Tumor Necrosis Factor-Mediated Survival of CD169⁺ Cells Promotes Immune Activation during Vesicular Stomatitis Virus Infection

TNF production by CD11b⁺ Ly6C⁺ Ly6G⁺ cells following VSV infection.

TNF can be detected during an infection with VSV (32, 37). Consistently, we found that TNF expression levels were higher in the spleen after infection with VSV than in uninfected controls (Fig. 1A). Backgating of intracellular-TNF-producing cells showed that TNF-producing cells are a heterogeneous CD11b⁺ CD19⁻ population (Fig. 1B and C). Therefore, we hypothesized that TNF was likely not expressed by B or T cells during infection. Accordingly, we observed TNF mRNA expression levels in Cd8^−/−, B-cell-deficient Jh^−/−, and Rag1^−/− mice that were comparable to those in wild-type (WT) mice (Fig. 1D). TNF-producing cells could be predominantly characterized as CD11b⁺ CD11c⁻ Ly6C⁺ Ly6G⁺ major histocompatibility complex class II negative (MHC-II⁻) (Fig. 1E). Consistent with reports that neutrophils (38, 39) and CD11b⁺ Ly6C⁺ Ly6G⁺ cells (40) are important during early defense against bacterial and viral infections via production of proinflammatory cytokines, such as interleukin 1b (IL-1b), IL-6, TNF, and IFN-I, we found a significant increase of TNF⁺ CD11b⁺ Ly6C⁺ Ly6G⁺ cells (Fig. 1F). Treatment with clodronate encapsulated in liposomes (clodronate liposomes) can deplete phagocytic cells in mice (Fig. 1G) (41, 42). Accordingly, clodronate depletion reduced TNF expression after VSV infection, suggesting a role of these phagocytic cells in the production of TNF (Fig. 1H). However, when we employed diphtheria toxin receptor (DTR)-induced specific depletion of CD169⁺ cells and CD11c⁺ cells, we did not observe a reduction in TNF production (Fig. 1H). Taken together, these findings indicate that TNF production following intravenous VSV infection is triggered by CD11b⁺ CD11c⁻ Ly6C⁺ Ly6G⁺ phagocytes.

TNF triggers the maintenance of CD169⁺ cells during viral infection to protect animals against the development of severe disease.

To determine whether TNF affects the outcome after VSV infection, we infected WT and TNF-deficient mice. TNF-deficient mice developed severe VSV infection in comparison to WT mice (Fig. 2A). A neutralizing antibody titer was achieved later in TNF-deficient mice than in WT mice after infection with low doses of VSV (Fig. 2B). Since IFN-I is critical to overcome an infection with VSV (15), we measured IFN-α and IFN-β in the sera of infected animals. IFN-α production was impaired in TNF-deficient mice compared to control animals (Fig. 2C). However, IFN-β was undetectable in the sera of animals infected with 10⁵ PFU VSV (Fig. 2C). Previous findings showed that CD169⁺ cells contribute to innate immune activation in mice, not only by allowing viral replication, but also by producing IFN-I (10, 43). When we depleted CD169⁺ cells expressing diphtheria toxin receptor (CD169-DTR cells) by administering diphtheria toxin (DT) (44), we observed reduced IFN-I concentrations in the sera of infected animals (Fig. 2D). To exclude the possibility of defective innate Toll-like receptor (TLR) activation, we administered the TLR3 agonist poly(I·C). We found that the IFN-I production was intact in both WT and TNF-deficient mice (Fig. 2E). Hence, we speculated that TNF promoted the function of CD169⁺ cells and thus contributed to IFN-I production following VSV infection. Shortly after infection with VSV, the number of CD169⁺ cells in spleen tissue decreased in TNF-deficient mice compared to spleen tissue harvested from WT animals (Fig. 2F to H). To understand the reduced production of IFN-I in the absence of TNF, we monitored the virus replication in spleen tissue of WT and Tnfa^−/− mice. The expression of VSV glycoprotein (VSV-G) was detected in smaller quantities in spleen tissue harvested from TNF-deficient animals than in spleen tissue harvested from WT mice after VSV infection (Fig. 2I and J). Consistently, early VSV titers after infection were lower in Tnfa^−/− mice than in control mice, a condition that negatively affected antiviral immune activation (Fig. 2K). Injection of UV light (UV)-inactivated virus could increase TNF mRNA expression in WT mice (Fig. 2L). However, the decrease of CD169⁺ cells was dependent on live virus, because UV-inactivated virus did not affect CD169⁺ cells in spleen tissue of Tnfa^−/− mice (Fig. 2M). These findings indicate that TNF is necessary to sustain virus replication in the early hours of infection but is dispensable for sterile innate immune activation. Notably, CD169^−/− mice exhibited VSV-G expression in spleen tissue, a finding indicating that downregulation of the protein CD169 would not cause absence of virus replication (Fig. 2N). Taken together, these findings indicate that the absence of TNF results in defective antiviral innate immune activation after infection with VSV.

