The role of the Th1 transcription factor T-bet in a mouse model of immune-mediated bone-marrow failure

Human aplastic anemia (AA) is characterized by pancytopenia and bone marrow (BM) hypoplasia.^1-4  In most cases, AA is an immune-mediated disease, with active destruction of hematopoietic stem and progenitor cells by T lymphocytes.^1,5  The aberrant immune response may be triggered by drugs, viruses, or chemical exposure, but in the majority of cases the etiology is unknown.^1-3  That most patients respond to treatments with antithymocyte globulin and other immunosuppressive agents has provided powerful evidence of the role of the immune system in the pathophysiology of AA.^6-9  Excessive production of interferon gamma (IFN-γ), tumor necrosis factor-α (TNF-α), and interleukin-2 (IL-2) from patients' T cells suggests that hematopoietic cells are destroyed through a Th1 T-cell response.^10,11 

We recently observed that the Th1 transcription factor T-bet was elevated in peripheral blood T cells from patients with AA,¹²  suggesting that the up-regulated T-bet signaling may contribute to the immune destruction of BM cells. T-bet is a member of the T-box family of transcription factors. Members of this family of proteins each contains a highly conserved DNA binding domain, the T-box, which binds to a specific sequence in the promoter region of different genes. T-bet is found in Th1 but not in Th2 cells and is the key regulator of Th1 development and function.¹³  Mice lacking T-bet failed to develop Th1 cells and were driven toward Th2-mediated diseases.¹⁴ 

We postulated that T-bet plays a critical role in the development of AA and BM failure. We have developed murine models that mimic pathologic features of human BM failure by infusing allogeneic lymph node (LN) cells into major histocompatibility complex (MHC) or minor histocompatibility antigen-mismatched recipients.^15-17  Donor lymphocytes infiltrate host BM and expand dramatically, accompanied by the development of severe pancytopenia and marrow hypoplasia. Increased serum IFN-γ concentration in affected animals and the effectiveness of immunosuppressive therapy with anti–IFN-γ antibody strongly suggest that marrow destruction in this model is mediated by Th1 immune responses.^15,16 

In the current study, we tested the impact of T-bet deficiency in BM failure using T-bet knockout (T-bet^−/−) mice as lymphocyte donors in the MHC-mismatch BM failure mouse model. Infusion of T-bet^−/− LN cells failed to produce fatal BM failure because the infused LN cells did not expand and there were fewer cells expressing IFN-γ, indicating that lack of T-bet and resultant Th1 immune responses abolish the ability for allogeneic LN cells to mediate BM destruction.

Inbred C57BL/6 (B6, H2^b/b), hybrid CByB6F1 (F1, H2^b/d), and induced mutant B6.129S6-Tbx21^tm1Glm (T-bet^−/−) mice were purchased from The Jackson Laboratory and were bred and maintained at the National Institutes of Health animal facility with standard care and nutrition. All animal study protocols were approved by the Animal Care and Use Committee of the National Heart, Lung and Blood Institute. Male and female mice were 6 to 16 weeks of age and sex-matched between donors and recipients in each specific experiment. Peripheral blood was collected from the retro-orbital sinus. BM cells were extracted from bilateral tibiae and femurs. Spleen, thymus and inguinal, axillary, and lateral axillary LNs were removed, homogenized with a tissue grinder (A. Daigger & Company) in Iscove modified Eagle media (Quality Biologicals), filtered through 100-μm nylon mesh (Small Parts), washed in Iscove modified Eagle media, and counted by the use of a ViCell XR Counter (Beckman Coulter Inc). Complete blood counts were performed with the use of a Hemavet 950 analyzer (Drew Scientific).

Monoclonal antibodies to murine CD3 (clone 145-2C11), CD4 (clone GK 1.5), CD8 (clone 53-6.72), CD11b (clone M1/70), CD19 (clone ID3), CD34 (clone RAM34), CD45R (B220; clone RA3-6B2), CD95 (Fas; clone Jo2), CD117 (c-Kit; clone 2B8), erythroid cells (clone Ter119), granulocytes (Gr1/Ly6-G; clone RB6-8C5), IFN-γ (clone XMG1.2), and stem cell antigen 1 (Sca1; clone E13-161) were all purchased from BD Biosciences. Each antibody was conjugated with fluorescein isothiocyanate, phycoerythrin, biotin, or allophycocyanin. Fluorescein isothiocyanate–conjugated annexin V was also from BD Biosciences.

