Lethal Effect of CD3-Specific Antibody in Mice Deficient in TGF-β1 by Uncontrolled Flu-Like Syndrome

Abstract

CD3-specific Ab therapy results in a transient, self-limiting, cytokine-associated, flu-like syndrome in experimental animals and in patients, but the underlying mechanism for this spontaneous resolution remains elusive. By using an in vivo model of CD3-specific Ab-induced flu-like syndrome, we show in this paper that a single injection of sublethal dose of the Ab killed all TGF-β1^−/− mice. The death of TGF-β1^−/− mice was associated with occurrence of this uncontrolled flu-like syndrome, as demonstrated by a sustained storm of systemic inflammatory TNF and IFN-γ cytokines. We present evidence that deficiency of professional phagocytes to produce TGF-β1 after apoptotic T cell clearance may be responsible, together with hypersensitivity of T cells to both activation and apoptosis, for the uncontrolled inflammation. These findings indicate a key role for TGF-β1 and phagocytes in protecting the recipients from lethal inflammation and resolving the flu-like syndrome after CD3-specific Ab treatment. The study may also provide a novel molecular mechanism explaining the early death in TGF-β1^−/− mice.

Treatment with CD3-specific Ab induces immune tolerance in experimental animals and was initially used in transplant patients, but its use today is limited due to deleterious side effects (1, 2). The most salient side effect is caused by transient T cell activation after the CD3-specific Ab injection (1), leading to the systemic release of inflammatory cytokines within the initial hours after the first injection of Ab (3– 8). This release of TNF, IL-6, and IFN-γ leads to a flu-like syndrome, although there might be more complex mechanisms in the patients. Among these involved cytokines, TNF plays a critical role, as its exclusive blockade with TNF-specific Abs was sufficient to abrogate the flulike syndrome (1). Intriguingly, this flu-like syndrome is transient and self-limiting and resolves by the second or third day of the treatment following the elimination of the systemic cytokines. However, the underlining mechanism responsible for this self-resolution of the syndrome is unknown. We hypothesize that TGF-β can be involved in this mechanism, because TGF-β plays a critical role in the regulation of immune responses (9–19) as well as in CD3-specific Ab mediated immune tolerance (20–24). To explore the function of TGF-β in the self-resolution of the flu-like syndrome, we used the TGF-β1 null (TGF-β1^−/−) mice. Early after birth, these mice have been shown to develop a wasting syndrome associating a multifocal mixed inflammatory response and leading to organ failure and death (25, 26). Although the lethal inflammation of TGF-β1^−/− mice has demonstrated an indispensable role for TGF-β1 in vivo (25, 26), the underlying mechanisms for the death of the null mice remain a mystery. The lack of TGF-β1 in vivo is not the sole explanation, because neither administration of exogenous TGF-β1 protein nor gene therapy with TGF-β1 plasmid (27) could rescue the TGF-β1^−/− mice from death. Intriguingly, even when TGF-β1^−/− mice were crossed with transgenic (TG)⁵ mice that specifically expressed TGF-β1 in the liver and secreted it into the blood, the resultant TGF-β1^−/−(TG) mice exhibited a survival profile similar to the TGF-β1^−/− mice. This occurred despite that serum levels of TGF-β1 in these TGF-β1^−/−(TG) mice were restored to normal levels with expression in all the tissues (28). This finding indicates that the TGF-β1 deficiency in immune cells may play a critical role in the uncontrolled systemic inflammation in TGF-β1^−/− mice. To add the complex situation further, TGF-β1^−/− T cells exhibit an increase in spontaneous apoptosis as well as TCR activation-induced cell death (29). In normal mice, T cell apoptosis, followed by apoptotic cell uptake by macrophages and some dendritic cell (DC) subsets releasing in turn TGF-β, is involved in the resolution of inflammation (22, 30–34). Thus, a salient question is why enhanced T cell death fails to lead to the resolution of the immune responses, but rather is accompanied by the uncontrolled inflammation and consequential demise of TGF-β1^−/− mice.

In this study, by using a CD3-specific Ab treatment model in TGF-β1^−/− mice, we show that a single sublethal dose of CD3-specific Ab injection killed all knockout mice due to uncontrolled inflammatory cytokine release. We presented evidence that deficiency of professional phagocytes to produce TGF-β1 after apoptotic T cell clearance may be responsible, together with hypersensitivity of T cell activation and increased T cell apoptosis, for lethal inflammation. The rescue from death of CD3-specific Ab-treated TGF-β1^−/− mice by depletion of their endogenous phagocytes or adoptive transfer of wild-type phagocytes suggests TGF-β production by phagocytes is sufficient to control inflammation. The findings provide a novel explanation for the lethal inflammation in TGF-β1^−/− mice and help resolve the paradoxical observation of increased T cell death accompanying uncontrolled inflammation in TGF-β1^−/− mice. This study also implicates phagocyte-derived TGF-β release as an underlying mechanism for the self-resolution of flu-like syndrome after CD3-specific Ab treatment.

Materials and Methods

Mice

TGF-β1^−/− (C57BL/6×SVJ129) and age-matched wild-type (TGF-β1^+/+) or heterozygous (TGF-β1^+/−) mice (29), as well as Tgfbr1^f/f-Lck-Cre⁺ or Tgfbr1^f/f-Lck-Cre⁻ (35) mice were housed in a specific pathogen-free rodent facility at the National Institute of Dental and Craniofacial Research, National Institutes of Health. All animal studies were performed according to National Institutes of Health guidelines for use and care of live animals and approved by Animal Care and Use Committee of National Institute of Dental and Craniofacial Research.

