The AIMp1 null allele in the C57BL/6 background was graciously provided by Dr. Sunghoon Kim at the Seoul National University. Influenza experiments were performed in 129Sv/Ev mice into which the null allele had been backcrossed to the F₇ generation. Tumor experiments were performed in C57BL/6×129Sv/Ev F₁ heterozygotes. All AIMp1 deficient animals were derived from heterozygous breeders, and wild-type (WT) control populations in all experiments were always derived from the littermates of those intercrosses. Other mice utilized included WT C57BL/6J, 129Sv/Ev and OT-II were purchased from the Jackson Laboratory (Barr Harbor, ME, USA). Experiments were performed utilizing NIH and United States Department of Agriculture guidelines, The Public Health Service Policy on Humane Care and Use of Laboratory Animals, and experimental protocols approved by the Baylor College of Medicine Investigational Animal Care and Use Committee (IACUC Protocol numbers AN-1428 and AN-2307).

p38 MAPK inhibitors SB202190 and SB203580 (Sigma-Aldrich, St. Louis, MO, USA), PP2A inhibitors Okadaic acid (Sodium Salt, EMD Millipore, Billerica, MA, USA) and Endothall (Sigma-Aldrich), DUSP1/6 Inhibitor (BCI, EMD Millipore); OVA (257–264)/SIINFEKL peptide (Anaspec, Fremont, CA, USA), recombinant endotoxin-free OVA protein (Invivogen, San Diego, CA, USA), Toll-like receptor (TLR) agonists LPS (O127:B8 E. coli strain, Sigma-Aldrich), and mouse IL-6 ELISA and mouse IL-1β ELISA sets (BD Pharmingen, San Diego, CA, USA). Other antibodies and protein standards were purchased from Biolegend including purified antimouse IL-12 (p70) antibody (C18.2), biotin antimouse IL-12/IL-23 p40 antibody (C17.8), recombinant mouse IL-12 (p70) (ELISA Std.), purified antimouse IFN-γ antibody (AN-18), biotin antimouse IFN-γ antibody (R4-6A2), recombinant mouse IFN-γ (ELISA Std.), purified antimouse IL-4 antibody (11B11), and biotin antimouse IL-4 antibody (BVD6-24G2).

Bone marrow leukocytes were flushed from mouse tibia and femur and cultured in RPMI-1640 (Lonza, Allendale, NJ, USA) containing 10% fetal bovine serum (Invitrogen, Carlsbad, CA, USA), 1% antibiotic–antimycotic (Invitrogen) and supplemented with 20 ng/mL mouse GM-CSF (R&D Systems, Minneapolis, MN, USA) and 10 ng/mL mouse IL-4 (R&D Systems). Cells were cultured in a humidified chamber at 37°C and 5% atmospheric CO₂. On day 3, half the culture medium was gently removed and replenished with fresh medium and cytokines. BMDC were harvested on day 6 with Cell Dissociation Buffer (Invitrogen). The immature BMDC were then treated with lipopolysaccharide (Sigma-Aldrich) or maturation cytokine cocktail comprising 10 ng/ml IL-1β (R&D Systems), 10 ng/ml TNF-α (R&D Systems), 15 ng/ml IL-6 (R&D Systems), and 1 µg/ml PGE₂ (Sigma-Aldrich).

Immature BMDC were tandemly loaded with peptide and protein or with tumor cell-derived mRNA and lysate as described previously (25). Briefly, cells were resuspended in Viaspan (Barr Laboratories, Pomona, NY, USA) buffer at 40 × 10⁶ cells/ml and incubated with mRNA (1 µg mRNA/10⁶ cells) or peptide (10 µg/ml) on ice for 10 min. Cells were then electroporated with exponential decay pulse (250 V, 125 μF, Ω = ∞) using a Gene Pulser-X cell (Bio-Rad, Hercules, CA, USA) in Gene Pulser 4 mm cuvettes (Bio-Rad). Shortly after electroporation, cells were supplemented with RPMI-1640 (5% FBS) and incubated with lysate or protein antigen (10 µg/ml). At this stage, peptide-loaded cells were also supplied with more peptide to reach a final concentration of 10 µg/ml. After 3 h of antigen loading, cells were gently washed with PBS and cultured in fresh RPMI-1640 (10% FBS) containing maturation cytokine cocktail as described above for either 24 h (in vivo experiments) or 48 h (in vitro experiments). All unloaded control groups received electroporation without antigens under identical culture conditions (mock-electroporation).