CD169⁺ cell maintenance via TNFR1 results in productive VSV replication and immune activation.

To further characterize the role of TNF during viral infection, we infected TNFR1- and TNFR2-deficient mice with VSV. In line with findings from TNF-deficient animals, the absence of TNFR1, but not that of TNFR2, resulted in a decrease in the number of CD169⁺ cells in spleen tissue (Fig. 3A and B). Furthermore, VSV-G production was lower in Tnfrsf1a^−/− animals than in WT or Tnfrsf1b^−/− mice (Fig. 3A). Consistently, VSV titers were reduced in spleen tissue shortly after infection in Tnfrsf1a^−/− animals, in sharp contrast to the findings in WT and Tnfrsf1b^−/− mice (Fig. 3C). Interestingly, IFN-I production was defective in Tnfrsf1a^−/− mice but was also lower in Tnfrsf1b^−/− animals than in WT control mice (Fig. 3D). IFN-I is necessary for the expression of antivirally active ISGs (1). Consistently, we found reduced expression of ISGs in the CNS of Tnfrsf1a^−/− mice after infection with VSV (Fig. 3E). Defective ISG expression was not found to the same extent in Tnfrsf1b^−/− CNS tissue (Fig. 3F). VSV can drive neuropathological symptoms by infecting the CNS (22). When we measured viral titers in the spinal cord and brain tissue of mice exhibiting hind leg paralysis, we found infectious VSV in tissue from TNFR1-deficient mice (Fig. 3G). Consequently, Tnfrsf1a^−/− mice developed clinical signs of CNS infection, unlike WT and Tnfrsf1b^−/− mice (Fig. 3H). Taken together, these findings suggest that TNFR1 drives antiviral defense by promoting CD169⁺ cell survival.

TNFR1 triggers the survival of CD169⁺ cells.

Next, we opted to determine which factors drive the maintenance of CD169⁺ cells and enforced viral replication after viral infection. B-cell-mediated Ltβ production is important for splenic CD169⁺ cells. Hence, we wondered whether the defects in the absence of TNF were triggered by B cells. Notably, we did not observe any major changes of B-cell subsets in TNF-, TNFR1-, or TNFR2-deficient mice (Fig. 4A). Consistently, we did not see differential expression of Ltα, Ltβ, or Ltβ receptor (LtbR) in TNFR1-deficient mice (Fig. 4B). Additionally, we found no major differences in B-cell subsets between WT and Tnfrsf1a^−/− mice after infection (Fig. 4C). Furthermore, we reconstituted lethally irradiated C57BL/6 mice with mixed bone marrow (BM) from Rag1^−/− and Tnfrsf1a^−/− and from Rag1^−/− and WT donors at a ratio of 1:1. Mice reconstituted with Rag1^−/−-Tnfrsf1a^−/− bone marrow exhibited no significant reduction in IFN-α in the serum compared to mice reconstituted with Rag1^−/−-WT bone marrow (Fig. 4D). Furthermore, there was no difference between these mice in neutralizing antibody production (Fig. 4E). To elucidate if TNFR1 deficiency specifically on CD169⁺ cells has a role in virus replication, we reconstituted lethally irradiated C57BL/6 mice with mixed bone marrow from CD169-DTR⁺ and Tnfrsf1a^−/− donors, as well as CD169-DTR⁺ and WT donors, at a ratio of 1:1. We observed that the production of IFN-α was lower in the mice reconstituted with CD169-DTR⁺ and Tnfrsf1a^−/− bone marrow compared to control mice reconstituted with CD169-DTR⁺ and WT bone marrow after infection with VSV and DT treatment (Fig. 4F). Furthermore, we found slightly delayed presence of VSV neutralizing antibody titers in CD169-DTR⁺–Tnfrsf1a^−/− recipients compared to corresponding CD169-DTR⁺–WT recipients (Fig. 4G). CD169⁺ cells can be depleted in CD11c-DTR mice, because CD169⁺ cells exhibit intermediate expression of CD11c (10, 45). Consistently, lethally irradiated mice reconstituted with mixed bone marrow from CD11c-DTR⁺ and Tnfrsf1a^−/− mice exhibited reduced concentrations of IFN-α after VSV infection compared to CD11c-DTR⁺–WT bone marrow recipients (Fig. 4H). These findings suggest that TNFR1 triggers cell-intrinsic effects on CD169⁺ cells.