Antibodies and reagents

The following Abs were purchased form BD Biosciences: purified anti-CD16/CD32 (clone 93), CD4 (clone GK1.5, NA/LE), CD8 (clone 53-6.7, NA/LE), CD80 (clone 16-10A1), CD86 (Clone GL1), MHC class II (Clone M5/114.15.2), CD45RB220 (Clone RA3–6B2), FITC-conjugated anti-CD3, -CD4, -CD11c, and -IFN-γ; PE-conjugated anti-CD8, -CD11c, -CD11b, -TNF, -IL-4, and -IFN-γ; PerCP-conjugated anti-CD4 and -CD8; allophycocyanin-conjugated anti-CD45_LCA and PE- or allophycocyanin-conjugated IL-2 Abs. CD83 (Clone Michel-17) was purchased from e-Biosciences, FITC-conjugated anti-F4/80 Ab and CFSE were purchased from Invitrogen. Annexin-V, 7-AAD kit, and collagenase were also purchased from BD Biosciences.

CD3-Ab injection

For CD3-specific Ab treatment, mice received a single i.p. injection of CD3-specific Ab (BD Bioscience; clone 145–2C11, NA/LE) as indicated. Some mice received a single dose of nonmitogenic CD3-specific IgG3 Ab (50 µg per mouse) (22). In some experiments, IFN-γ- or TNF-specific Abs (20–100 µg per mouse i.p.; BD Biosciences, NA/LE) or purified IL-6- or IL-17-specific Abs (100 µg per mouse i.p.; eBioscience) were injected before CD3-specific Ab treatment.

Cytokine assays

Cytokines in plasma and culture supernatants were quantified by ELISA using commercial kits following manufacturer’s instructions (kits: IFN-γ, BioLegend; TNF, eBioscience; TGF-β1, Promega; IL-2, BioSource International). Plasma was obtained after centrifugation of blood collected on EDTA-coated tubes after retro-orbital bleeding of isoflurane-anesthetized mice (22).

Isolation and culture of DCs and macrophages

DCs and macrophages were isolated from spleens using Miltenyi Biotec isolation kits following manufacturer’s instructions (22). For culture and apoptotic cell uptake assay, DCs or macrophages were plated into 24-well plates (0.1–1 × 10⁶ cells/well) and incubated in the absence or presence of apoptotic cells (5:1 ratio) with or without LPS (10 ng/ml; Sigma-Aldrich) overnight in complete DMEM. Supernatants were then collected for cytokine quantification. Apoptotic cells were obtained from thymocytes of naive wild-type or TGF-β1^−/− mice after gamma-irradiation (1500 Rad) followed by a 4–8 h incubation in complete DMEM at 37°C, 5% CO₂. In some experiments, thymocytes were labeled with CFSE at 2.5 µM at 37°C for 10 min and then washed three times in cold DMEM complete medium. The thymocytes were then irradiated and cultured at 37°C 5% CO₂ overnight. For CFSE-labeled apoptotic cell-uptake assay, the freshly isolated splenic DCs and macrophages were incubated with apoptotic cells for 4–5 h. The cells were then stained with CD11c- or F4/80-Abs and analyzed with flow cytometry.

Phagocyte depletion and adoptive transfer

Dichloromethylenediphosphonic acid (clodronate)-loaded or PBS-loaded liposomes (Encapsula NanoSciences) were injected i.p. or i.v. into mice (200 –300 µl/mouse) as previously described (22, 36). For adoptive transfer experiments, 10 –20 × 10⁶ spleen-isolated and enriched wild-type F4/80⁺ and CD11c⁺ cells were washed and injected in PBS (i.p.) into TGF-β1^−/− mice one night before anti-CD3 Ab injection. Survival after CD3-specific Ab injection was assessed in the wild-type phagocytes-reconstituted TGF-β1^−/− mice.

Flow cytometry analysis

Flow cytometry staining was performed as described before (22, 37). Spleen, lymph nodes, and whole blood cells were isolated at the indicated time points post-CD3-specific Ab injection and subjected to flow cytometry staining. For isolation of leukocytes from lungs, the lungs were cut into small pieces and placed in 5-ml digestion solution containing 2 mg/ml collagenase in PBS. Following digestion at 37°C for 35 min, the solution was passed through a 70-µm cell strainer. The elucidated cells were washed once in PBS, placed in 38% Percoll solution and centrifuged for 20 min at 2000 rpm. Intracellular expression of TNF, IFN-γ, and IL-2 was determined as described before (22, 37, 38), together with staining of surface markers in the indicated cell types. For analysis of apoptotic cells, spleen, lymph nodes, and blood were collected from the CD3-Ab-treated mice and the mononuclear cells were stained with CD4, CD8, or B220 surface markers, followed by apoptotic cell staining with Annexin-V/7-AAD kit (BD Biosciences). The stained cells were analyzed by flow cytometry within 1 h.

Statistical analysis

Group comparisons of parametric data were made by two-tail Student’s t test unless otherwise stated. A p value < 0.05 was considered significant.

