC were XL as described above. Cells were incubated in Click's medium containing 10% AIM-V medium (GIBCO), 2% normal monkey serum, and 25 U/ml recombinant human interleukin-2 (IL-2). Cells were harvested at the indicated times and stained for intracellular TGF-β1. Briefly, cells were fixed and permeabilized using a Cytostain kit (PharMingen) according to the manufacturer's instructions. Fixed and permeabilized cells (1×106) were incubated with 15 µl of undiluted antihuman TGF-β1 pAb or an equivalent amount of normal rabbit serum. The cells were washed and then stained with 4 µl of undiluted goat anti-rabbit IgG-PE (Sigma). After the final wash, cells were analyzed by flow cytometry (Coulter Epics Elite, Coulter, Miami, FL), gating on the CD8+ subset. One parameter histograms obtained were subjected to immuno-4 processing (Coulter Epics Elite software) to detect changes in intracellular TGF-β1 levels.

Detection of DNA fragmentation. PBMC were isolated from defibrinated rhesus blood as described above. PHA-blasts were prepared by culturing cells at 2-3×106/ml in complete RPMI 1640 medium supplemented with 5 µg/ml PHA (Murex Biotech, Dartford, England). After 2 days, 1×106 cells were incubated in a volume of 1 ml with or without 10 ng/ml TGF-β1. In a second set of experiments, PHA-activated cells were added to COS cells that were transduced with recombinant adenovirus expressing constitutively active porcine TGF-β1 or luciferase, an irrelevant protein (generously provided by Dr. David Curiel, University of Alabama at Birmingham). Transduction with adenovirus was performed at a multiplicity of infection of 10. Neutralizing chicken anti-TGF-β1 pAb (R&D Systems, Minneapolis, MN) was included in some experiments to a final concentration of 30 µg/ml. Apoptosis was detected by a TdT-mediated end-labeling (TUNEL) assay (in situ cell death kit; Boehringer Mannheim Biochemical, Indianapolis, IN) or annexin V-fluorescein isothiocyanate (FITC) staining (Genzyme, Cambridge, MA) according to the manufacturer's instructions. Green fluorescence (FL1-dUTP fluorescein) was detected and quantified by flow cytometry. The specific percentage apoptosis was calculated as described by Genestier et al (15).

In vitro veto cell assay. Allogeneic MLR-induced cytotoxic T lymphocyte (CTL) veto assays were performed using BMC from the stimulator cell donor as previously described (2). Briefly, triplicate one-way MLR cultures were initiated using 2×105 responder cells and 7×104 irradiated stimulator cells (2000 cGy) in Click's medium supplemented with antibiotics, 2% fresh autologous serum, and 25 U/ml recombinant human IL-2 (Genzyme, Cambridge, MA). BMC (1.5×105, 3×104, or 6×103 for veto cell to responder cell ratio of 0.75, 0.15, and 0.03, respectively) were added on day +1. Target cells, 3-day PHA lymphoblasts labeled with 51Cr, were added to washed effectors on day 6. Data are presented as median percentage specific lysis or median percent inhibition of control lysis of target cells obtained without addition of BMC.

Statistical analysis. The significance of median differences in the apoptosis study and the median differences in cell-mediated lymphocytotoxicity (CML) assays was examined by Mann-Whitney testing using Statgraphics software (Manugistics, Inc., Rockville, MD).


CD8 and CD16 XL induce accumulation of TGF-β1 mRNA. Our previous studies (8) showed that CD8 XL of BMC induced secretion of active TGF-β1 into the culture supernatant. As active TGF-β1 could result from secretion of prestored TGF-β1 precursors or from activation of gene expression, we used semiquantitative RT-PCR to analyze TGF-β1 mRNA levels after XL CD8 and/or CD16 surface molecules on BMC. CD8 or CD16 XL induced an ∼2-fold increase in the steady state amounts of TGF-β1 mRNA within 4 hr (Fig. 1A). As both of the enriched CD8+ cell populations, CD16+ and CD16-, respectively, were already coated with anti-CD8 during purification, we used an isotype control normal mouse IgG as a baseline control for CD8 XL. For CD16 XL, we used anti -CD16 mAb without XL as a baseline control rather than the normal IgG. Other experiments (data not shown) demonstrated consistently that addition of anti-CD16 antibody without adding beads gave similar background TGF-β1 mRNA levels as XL with isotype control antibody. Perturbation of both CD8 and CD16 surface molecules had an additive effect on TGF-β1 mRNA accumulation. To determine the kinetics of mRNA accumulation, we measured TGF-β1 steady state mRNA levels at times 0, 4, 8, and 24 hr after CD8 XL of BMC. The accumulation of TGF-β1 mRNA peaked at 4 hr, decreased to baseline levels by 8 hr, and remained at baseline levels at 24 hr (Fig. 1B). CD16 XL induced accumulation of TGF-β1 mRNA with similar kinetics (data not shown). Thus, CD8 or CD16 XL induced a transient and rapid accumulation of TGF-β1 mRNA.