Naive splenic CD4⁺ T-cells were isolated using anti-CD4-conjugated magnetic beads (Miltenyi Biotec, San Diego CA, USA) with an autoMACS cell separator. CD4⁺ splenic T-cells were differentiated under T_H1 polarizing conditions. Briefly, 2.0–2.5 × 10⁶/ml cells were activated with 1.5 µg/ml plate-coated anti-CD3 (BD Pharmingen) and 1.5 µg/ml soluble anti-CD28 antibodies (BD Pharmingen) in addition to 10 µg/ml anti-IL-4 blocking antibody, 50 U/ml IL-2 and 20 ng/ml IL-12. Cells were cultured for 3–5 days, then harvested and washed for intracellular staining of IFN-γ.

OT-II or naive T-cells from mouse splenocytes were enriched by Pan T Cell Negative Isolation Kit (Miltenyi Biotec). Purified T-cells were cultured together with BMDC at a ratio of 10:1 unless specified otherwise in 96-well U-bottomed tissue culture plates with RPMI-1640 supplemented with 10% FBS. The medium and cells were collected on day 3 for further analysis.

Total RNA from cell lines or mouse bronchoalveolar lavage fluid (BALF) was extracted with QIAzol lysis reagent (Qiagen, Germantown, MD, USA); total RNA from mouse lung tissue was extracted with RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. cDNA was synthesized with the high-capacity cDNA reverse transcription kit (Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s instructions. qRT-PCR was performed using a 7500 Realtime PCR system (Applied Biosystems, Foster City, CA, USA) with the Taqman Realtime PCR assay (ThermoFisher Scientific, Waltham MA, USA) according to the manufacturer’s instructions. Primers: il6 (Mm00446190_m1, FAM), il12a (Mm00434165_m1, FAM), il12b (Mm01288989_m1, FAM), ifng (Mm01168134_m1, FAM), ifit1 (Mm00515153_m1, FAM), mx2 (Mm00488995_m1, FAM), oasl1 (Mm00455081_m1, FAM), irf7 (Mm00516793_g1, FAM), ifna1 (Mm03030145_gH, FAM), and 18s rRNA (4319413E, VIC).

Preparation of whole cell lysate: cells were lysed in 1% NP-40 lysis buffer containing protease inhibitor cocktail, phosphatase inhibitor cocktail 2 and phosphatase inhibitor cocktail 3 (all purchased from Sigma-Aldrich) on ice with vortexing every 10–15 min. Cell lysate was centrifuged at 14,000 g for 15 min and the cleared lysate was denatured with laemmli buffer (Bio-Rad) containing 5% β-mercaptoethanol (Bio-Rad) for 10 min. Denatured whole cell lysate samples were stored at −20°C for further analysis. Electrophoreses and blotting: proteins samples were separated by SDS-gel electrophoreses (Invitrogen) with subsequent transfer to a 0.45 µm nitrocellulose membrane (Bio-Rad) for antibody probing. All blocking and antibody staining steps were carried out in 5% BSA (RPI, Grainger) in 1 × TBST buffer (0.05% Tween-20). Western blotting chemiluminescent signal was detected with SuperSignal West Femto Maximum Sensitivity Substrate (ThermoFisher Scientific) using a ChemiDoc XRS digital imaging system supported by Image Lab software version 2.0.1 (Bio-Rad). Densitometry was performed using Image Lab software. Antibodies used for Western blotting: antihuman/mouse AIMp1 (Lifespan Biosciences Inc., Seattle, WA, USA); phospho-STAT1(Tyr701) (58D6), STAT1, phospho-STAT4(Tyr693) (D2E4), T-bet/TBX21 (V365), phosphor-IKKα/β (Ser176/180) (16A6), phospho-p38 MAPK(Thr180/Tyr182) (D3F9), p38, phospho-SAPK/JNK (Thr183/Tyr185) (G9), phospho-p44/42 MAPK (ERK1/2)(Thr202/Tyr204) (D13.14.4E), phospho-MKK3(Ser189)/MKK6 (Ser207) (D8E9), MAPKAPK-2 Antibody Sampler Kit, and the PP2A Antibody Sampler Kit were all purchased from Cell Signaling Technology (Danvers, MA, USA); DUSP1/MKP1 (V-15, Santa Cruz, Dallas, TX, USA); antihuman/mouse β-actin (Santa Cruz). Unedited Western blotting gel images can be found in supplementary figures (Figures S7 and S8 in Supplementary Material).