Results

A single injection of sublethal dose of CD3-specific Ab kills TGF-β1^−/− mice

One of the major side effects in CD3-specific Ab therapy in patients and experimental animals is the flu-like syndrome caused by the transient increase in systemic proinflammatory cytokines. However, the syndrome is self-limited and resolves in a few days. To test whether TGF-β was involved in this spontaneous resolution of the syndrome, we injected intact CD3-specific Ab into TGF-β1^−/− mice. Unexpectedly, i.p. injection of a sublethal dose (50 µg) of intact CD3-specific Ab killed all TGF-β1^−/− mice within 24 h, whereas none of the age-matched wild-type mice died (Table I). The death was specifically associated with TCR engagement, as injection of the same amount of isotype control Ab (hamster IgG) failed to kill any TGF-β1^−/− mice (Table I). Moreover, injection of CD4- plus CD8-specific Abs that depleted T cells without stimulation of the TCR failed to kill TGF-β1^−/− mice (Table I). The death was also not associated with potential contamination of endotoxin in the Ab preparation, as the CD3-specific Ab is in the low endotoxin format and injection of the TGF-β1^−/− mice with 500 ng of LPS alone failed to kill the mice (data not shown). Because TGF-β1^−/− mice develop a rapid T cell activation and inflammation around 3– 4 wk of age (25, 26, 29, 39), we injected CD3-specific Ab into 10-day-old mice before any symptoms, which again killed all TGF-β1^−/− mice (Table I). Even more strikingly, as little as 10 µg of CD3-specific Ab was lethal to TGF-β1^−/− mice (Table I). Interestingly, injection of the nonmitogenic/low T cell-depleting form of CD3-specific Ab (CD3-IgG3, 50 µg per mouse) failed to kill TGF-β1^−/− mice (Table I). Overall, a sublethal dose of intact (mitogenic) CD3-specific Ab kills TGF-β1^−/− mice irrespective of age and preexisting inflammation. This indicates a critical role of TGF-β1 in preventing the lethal effect associated with CD3-specific Ab treatment, which may involve control of the increase in systemic inflammatory cytokines.

Table I.

Lethal effect of CD3-specific Ab on TGF-β1^−/− mice^a

Mice (TGF-β1)	Age (days)	Injection (i.p.)	Ab (µg/mouse)	Survive/Total (mice)
Expt. 1
+/+	14	PBS	–	2/2
−/−	14	PBS	–	2/2
+/+	14	CD3	50	2/2
−/−	14	CD3	50	0/2
Expt. 2
+/+	10	Contrl-Ab	50	2/2
−/−	10	Contrl-Ab	50	2/2
+/+	10	CD3	50	4/4
−/−	10	CD3	50	0/4
Expt. 3
+/+	10–12	CD3	10	3/3
−/−	10–12	CD3	10	0/4
Expt. 4
+/+	11–12	CD3	10	2/2
−/−	11–12	CD3	10	0/2
+/+	11–12	CD4 + CD8	100 + 100	2/2
−/−	11–12	CD4 + CD8	100 + 100	2/2
Expt. 5
+/+	14–16	CD3-IgG3	50	4/4b
−/−	14–16	CD3-IgG3	50	4/5b

^aAbs were injected into each mouse (i.p.) in total 50 µl PBS, and mice were monitored for 24 h. +/+, TGF-β1^+/+; −/−, TGF-β1^−/− mice; Contrl-Ab, hamster IgG.

^bThe mice lived for 10 days after the experiment was terminated.

Sustained high levels of circulating proinflammatory cytokines are responsible for the death of CD3-specific Ab-injected TGF-β1^−/− mice

We investigated the reason for the death of the TGF-β1^−/− mice after CD3-specific Ab treatment. The lethal effect did not result directly from the rapid massive death of the liver cells as previously observed in Fas-specific Ab injection in normal mice (40), because the majority of liver cells were viable and functional, as demonstrated by normal levels of a panel of enzymatic activity including GOT/AST and GPT/ALT (data not shown). We then examined the circulating levels of proinflammatory cytokines in TGF-β1^−/− mice following CD3-specific Ab injection. In contrast to the wild-type control mice (7, 22, 41), CD3-specific Ab injection resulted in much higher levels of inflammatory cytokines, including IL-2, IFN-γ, and TNF in TGF-β1^−/− mice, as quickly as 90 min after infusion (Fig. 1, a–c), supporting the notion of TCR hypersensitivity in T cells lacking TGF-β1 (29, 39, 42). Unexpectedly, however, the levels of circulating cytokines, especially TNF, remained elevated and failed to diminish in the TGF-β1^−/− mice (Fig. 1, a–c). By 6 h after Ab injection, the TNF level was still elevated in the circulation of TGF-β1^−/− mice, whereas it was almost undetectable in the control mice (Fig. 1c) (6). This suggests the absence of a regulatory loop leading to the resolution of inflammation. To determine whether the sustained high levels of inflammatory cytokines were responsible for the death of TGF-β1^−/− mice, neutralizing Abs against TNF and/or IFN-γ were co-administered with CD3-specific Ab. Interestingly, neither anti-TNF nor anti-IFN-γ Abs alone rescued the TGF-β1^−/− mice from death, but a combination of the two Abs protected TGF-β1^−/− mice from death (Fig. 1d). Intriguingly, neutralization of IL-6 and IL-17, two inflammatory cytokines recently suggested as being critical mediators for several autoimmune diseases and inflammation (38, 43, 44), with specific Abs failed to prevent the CD3-induced death of TGF-β1^−/− mice (100% death within 24 h, n = 3 mice, data not shown). Consistent with the critical role of TNF in the death of the TGF-β1^−/− mice, there was no reduction of systemic TNF in the TGF-β1^−/− mice treated with CD3-specific Ab together with neutralizing IL-6 and IL-17 Abs (TNF: 2829 ± 2193 pg/ml, n = 3 mice, 6 h) compared with those mice injected with CD3-Ab alone (Fig. 1c). In addition, treatment of TGF-β1^−/− mice with control hamster IgG did not result in a significant upregulation of activation-associated markers (e.g., CD80, CD86, and MHC class II) on macrophages and DCs 6 h after injection compared with PBS injection (data not shown). The data collectively indicate that CD3-specific Ab causes sustained high levels of proinflammatory cytokines in the absence of TGF-β1 in vivo, particularly TNF, which are required for the death of the TGF-β1^−/− mice. The data also suggests TGF-β1 as a critical cytokine in limiting the uncontrolled release of the cytokines in normal mice treated with CD3-specific Ab.