Figure 1:
CD8 and CD16 XL up-regulates TGF-β1 mRNA. (A) Semiquantitative RT-PCR was performed on total RNA isolated from normal BMC XL on CD8 and CD16. (B) CD8 XL induces maximal accumulation of TGF-β1 mRNA at 4 hr. Rhesus macaque BMC were XL on CD8, incubated for the indicated times, and TGF-β1 mRNA levels were evaluated. cDNA from phorbol myristate acetate (PMA)-treated monkey PBMC cells served as positive control, and NC denotes controls in which no template was added to the PCR reactions. The 4-hr time point was used in subsequent experiments. These results were reproduced in four experiments. (C) Actinomycin D suppresses the induction of TGF-β1 mRNA by CD8 XL. BMC were XL on CD8 and incubated in medium with diluent or actinomycin D. TGF-β1 levels were analyzed by semiquantitative RT-PCR. The positive control for the PCR was cDNA from PMA/ionomycin-treated rhesus macaque PBMC. NC denotes the no template control. This result was reproduced in three independent experiments.

Induction of TGF-β1 mRNA accumulation by CD8 XL involves active transcription. Elevation of steady state TGF-β1 mRNA levels induced by CD8 or CD16 XL may be mediated by transcriptional or posttranscriptional mechanisms. To determine whether the accumulation of TGF-β1 mRNA involves increased transcription we used actinomycin D to inhibit active transcription (16). Normal BMC were subjected to CD8 XL and incubated for 4 h in the presence of medium supplemented with DMSO (vehicle) or 10 µg/ml actinomycin D. Actinomycin D suppressed the up-regulation of steady state TGF-β1 mRNA induced by CD8 ligation by almost 70% (Fig. 1C). This result indicates that the elevation of steady state TGF-β1 mRNA levels after CD8 XL involved active transcription.

Two subsets of CD8+ BMC up-regulate TGF-β1 mRNA in response to CD8 XL. The CD8+ BMC used in the above experiments included CD3-CD16+ and CD3+CD16- subsets. Our previous studies showed that depletion of CD8+ or CD16+ cells from donor BMC resulted in loss of the tolerance promoting effect of donor BMC in vivo as well as loss of the veto effect of donor BMC in vitro. In contrast, removal of CD3+ cells from donor BMC had no apparent effect (2,8). Therefore, we examined whether or not both CD3-CD16+ and CD3+CD16- subsets of CD8+ BMC up-regulate TGF-β1 mRNA in response to CD8 perturbation. To address this question, CD8+ cells were purified from normal BMC and sorted into CD3-CD16+ and CD3+CD16- subsets by fluorescence-activated cell sorting. Both subsets were subjected to CD8 XL and semiquantitative RT-PCR to evaluate TGF-β1 mNRA levels. Raw data are shown in Figure 2A as computer-generated images after capturing and processing of Polaroid photos of the gels using an AlphaImager 2000 documentation and analysis system (BioRad). The upper and lower panels represent β-actin and TGF-β1 RT-PCR, respectively. Bands were scanned, and calculation of arbitrary units was performed by normalizing the intensity of the TGF-β1 bands to the intensity of the actin bands. The intensity of the actin band was first corrected subtracting the intensity of the no RT actin control as described in Material and Methods. The normalized results are expressed as arbitrary units of TGF-β1 mNRA in Figure 2B. Although "equivalent amounts" of cDNA were used for PCR, the bands in lanes 1 and 3 are less intense compared to the bands in lanes 5 and 7. This implies underloading of cDNA in lanes 1 and 3. After correcting for the differences in cDNA amounts by normalization, the TGF-β1 band intensities for lanes 1 and 3 give higher arbitrary units of TGF-β1 mNRA than the bands in lanes 5 and 7. The CD3-CD8+CD16+ subset exhibited higher baseline levels of TGF-β1 mNRA after incubation with normal mouse IgG, and CD8 XL resulted in a 2-3-fold increase in steady state TGF-β1 mNRA (Fig. 2B). Although the CD16+ subset appeared to be the greater producer, both subsets of CD8+ BMC up-regulated TGF-β1 mNRA after CD8 XL.