All flow cytometric analysis was performed using an LSR II flow cytometer (BD Biosciences) and analyzed with FlowJo version 10.0.7 (Tree Star Inc., Ashland, OR, USA). Antibodies used for flow cytometry: eFluor^® 450 antimouse CD3ε (145-2C11, eBioscience, San Diego, CA, USA), PE-Cy7 antimouse NK1.1 (PK136, BD Pharmingen), PE-Cy7 antimouse CD19 (1-D3, BD Pharmingen), PE antimouse CD4 (GK1.5, TONBO Biosciences, San Diego, CA, USA), FITC antimouse CD8α (53-6.7, BD Pharmingen), APC antimouse IFN-γ (XMG1.2, TONBO Biosciences), Pacific Blue antimouse CD11c (N418, Biolegend, San Diego, CA, USA), FITC antimouse CD86 (GL1, BD Pharmingen), PE antimouse CD40 (3/23, BD Pharmingen), PE antimouse I-A^b (AF6-120.1, BD Pharmingen), PE-Cy5 antimouse CD11b (M1/70, TONBO Biosciences), PE-Cy7 antimouse CD8α (53-6.7, TONBO Biosciences), and APC-Cy7 antimouse CD103 (2E7, Biolegend).

Mice were subcutaneously inoculated with 50,000 WT B16F10 melanoma tumor cells (American Type Culture Collection, Manassas, VA, USA) or 200,000 B16F0-OVA melanoma tumor cells (26) resuspended in 100 µl PBS on day 0. On the day of vaccination, mice were allocated based on tumor size. Tumor sizes of each animal were recorded and randomized so that each group possessed similar average tumor sizes and standard errors. 200,000 BMDC loaded with tumor antigen resuspended in 50 µl PBS were injected in the footpad. On boost days, the same number of cryopreserved BMDC were washed and injected in the same fashion as the primary vaccination. Tumor antigens used to vaccinated against B16F10 tumor were B16F10 melanoma tumor-derived mRNA and lysate (as described above). Tumor antigens used to vaccinate against B16F0-OVA tumor was SIINFEKL peptide + recombinant OVA protein (also described above). Tumor size was determined by external caliper measurement and calculated by means of the formula (Length × Width²) × π/6.

Mice inoculated with B16F0-OVA tumor cells and vaccinated/boosted with BMDCs on postinoculation day 19 were sacrificed, and tumor tissues were harvested for dissociation with the Tumor Dissociation Kit (mouse, Miltenyi Biotec) according to the manufacturer’s instructions. Briefly, melanoma tumors were cut into small pieces and transferred to the gentle MACS C-tube (Miltenyi Biotec) containing proper enzyme mix (enzyme D, R, A in RPMI serum-free media). C-tubes were loaded onto the gentleMACS Dissociator (Miltenyi Biotec) and processed by programs “m_impTumor_02,” 40 min at 37°C shaker, and “m_impTumor_03.” Cells were collected via 70-µm cell strainer followed by red blood cell (RBC) lysis.

Mice were challenged with influenza A/HongKong/8/68 (H3N2) Swiss mouse lung adapted strain of H3N2 influenza A virus graciously provided by Dr. Brian Gilbert (27). Infection was performed by a nebulized 20-min aerosol (Aerotech II nebulizer flowing at 10 L/min of room air generated from an Aridyne 2000 compressor) exposure of influenza virus diluted in MEM media containing 0.05% gelatin. All mice infected for any given experiment were infected simultaneously in a single exposure chamber. Following infection, mice were housed in Baylor College of Medicine’s biohazard facility.

Mice were anesthetized and BAL fluid was collected by instilling and withdrawing 0.8 ml of sterile PBS twice through the trachea. Total and differential cell counts in the BAL were determined with the standard hemocytometer and HEMA3 staining (Biochemical Sciences Inc., Swedesboro, NJ, USA) of BAL cells on cytospin slides.

Mouse lung RBC-free single cell suspensions were stimulated with 10 ng/ml phorbol 12-myristate 13-acetate (PMA; EMD Millipore) and 1 µg/ml ionomycin (Calcium Salt; ThermoFisher Scientific) overnight and supplemented with 10 µg/ml brefeldin A (eBioscience) for 5 h the following morning. Cells were stained for surface markers with anti-CD3, anti-CD4, and anti-CD8 antibodies and then fixed/permeabilized for intracellular staining of IFN-γ using the Cytofix/Cytoperm Kit (BD Biosciences) according to the manufacturer’s instructions.