FIGURE 1.

CD3-specific Ab injection induced higher and sustained levels of circulating inflammatory cytokines in TGF-β1^−/− mice, leading to their death. a–c, IL-2 (0 h, n = 3 wild-type and 3 knockout; 1.5 h, n = 8 wild-type and 5 knockout; 6 h, n = 11 wild-type and 8 knockout; 24 h, n = 4 wild-type mice), IFN-γ (0 h, n = 4 wild-type and 3 knockout; 1.5 h, n = 8 wild-type and 5 knockout; 6 h, n = 11 wild-type and 8 knockout; 24 h n = 3 wild-type mice) and TNF (0 h, n = 3 wild-type and 2 knockout; 1.5 h, n = 7 wild-type and 5 knockout; 6 h, n = 11 wild-type and 7 knockout; 24 h, n = 3 wild-type mice) cytokines were quantified (mean ± SD) by ELISA in the plasma of wild-type (+/+,□) or TGF-β^−/− (−/−,■) mice before (0 h) and after CD3-specific Ab (i.p.) injection (1.5, 6, and 24 h).*, p < 0.05 vs wild-type mice. d, Survival of wild-type (+/+) control mice (n = 9) compared with survival of TGF-β1^−/− (−/−) mice 24 h after receiving CD3-specific Ab injection with or without anti-TNF and/or anti-IFN-γ neutralizing Abs (n = 3 to 10 mice per group). **, p < 0.01 (two tail Fisher’s exact test).

Differential requirement for TGF-β regulation of IFN-γ production between CD4⁺ and CD8⁺ T cells in vivo

The next question to address was the cellular origin of inflammatory cytokines. We first studied T cell cytokines in vivo in response to CD3-specific Ab treatment in TGF-β1^−/− mice. We focused on IFN-γ and TNF production by T cells, because these cytokines were responsible for the death of the null mice (Fig. 1). Indeed, T cells in CD3-specific Ab-treated TGF-β1^−/− mice produced much higher levels of IFN-γ and TNF than did T cells in the wild-type control mice (Fig. 2). Unexpectedly, however, we observed that CD4⁺ and CD8⁺ T cells exhibited distinct profile of IFN-γ expression between TGF-β1^−/− and wild-type mice after in vivo injection of CD3-specific Ab. When analyzed directly, ex vivo 2 h after CD3-specfic Ab injection without other restimulation, CD4⁺ T cells in the spleens of the wild-type control mice showed minimal number of IFN-γ-producing cells (1–2%). However, CD8⁺ T cells in the same mice exhibited substantially higher number of IFN-γ⁺ cells (20–30%) (Fig. 2a, left, and b). In marked contrast, both IFN-γ⁺ CD4⁺ (20–30%) and IFN-γ⁺ CD8⁺ (40–50%) T cells were detected in the spleens of TGF-β1^−/− mice (Fig. 2a, right, and b). This corresponds to a 20–30- and a 2–3-fold increase in IFN-γ⁺ cells in TGF-β1 null CD4⁺ and CD8⁺ T cells, respectively. As a negative control for TCR stimulation, both CD4⁺ and CD8⁺ T cells from wild-type control and TGF-β1^−/− mice treated with PBS had only background IFN-γ positive cells (≤1%, data not shown). A similar trend was also observed for TNF (Fig. 2, a and b) and IL-2 (Fig. 2b) expression in peripheral CD4⁺ and CD8⁺ T cells between TGF-β1^−/− and control mice, but the extent of the difference was less dramatic than that observed for IFN-γ. This phenomenon of distinct expression of IFN-γ between CD4⁺ and CD8⁺ T cells occurred even more notably when the mice were examined 6 h after Ab treatment (Fig. 2, c and d). Although CD8⁺ T cells in the wild-type mice showed similar numbers of IFN-γ⁺ cells as in TGF-β1^−/− mice, CD4⁺ T cells in the same wild-type mice still exhibited much less IFN-γ⁺ (2–3%) than that in the TGF-β1^−/− mice (40–50%) (Fig. 2, c and d). Consistent with the published data (6), neither CD4⁺ nor CD8⁺ T cells in the wild-type control mice had detectable TNF staining 6 h after CD3-specific Ab injection (Fig. 2c). Notably, TGF-β1^−/− T cells in the spleen also showed no TNF staining at this time (Fig. 2c). Similar trends were also observed in blood and lymph nodes (data not shown). To address whether the differential response between CD4⁺ and CD8⁺ T cells in response to CD3-specific Ab in vivo was due to a different frequency of memory CD4⁺ vs CD8⁺ T cells, we examined the effector/memory T cells in the wild-type and TGF-β1^−/− mice. As expected, more of the TGFβ1^−/− CD4⁺ and TGFβ1^−/− CD8⁺ T cells exhibited an effector/memory cell phenotype than did wild-type control T cells. These results might be responsible for the overall high levels of IFN-γl production in the knockout mice in response to CD3 Ab injection. In addition, we observed that there were comparable effector/memory cells (CD44^highCD62L⁻) among the CD4⁺ and CD8⁺ T cells within the wild-type or TGF-β1^−/− mice (data not shown). To exclude the possibility that this differential profile between CD4⁺ and CD8⁺ T cell response to TCR was ascribed to their difference in migration to other tissues, we analyzed the cytokine profile of CD4⁺ and CD8⁺ T cells in the lungs (Fig. 2, e and f) and livers (data not shown) and observed a similar trend. To further validate this observation, we injected CD3-specific Ab into conditional knockout mice with T cell-specific deletion of TGF-β receptor I (Tgfbr^f/f-Lck-Cre⁺) (35). Similar results were observed (Fig. 2g). Thus, these data reveal a previously unrecognized requirement for TGF-β regulation of IFN-γ production between CD4⁺ and CD8⁺ T cells in response to TCR stimulation in vivo, i.e., production of IFN-γ after TCR signaling is less dependent on TGF-β regulation in CD8⁺ T cell than in CD4⁺ T cells.