Figure 2:
The two subsets of CD8+ BMC up-regulate TGF-β1 mRNA in response to CD8 XL. CD8-enriched BMC were sorted into CD3-CD16+ and CD3+CD16- subset. The sorted cells were XL on CD8 and analyzed for TGF-β1 mRNA. (A) Computer-generated images of agarose gels of β-actin (upper panel) and TGF-β1 (lower panel) RT-PCR products. M and PC in both panels denote molecular weight markers (1-kb DNA ladder, GIBCO) and positive control (cDNA from PMA/ionomycin-treated rhesus PBMC), respectively. Upper panel: lanes 1-4 and 5-8 represent cDNA from CD3-CD16+ and CD3+CD16- subsets of CD8+ cells, respectively. Lanes 1 and 5 represent isotype control XL, and lanes 3 and 7 represent CD8 XL. Lanes 2, 4, 6, and 8 are corresponding samples without reverse transcriptase added during cDNA synthesis. Lower panel: lanes 1, 3, and 5, 7, correspond to those in the upper panel. (B) Graphic representation of panel A. Bands were scanned and processed as described in Materials and Methods to generate the bar chart in B.

CD8 XL induces secretion of TGF-β1. To determine whether CD8 XL induces accumulation of intracellular TGF-β1 protein, rhesus BMC were XL as above and stained for intracellular TGF-β1 at the indicated times. Stained samples were analyzed by flow cytometry (Coulter Epics Elite) and subjected to immuno-4 processing (Coulter Epics Elite software). The immuno-4 program subtracts the negative control (isotype-matched mAb XL) histogram channel by channel, from the histogram for the test sample (CD8 XL), and normalizes the negative control histogram to obtain the positives histogram (insets in Fig. 3). At 0 hr (immediately after XL) 9% of the gated CD+ cells stained positive for intracellular TGF-β1. By 8 hr after XL, 4% of the gated cells were positive for intracellular TGF-β1. Thereafter, the percentage of gated cells expressing TGF-B1 in response to CD8 XL increased, to 21% and 32% at 24 and 48 hr after XL, respectively. Thus, CD8 XL induced accumulation of intracellular TGF-β1 relative to the isotype control (Fig. 3).

Figure 3:
CD8 XL induces accumulation of intracellular TGF-β1 protein. Rhesus BMC were subjected to XL, incubated for the indicated times, followed by staining for intracellular TGF-β1 as described in Materials and Methods. The gate was set on the CD8+ subset, and the isotype-matched and CD8 XL single parameter histograms were analyzed by immuno-4 processing (Coulter Epics Elite Software). Solid and dotted lines represent IgG1 and CD8 XL, respectively. The inset in each panel shows the immuno-4-derived positives after normalization and subtraction of the IgG1 histogram. Figure is representative of three experiments.

We demonstrated previously by a bioassay that BMC releases active TGF-β1 into the culture supernatant after CD8 XL (8). In view of the suggestion that the bioassay is only partly specific for TGF-β1 (17), here we measured active TGF-β1 directly using a highly specific and sensitive ELISA kit (Emax™ TGF-β1 immunoassay system; Promega). CD8+ cells were isolated from two normal rhesus donors and subjected to XL as described above. After culturing for 48 hr, supernatants were harvested and tested for TGF-β1 protein as described in Materials and Methods. CD8 XL induced a 30% (62-78 pg/ml) and 50% (70-104 pg/ml) increase in the amount of TGF-β1 detected in the culture supernatant at 48 hr compared to the isotype control (Fig. 4).

Figure 4:
CD8 XL induces secretion of TGF-β1. CD8+ cells were XL with CD8 mAb or isotype control mAb. Supernatants were harvested at 48 hr and secreted TGF-β1 was measured by ELISA as described in Materials and Methods.

CD8-enriched BMC exhibit veto activity in vitro. Previous studies demonstrated indirectly by depletion that the veto activity of rhesus macaque allo-BMC is mediated by a subset that expresses a CD2+CD3-CD8+CD16+ phenotype (2). To determine directly if CD8+ donor BMC exhibit veto activity in vitro, we added purified CD8+ donor BMC to MLR-induced CTL cultures 24 hr after culture initiation. Addition of unfractionated BMC or the CD8+ fraction of BMC from the MLR stimulator cell donor to allogeneic MLR cultures consistently inhibited the control CTL response. In the representative experiment shown (Fig. 5A), CD8+ BMC significantly reduced the control specific lysis from a median 37.6% lysis (n=4) to 13.6% lysis (n=4, P=0.03 by Mann-Whitney test) at the 0.75 ratio of veto cells to responder cells. The inhibition of median percent specific lysis by intact BMC had borderline significance (P=0.08) compared to the median percent specific lysis in the absence of BMC. In contrast, unrelated third-party intact or CD8+ BMC consistently resulted in minimal inhibition (data not shown). Thus veto activity was enriched in the CD8+ fraction of rhesus BMC, selectivity inhibiting the response of allogeneic CTLp to the BMC donor's PBL stimulator cells.