Serum anti-influenza HA antibody isotypes were quantitated with the Mouse Monoclonal Antibody Isotyping Reagents (Sigma-Aldrich) according to the manufacturer’s instructions. Briefly, polystyrene multiwell plates were coated with 1 µg/ml influenza HA protein (A/Hongkong/1968 hemagglutinin protein, Sino Biological, Beijing, China) and incubated in immune sera diluted 10,000-fold in PBS for 2 h at room temperature. After stringent washing, immune complexes were incubated with the provided anti-isotype specific reagents and detected with peroxidase conjugated rabbit antigoat IgG.

To analyze human AIMP1 expression across multiple cancer types, we used the TCGA melanoma RNA-seq dataset (dataset ID: TCGA_SKCM_exp_HiSeqV2), TCGA ovarian cancer RNA-seq dataset (dataset ID: TCGA_OV_exp_HiSeq), and pan-cancer RNA-seq dataset (dataset ID: HiSeqV2_PANCAN). Patient sample characteristics were downloaded from the UCSC cancer browser (https://genome-cancer.ucsc.edu/proj/site/hgHeatmap/) (28). Cox regression p-values and log rank p-values were calculated using “survdiff” command in the “survival” package of R. Kaplan–Meier curves were drawn with the “survfit” command in the same R package by tertiles of AIMP1 expression. The numbers in parentheses indicate the fraction of the patient samples that could be linked to outcomes data. Analysis of IFN-γ expression was performed using the same methodology.

To predict the relative percentages of 22 different immune cells in basal-like breast cancer samples from GSE76275 (29), we downloaded data using the “GEOquery” package of R and used the gene expression profiling as the input for CIBERSORT (30, 31) enabling identification of 22 different immune cell signatures as previously described (30). The quantile normalization was disabled as the samples were from the same GEO data set. To generate the signaling score for T-cell subsets, the subset T-cell signature genes were summed (32).

Reverse phase protein array was performed at the antibody-based Proteomics Core of the Dan L. Duncan Comprehensive Cancer Center, Baylor College of Medicine. Cell lysate from BMDCs was isolated according to core facility protocol by the investigators. A detailed description of sample processing, antibody validation, and data analysis are available on the facility’s website (https://www.bcm.edu/centers/cancer-center/research/shared-resources/cprit-cancer-proteomics-and-metabolomics/reverse-phase-proteinarray). RPPA is performed as described (33). Heatmaps of log2-transformed data were generated with Cluster 3.0 (http://bonsai.hgc.jp/~mdehoon/software/cluster/) as a hierarchical cluster using Pearson correlation and a centered metric visualized and presented with Treeview software (http://taxonomy.zoology.gla.ac.uk/rod/treeview.html).

Gene expression profiling of BMDC was performed with Affymetrix Mouse Transcriptome Array 1.0 Chip (Affymetrix, Santa Clara, CA, USA) by the Sequencing and Microarray Facility, at the University of Texas MD Anderson Cancer Center (Houston, TX, USA). Total mRNA from BMDCs was isolated by RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. A detailed description of sample requirements and data preanalysis is available on the facility’s website (https://www.mdanderson.org/research/research-resources/core-facilities/sequencing-and-microarray-facility-smf/services-and-fees/microarray-services-overview.html). Data were analyzed and visualized with Transcriptome Analysis Console v3.0 (Affymetrix). Gene ontology analysis was performed with WEBGESTALT online resource. Information and raw array data can be found at GEO (GSE102282).

Significance of differences was determined by two-way analysis of variance (ANOVA) or one-way ANOVA using the Bonferroni post hoc test for multiple comparisons unless indicated otherwise. Kaplan–Meier survival curve significance was determined by the log rank (Mantel-Cox) test. Bioinformatic statistics were described in above method section accordingly (Cancer Database Survival Analysis and Immune Signaling Prediction). All data are displayed as the mean ± SEM, and all analyses were performed using the Prism software (GraphPad Software) unless indicated otherwise. Statistical significance was defined as p ≤ 0.05.