FIGURE 2.

Different requirement of TGF-β1 for IFN-γ production between CD4⁺ and CD8⁺ T cells in response to CD3-specific Ab treatment in vivo. a, Flow cytometry of splenic CD4⁺ (top row) and CD8⁺ (bottom row) T cells in wild-type control (+/+) and TGF-β1^−/− (−/−) mice (2 h after Ab injection). Numbers in quadrants indicate percent TNF⁺ (top left), TNF⁺IFN-γ⁺ (top right), and IFN-γ⁺ (bottom right) cells. Each plot represents one of two mice. b, Percent IL-2⁺, IFN-γ⁺, or TNF⁺ cells (mean ± SEM) in the indicated CD4⁺ or CD8⁺ T cells in the mice in a. **, p < 0.01. The experiment in a and b was repeated twice with similar results with a total of four mice in each group). c, Flow cytometry of splenic CD4⁺ (top row) and CD8⁺ (bottom row) T cells in wild-type control and TGF-β1^−/− mice (6 h). Numbers in quadrants indicate the percentage of positive cells, as in a. Each plot represents one of three to four mice. d, Percentage of IFN-γ⁺ cells (mean ± SEM) in the mice in c. *, p < 0.05. Data in c and d represent three independent experiments. e, Flow cytometry of CD4⁺ (top row) and CD8⁺ (bottom row) T cells in lungs (6 h). Each plot represents one of four mice. f, Percentage of IFN-γ⁺ cells (mean ± SEM) of the mice in e. ***, p < 0.001. Data represent two independent experiments. g, Percentage of IFN-γ⁺ cells (mean ± SEM) in CD4⁺ or CD8⁺ T cells in the spleens of Tgfbr^f/f-Lck-Cre⁺ (open bar, n = 2) and Tgfbr^f/f-Lck-Cre⁻ (black bar, n = 2) mice 6 h post CD3-specific Ab injection. *, p < 0.05.

TGF-β1^−/− macrophages and DCs produce TNF in vivo upon CD3-specific Ab injection

Because circulating TNF still remained at high levels in TGF-β1^−/− mice 6 h after CD3-specific Ab injection (Fig. 1c), yet the TGF-β1^−/− T cells showed undetectable TNF⁺ cells in spleens (Fig. 2c) and minimal TNF⁺ cells in lungs (Fig. 2e), we reasoned that other non-T immune cells might also secrete TNF in CD3-treated TGF-β1^−/− mice. F4/80⁺ macrophages and CD11c⁺ DCs were isolated from spleens of TGF-β1^−/− and wild-type control mice injected with CD3-specific Ab 6 h earlier. TNF protein in single cells was immediately examined by flow cytometry ex vivo without further stimulation. F4/80⁺ macrophages and CD11c⁺ DCs in the wild-type control mice expressed undetectable levels of TNF protein (data not shown), consistent with published results that TNF was exclusively produced in T cells in normal mice in response to CD3-specific Ab injection (1). Even when TGF-β heterozygous littermates (TGF-β1^+/−) were injected with CD3-specific Ab (50 µg), all mice survived and macrophages showed marginal levels of intracellular TNF protein (Fig. 3a). In marked contrast, both F4/80⁺ macrophages and CD11c⁺ DCs exhibited higher levels of intracellular TNF protein in the spleens of CD3-specific Ab-treated TGF-β1^−/− mice (Fig. 3a). TGF-β1^−/− macrophages and DCs in the lungs also produced substantially higher levels of TNF than did the macrophages and DCs in the wild-type control mice (Fig. 3b). Thus, CD3-specific Ab treatment not only stimulates T cells immediately, but also drives DCs and macrophages to produce TNF at late stage in TGF-β1^−/− mice. This suggests that the lethal inflammation observed in the TGF-β1^−/− mice after CD3-specific Ab treatment may also result from macrophage and DC dysregulation.

FIGURE 3.

Phagocytes play a crucial role in CD3-specific Ab-induced death of TGF-β1^−/− mice. a, Flow cytometry of TNF⁺ cells in splenic macrophages (F4/80⁺) and DCs (CD11c⁺) in TGF-β1^+/− (+/−) and TGF-β1^−/− (−/−) mice (6 h). A representative profile of one mouse from each group is shown. The experiment was repeated five times with similar results. b, Percentage of TNF⁺ cells (mean ± SEM) in macrophages (F4/80⁺) and DCs (CD11c⁺) in lungs (6 h after Ab injection). Results are pooled from two independent experiments (n = 4 mice/group). c, Efficient depletion of CD11c⁺ DCs and F4/80⁺ macrophages was determined by flow cytometry in the spleen of TGF-β1^−/− mice after clodronate- (lower panels) or PBS-loaded (upper panels) liposome injection. A representative of three independent experiments is shown. d, Survival of the wild-type (+/+) control or TGF-β1^−/− (−/−) mice injected with clodronate- or PBS-loaded liposomes followed by CD3-specific Ab (i.p.) injection. e, Circulating TNF levels (mean ± SEM) in the mice (n = 3–5) injected with CD3-specific Ab with (clodronate) or without (PBS) depletion of phagocytes (6 h after Ab injection). *, p < 0.05 (one tailed t test). Data represent two independent experiments. f, Adoptive transfer of wild-type phagocytes (APC +/+, n = 3) or PBS (n = 3) was performed a night before TGF-β1^−/− mice were injected with CD3-specific Ab. g, Tgfbr1^f/f-Lck-Cre⁺ and Tgfb1^f/f-Lck-Cre⁻ mice (5–6 wk old, n = 2) were injected i.p. with CD3-specific Ab, 50 µg per mouse. Results in d, f, and g are expressed in percentage of surviving mice 24 h after CD3-specific Ab injection. Numbers of mice are given in each graph.