Figure 5:
CD8-enriched BMC exhibit veto activity. (A) CD8-enriched BMC were tested for veto activity in vitro by the MLR-induced CTL assay. For each experiment, the number or responding cells was 2×105 per well, and the number of "veto cells" added per well is indicated by the ratio. Results are presented as median percent specific lysis. (B) Both subsets of CD8+ cells demonstrate veto activity. CD8-enriched cells were labeled with anti-CD16 mAb and sorted into CD3-CD16+ and CD3+CD16- subsets, which were then tested for veto activity in vitro by the MLR-induced CTL assay. Results are presented as median percent inhibition of control CTL lysis, and were duplicated in a second independent experiment.

CD8+ cells were further fractionated into CD3-CD8+CD16+ and CD3+CD8+CD16- subsets and tested for veto activity in the MLR-induced CTL system. Both subsets of CD8+ cells were inhibitory in the in vitro veto assay (Fig. 5B). However, at the lowest veto cell/responder cell ratio (0.03), the median percent inhibition was reduced to baseline levels (33%) in the CD3+CD8+CD16- fraction while the CD3-CD8+CD16+ fraction became maximally inhibitory (100%, P=0.002 by Mann-Whitney test). Thus, the CD16+ subset of CD8+ cells is clearly enriched for veto activity.

Recombinant TGF-β1 induces apoptosis of activated T cells. On the basis of our finding that perturbation of CD8 by XL induces TGF-β1 mRNA and protein accumulation and that secretion of active TGF-β1 correlated with veto activity in a previous study (8), we hypothesized that paracrine TGF-β1 might facilitate deletion of responding T cells by directly inducing apoptosis. Our hypothesis is consistent with the finding that murine BMC mediated veto activity by clonal deletion of responding CTLp (18). To examine this possibility, we first determined whether human recombinant TGF-β1 (rTGF-β1) can induce apoptosis in activated rhesus macaque T cells. In these experiments, T cells were preactivated with PHA for 2 days and incubated for an additional ∼20 hr with or without 10 ng/ml rTGF-β1. After harvesting, the T cells were examined for apoptosis by a flow cytometry-based TUNEL assay. Human rTGF-β1 induced moderate median percent apoptosis (32.9, n=6, P=0.02) over background median percent apoptosis (24.4%, n=6) in activated T cells (Fig. 6A). The percent specific TGF-β1-induced apoptosis (the difference between the treated and untreated cells) was 13.2 ± 9.9% SD.

Figure 6:
TGF-β1 induces apoptosis of activated rhesus macaque T cells. (A) PHA-activated rhesus T cells were incubated for ∼20 hr with 10 ng/ml human rTGF-β1. Apoptosis in the T cells was assessed by a flow cytometry-based TUNEL assay. Data are from seven independent experiments. (B) Activated rhesus T cells were co-cultured with COS cells expressing luciferase (Ad/luciferase, upper panel) or constitutively active TGF-β1 (Ad/TGF-β1, lower panel). T cells were harvested after ∼20 hr and stained with annexin V-FITC to detect cells undergoing apoptosis. M1 denotes percentage of gated cells stained with annexin V-FITC, indicative of apoptosis. These results are representative of four experiments. (C) Activated T cells were co-cultured with COS cells expressing constitutively active TGF-β1 as above, with or without neutralizing anti-TGF-β1 pAb. Apoptosis was determined by annexin V-FITC staining. Results are presented as median percent specific apoptosis in gated cells or percent specific apoptosis (calculated as stated in Materials and Methods).

In a separate experiment, activated T cells were co-cultured with COS cells transduced with recombinant adenovirus expressing constitutively active TGF-β1 or a control recombinant adenovirus expressing an irrelevant (luciferase) gene. Apoptosis was detected in 23% of the activated T cells co-cultured with control COS cells expressing luciferase (Fig. 6B, upper panel, Ad/luciferase). In contrast, 54% of the activated T cells cocultured with COS cells expressing TGF-β1 were apoptotic (Fig. 6B, lower panel, Ad/TGF-β1). Thus, exposure of the activated T cells to continuous secretion of active TGF-β1 increased the frequency of apoptotic cells nearly 2.5-fold. Chicken anti-TGF-β1 neutralizing pAb (designated αTGF-β1 NAb, R&D Systems) suppressed the specific apoptosis induced by COS cells expressing simian TGFβ1 by 80% (Fig. 6C).