For flow cytometry, cells were incubated in Geys solution (130.68mM NH₄Cl, 4.96mM KCl, 0.82mM Na₂HPO₄, 0.16mM KH₂PO₄, 5.55mM dextrose, 1.03mM MgCl₂, 0.28mM MgSO₄, 1.53mM CaCl₂, and 13.39mM NaHCO₃) twice for 10 minutes each to lyse red blood cells. After washing in flow buffer (2.68mM KCl, 1.62mM Na₂HPO₄, 1.47mM KH₂PO₄, 137mM NaCl, 7.69mM NaN₃, and 1% bovine serum albumin), cells were incubated with a premixed antibody cocktail for 30 minutes on ice, washed, resuspended in flow buffer, and analyzed by the use of a LSR II flow cytometer (Becton Dickinson).

For intracellular staining, cells were resuspended in a fixation/permeabilization mixture (Becton Dickinson) and incubated on ice for 30 minutes, then washed with permeabilization buffer (Becton Dickinson), and stained for 30 minutes on ice with antibodies specific for intracellular proteins. Cells were washed in permeabilization buffer and resuspended in flow buffer for flow cytometry.

Recipient F1 mice were preirradiated with 5 Gy total body irradiation (TBI, Shepherd Mark 1 ¹³⁷Cesium gamma source from J. L. Shepherd & Associates) and were injected with MHC-mismatched LN cells (5 × 10⁶ B6 LN or 5-10 × 10⁶ T-bet^−/− LN) through the lateral tail vein 4 to 6 hours after irradiation. All mice were euthanized at 12 to 13 days after LN-cell infusion to collect tissues for analyses.

So that we could specifically study the role of exogenous IFN-γ in rescuing the phenotypes of T-bet^−/−, some animals received TBI + IFN-γ or TBI + T-bet^−/− LN + IFN-γ treatments. In 2 experiments, recombinant mouse IFN-γ (R&D Systems) was injected intravenously at 0.2 μg/mouse at the time of T-bet^−/− LN-cell infusion and then intraperitoneally once per day at 0.2 μg/mouse for 11 days. In 2 further experiments, sublethally irradiated F1 recipients received continuous IFN-γ infusion in addition to the injection of 5 × 10⁶ T-bet^−/− LN through an Osmotic pump (DURECT Corporation), which was implanted under the skin for continuous injection at 0.5 μL/hour for 12 days, rounding an injection of IFN-γ at 300 ng/g per day. Mice were bled and euthanized at day 12 to evaluate marrow damage.

Blood was collected by orbital sinus bleeding into tubes containing 50 μL of 0.5M EDTA (ethylenediaminetetraacetic acid) and centrifuged at 1000g for 10 minutes. Plasma was removed and stored at −20°C. Plasma cytokine concentrations were measured by use of the Multi-Analyte ELISArray Kits (SABiosciences). In brief, 50 μL of plasma or antigen standard (IL-2, -4, -5, -6, -10, -12, -13, -17A, -23, IFN-γ, TNF-α, and transforming growth factor [TGF]-β) was added to each well in a 96-well enzyme-linked immunoassay plate preadded with 50 μL of assay buffer and was incubated for 2 hours at room temperature. Then 100 μL of each detecting antibody was added and incubated at room temperature for 1 hour. One hundred μL/well of avidin–horseradish peroxidase solution was added to the well, and the plate was incubated for 30 minutes at room temperature. The plate was washed 3 times with washing buffer between each step. Afterward, 100 μL of developing solution was added to each well, and the plate was incubated in the dark for 30 minutes at room temperature. The reaction was stopped by the addition of 50 μL/well stopping buffer and analyzed on the Wallac1420 Victor 3 reader (Perkin Elmer) at 450- and 570-nm wavelengths.

All collected data were analyzed by the use of JMP Statistical Discovery Software (SAS Institute Inc) on the “fit Y by X” platform. Statistical significance was claimed at P less than .05 and P less than .01 levels, respectively. Mean comparisons were carried out by use of the “each pair, student t” option after general variance analysis.