Antitumor immunity is broadly dependent upon T_H1 immune processes for destruction of neoplastic cells (34–36), and immunity to B16 melanoma is specifically known to be dependent upon T_H1 T-cell generation (37). Administration of recombinant E. coli-expressed human AIMp1 protein to tumor-bearing mice has been shown to have tumor inhibitory effects in a human xenograft model of stomach cancer (38, 39) as well as EG7 lymphoma (40), breast cancer (41), and others. To determine if intrinsic AIMp1 might possess inherent antitumor properties under conditions of physiologic homeostasis, we inoculated AIMp1^−/− mice with B16F10 melanoma and observed a tumor growth rate nearly double that exhibited in WT animals (Figure 1A). To investigate if the antitumor effect of AIMp1 relied upon activation of certain immune cell subsets (i.e., DC), we took advantage of the well-characterized B16-OVA tumor cell line which expresses the model antigen ovalbumin (OVA) (26, 42). Cohorts of WT and AIMp1^−/− mice were inoculated subcutaneously (s.c.) with 200,000 tumor cells and subsequently vaccinated with BMDC loaded with recombinant OVA protein and its immunodominant MHC class I peptide epitope SIINFEKL (Figure 1B). One WT and one AIMp1^−/− cohort were vaccinated with WT BMDC, while one WT and one AIMp1^−/− cohort were vaccinated with AIMp1^−/− BMDC. A non-vaccinated WT cohort was included as a control. Tumor growth was equally well controlled in both WT and AIMp1^−/− cohorts so long as the antigen-loaded BMDC were derived from WT mice. In contrast, tumor growth was poorly controlled in both WT and AIMp1^−/− cohorts vaccinated with AIMp1^−/− BMDC, ultimately reaching comparable size as non-vaccinated controls with only a short delay (Figure 1C). The trend was well reflected by Kaplan–Meier survival analysis (Figure 1D). These results indicated that BMDC-expressed AIMp1 is critical for vaccine-mediated rejection of immunogenic melanoma tumor, whereas the impact of AIMp1 in host effector cells appeared to have little relevance in this regard. WT tumor recipient mice treated as in this experiment were sacrificed on day 19 and tumor tissues were dissociated for immune cell analysis. Administration of WT BMDC vaccines upregulated the infiltration of IFN-γ⁺ T-cells and NK cells within the tumor, whereas infiltration of IFN-γ⁺ cells among animals vaccinated with AIMp1^−/− BMDC was identical to that of unvaccinated mice (Figure 1E). The data indicated that AIMp1 presence within the BMDC vaccine promotes type 1 polarization of tumor infiltrating lymphocytes in vivo.

AIMp1 is known to be released under conditions of cell stress, and recombinant AIMp1 protein exerts function directly or indirectly on multiple cell types including APCs, NK cells, and T-cells. Thus, it is possible that any AIMp1 released from BMDC vaccines during the antigen loading and maturation process also exerted pro-inflammatory effects in vivo that accounted for its antitumor potential. To control for this concern, mice were vaccinated with mock-electroporated WT DC treated identically to loaded BMDC but without exposure to antigen. In this experiment melanoma tumor growth in recipient mice vaccinated with WT unloaded BMDCs (mock-electroporated) was nearly identical to that of mice vaccinated with AIMp1^−/− antigen-loaded BMDC (Figure S1 in Supplementary Material). These data indicated that AIMp1 released from BMDC during vaccine preparation and injection could not account for any observed differential in antitumor effects. To further determine whether results could be due to differences in BMDC lineage development in the absence of AIMp1, WT and AIMp1^−/− BMDC populations were carefully characterized by flow cytometry. This analysis indicated that both populations of BMDC were almost uniformly CD11c⁺ with similar numbers of CD11b^mediumMHC Class II^high DC (43) as well as CD8α⁺CD103⁺ cross-presenting DC (44) whether immature, or matured (Figure S2 in Supplementary Material). The absence of observed differences in lineage marker expression suggested that AIMp1^−/− BMDC were the lineage equivalent of WT BMDC and that AIMp1 function was more likely to be upon immune regulation rather than upon BMDC development.