In this study, we have demonstrated that ligation of the CD8 and CD16 surface molecules on rhesus macaque BMC induces elevation of steady state TGF-β1 mRNA levels with maximal accumulation at 4 hr. The effect of actinomycin D observed in our study suggests that the accumulation of TGF-β1 mRNA after CD8 ligation involves active transcription. Experiments are in progress to show directly by nuclear run-off assays the role of active transcription in the accumulation of TGF-β1 mRNA after CD8 ligation. These results support our hypothesis that activation of CD8 and CD16 pathways triggers signaling events within BMC, presumably through p56lck tyrosine kinase, leading to increased TGF-β1 mNRA steady state levels. These data provide a possible explanation for the finding that CD8+ cells lacking the cytoplasmic domain of the CD8 molecule were unable to veto, i.e., induce deletion of responding T cells (12). CD8+ cells lacking the cytoplasmic domain of CD8 would not be expected to transduce activation signals for TGF-β1 up-regulation in response to CD8 ligation. We demonstrated previously that TGF-β1 accumulated in supernatants at 24 hr and later (8), and in this study we showed that TGF-β1 accumulated inside cells at 24 hr and in the supernatant 48 hr after CD8 was XL on rhesus monkey cells. There was a decrease in intracellular TGF-β1 protein at 8 hr after CD8 XL, compared to 0 hr. We speculate that CD8 XL may cause release of preexisting intracellular TGF-β1. Maximal levels of TGF-β1 mRNA occurred at 4 hr, whereas TGF-β1 protein began to accumulate at 24 hr. As we observed only transient increases in the steady state levels of TGF-β1 mNRA, we cannot exclude the possibility that secretion of active TGF-β1 in response to CD8 perturbation may also involve posttranscriptional mechanisms (Fig. 7). Indeed, Ahuja et al (17). noted the involvement of transcriptional as well as posttranscriptional mechanisms in the elevation of TGF-β1 mRNA that occurred when human T cells were activated in the presence of cyclosporine.

Figure 7:
Schematic representation of potential mechanisms used by CD8+ donor veto cells to delete recipient CTLp by apoptosis.

Our previous studies showed that depleting donor BMC of CD8+ or CD16+ cells resulted in loss of veto activity in vitro, and loss of the tolerogenic effect of donor BMC in vivo (8). Here, we have demonstrated directly that CD8-enriched BMC exhibit veto activity in vitro. Although both CD3-CD8+CD16+ and CD3+CD8+CD16- subsets suppress CTL lysis in vitro, the CD8+CD16+ subset was most effective at the lowest veto cell/responder cell ratio at which the CD8+CD16- subset was ineffective. Interestingly, the CD8+CD16+ subset had a higher TGF-β1 mNRA baseline and maximal level after CD8 XL relative to the CD8+CD16- subset (Fig. 2B). In addition, our preliminary data (not shown) suggest that both CD3-CD16+ and CD3+CD16- subsets of CD8+ cells stain positive for intracellular TGF-β1 protein. That both CD8+ populations are capable of producing TGF-β1 is consistent with the broad ability of leukocyte populations to produce TGF-β1 (reviewed in 19).

Both CD8 and CD16 are expressed on rhesus cells (1), and natural killer cells have been demonstrated to have veto activity (20,21). These findings are consistent with the suggestion that multiple types of cells can serve as veto cells (5). Our results are in agreement with murine studies suggesting a pivotal role for CD8+ cells in veto effect (reviewed in 21). In contrast to a recent report on human CD8+ veto cells (22), the CD8-enriched BMC-mediated veto activity described in the rhesus macaque MLR-induced CML model is relatively specific for the stimulator, as addition of third-party, unrelated BMC produced only background suppression of CTL lysis. This difference may be related to the addition of recombinant IL-2 in the rhesus system during the MLR-induction phase to avoid limitation in IL-2 availability.

TGF-β1 induces apoptosis in a variety of cell types (23-28), including human (29) and murine IL-2-dependent (26,30) T cell clones, and human B cells (28). More recently, it was reported that Ca2+ synergized with cyclosporine to induce secretion of TGF-β1 from B cells, and the released TGF-β1 induced apoptosis in T and B lymphocytes in the presence of elevated intracellular Ca2+(27). Here, we have demonstrated that human and porcine recombinant TGF-β1 alone induce apoptosis in activated rhesus T cells. Greater apoptosis was observed when activated T cells were exposed to a continuous cellular source of active TGF-β1, which suggests that the paracrine effect of TGF-β1 may be important for the induction of apoptosis. The ability of neutralizing anti-TGF-β1 pAb to abrogate most of the specific apoptosis induced by the COS cells expressing constitutively active simian TGF-β1 indicates that the apoptosis observed is due specifically to TGF-β1. The moderate levels of apoptosis induced by rTGF-β1 in this study may be a result of the dose used, to efficient inactivation of rTGF-β1 by serum factors in the FBS used to supplement culture medium, or possibly to the fact that the human recombinant protein was used. The latter is unlikely as a result of the highly conserved nature of this protein across species (31). Others have reported that TGF-β1-mediated apoptosis was maximal at 48 hr (26). In our studies, rTGF-β1 was incubated with activated T cells for ∼20 hr; thus, the shorter incubation could account for the moderate levels of apoptosis we observed.