We first investigated the effect of T-bet deficiency on mouse lymphohematopoiesis by comparing the cellular composition in blood, BM, spleen, and LN between T-bet^−/− and normal B6 mice. In peripheral blood, T-bet^−/− mice showed a 23% reduction (P < .05) in white blood cells (WBCs), a 31% reduction (P < .01) in lymphocytes, a mild 9% reduction (P < .01) in red blood cells, and no change in neutrophils and platelets in comparison with normal B6 mice (Table 1). In the BM, total number of BM cells and BM Lin⁻Sca1⁺CD117⁺CD34⁻ (KSLCD34⁻) hematopoietic stem and progenitor cells were similar between T-bet^−/− and B6 mice (Table 1). We specifically analyzed the proportions of CD4 and CD8 T cells and found no significant difference between T-bet^−/− and B6 mice in peripheral blood, BM, and spleen. In LN, the CD4-cell percentage was 19% lower (P < .05) in T-bet^−/− (27 ± 1.0%) than in B6 (33 ± 1.7%) mice, whereas the CD8-cell percentage was the same for both genotypes (29 ± 0.9% and 29 ± 1.6%, respectively). Deletion of T-bet appeared to have resulted in a mild decrease in lymphocytes, especially CD4 T cells in the LN, but had no major effect on other lymphohematopoietic cell types.

We then studied the effects of T-bet on lymphocyte function using normal B6 or T-bet^−/− mice as lymphocyte donors in a mouse model of infusion-induced BM failure. Infusion of 5 × 10⁶ normal B6 LN cells into sublethally irradiated, MHC-mismatched F1 recipients produced severe pancytopenia (Figure 1A; Table 2) and BM hypoplasia (Figure 1B) when analyzed at 12 to 13 days after LN cell infusion. However, infusion of the same number of LN cells from T-bet^−/− donors induced no marrow failure: recipient blood and marrow cell numbers were similar to mice that had received TBI only (Figure 1; Table 2). Increasing the number of T-bet^−/− LN-cell infusion to 10 × 10⁶ cells/recipient produced only mild marrow hypoplasia (Table 2; Figure 1). We infused 23 × 10⁶ T-bet^−/− LN cells into one 5-Gy TBI-treated F1 mouse and found no sign of cytopenia in this animal at 12 days after cell infusion: neutrophils (7.2 × 10⁶/mL), red blood cells (6.7 × 10⁹/mL), platelets (550 × 10⁶/mL), and total BM cells (197.5 × 10⁶). Thus, T-bet^−/− LN cells were defective for the induction of marrow failure even at much greater cell doses.

Because a characteristic feature of immune-mediated BM failure is the destruction of hematopoietic cells, we measured hematopoietic stem and progenitor cells in recipient BM using the previously defined KSLCD34⁻ markers. Infusion of B6 LN cells resulted a significant decrease (P < .05) in the number of total BM KSLCD34⁻ cells in comparison with mice that received TBI only (Figure 2), concurrent with the development of severe pancytopenia (Figure 1A). In contrast, recipients of T-bet^−/− LN cells showed significant increases (P < .01) in the proportion and total number of KSLCD34⁻ cells in the BM compared with mice that received TBI only. This observation was consistent in all 16 F1 recipients that received T-bet^−/− LN-cell infusion.

In our mouse model, infused B6 LN cells, but not T-bet^−/− LN cells, expanded markedly in F1 recipient BMs (Figure 3A-B). Average proportions of CD4 (0.4 ± 0.7%, 13 ± 0.7%, 0.4 ± 0.4%, P < .01) and CD8 (0.5 ± 1.7%, 32 ± 1.5%, 1.1 ± 1.4%, P < .01) T cells in recipient BM for TBI only, TBI + B6 LN, and TBI + T-bet^−/− LN treatments showed significant (P < .01) T-cell expansion in recipients of B6 LN cells but not in recipients of T-bet^−/− LN cells. Total BM CD4 and CD8 T cells in the residual BM increased 14-fold (P < .01) and 29-fold (P < .01), respectively, in TBI + B6 LN-cell recipients compared with F1 mice receiving TBI only (Figure 3B). In contrast, there was no expansion of CD4 or CD8 T cells in the recipients of T-bet^−/− LN cells. Both the percentage and the total number of CD4 and CD8 T cells were similar between mice treated with TBI only and mice treated with TBI only + T-bet^−/− LN, indicating that T-bet deficiency abolished the ability of T cells to proliferate in an MHC mismatch recipient environment.