Phagocytes play a critical role in CD3-specific Ab-induced death of TGF-β1^−/− mice

Having determined that TGF-β1^−/− macrophages and DCs produced TNF after CD3-specific Ab injection, we then studied the role of these phagocytes in the death of the TGF-β1^−/− mice. Several sets of experiments were performed to address this question. Firstly, we depleted macrophages and DCs with clodronate-loaded liposomes in TGF-β1^−/− mice before CD3-specific Ab administration (Fig. 3c). Efficient depletion of endogenous phagocytes in TGF-β1^−/− mice surprisingly rescued them from the immediate killing by CD3-specific Ab; the treated TGF-β1 null mice (three of five vs zero of six mice treated with PBS) survived for at least 24 h (Fig. 3d). No death was observed in TGF-β1^−/− or the wild-type control mice receiving clodronate-loaded liposomes alone (i.e., without CD3-specific Ab injection, data not shown). Importantly, deletion of phagocytes before CD3-specific Ab injection significantly decreased the systemic TNF levels in the TGF-β1^−/− mice compared with those mice treated with CD3-specific Ab alone (Fig. 3e), further confirming that phagocytes are a critical cellular source of TNF. To further assess the role of TGF-β1^−/− innate immune cells in the death of the knockout mice, we injected recombinant bioactive TGF-β1 (1 ng per mouse) immediately before CD3-specific Ab administration and observed that exogenous active TGF-β1 failed to protect the TGF-β1^−/− mice from death (data not shown). The data are consistent with the failure of prolonging the life span of TGF-β1^−/−(TG) mice and emphasize the importance of TGF-β production (capacity) by immune cells in protecting the mice from lethal cytokine storm.

We next investigated whether the wild-type phagocytes had a protective function for the null mice receiving CD3-specific Ab. We adoptively transferred splenic macrophages and DCs cells enriched from naive wild-type mice into the TGF-β1^−/− mice and injected these mice with CD3-specific Ab 10 h later. Intriguingly, the TGF-β1^−/− mice reconstituted with wild-type phagocytes survived for at least 24 h after receiving CD3-specific Ab (Fig. 3f). Moreover, we injected intact CD3-specific Ab (50 µg/per mouse) into Tgfbr1^f/f-Lck-Cre⁺ mice (35), in which TGF-β signaling was deleted only in T cells without affecting their macrophages and DCs. We observed that CD3-specific Ab failed to kill the Tgfbr1^f/f-Lck-Cre⁺ mice (Fig. 3g). Finally, when neutralization Ab against active TGF-β1,2,3 was injected together with CD3-specific Ab into the wild-type mice, no mice died (data not shown). These data are collectively indicate a critical function of TGF-β1^−/− phagocytes in contributing to the uncontrolled cytokine storm and consequent death of TGF-β1^−/− mice injected with CD3-specific Ab.

Clearance of apoptotic T cells fails to inhibit TNF in TGF-β1^−/− macrophages and DCs

We investigated how TGF-β1^−/− macrophages and DCs produced TNF in response to CD3-antibody injection. In normal mice, CD3-specific Ab induces apoptotic T cells that are engulfed and ingested by macrophages and immature DCs to produce TGF-β. This production of TGF-β subsequently inhibits autocrine and/or paracrine TNF production in cultures (30, 31) and contributes to CD3-specific Ab-induced tolerance in vivo (22). We hypothesized that the TNF production in TGF-β1^−/− macrophages and DCs might be associated with their inability to suppress this inflammatory cytokine despite apoptotic cell clearance. To test this hypothesis, we first examined T cell apoptosis in CD3-specific Ab-treated TGF-β1^−/− mice. Consistent with our previous report (29), intact CD3-specific Ab treatment of TGF-β1^−/− mice resulted in considerably more detectable T cell apoptosis in vivo than that in the control mice (Fig. 4a). Thus, in the absence of TGF-β1, CD3-specific Ab treatment induces more T cell apoptosis in vivo.

FIGURE 4.

TGF-β1^−/− macrophages and DCs fail to decrease LPS-induced TNF after contact with apoptotic T cells in cultures. a, Flow cytometry of splenic T cells isolated from TGF-β1^+/− (+/−, top row) and TGF-β1^−/− (−/−, bottom row) mice 6 h after CD3-specific Ab injection (n = 2–3 mice). Left column, profile of CD4 vs CD8 in spleen cells; right columns, profile of Annexin-V vs 7-AAD positive cells in gated CD4⁺ or CD8⁺ T cells. The percentage of earlier apoptotic cells (Annexin-V⁺/7-AAD−) or late apoptotic/dead cells (Annexin-V⁺/7-AAD⁺) are given on each graph. Each plot represents one of two to three mice. b, Flow cytometry of macrophages (top row) and DCs (bottom row) incubated with apoptotic thymocytes (prelabeled with CFSE) in cultures for 5 h. Numbers in quadrants indicate percentage of F4/80⁺CFSE⁺ (macrophages containing apoptotic cells) or CD11c⁺CFSE⁺ (DCs containing apoptotic cells). Each plot represents the cells pooled from two mice before culture. c, Supernatants from DCs and macrophages (MØ) isolated from spleens of wild-type (+/+) and TGF-β1^−/− (−/−) mice cultured overnight in medium (Med), apoptotic cells (Apo), LPS, or apoptotic cells and LPS (Apo/LPS) were quantified for the presence of TNF by ELISA (mean ± SD). The experiments in this figure were repeated at least twice with similar results. Note: the scales are different between panels.