Our demonstration that TGF-β1 induced apoptosis in rhesus PHA-activated T cells contrasts with the report that TGF-β1 enhanced survival of PHA-activated human T cells. There were, however, two methodical differences between the human study and our study. In the human study, T cells were grown in serum-free medium, and TGF-β1 was present during initial activation of the T cells (32). In our study, rhesus PBMC were grown in RPMI 1640 containing 10% FBS, and TGF-β1 was added 48 hr after PHA activation of the T cells. This timing is more relevant to our concept of the veto effect, as the responding CTLp are optimally inactivated when antigen-expressing CD8+ cells are added 1-2 days after activation (33,34). Thus, inactivation of responding CTLp by veto cells, as well as TGF-β1-induced apoptosis, requires prior activation of the responding CTLp.

The molecular basis of TGF-β1-mediated apoptosis remains unclear. TGF-β1 effects are mediated by binding to transmembrane receptors with serine/threonine kinase activity (reviewed in 31). Negative growth regulatory effects of TGF-β1 have been associated with activation of Ras and extracellular signal-regulated kinases 1 and 2, and stress-activated protein kinase/Jun N-terminal kinase (SAPK/JNK) (35). Activation of SAPK/JNK signaling pathways has been reported to induce apoptosis (36). TGF-β1-induced cell cycle arrest and apoptosis was shown to correlate with inhibition of tyrosine phosphorylation and activation of Jak-1 and Stat 5 in murine T lymphocytes (37). Thus, TGF-β1 may induce apoptosis of activated T cells by simultaneously activating the Ras, ERKs 1 and 2, SAPK/JNK pathway, and inhibiting the IL-2-dependent activation of Jak-1 and Stat 5. In terms of more distal events, TGF-β2 induced apoptosis of murine T cells without altering Bcl-2 mRNA levels (26). Studies are in progress to determine the effect of TGF-β1 on Bcl-2, Bcl-x, and Bax levels in activated rhesus macaque T cells. In B cell lymphoma cell lines, TGF-β1 reduced NF-κB/Rel activity, leading to decreased c-Myc expression, with resultant apoptosis (38). Another potential mechanism by which TGF-β1 might facilitate veto activity is through an autocrine positive feedback loop to increase CD8 expression. The CD8/class I receptor-ligand interaction between the antigen-expressing CD8 cells and responding T cells is obligatory for veto activity (8,12), and TGF-β1 may enhance the CD8/class Iα3 domain interaction by up-regulating expression of CD8 (39). This loop could in turn lead to increased signal transmission via CD8 that results in augmentation of TGF-β1 transcripts and also transcription of FasL, a complementary apoptosis pathway (W. Wang, C. Asiedu, and J. Thomas, unpublished data).

We have not examined the role for TGF-β1-mediated apoptosis in veto activity in vivo, but experiments are currently in progress to knock out TGF-β1 expression in CD8+ BMC and test veto activity in vitro and in vivo. It has been suggested that veto cells may normally function to maintain autotolerance (21). A role for TGF-β1 in the veto effect in vivo is consistent with this view, as TGF-β1 knockout mice are unable to maintain self-tolerance and die within 3 weeks after birth with diverse autoimmune pathologies (40,41). Furthermore, there is evidence that treatment with exogenous TGF-β1 is beneficial in autoimmune disease (42).

Deletion of responding T cells may be mediated by induction of apoptosis by restimulation through the TCR/CD3 complex (43) or CD2 (44), or simultaneous stimulation of TCR and class Iα3 domain (10), or Fas/FasL interaction (45,46). We postulate that TGF-β1-mediated apoptosis may be added to the list of activation-induced T cell death pathways that contribute to immune regulation (Fig. 7). Consistent with this hypothesis, increased secretion of TGF-β1 accompanies T cell activation (17,47).

In summary, these studies confirm and extend our earlier studies implicating TGF-β1 in the tolerogenic veto mechanisms of rhesus monkey allo-BMC. Although TGF-β1 may not be the only mechanism of functional CTLp deletion and tolerance induction, the data are consistent with a hypothesis that TGF-β1 induction and paracrine secretion by donor CD8+ cells is an important component of the process.

Acknowledgments. The authors thank Dr. Robert Knowles for providing anti-CD8 mAb, and Drs. David T. Curiel and Jesus Navarro for generously supplying recombinant adenoviruses. The authors are grateful to Scott D. Sweeney for help with fluorescence-activated cell sorting analysis, to Dana Willis and Dr. Sam Cartner for expert primate care and assistance, and Dr. William J. Hubbard for helpful discussions.