We measured the number of activated T cells producing the inflammatory cytokine IFN-γ. Although the proportions of IFN-γ⁺CD4 and CD8 T cells were not significantly different (Figure 3C), total number of IFN-γ⁺CD4 T cells increased 4.3-fold (9.1 ± 3.7 × 10⁶ vs 2.1 ± 4.0 × 10⁶, P < .05), and total number of IFN-γ⁺CD8 T cells increased 15-fold (7.39 ± 0.63 × 10⁶ vs 0.49 ± 0.68 × 10⁶, P < .01), in TBI + B6 LN recipients in comparison with TBI only mice. Total numbers of IFN-γ⁺CD4 (4.7 ± 3.3 × 10⁶) and IFN-γ⁺CD8 (0.67 ± 0.56 × 10⁶) T cells for mice that received TBI + T-bet^−/− LN were similar to those of mice that received TBI only (Figure 3D), indicating that T-bet deficiency impaired the ability for T cells to activate and produce the inflammatory cytokine IFN-γ.

Although T-bet^−/− LN cells were largely ineffective, at a high dose they did induce very mild BM failure in recipients, with reduced residual BM cells in comparison with TBI-only control mice (Figure 1B), suggesting that T-bet deletion might not completely abolish the ability of these LN cells to induce BM damage. To study the underlying mechanism, we analyzed plasma concentrations (in pg/mL) of key cytokines for Th1 and Th17 responses. Plasma IFN-γ, a key cytokine for Th1 immune response, was significantly greater (P < .05) in TBI + B6 LN-treated mice (142.4 ± 13.9) than in TBI + T-bet^−/− LN-treated mice (77.3 ± 42.9) or TBI-only control mice (57.7 ± 14.2; Figure 4A). Conversely, plasma TGF-β, an important cytokine for Th17 immune response, was significant lower (P < .01) in TBI + B6 LN-treated mice (136.0 ± 44.3), but significantly greater (P < .01) in TBI + T-bet^−/− LN-treated mice (3360.1 ± 432.9), in comparison with TBI-only control mice (1580.9 ± 635.1; Figure 4B). Another key cytokine for Th17 immune response, IL-17A, was elevated in TBI + T-bet^−/− LN-treated mice (239.5 ± 183.9) compared with TBI + B6 LN-treated (107.6 ± 39.4) and TBI-only (20.2 ± 8.7) animals, although these differences were not statistically significant (Figure 4B).

Furthermore, we also found that TBI + B6 LN-treated mice had greater concentrations (pg/mL) of IL-2 (72.5 ± 28.4 vs 19.6 ± 6.4, P < .05), IL10 (288.9 ± 49.7 vs 135.4 ± 77.3, P < .05), IL-13 (191.5 ± 87.3 vs 37.3 ± 14.3, P < .05), IL-4 (148.9 ± 60.3 vs 51.4 ± 27.4, P > .05), IL-6 (483.2 ± 148.2 vs 207.4 ± 99.6, P > .05), IL-12 (338.2 ± 133.8 vs 167.9 ± 107.9, P > .05), and TNF-α (37.9 ± 8.1 vs 34.7 ± 8.1, P > .05) than TBI + T-bet^−/− LN-treated animals. Conversely, plasma IL-5 and IL-23 concentration for recipient of TBI + T-bet^−/− LN cells (299.2 ± 101.4 and 169.4 ± 123.0) were greater than those for recipients of TBI + B6 LN cells (212.6 ± 72.4 and 145.6 ± 49.6), although the differences did not reach statistic significance. These observations suggested that the reduced Th1 immune response caused by T-bet deficiency was compensated by up-regulation in cytokines of the typical Th17 response after the infusion of T-bet^−/− LN cells.