Previous literature has suggested a potential role for AIMp1 in T_H1 immunity (16–22, 45–47), and our work demonstrated BMDC-expressed AIMp1 is required for vaccine-mediated melanoma tumor rejection. To understand in greater detail whether BMDC-expressed AIMp1 can directly promote type 1 T-cell polarization, we began by investigating the function of AIMp1 in DC and T-cell crosstalk in vitro. Matured WT and AIMp1^−/− BMDC were cocultured with purified T-cells to test if the general activation of T-cells was affected by AIMp1 deficiency. WT CD3⁺T-cells cocultured with AIMp1^−/− BMDC for 72 h produced significantly less IFN-γ than ones cocultured with WT BMDC (Figure 2A). Additionally, OT-II T-cells which are transgenic CD4⁺T-cells specific for OVA antigen also produced significantly less IFN-γ when activated by OVA antigen-loaded AIMp1^−/− BMDC (Figure 2B), indicating antigen-specificity of the response. The production of IFN-γ from T-cells is a hallmark of T_H1 polarization and is regulated by the intracellular signaling and transcriptional activities of STAT1, STAT4, and T-bet. Western blot analysis revealed significant reductions in STAT1 (Y701) and STAT4 (Y693) phosphorylation along with concomitant impaired upregulation of T-bet expression among T-cells cocultured for 6 h with AIMp1^−/− BMDC (Figure 2C). Given that WT and AIMp1^−/− mice exhibited no discernable differences in lymphocyte subset numbers within secondary lymphoid organs (Figure S3A in Supplementary Material) and that in vitro APC-free T-cell polarization assays indicated no a priori T_H1 polarization deficit among AIMp1^−/− CD4⁺ T-cells (Figures S3B,C in Supplementary Material), the data indicated a direct role for BMDC-expressed AIMp1 in the observed T_H1 polarization defect imparted during DC/T-cell interaction.

In addition to the presentation of antigen, DCs also express cytokines and costimulatory molecules necessary for T-helper cell polarization. Secretion of the master T_H1-polarizing cytokine IL-12p70 (48) was dramatically impaired following LPS treatment of AIMp1^−/− BMDC in comparison to that secreted from WT BMDC, yet no deficits in other proinflammatory cytokines including IL-6 and IL-1β were observed (Figure 2D). Matured AIMp1^−/− BMDC also expressed lower levels of costimulatory molecules CD86 and CD40 as well as I-A^b (MHC class II) (Figure 2E). Using conditioned BMDC supernatants, we further demonstrated that supernatants derived from WT but not AIMp1^−/− OVA/SIINFEKL-loaded BMDC could induce upregulation of il12a and il12b mRNA transcripts in WT BMDC whereas any differential induction of other proinflammatory mRNA transcripts like il6 was not observed (Figure 2F). Interestingly, AIMp1^−/− recipient BMDC failed to upregulate il12a and il12b transcripts even in the presence of WT BMDC-conditioned supernatants. These results showed that BMDC-expressed AIMp1 can promote T_H1-specific programming via the regulation of specific cytokines and costimulatory molecules.

The observation that AIMp1 in BMDC specifically promotes downstream T_H1 T-cell polarization led us to investigate the underlying immune signaling pathways responsible. Previous work has demonstrated that the impact of AIMp1 on cell signaling can be complex, multifaceted, and frequently cell-type specific. Recombinant AIMp1 was reported to induce phosphorylation of ERK1/2, p38 MAPK, and c-Jun and to be critical for the production of TNF-α in THP-1 cells (16, 49). Activation of these MAP kinase signaling pathways was also induced by recombinant AIMp1 in skin fibroblasts; however, only ERK1/2 phosphorylation was responsible for AIMp1-mediated wound repair (15). Conversely, IL-12 production in primary macrophages induced by recombinant AIMp1 was reported to involve both NF-κB and p38 MAPK (22). To determine which of these various signaling pathways could be involved in BMDC-expressed AIMp1 signaling, we performed RPPA analysis of over 200 signaling proteins among WT and AIMp1^−/− BMDC at 30 min post-LPS treatment. This comprehensive analysis implicated significant impairment of early p38 MAPK signaling in AIMp1^−/− BMDC but no observed deficits in ERK, JNK, nor NF-κB signaling (Figure 3A). By Western blot analysis, we confirmed that AIMp1 deficiency did not impact the activation of NF-κB signaling in response to LPS treatment as measured by IKK phosphorylation (Figure 3B) and IκB degradation (data not shown). Similarly, there were no observed differences in phosphorylation levels of ERK1/2 or SAPK/JNK MAPK (Figure 3B). In contrast, phosphorylation of p38 MAPK and downstream substrate MAPKAPK2 were significantly impaired in AIMp1^−/− BMDC, whereas activation of MKK3/6 directly upstream of p38 MAPK was not (Figures 3C,D), suggesting that AIMp1 might exert its regulatory effects specifically at the level of p38 MAPK.