1. Thomas J, Carver M, Cunningham P, Park K, Gonder J. Promotion of incompatible allograft acceptance in rhesus monkeys given posttransplant antithymocyte globulin and donor bone marrow: in vivo parameters and immunohistologic evidence suggesting michrochimerism. Transplantation 1987; 43: 332.
2. Thomas JM, Carver FM, Cunningham PRG, Olson LC, Thomas FT. Kidney allograft tolerance in primates without chronic immunosuppression: the role of veto cells. Transplantation 1991; 51: 198.
3. Thomas JM, Verbanac KM, Thomas FT. The veto mechanism in transplant tolerance. Transplant Rev 1991; 5: 209.
4. Thomas J, Carver M, Foil B. Transplantation 1983; 36: 104.
5. Fink PJ, Shimonkevitz RP, Bevan MJ. Veto cells. Annu Rev Immunol 1988; 6: 115.
6. Miller RG. The veto phenomenon and transplantation tolerance. In Auchincloss H, ed. Transplantation immunology. Wiley-Liss, Inc., New York, 1995; 267.
7. Heeg K, Wagner H. Induction of peripheral tolerance to class I major histocompatibility complex (MHC) alloantigens in adult mice: transfused class I MHC-incompatible splenocytes veto clonal responses of antigen-reactive Lyt-2+ T cells. J Exp Med 1990; 172: 719.
8. Verbanac KM, Carver M, Haisch CE, Thomas JM. A role for transforming growth factor-beta in the veto mechanism in transplant tolerance. Transplantation 1994; 57: 893.
9. Kaplan DR, Hambor JE, Tykocinski, ML. An immuno-regulatory function for the CD8 molecule. Proc Natl Acad Sci USA 1989; 86: 8512.
10. Sambhara SR, Miller RG. Programmed cell death of T cells signaled through the T cell receptor and the α3 domain of class I MHC. Science 1991; 252: 1424.
11. Hambor JE, Kaplan DR, Tykocinski ML. CD8 functions as an inhibitory ligand in mediating the immunoregulatory activity of CD8+ cells. J Immunol 1990; 145: 1646.
12. Zhang L, Fung-Leung W, Miller RG. Down-regulation of CD8 on mature antigen-reactive T cells as a mechanism of peripheral tolerance. J Immunol 1995; 155: 3464.
13. Woodle SE, Smith DM, Bluestone JA, Kirkman WM III, Green DR, Skowronski EW. Anti-human class I MHC antibodies induce apoptosis by a pathway that is distinct from the Fas antigen-mediated pathway. J Immunol 1997; 158: 2156.
14. Miyazono K, Hellman U, Wernstedt C, Heldin CH. Latent high molecular weight complex of transforming growth factor-β1. J Biol Chem 1988; 263: 6408.
15. Genestier L, Paillot R, Bonnefoy-Berard N, Waldmann H, Revillard JP. T cell sensitivity to HLA class I-mediated apoptosis is dependent on interleuken-2 and interleuken-4. Eur J Immunol 1997; 27: 495.
16. Yang L, Steussy CN, Fuhrer DK, Hamilton J, Yang Y-C. Interleukin-11 mRNA stabilization in phorbol ester-stimulated primate bone marrow stromal cells. Mol Cell Biol 1996; 16: 3300.
17. Ahuja SS, Shrivastav S, Danielpour D, Balow JE, Boumpas DT. Regulation of transforming growth factor-β1 and its receptor by cyclosporine in human T cells. Transplantation 1995; 60: 718.
18. Hiruma K, Gress RE. Cyclosporine A and peripheral tolerance. J Immunol 1992; 149: 1539.
19. Letterio J, Roberts A. Regulation of immune responses by TGF-β1. Annu Rev Immunol 1998; 16: 137.
20. Azuma E, Kaplan J. Role of lymphokine-activated killer cells as mediators of veto and natural suppression. J Immunol 1988; 141: 2601.
21. Miller RG, Zhang L, Sheng-Tanner X-F, Hammad N, Radvanyi L. The veto hypothesis and transplantation tolerance. Transplant Sci 1993; 3: 206.
22. Mathew JM, Carreno M, Fuller L, et al. Modulatory effects of human donor bone marrow cells on allogeneic cellular immune responses. Transplantation 1997; 63: 686.
23. Oberhammer FA, Pavelka M, Sharma S, et al. Induction of apoptosis in cultured hepatocytes and in regressing liver by transforming growth factor β1. Proc Natl Acad Sci USA 1992; 89: 5408.
24. Bursch W, Oberhammer F, Jirtle RL, et al. Transforming growth factor β1 as a signal for induction of cell death by apoptosis. Br J Cancer 1993; 67: 531.
25. Takiya S, Tagaya T, Takahashi K, et al. Role of transforming growth factor β1 on hepatic regeneration and apoptosis in liver diseases. J Clin Pathol. 1995; 48: 1093.