We speculated that TNF delivers an important survival signal for CD169⁺ cells. To determine if TNF is involved in protection against VSV-induced apoptosis, we measured caspase 3 activity on whole spleen tissue lysates. After VSV infection, caspase 3 activity was significantly higher in Tnfa^−/− mice than in control animals (Fig. 5A). VSV is known to induce apoptosis and inactivates Mcl-1 and Bcl-Xl (46). To elucidate if TNF plays a role in promoting expression of antiapoptotic genes, we measured mRNA expression of Bcl2, Bcl-Xl, and xIAP in spleen tissue of mice after VSV infection (Fig. 5B). After VSV infection, Bcl2 and Bcl-Xl expression was significantly reduced in Tnfa^−/− mice compared to WT mice (Fig. 5B). To enumerate the mechanism that reduces CD169⁺ cells in TNF-deficient mice after infection, we made use of terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) assays. The number, as well as the mean fluorescence intensity, of TUNEL-positive CD169⁺ cells was higher in spleen tissue from TNF-deficient mice than in tissue from corresponding WT control mice (Fig. 5C and D). The proportion of CD169⁺ cells that stained positive for 7-aminoactinomycin D (7-AAD) was higher in TNFR1-deficient mice than in WT control mice 8 h after infection (Fig. 5E). Next, we wondered if we could rescue the CD169⁺ cells by injecting the pancaspase inhibitor Z-Val-Ala-Asp-fluoromethylketone (Z-VAD). Z-VAD treatment restored the presence of CD169⁺ cells in TNF-deficient animals, a finding indicating that CD169⁺ cells depend on TNF-mediated survival (Fig. 5F and G). Although treatment of TNF-deficient mice with Z-VAD-rescued CD169⁺ cells, it failed to rescue the IFN-I response, suggesting the role of TNF signaling is essential not only to prevent apoptosis, but also for IFN-I production (Fig. 5H). In summary, these findings indicate that TNF delivers a survival signal that is important for the maintenance of CD169⁺ cells in the spleen after viral infection and for IFN-I production.

The NF-κB regulator MALT1 promotes canonical NF-κB expression, VSV replication in CD169⁺ cells, and immune activation during viral infection.

TNF can induce NF-κB activation via TNFR1 and can promote the expression of genes driving survival and of proinflammatory cytokines (47). Furthermore, TNF is known to promote IFN-I production (48). Consistently, RelA expression was increased in the marginal zone of spleen tissue after VSV infection (Fig. 6A). We quantified cytoplasmic and nuclear expression of RelA in CD169⁺ cells. The nuclear presence of RelA in CD169⁺ cells was higher in VSV-infected mice than in naive controls (Fig. 6B). We wondered whether nuclear RelA protein expression was dependent on TNF. As expected, compared with WT control mice, VSV-infected mice exhibited reduced expression of RelA in the nuclear compartments of CD169⁺ cells in the absence of TNF (Fig. 6C). Notably, the presence of RelA was reduced in TNFR1-deficient mice, but we observed no difference in RelA expression between TNFR2-deficient mice and corresponding control mice (Fig. 6D and E). It has been reported that one of the major regulators of RelA signaling is RelB, which acts through sequestration of RelA in the cytoplasm and competitive binding of DNA (49). It has also been reported that the paracaspase MALT1 can promote canonical NF-κB signaling by cleaving RelB (50, 51). Hence, we stained spleen sections of Malt1^+/− and Malt1^−/− mice for RelB. Ablation of MALT1 resulted in increased levels of RelB in CD169⁺ cells in the marginal zone of the spleen (Fig. 7A and B). In turn, nuclear RelA levels were lower in CD169⁺ cells in Malt1^−/− spleen tissue than in control tissue (Fig. 7C). Consistently, mouse embryonic fibroblasts (MEFs) derived from Malt1^−/− mice showed reduced translocation of p65 into the nucleus after stimulation with TNF but higher expression of RelB in the nucleus (Fig. 7D and E). These findings indicate that MALT1 destabilizes RelB in marginal-zone macrophages to promote canonical NF-κB signaling. The presence of CD169⁺ cells in spleen tissue was not affected by Malt1 before or after infection with VSV (Fig. 8A). However, the expression of VSV-G was lower in Malt1^−/− mice than in control mice (Fig. 8B and C). Consistently, the numbers of infectious VSV particles were lower in spleen tissue harvested from Malt1^−/− mice than in spleen tissue from control mice (Fig. 8D). Hence, serum IFN-I concentrations after VSV infection were lower in MALT1-deficient mice than in control mice (Fig. 8E). A previous report suggested that MALT1 is not required for RIG-I activation (52). Consistently, when we injected poly(I·C) into Malt1^+/− and Malt1^−/− mice, we found similar serum IFN-I levels in both groups (Fig. 8F). Hence, we concluded that defective IFN-I production during VSV infection was caused by reduced VSV replication early during infection. Consequently, MALT1-deficient mice succumbed to the infection, in sharp contrast to control animals (Fig. 8G).