The p38 MAPK signaling pathway within BMDC has previously been linked to inflammatory function and induction of T_H1 polarization (50, 51). Treatment with p38 MAPK small molecule inhibitors (SB202190 and SB203580) led to downregulated phosphorylation of the p38 MAPK downstream substrate MAPKAPK2 (Figures S4A–D in Supplementary Material). In the presence of p38 MAPK inhibitors, IL-12 protein and mRNA transcripts were also downregulated in LPS-treated WT BMDC to levels comparable to those of AIMp1^−/− BMDC (SB202190 shown Figures 3E,F). Furthermore, T-cells cocultured with p38-inhibited WT BMDC produced IFN-γ at levels similar to those cocultured with AIMp1^−/− BMDC in a dose-dependent fashion (Figure 3G). All inhibitor-treated BMDC were thoroughly washed prior to T-cell coculture, mitigating any potential effects of the inhibitors on the T-cells themselves. Further investigation will be required to explain how specific impairment of p38 MAPK mediates downregulation of BMDC IL-12 production and IFN-γ from cocultured T-cells. Nonetheless, these results demonstrated that p38 MAPK inhibition in WT BMDC mimics the AIMp1^−/− phenotype with regard to T_H1 polarization.

p38 MAPK signaling is regulated by both kinases and phosphatases. PP2A and DUSP1/MKP1 are among several key regulatory phosphatases known to control the activity of p38 MAPK (52, 53). While there were no observed differences in levels of DUSP1 nor regulatory subunits A and B of PP2A following LPS treatment, levels of PP2A catalytic subunit C were significantly elevated by 60% among AIMp1^−/− BMDC (Figure S5 in Supplementary Material). Direct binding of AIMp1 to PP2A was also suggested by mass spectrometry analysis of proteins that coimmunoprecipitated with AIMp1 (data not shown). Therefore, we used small molecule inhibitors of DUSP1 and PP2A on AIMp1^−/− BMDC to determine if either of these could rescue T_H1 responses. While inhibition of DUSP1 had no impact on T_H1 impairment (data not shown), inhibition of PP2A in AIMp1^−/− BMDC was sufficient to restore p38 MAPK phosphorylation (Figures S4E–H in Supplementary Material) and downstream IFN-γ secretion of cocultured lymphocytes to WT levels (Figure 3H). These results suggested that upregulated PP2A phosphatase activity in AIMp1^−/− BMDC could account for decreased p38 MAPK function, though delineation of the complete mechanism may require additional investigation. Nonetheless, the data indicated that impaired p38 MAPK signaling is at least partially responsible for the reduction of downstream T_H1 polarization in AIMp1^−/− BMDC.

To study the global effects of AIMp1 deficiency on DC immune responses, we compared the transcriptional profile of WT and AIMp1^−/− BMDC, both mock loaded and antigen loaded, using the Mouse Exon 1.0 ST Array, comprised of over 1.2 million unique probe sets that correspond to 554,000 unique ESTs. This analysis identified 1,923 protein-encoding genes differentially expressed in a statistically significant fashion. Of 820 upregulated genes, the absence of AIMp1 impaired or abolished upregulation of 261; and of 1,103 downregulated genes, the absence of AIMp1 impaired or abolished downregulation of 228 (Figure 4A). Identification and in silico analysis of the 489 genes dependent upon AIMp1 expression suggested prominent AIMp1 regulation in a variety of immune processes, including the innate immune response and antiviral defense (Figure 4B). The top 50 most differentially regulated genes are shown in Table 1. Although not among the top 50, genes from the mouse major histocompatibility complex H-2 gene cluster which encodes proteins associated with antigen presentation were also identified. Important antiviral genes including interferon-activated genes (Ifit1, Ifit2, Ifit3, etc.) or interferon regulatory factor genes (Irf7), as well as innate immune sensors for viral RNA such as Oas family genes and Ddx58, were also shown to be critically regulated. The upregulation of such antiviral programming is known to be closely related to the antitumor potential of DCs (54), explaining the inability of AIMp1^−/− BMDC vaccines to mediate the rejection of B16-OVA tumors. The analysis also identified 124 differentially regulated miRNAs and hundreds of long non-coding (lnc) RNAs (data not shown), the biological significance of which may be delineated in future studies.