26. Weller M, Constam DB, Malipiero U, Fontana A. Transforming growth factor-β2 induces apoptosis of murine T cell clones without down-regulating bcl-2 mRNA expression. Eur J Immunol 1994; 24: 1293.
27. Andjelic S, Khana A, Suthanthiran M, Nikolic-Zugic J. Intracellular Ca2+ elevation and cyclosporin A synergistically induce TGF-β1-mediated apoptosis in lymphocytes. J Immunol 1997; 158: 2527.
28. Lomo J, Blomhoff HK, Beiske K, Stokke T, Smeland EB. TGF-β1 and cyclic AMP promote apoptosis in resting human B lymphocytes. J Immunol 1995; 154: 1634.
29. AyanlarBatuman O, Ferrero AP, Diaz A, Jimenez S. Regulation of transforming growth factor-β1 gene expression by glucocorticoids in normal human T lymphocytes. J Clin Invest 1991; 88: 1574.
30. Weller M, Malipiero U, Groscurth P, Fontana A. T cell apoptosis induced by IL-2 deprivation or transforming growth factor-β2: modulation by the phosphatase inhibitors okadaic acid and calyculin A. Exp Cell Res 1995; 221: 395.
31. Kingsley DM. The TGF-β superfamily: new members, new receptors, and new genetic tests of function in different organisms. Genes Dev 1994; 8: 133.
32. Cerwenka A, Kovar H, Majdic O, Holter W. Fas- and activation-induced apoptosis are reduced in human T cells preactivated in the presence of TGF-β1. J Immunol 1996; 156: 459.
33. Fink PJ, Rammensee HG, Benedetto JD, Staerz UD, Lefrancois L, Bevan MJ. Studies on the mechanism of suppression of primary cytotoxic responses by cloned cytotoxic T lymphocytes. J Immunol 1984; 133: 1769.
34. Thomas JM, Carver FM, Kasten-Jolly J, et al. Further studies of veto activity in rhesus monkey bone marrow in relation to allograft tolerance and chimerism. Transplantation 1994; 57: 101.
35. Frey RS, Mulder KM. Involvement of extracellular signal-regulated kinase 2 and stress-activated protein kinase/Jun N-terminal kinase activation by transforming growth factor β in the negative growth control of breast cancer cells. Cancer Res 1997; 57: 628.
36. Ichijo H, Nishida E, Irie K, et al. Induction of apoptosis by ASK1, a mammalian MAPKKK that activates SAPK/JNK and p38 signaling pathways. Science 1997; 275: 90.
37. Bright JM, Kerr LD, Sriram S. TGF-β inhibits IL-2-induced tyrosine phosphorylation and activation of Jak-1 and Stat 5 in T lymphocytes. J Immunol 1997; 159: 175.
38. Arsura M, Wu M, Sonenshein GE. TGFβ1 inhibits NF-κB/Rel activity inducing apoptosis of B cells: transcriptional activation of IκBα. Immunity 1996; 5: 31.
39. Inge TH, McCoy KM, Susskind BM, Barrett SK, Zhao G, Bear HD. Immunomodulatory effects of transforming factor-β on T lymphocytes. J Immunol 1992; 148: 3847.
40. Kulkarni AB, Ward JM, Yaswen L, et al. Transforming growth factor-β1 null mice: an animal model for inflammatory disorders. Am J Pathol 1995; 146: 264.
41. Yasmen L, Kulkarni AB, Fredrickson T, et al. Autoimmune manifestations in the transforming growth factor-β1 knock-out mouse. Blood 1996; 87: 1439.
42. Ruscetti F, Varesio L, Ochoa A, Ortaldo J. Pleiotropic effects of transforming growth factor-β on cells of the immune system. Ann NY Acad Sci 1993; 685: 488.
43. Wesselborg S, Janseen SO, Kabelitz D. Induction of activation-driven death (apoptosis) in activated but not resting peripheral blood T cells. J Immunol 1993; 150: 4338.
44. Wesselborg SU, Pruffer M, Wild B, Schrave S, Meuer C, Kabelitz D. Triggering via the alternative CD2 pathway induces apoptosis in activated human T lymphocytes. Eur J Immunol 1993; 23: 2707.
45. Nagata S, Golstein P. The Fas death factor. Science 1995; 267: 1449.
46. Kramer PH, Dhein J, Knipping E, et al. The role of APO-1 mediated apoptosis in the immune system. Immunol Rev 1994; 142: 175.
47. Li B, Sehajpal PK, Khanna A, Vlassara H et al. Differential regulation of transforming growth factor beta and interleukin 2 genes in human T cells: demonstration by usage of novel competitor DNA constructs in the quantitative polymerase chain reaction. J Exp Med 1991; 174: 1259.
© 1999 Lippincott Williams & Wilkins, Inc.