Because IFN-γ is the most important Th1 cytokine regulated by T-bet signaling, we postulated that the impaired immune activity of T-bet^−/− LN cells in the induction of BM failure might be restored by administration of exogenous IFN-γ. We first conducted a pharmacokinetic study of the time-course plasma IFN-γ concentration after intravenous or intraperitoneal injection of IFN-γ and found that plasma IFN-γ concentration was 6.8-fold greater at 4 hours (772.0 ± 31.9 vs 113.8 ± 27.6, pg/mL) and 4.4-fold greater at 8 hours (149 ± 23.9 vs 33.8 ± 20.7, pg/mL) after intraperitoneal injection in comparison with intravenous injection. We thus injected IFN-γ (intravenously for day 1 and intraperitoneally for days 2-12) into recipient mice that had received TBI + T-bet^−/− LN at 0.2 μg/mouse. Under these conditions, exogenous IFN-γ failed to induce BM failure in mice that had received TBI + T-bet^−/− LN (Figure 5A).

Because the half-life for IFN-γ is very short, ineffectiveness of exogenous IFN-γ treatment delivered by intravenous and intraperitoneal injection might be related to a failure to sustain an effective cytokine concentration. Toward this end, we conducted 2 more experiments in which sublethally irradiated, T-bet^−/− LN cell–injected F1 mice received continuous IFN-γ infusion through an osmotic pump for 12 days. We first compared recipient animals that received TBI and T-bet^−/− LN-cell infusion with or without exogenous IFN-γ: the addition of IFN-γ caused moderate level of marrow suppression but did not result in T-cell expansion. Because IFN-γ itself is known to be marrow-suppressive, we then compared F1 mice that had received TBI and exogenous IFN-γ, with or without the infusion of T-bet^−/− LN cells. Results from 2 experiments were combined. The osmotic pump delivery maintained plasma IFN-γ concentrations 40 to 60 times greater than those measured in TBI-only animals and 15 to 25 times greater than those in B6 LN cell–injected animals (Figure 5B). The high-level IFN-γ was associated with lower numbers of neutrophils (P < .05) and total BM cells (P < .05, Figure 5C) in comparison with mice that received T-bet^−/− LN cells without IFN-γ. Mice that received TBI + IFN-γ without T-bet^−/− LN-cell infusion had similar evidence of hematopoietic suppression (Figure 5C).

We further analyzed T cells in the BM of experimental animals. Mice that received B6 LN-cell infusion had drastic expansion of both CD4 and CD8 T cells (P < .01, Figure 6A). Recipients of T-bet^−/− LN cells, in contrast, showed no T-cell expansion in comparison with TBI-only control mice. Continuous infusion of IFN-γ caused no T-cell expansion, with or without the T-bet^−/− LN cells (Figure 6A). Similarly, the number of IFN-γ⁺ CD4 and IFN-γ⁺ CD8 T cells showed no expansion as a result of continuous IFN-γ infusion (Figure 6B). IFN-γ infusion resulted in more activated T cells, distinguished as CD11a⁺ and especially CD11a⁺CD8 T cells (P < .05, Figure 6C), but the increase in CD11a⁺ T cells was relatively small and associated with IFN-γ itself and not specifically with IFN-γ rescue of T-bet gene–deleted animals (Figure 6C).

Laboratory studies and clinical observations have suggested that the pathophysiology of AA is largely immune mediated.^1,2  Alterations in T-cell function and cytokine production play critical roles in this disease, evidenced by the significant expansion of pathogenic T cells and up-regulation in cytokines IFN-γ and TNF-α in AA patients.¹⁸  Increased T-bet levels in circulating CD4 and CD8 T lymphocytes from AA patients further supports an abnormally up-regulated Th1 immune response in this disease.¹²  The previously developed murine model for BM failure was analogous in these characteristic features because infusion of MHC-mismatched allogeneic LN cells induces dramatic T-cell expansion and activation, with increased IFN-γ and TNF-α production, leading to massive BM-cell destruction and fatal pancytopenia.^15,16 

In the current study, we provide direct evidence of the role of T-bet in our immune-mediated BM failure mouse model when mice carrying a T-bet gene deletion were used as LN-cell donors. Lack of T-bet gene expression prevented T-cell expansion, T-cell activation, and inflammatory cytokine production when these cells were infused into MHC-mismatched F1 recipients. Genetically engineered T-bet deficiency thus essentially abolished the ability of allogeneic lymphocytes to attack and destroy recipient BM hematopoietic cells.

