Macrophage TNF-α licenses donor T cells in murine bone marrow failure and can be implicated in human aplastic anemia

Abstract

Interferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α) have been implicated historically in the immune pathophysiology of aplastic anemia (AA) and other bone marrow (BM) failure syndromes. We recently defined the essential roles of IFN-γ produced by donor T cells and the IFN-γ receptor in the host in murine immune-mediated BM failure models. TNF-α has been assumed to function similarly to IFN-γ. We used our murine models and mice genetically deficient in TNF-α or TNF-α receptors (TNF-αRs) to establish an analogous mechanism. Unexpectedly, infusion of TNF-α^-/- donor lymph node (LN) cells into CByB6F1 recipients or injection of FVB LN cells into TNF-αR^-/- recipients both induced BM failure, with concurrent marked increases in plasma IFN-γ and TNF-α levels. Surprisingly, in TNF-α^-/- recipients, BM damage was attenuated, suggesting that TNF-α of host origin was essential for immune destruction of hematopoiesis. Depletion of host macrophages before LN injection reduced T-cell IFN-γ levels and reduced BM damage, whereas injection of recombinant TNF-α into FVB-LN cell-infused TNF-α^-/- recipients increased T-cell IFN-γ expression and accelerated BM damage. Furthermore, infusion of TNF-αR^-/- donor LN cells into CByB6F1 recipients reduced BM T-cell infiltration, suppressed T-cell IFN-γ production, and alleviated BM destruction. Thus, TNF-α from host macrophages and TNF-αR expressed on donor effector T cells were critical in the pathogenesis of murine immune-mediated BM failure, acting by modulation of IFN-γ secretion. In AA patients, TNF-α-producing macrophages in the BM were more frequent than in healthy controls, suggesting the involvement of this cytokine and these cells in human disease.

Graphical abstract

Figure 1.

TNF-α deficiency in donor effector cells does not block T-cell–mediated BM destruction. (A) CByB6F1 mice were irradiated with 5 Gy TBI without (TBI; n = 4) or with infusion of 5 × 10⁶ LN cells from WT C57BL/6 (B6 LN; n = 5) or TNF-α^−/− (TNF-α^−/− LN; n = 5) donors to induce BM failure. Animals were euthanized and evaluated at day 13. (B) Histology of sternums from CByB6F1 mice that received TBI, TBI plus B6 LN, or TBI plus TNF-α^−/− LN treatments shown as representative images (original magnification ×200; hematoxylin and eosin stain). (C) CByB6F1 mice that received LN infusion from either WT B6 or TNF-α^−/− donors developed BM failure showing reduction in BM cells, WBCs, and platelets. (D) CByB6F1 mice that received B6 LN or TNF-α^−/− LN infusion expressed high levels of Fas on whole BM cells. (E) BM cells from CByB6F1 mice that received B6 LN or TNF-α^−/− LN infusion showed reduced total CFU, CFU-GM, CFU-G, and CFU-M, relative to control mice treated with TBI only. (F) CByB6F1 mice that received B6 LN or TNF-α^−/− LN infusion had high levels of IFN-γ and TNF-α in the plasma. *P < .05; **P < .01; ***P < .001; ****P < .0001. SSC, side scatter.

Figure 2.

TNF-α–deficient mice are resistant to immune-mediated BM failure. (A) WT C57BL/6 (B6) mice or B6 mice carrying germline deletion of TNF-α (TNF-α^−/−) were irradiated with 6.5 Gy TBI without (TBI) or with infusion of 5 × 10⁶ LN cells (TBI+LN) from FVB donors to induce BM failure. Animals were evaluated at day 10. (B) Histology of sternums from B6 or TNF-α^−/− mice that received TBI or TBI+LN infusion (original magnification ×200; hematoxylin and eosin stain). (C) B6 mice that received TBI+LN infusion (n = 10) showed declines in WBCs, platelets, and total BM cells relative to TBI controls (n = 8), whereas TNF-α^−/− animals that received TBI+LN infusion (n = 5) showed no decline in blood or BM cells relative to TBI controls (n = 4). (D) TBI+LN infusion caused significant declines in BM total CFU, CFU-G, CFU-M, and CFU-GM in B6 mice, whereas no such decline was detected in TBI+LN-infused TNF-α^−/− mice. (E) Fas expression on total BM cells was increased in B6, but not in TNF-α^−/−, recipient mice following FVB LN-cell injection. (F) Blood plasma concentrations of IFN-γ and TNF-α increased drastically in B6, but not in TNF-α^−/−, recipient mice following FVB LN-cell infusion, relative to their respective TBI controls. *P < .05; **P < .01; ***P < .001; ****P < .0001.

Figure 3.