Taken together, these findings indicate that absence of MALT1 results in reduced canonical NF-κB signaling in response to VSV infection. Malt1-deficient mice exhibit reduced VSV replication and immune activation.

Mice, viruses, and virus titration.

Tnfa^−/−, Tnfrsf1b^−/−, CD8^−/−, and Rag1^−/− mice were purchased from Jackson Laboratories (United States). Tnfrsf1a^−/− mice have been previously described (34). Malt1^−/−, CD169^−/−, CD169-DTR, and CD11c-DTR mice have also been previously described (71–74). All the mice were maintained on a C57BL/6 genetic background. VSV Indiana serotype (Mudd-Summers strain) was originally obtained from D. Kolakofsky (University of Geneva, Geneva, Switzerland). VSV was propagated and titrated as previously described (13). Mice were infected with VSV via tail vein injection. In survival experiments, mice exhibiting symptoms of hind leg paralysis were considered “severe,” taken out of the experiment, and counted as dead. Blood was collected at the indicated time points after infection. VSV neutralizing antibody titers were measured by plaque reduction neutralization test (PRNT) as previously described (9, 13). Briefly, serum samples were diluted 1:40 and incubated at 56°C for 30 min. To evaluate immunoglobulin G (IgG), the serum was pretreated with 0.1 M β-mercaptoethanol. Serial 2-fold dilutions were performed for 12 steps, and the serum was incubated with 5,000 PFU of VSV. The virus and serum mixture was incubated on a Vero cell monolayer. The plates were stained with crystal violet after 24 h. To inhibit caspase activity in vivo, we administered three doses (2 μg/g of body weight each) of Z-VAD (Abcam, Cambridge, United Kingdom) (75, 76). For chimera experiments, mice were lethally irradiated with 10.2 Gy. After 24 h, mixed bone marrow from WT or Tnfrsf1a^−/− and CD169-DTR, CD11c-DTR, and Rag1^−/− mice was transplanted into the irradiated mice as indicated. All mice were maintained under specific-pathogen-free conditions under the authorization of the Landesamt für Natur, Umwelt und Verbraucherschutz of North Rhine-Westphalia (LANUV NRW) in accordance with the German laws for animal protection.

Depletion of cells.

To deplete macrophages, 200 μl clodronate liposomes was injected intravenously, and 24 h later, the mice were infected with VSV. Liposomes containing phosphate-buffered saline (PBS; PBS liposomes) were used as a control. Clodronate and PBS liposomes were provided by Nico van Rooijen (Vrije University Medical Center, Netherlands) and used as previously described (41, 42). CD169- and CD11c-expressing cells in CD169-DTR and CD11c-DTR mice were depleted by injecting 2 doses of 100 ng DT (Sigma) 1 day before and on the day of infection.

Histology and ELISA.