An unexpected observation from our current study was a much increased proportion and absolute number of BM KSLCD34⁻ cells, which have been characterized as containing hematopoietic progenitor and stem cells,^19,20  after the infusion of T-bet^−/− LN cells. The enlarged KSLCD34⁻-cell pool may be attributable to the expansion of this cell population because total BM cells were slightly decreased in TBI + T-bet^−/− LN-treated animals than in TBI-only control mice. It remains to be tested whether these expanded KSLCD34⁻ cells have the functional characteristics of hematopoietic stem and progenitor cells. In addition, the molecular mechanism responsible for increased KSLCD34⁻ BM cells after T-bet^−/− LN-cell infusion would need to be elucidated in future studies.

We were unable to restore the T-bet^−/− LN cells' functional ability to induce BM failure by exogenous IFN-γ, either through intraperitoneal injection or micro-pump infusion, despite that IFN-γ antibody effectively blocked BM failure in our mouse model.¹⁵  When we used the osmotic pump to inject IFN-γ continuously for 12 days, exogenous IFN-γ suppressed hematopoiesis with or without the infusion of T-bet^−/− LN cells, consistent with many reports of IFN-γ negative modulation of self-renewal of repopulating human hematopoietic stem cells and as a mediator of hematopoietis suppression in aplastic anemia.^21,22  In our study, IFN-γ delivered even by this efficient method did not rescue the phenotype of T-bet deficiency: infusion of exogenous IFN-γ did not cause infused T-bet^−/− to expand in the BM of recipient animals, neither did it produce obvious BM failure. We offer 2 possible reasons for the observed ineffectiveness of exogenous IFN-γ. First, the IFN-γ concentration may not have been sufficient in BM tissue to cause hematopoietic cell destruction, despite high IFN-γ levels in plasma. Second, factors other than IFN-γ that are deficient in T-bet^−/− mice may play an important role in driving T-cell expansion and BM destruction. Fas-mediated apoptosis plays a major role, whereas perforin-mediated cell death plays a minor role in immune-mediated BM destruction.^18,23-26  However, in the case of T-bet^−/− LN-cell infusion, lack of T-cell expansion appears most likely responsible for lack of BM failure induction.

Observations from our current study are consistent with findings from other disease models in animals, such as atherosclerosis, type 1 diabetes, inflammatory arthritis, ulcerative colitis, and melanoma metastatic disease, in which T-bet–dependent Th1 immune responses are critical.^18,23-32  In experimental autoimmune encephalomyelitis, the production of IFN-γ in the central nervous system was dramatically reduced in T-bet^−/− mice in comparison with wild-type T-bet^+/+ mice.³³  There was an increased IL-10 production in T-bet^−/− mice in the experimental autoimmune encephalomyelitis model, suggesting compensatory up-regulation in Th2 immune response. T-bet^−/−–deficient mice developed spontaneous airway changes consistent with human asthma due to up-regulation of Th2 response as a compensation of defective Th1 response caused by T-bet deficiency.^14,34  In our model of BM failure, deficiency of T-bet and Th1 immune response did not appear to have resulted in an obvious up-regulation of Th2 responses because Th2-related cytokines were not increased in the plasma. Instead, we observed an up-regulation in TGF-β, IL17A, and IL23, all of which are cytokines related to Th17 immune responses,^34-37  indicating that the lack of T-bet and Th1 immune response may have resulted in an up-regulation in Th17 immune response that might be responsible for the mild marrow failure observed in mice that received T-bet^−/− LN cells.

Contribution: Y.T. designed the study, executed the research plan, collected and analyzed the data, and wrote the paper; M.J.D. performed experiments and collected data; J.C. designed the study, analyzed the data, and wrote the paper; and N.S.Y. designed the study and edited the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Neal S. Young, MD, Hematology Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Bldg 10 Clinical Research Center Rm 3E-5140, 10 Center Dr, Bethesda, MD 20892-1202; e-mail: youngn@nhlbi.nih.gov.

National Institutes of Health