Macrophage depletion ameliorates BM failure. (A) Macrophages from CByB6F1 recipient mice were depleted with Clodrosome before induction of BM failure; Encapsome was used as vehicle control. Severity of BM failure was evaluated at day 13. (B) Macrophage depletion with Clodrosome (n = 9) attenuated immune-mediated BM failure in CByB6F1 mice with improved BM structure, and increased numbers of BM cells, RBCs, and platelets, when compared with mice treated with Encapsome as vehicle control (n = 5) without macrophage depletion (original magnification ×200; hematoxylin and eosin stain). (C) Macrophage depletion by Clodrosome suppressed expansion of T cells, especially CD8⁺ T cells, in the BM relative to vehicle control without macrophage depletion. (D) Macrophage depletion by Clodrosome reduced plasma levels of IFN-γ and TNF-α. (E) Macrophage depletion by Clodrosome suppressed IFN-γ levels in both CD8⁺ and CD4⁺ T cells in the BM. *P < .05; **P < .01; ***P < .001; ****P < .0001.

Figure 4.

Exogenous TNF-α accelerates BM destruction and promotes T-cell IFN-γ secretion in TNF-α^−/−mice. (A) TNF-α^−/− mice received 6.5 Gy TBI and infusion of 5 × 10⁶ FVB LN cells were either untreated (n = 4), or were injected with recombinant TNF-α protein at 100 ng per day IV for 7 days (n = 5). Mice were euthanized and analyzed on day 10. Injection of recombinant TNF-α to LN-cell–infused TNF-α^−/− mice reduced BM cellularity (n = 4; original magnification ×200; hematoxylin and eosin stain) (B), significantly increased intracellular IFN-γ levels in BM CD4⁺ and CD8⁺ T cells (C), but did not change intracellular TNF-α levels in BM CD4⁺ and CD8⁺ T cells (D), when compared with LN-cell–infused TNF-α^−/− mice without TNF-α injection. *P < .05; **P < .01.

Figure 5.

TNF-α modulates T-cell function through engagement with TNF-αRs on T cells. (A) CByB6F1 recipients were preirradiated at 5 Gy TBI and were infused with 5 × 10⁶ LN cells from WT B6 (n = 5) or TNFrsf1a^−/−1b^−/− (n = 4) donors. Recipient mice were bled and analyzed on day 13. Relative to recipients of WT B6 LN cells, recipients of TNFrsf1a^−/−1b^−/− LN cells had: (B) higher levels of BM cells, RBCs, WBCs, and platelets (PLT) (original magnification ×200; hematoxylin and eosin stain); (C) lower-level infiltration of T cells, especially CD8⁺ T cells, in recipient BM; (D) reduced intracellular IFN-γ expression in BM-infiltrated CD4⁺ and CD8⁺ T cells; (E) reduced BM proportion of effector memory (EM) CD8⁺ T cells and increased BM proportion of central memory (CM) CD8⁺ T cells; (F) reduced Fas expression on residual BM cells. *P < .05; **P < .01; ***P < .001; ****P < .0001.

Figure 6.

Effects of TNF-α on murine T-cell IFN-γ secretion mediated by TNF-αRs. Recombinant TNF-α (50 ng/mL) were added to LN cells (LN) from WT B6 mice or TNFrsf1a^−/−1b^−/− mice under resting condition or activated by phorbol 12-myristate-13-acetate (PMA; 5 ng/mL) plus ionomycin (1 μM) overnight. Intracellular IFN-γ levels in CD4⁺ T cells (A) and CD8⁺ T cells (B) were examined by flow cytometry. Representative flow cytometry plots from 3 separate experiments are shown. (A-B) Top panels, Under resting condition; bottom panels, after stimulation with PMA plus ionomycin. (C) Transcriptome changes related to the impact of TNF-α on activated T cells. CD8⁺ T cells from B6 or TNFrsf1a^−/−1b^−/− mice were stimulated with PMA plus ionomycin in the presence of TNF-α (50 ng/mL) for 4 hours, and cells were then subjected to RNA extraction and complementary DNA (cDNA) synthesis. A PCR array focused on genes related to T-cell activation and anergy was performed. Genes with at least twofold differences between CD8⁺ T cells from B6 and TNFrsf1a^−/−1b^−/− mice are shown as a heatmap. Arrays were replicated from 2 different pools of CD8⁺ T cells. Red indicates high expression; blue, low expression. *P < .05.

Figure 7.

TNF-α expression by BM macrophages from AA patients and impact of TNF-α on human T-cell IFN-γ expression. (A) Comparison of the frequencies of BM macrophages (CD16⁺CD68⁺ in viable CD3⁻ population) in AA patients (n = 8) and healthy controls (HC; n = 7). (B) CD14 and CD16 expression in CD16⁺CD68⁺ population. (C) Frequencies of TNF-α–producing cells in CD16⁺CD68⁺, CD4⁺, and CD8⁺ populations in AA patients detected by intracellular staining. Representative dot plots are shown. (D) Peripheral blood mononuclear cells (2 × 10⁶/mL) from AA patients or healthy controls were stimulated with PMA (5 ng/mL) plus ionomycin (1 μM) in the presence or absence of human recombinant TNF-α (100 ng/mL) overnight. Intracellular IFN-γ levels in CD4⁺ T cells and CD8⁺ T cells from healthy controls (n = 7) and AA patients (n = 8) were examined by flow cytometry. *P < .05; **P < .01.