We then tested the TNF production from TGF-β1^−/− macrophages and DCs after exposure to apoptotic cells. We first examined the ability of phagocytes to clear apoptotic cells and observed that both TGF-β1^−/− and wild-type macrophages and DCs have a similar capacity of apoptotic cell uptake (Fig. 4b). We then incubated splenic macrophages and CD11c⁺ DCs from TGF-β1^−/− and wild-type mice with gamma-irradiation-induced apoptotic thymocytes and stimulated them with LPS. Irradiated neonatal TGF-β1^−/− thymocytes were used as apoptotic cell source to eliminate an endogenous TGF-β release from apoptotic T cells (30). As expected and consistent with the published studies (22, 30, 31), LPS-induced TNF production was substantially inhibited in the wild-type macrophages and DCs upon ingestion of apoptotic cells (Fig. 4c). This is associated with the up-regulation of TGF-β after apoptotic cell uptake (22, 30, 31, and our unpublished data). In marked contrast, TGF-β1^−/− macrophages and DCs showed no decrease or increase in TNF in response to LPS stimulation when exposed to apoptotic cells (Fig. 4c). Of note, substantially higher levels of TNF were observed in TGF-β1^−/− phagocytes than in the wild-type phagocytes (Fig. 4c). We then studied a possible link between contact and uptake of apoptotic cells by TGF-β1^−/− phagocytes and their activation state. We incubated freshly isolated splenic CD11c⁺ DCs with apoptotic thymocytes for 5 h in the absence of additional stimuli and analyzed the expression of their activation markers. Intriguingly, TGF-β1^−/− DCs that were exposed to apoptotic cells increased the expression of CD80 (mean fluorescence intensity (MFI): Medium 457 vs plus apoptotic cells 495), CD86 (MFI: Medium 419 vs plus apoptotic cells 527), and CD83 (percent positive: Medium 12.1% vs plus apoptotic cells 26.5%) molecules compared with those DCs cultured without apoptotic cells, whereas the wild-type DCs decreased the levels of these molecules after contact with apoptotic cells (CD80 MFI: Medium 459 vs plus apoptotic cells 355; CD86 MFI: Medium 432 vs plus apoptotic cells 386; CD83 positive cells: Medium 17.7% vs plus apoptotic cells: 10.6%). These data collectively indicate that TGF-β1^−/− phagocytes, after contact with apoptotic cells, increase rather than inhibit their activation status and consequently fail to counter TNF production. This may provide a link between the increase in apoptotic T cells and sustained high levels of TNF observed in the TGF-β1^−/− mice injected with CD3-specific Ab.

Discussion

In this paper, we show that a single sublethal dose of intact CD3-specific Ab injection into TGF-β1^−/− mice kill all the null mice. This accelerated death required optimal CD3-induced TCR/CD3 complex stimulation, because CD4- plus CD8-specific Abs or non-mitogenic CD3-specific Ab did not induce the death of TGF-β1^−/− mice. We demonstrated that the death of TGF-β1^−/− mice was related to sustained high levels of systemic inflammatory cytokines, TNF and IFN-γ. Significantly, TNF was found to be produced not only by T cells as observed in normal mice, but also by macrophages and DCs in TGF-β1^−/− mice. High levels of TNF are associated with the lack of TGF-β regulation in TGF-β1^−/− macrophages and DCs during the clearance of a high number of apoptotic T cells caused by CD3-specific Ab injection. In addition, we demonstrated a previously unrecognized distinct requirement for TGF-β in regulation of IFN-γ production between CD4⁺ and CD8⁺ T cells in response to TCR stimulation in vivo. These findings not only emphasize a key role for TGF-β1 production in immune cells in preventing uncontrolled inflammatory cytokine secretion, but also provide a novel cellular and molecular explanation for the early onset of lethal inflammation in TGF-β1^−/− mice. Moreover, these data also support a critical function of TGF-β in the resolution of the flu-like syndrome after CD3-specific Ab treatment in mice, although effects of the Ab treatment in humans are likely to be more complex.

Several conclusions can be drawn from the present study. Firstly, the unexpected lethal effect in TGF-β1^−/− mice after a sublethal dose of intact CD3-specific Ab has proven a key role of TGF-β in preventing and/or suppressing potential T cell activation and consequent lethal inflammatory storm in response to CD3-specific Ab-treatment. The data indicates that TGF-β plays a critical role in the spontaneous resolution of flu-like syndrome in CD3-Ab-treated normal animals. The sudden over-produced and persistent inflammatory cytokines, including TNF and IFN-γ, in circulation induced by anti-CD3 injection is the main cause for the death of TGF-β1^−/− mice, because neutralization of TNF and IFN-γ could rescue the TGF-β1^−/− mice.