We have demonstrated the importance of AIMp1 in T_H1 polarization during antitumor immunity both in vivo and in vitro; however, the microarray data indicated that AIMp1 function might also be highly relevant to immunity against intracellular virus infection. Moreover, recent studies have indicated that AIMp1 is physiologically upregulated in bronchial epithelial cells in response to influenza virus infection (18). To study the function of AIMp1 in antiviral immunity, we challenged WT and AIMp1^−/− mice with varying titers of aerosolized A/Hong Kong/8/68 (H3N2) influenza A virus. A high infectious dose of 20 TCID₅₀ per mouse was sufficient to cause 100% lethality among AIMp1^−/− mice yet only resulted in 20% lethality among WT mice (Figure 5A). A sub-lethal dose of 7.50 TCID₅₀ per mouse still caused 60% mortality among AIMp1^−/− mice versus 0% among WT animals (Figure 5A). Survivors in both cohorts exhibited similar levels of weight loss and peak virus titers (data not shown) following the sub-lethal dose. Despite no overt clinical phenotype at the earliest stages of infection amongst the AIMp1^−/− survivors, AIMp1 deficiency mediated significant effects on innate and adaptive immune responses on days 7 and 15 postinfection. On day 7, we observed elevated levels of infiltrating neutrophils and macrophages in BALF derived from WT mice whereas these were significantly reduced in BALF derived from AIMp1^−/− mice at this time point (Figure 5B). On day 7, there were similar levels of lung-infiltrating IFN-γ⁺ T-cells observed in both WT and AIMp1^−/− mice; however, by day 15, IFN-γ⁺ lung-infiltrating T-cells were substantially reduced in AIMp1^−/− mice (Figures 5C–E). Analysis of anti-HA isotype-specific antibody titers in serum collected from mice on days 8 and 15 postinfection demonstrated that IgG_2a, the T_H1-specific antibody isotype in the mouse (55), was the predominant anti-HA isotype produced in WT mice at day 15 postinfection. In contrast, elevation of this isotype was highly impaired in AIMp1^−/− animals (Figures 5F,G). No statistical differences were observed among any other anti-HA antibody isotype. We performed real-time RT-PCR on BALF samples to validate the in vivo expression of selected differentially expressed antiviral genes identified by the microarray analysis above. Ifit1, mx2, oasl1, irf7, ifna1, and ifng were significantly upregulated following infection in WT mice but not in AIMp1^−/− mice (Figures 5H–M). Collectively, these results suggested that AIMp1 deficiency leads to insufficient innate and adaptive antiviral immune responses, correlating with the higher levels of mortality observed in these animals.

After decades of retrospective study, immune response signatures are now utilized as prospective prognostic indicators of patient outcome in cancer. Recently, analysis of 297 glioma samples from the Chinese Glioma Genome Atlas for local immune signature genes identified AIMP1, FOXO3, and ZBTB16 as positive prognostic indicators well correlated with patient overall survival (OS) (20). Based upon this observation, we sought to expand further upon the relationship between tumor-associated AIMp1 and good outcomes in cancer. First, we utilized CIBERSORT (30, 31) with the previously validated GSE76275 basal-like breast cancer and TGCA serious ovarian cystadenocarcinoma databases to examine the relationship between the relative percentage of 22 different immune cell subset signatures and AIMp1 expression levels in primary basal-like breast and ovarian tumors. In basal-like breast cancer, elevated levels of AIMP1 expression were found to be positively correlated with both increased numbers of activated tumor-infiltrating DCs and a T_H1 T-cell signature (Figure 6A). There were no significant correlations between AIMP1 expression and any other T_H signature including T_H2, T_H17, T_FH, and T_REG (Figure 6B). A similar observation was made using the TCGA ovarian cancer dataset (Figures 6C,D). While these data could not definitively identify DCs as the primary source of AIMP1 expression within primary tumors, they were highly suggestive and nonetheless affirmed the impactful relationship between AIMp1 and T_H1 immunity in a relevant, real world setting of human disease. To better correlate AIMP1 expression with outcomes data, patients in individual tumor databases were stratified into high, medium, and low tertiles by AIMP1 expression level, and Kaplan–Meier survival analysis was performed. Strong trends toward enhanced survival among patients with high AIMP1 expression were observed in ovarian cancer, breast cancer (data not shown) and melanoma. IFN-γ expression was shown to be critically important to survival in melanoma as reported previously (Figure S6 in Supplementary Material). A similar survival analysis was then performed using all 8,901 patient samples in the pan-cancer TCGA database for which outcomes data could be determined. This large-scale analysis demonstrated a highly significant 70% survival advantage among patients in medium or high level AIMP1 expression tertiles at 15 years postdiagnosis (Figure 6E). Interestingly, IFN-γ gene expression in these tumors correlated with only a non-significant 20% survival advantage among patients in the highest tertile in comparison to those in the medium and low tertiles (Figure 6F). Hence for tumors already thought to be immunogenic (i.e., melanoma), the data validated the importance of IFN-γ expression to survival. However, for the sum total of all tumors, the data indicated that primary tumor AIMp1 expression was at least as important as that of IFN-γ to long-term survival in cancer.