Histological analysis of snap-frozen tissue was performed as previously described (9). Briefly, snap-frozen tissue sections were cut at 7-μm thickness, air dried, and fixed with acetone for 10 min. The sections were blocked with 2% fetal calf serum in phosphate-buffered saline (PBS) for 1 h and stained with anti-CD169 (final concentration, 4 μg/ml; Acris, Germany; clone MOMA-1), anti-VSV-G (final concentration, 1 μg/ml; produced in house; clone Vi10), anti-RelA (final concentration, 1 μg/ml; Santa Cruz Biotechnology, USA), anti-F4/80 (final concentration, 2 μg/ml; eBioscience; clone BM8), and anti-RelB (final concentration, 1 μg/ml; Cell Signaling, USA; polyclonal) for 1 h. Then, the sections were washed with PBS containing 0.05% Tween 20 (Sigma). The secondary antibodies, phycoerythrin (PE) streptavidin (final concentration, 1 μg/ml; eBioscience), anti-rabbit fluorescein isothiocyanate (FITC) (final concentration, 1 μg/ml; Thermo Fisher), and anti-goat FITC (final concentration, 1 μg/ml; Santa Cruz Biotechnology, USA), were incubated for 1 h. Then, the sections were washed with PBS containing 0.05% Tween 20 (Sigma) and mounted using fluorescence mounting medium (Dako). A caspase 3 activity assay was performed with a fluorescence assay (Cell Signaling) according to the manufacturer's instructions. TUNEL staining was performed on formalin-fixed spleen sections according to the manufacturer's instructions (Thermo Scientific, USA). Images were obtained with an LSM510 confocal microscope and an Axio Observer Z1 fluorescence microscope (Zeiss, Germany). Analysis of the fluorescence images was performed with ImageJ software. IFN-α and IFN-β (PBL Biosciences, New Jersey, USA) concentrations were determined using an enzyme-linked immunosorbent assay (ELISA) according to the manufacturers' instructions.

Reverse transcription (RT)-PCR analyses.

RNA purification (Qiagen RNeasy kit or TRIzol) was performed according to the manufacturer's instructions. Gene expression of Bcl2, Bcl-xl, Xiap, Lta, Ltb, Ifit1, Ifit2, Ifit3, Irf7, Isg15, Oasl1, and Tnfa was performed using 6-carboxyfluorescein (FAM) and VIC probes (Applied Biosystems) and an iTAQ one-step PCR kit (Bio-Rad). For analysis, the expression levels of all the target genes were normalized to β-actin/GAPDH (glyceraldehyde-3-phosphate dehydrogenase) expression (ΔC_T). Gene expression values were then calculated based on the ΔΔC_T method, using naive WT mice as a control to which all other samples were compared. Relative quantities (RQ) were determined using the following equation: RQ = 2^−ΔΔCT.

Immunoblotting.

Malt1^+/− and Malt1^−/− MEFs were obtained from Jürgen Ruland (Technische Universität Munich, Munich, Germany). Malt1^+/− and Malt1^−/− MEFs were stimulated with 100 ng/ml murine soluble TNF (mTNF) (R&D Systems). Cytoplasmic and nuclear extracts were prepared using a nuclear protein extraction kit according to the manufacturer's instructions (Active Motif, Belgium). Immunoblots were probed with primary anti-p65 (Santa Cruz Biotechnology), anti-RelB (Cell Signaling), and anti-p100/p52 (Cell Signaling).

Flow cytometry.

For intracellular-cytokine staining, single-cell suspended splenocytes were incubated with brefeldin A (eBioscience), followed by an additional 5 h of incubation at 37°C. After surface staining with anti-CD3, anti-CD8, anti-CD11b, anti-CD11c, anti-CD19, anti-CD115, anti-F4/80, anti-Ly6C, anti-Ly6G, anti-MHC-II, and anti-NK1.1 antibodies (all from eBioscience), the cells were fixed with 2% formalin, permeabilized with 0.1% saponin, and stained with anti-TNF antibodies (eBioscience) for 30 min at 4°C. B-cell subsets were detected in single-cell suspensions of splenocytes with anti-CD5, anti-CD19, anti-CD21, anti-CD23, and anti-IgM antibodies (all from eBioscience). BD Calibrite (BD Biosciences, USA) beads were added to the samples before acquisition with a BD LSRFortessa.

Statistical analyses.

Data are represented with standard errors of the mean (SEM). Statistically significant differences between two groups were determined with Student's t test. Statistically significant differences between several groups were determined by one-way analysis of variance (ANOVA) with additional Bonferroni or Dunnett post hoc tests. Statistically significant differences between groups in experiments involving more than one time point were determined with two-way ANOVA (repeated measurements).