Secondly, this reveals that CD4⁺ and CD8⁺ T cells may have a distinct response to TGF-β1 regulation in vivo. Although CD8⁺ T cells may readily produce IFN-γ in response to TCR signaling despite the presence of TGF-β1, CD4⁺ T cells are more restrained by the TGF-β1 regulation. The different profiles of IFN-γ production between CD4⁺ and CD8⁺ T cells in CD3-specific Ab-treated mice is surprising because the current paradigm believes that both subsets of T cells are indistinguishable in response to TGF-β signaling, at least in vitro (12, 15). Although mechanistically elusive, this data may reveal a previously unrecognized different sensitivity to TGF-β regulation between CD4⁺ and CD8⁺ T cells, in which TGF-β restrains CD4⁺ rather than CD8⁺ T cells. These data provide a link between two independent reports demonstrating that inflammation in TGF-β1^−/− mice is both CD4- and IFN-γ-dependent (45, 46). Alternatively, CD3-specific Ab treatment in the presence of endogenous TGF-β1 converts CD4⁺ toward Foxp3⁺ regulatory T cells (20, 22, 23, 37) and TGF-β has been shown to prevent Th1 CD4⁺ T cell differentiation (IFN-γ is a major Th1 cytokine).

Thirdly, TGF-β1^−/− macrophages and DCs produce TNF and contribute to the inflammatory storm in response to intact CD3-specific Ab injection. In wild-type mice, intact CD3-specific Ab treatment results in a transient elevation of inflammatory cytokines, including IL-2, IFN-γ, TNF, and IL-6, in the circulation, which are exclusively produced from T cells (6). Nevertheless, these T cell-derived inflammatory cytokines are rapidly down-regulated a few hours after CD3-antibody injection (2). Instead, systemic TGF-β is increased in CD3-Ab-treated normal mice by 24 h (22). It is known that TGF-β1^−/− T cells are overly responsive to TCR stimulation (15, 29, 39, 47), which may account for the first wave of 20-to 70-fold increase in the aforementioned inflammatory cytokines in circulation 1–3 h post-CD3-specific Ab injection in TGF-β1^−/− mice. However, the sustained high levels of systemic TNF for more than 6 h after CD3-specific Ab treatment in TGF-β1^−/− mice indicate that other non-T cells may also contribute to this persistent TNF production. In support of this idea, we have observed that both macrophages and DCs in TGF-β1^−/− mice produce TNF in vivo when directly tested 6–8 h after CD3-antibody injection. In contrast, wild-type phagocytes show undetectable TNF. Significantly, deletion of endogenous phagocytes in TGF-β1^−/− mice rescues them from CD3-specific Ab-induced death with a dramatic decrease in the circulating levels of TNF. Moreover, adoptive transfer of wild-type phagocytes into TGF-β1^−/− mice before CD3-specific Ab injection also rescues TGF-β1^−/− mice. Finally, the conditional knockout mice with T cell-specific deletion of TGF-β signaling were resistant to CD3-specific Ab-induced death, because the phagocytes in these mice were intact in response to TGF-β regulation. Taken together, the data strongly suggest that the production and the process of TGF-β1 secretion by macrophages and DCs are indispensable in preventing and controlling the potential inflammation initiated by CD3-Ab injection, albeit the underlying mechanisms remain to be elucidated. These findings may also provide an explanation why delivery of exogenous active TGF-β protein or even normal levels of active TGF-β1 produced in liver cells in TGF-β1^−/−(TG) mice all fail to prevent aggressive, lethal inflammation and prolong their lifespan (28, 48).

The next immediate question is how macrophages and DCs produce high levels of TNF in TGF-β1^−/− mice after CD3-Ab injection. Our data suggest that the lack of TGF-β1 production by macrophages and DCs in TGF-β1^−/− mice upon exposure to apoptotic T cells is a major cause for the failure of TNF production down-regulation in the TGF-β1^−/− phagocytes in response to exogenous stimuli, although an inherent defect of TGF-β1^−/− phagocytes cannot be completely excluded. Firstly, mitogenic CD3-specific Ab injection results in elevated production of inflammatory cytokines, such as IFN-γ, in TGF-β1^−/− T cells that help activate macrophages and DCs. Secondly, the sudden accumulation of overwhelming apoptotic cells induced by intact CD3-specific Ab in TGF-β1^−/− mice makes both the TGF-β1-deficient macrophages and DCs overloaded in engulfing and digesting them. It should be pointed out that T cell apoptosis alone is insufficient to initiate this lethal cascade, as injection of CD4⁻- and/or CD8-specific Abs that deplete T cells but do not engage TCR also fails to kill TGF-β1^−/− mice. In contrast to the normal phagocytes producing TGF-β that restrain TNF production (22, 30, 31, 49–51) after ingestion of apoptotic cells, TGF-β1^−/− phagocytes lack this regulatory force, and thus fail to prevent up-regulation of their costimulatory molecules or to inhibit TNF secretion. This idea is supported by our observation that TGF-β1^−/− macrophages and DCs cannot be suppressed to produce LPS-induced TNF in cultures after exposure to apoptotic cells. In addition, injection of the same amounts of nonmitogenic CD3-IgG3 Ab that is known to suboptimally activate the TCR and induce significantly lower numbers of T cell apoptosis (1, 2, our unpublished data) fails to kill the TGF-β1^−/− mice. Taken together, we propose that three major factors are responsible for the systemic inflammatory cytokine storm that causes the death of TGF-β1^−/− mice following CD3-specific Ab treatment: TCR hypersensitivity (initiating factor and effector force) (52), increase in T cell apoptosis (driving force), and failure of TGF-β1 production by phagocytes after exposure to apoptotic T cells leading to a sustained secretion of TNF (effector force). The data obtained with CD3-specific Ab injection in the TGF-β1^−/− mice thus help envision that similar events and consequences may occur in the unmanipulated TGF-β1^−/− mice that are exposed to foreign or self-Ags in an accumulated and accelerated manner after they are born.