Abstract
During RBC transfusion, production of alloantibodies against RBC non-ABO Ags can cause hemolytic transfusion reactions and limit availability of compatible blood products, resulting in anemia-associated morbidity and mortality. Multiple studies have established that certain inflammatory disorders and inflammatory stimuli promote alloimmune responses to RBC Ags. However, the molecular mechanisms underlying these findings are poorly understood. Type I IFNs (IFN-α/β) are induced in inflammatory conditions associated with increased alloimmunization. By developing a new transgenic murine model, we demonstrate that signaling through the IFN-α/β receptor is required for inflammation-induced alloimmunization. Additionally, mitochondrial antiviral signaling protein–mediated signaling through cytosolic pattern recognition receptors was required for polyinosinic-polycytidylic acid–induced IFN-α/β production and alloimmunization. We further report that IFN-α, in the absence of an adjuvant, is sufficient to induce RBC alloimmunization. These findings raise the possibility that patients with IFN-α/β–mediated conditions, including autoimmunity and viral infections, may have an increased risk of RBC alloimmunization and may benefit from personalized transfusion protocols and/or targeted therapies.
Introduction
During allogeneic RBC transfusion, most non-ABO Ags on RBCs are not routinely matched between donor units and recipients. Hence, transfusion exposes recipients to numerous RBC alloantigens, including Kell, Duffy, and Kidd, which induce formation of Ag-specific IgG alloantibodies in 3–10% of all recipients and as many as 30–50% of transfusion-dependent patients with sickle cell disease (1–7). Such responses can cause potentially fatal hemolytic transfusion reactions (8, 9). Moreover, substantial morbidity and mortality can occur for patients who have alloantibodies against multiple Ags. Units of compatible RBCs are more difficult to locate, resulting in delays in therapy, and in extreme cases, death from lack of compatible blood (10, 11). However, the majority of transfusion recipients do not form alloantibodies and tools allowing prediction of such responses are not currently available. Thus, characterization of factors that promote RBC alloantibody responses could allow for identification of at-risk patients and intervention to mitigate alloimmunization and its detrimental effects.
Recent human studies have confirmed earlier mouse experiments indicating that the inflammatory state of transfusion recipients can influence the frequency of RBC alloantibody responses. Elevated alloimmunization rates have been reported for patients with acute chest syndrome, febrile transfusion reactions, and autoimmune diseases, including systemic lupus erythematosus (SLE) and inflammatory bowel disease (12–15). Also, similar to prior murine studies, a recent human study reported that inflammation associated with different infections can have distinct effects. Patients with viremia were reported to have increased rates of alloimmunization, whereas patients with Gram-negative bacterial infections had lower rates (16). These associations indicate that specific pathways activated in some inflammatory conditions, such as autoimmunity and viral infections, promote RBC alloimmunization. However, examination of these pathways, including inflammatory cytokine signaling, has only just begun (17, 18).
Given the large number of antigenic differences between human donors and recipients, multiple groups have used murine transfusions models that allow examination of alloimmune responses to a single donor RBC Ag (19–21). In the absence of an adjuvant, transfusion of mouse RBCs expressing human or model RBC Ags results in low-level alloimmune responses in some donor models and no response in others. However, treatment of transfusion recipients with inflammatory pathogen-associated molecular patterns has been shown to induce or enhance RBC alloimmune responses (22). Cotransfusion with CpG DNA or pretreatment with polyinosinic-polycytidylic acid [poly(I:C)], a mimetic of viral dsRNA, was shown to induce alloimmunization to human glycophorin A expressed on mouse RBCs (21, 23). In addition, poly(I:C) has been shown to enhance the magnitude of alloimmune responses in all models studied to date, including donor RBCs expressing hen egg lysozyme (HEL), the human KEL2 Ag (K2), and a chimeric protein containing HEL, OVA, and the Duffy Ag (HOD) (20, 24, 25).
Poly(I:C) promotes innate and adaptive immune responses through multiple pathways. Poly(I:C) is recognized by multiple pattern recognition receptors, including TLR3, and the retinoic acid inducible gene-I (RIG-I)–like receptors (RLRs), which include melanoma differentiation–associated gene 5 (MDA5) and RIG-I (26–28). Signaling through TLR3 and RLRs use distinct signaling adapter proteins to induce type I IFNs (IFN-α/β) and numerous NFҡB-regulated cytokines, including IL-6, IL-12, macrophage chemoattractant protein, and TNF-α (29, 30). Thus, poly(I:C) may augment RBC alloimmune responses by activating multiple pathways that induce critical inflammatory cytokines. However, the relative roles of each pathway and cytokine in inflammation-induced RBC alloimmunization have not been investigated.
Of the many inflammatory cytokines that may regulate inflammation-induced alloimmunization, IFN-α/β stands out as an important regulator of antiviral immunity and autoimmune pathology. IFN-α/β includes a single IFN-β and 13 IFN-α proteins that signal through a ubiquitously expressed dimeric receptor, consisting of IFN-α and -β receptor 1 (IFNAR1) and IFNAR2. Signal transduction results in the expression of numerous IFN-stimulated genes that inhibit viral replication and dissemination (31). Studies have also demonstrated that IFNAR signaling can promote antiviral neutralizing Ab responses (32, 33). In addition, IFN-α/β has been implicated in the pathogenesis of multiple autoimmune diseases, including rheumatoid arthritis, dermatomyositis, scleroderma, and Sjögren syndrome (34–38). In particular, many patients with SLE have elevated serum IFN-α/β and IFN-stimulated gene expression, which correlate with increased autoantibody production and disease severity (39–42). More than 50% of SLE-associated genetic variants have been linked to the IFN-α/β pathway (43), and clinical trials of anti–IFN-α therapy are in progress (44, 45).
Given that multiple inflammatory conditions are associated with elevated IFN-α/β production and RBC alloimmunization, we hypothesize that IFN-α/β may regulate inflammation-induced alloimmune responses to transfused RBC Ags. The antithetical K1 and K2 Ags in the Kell system are defined by a methionine or threonine at position 193, respectively. K1 is the most immunogenic RBC Ag that is not routinely matched between donor units and recipients (46, 47). With the exception of phenotypic matching or antigenic avoidance, there are no therapies that prevent K1 alloimmunization in humans (48). In this study, we introduce a new transgenic donor mouse that expresses the human K1 Ag on RBCs and formally test the role of IFN-α/β production and IFNAR signaling in a model of inflammation-induced RBC alloimmunization.
Materials and Methods
Mice
C57BL/6 and congenic C57BL/6-Ly5.1 wild-type (WT) mice were purchased from Charles River Laboratories (Wilmington, MA). IFNbmob/mob mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Ifnar1−/−, Mavs−/−, Irf3−/−, Irf7−/−, MyD88−/−, and Trif−/− mice were previously described (49–54). Appropriate gene-deficient mice were bred to produce Irf3/7−/− and MyD88−/−Trif−/− double knockout mice. All mice were 8–12 wk of age and were backcrossed to the C57BL/6 background for more than eight generations. All animal protocols were approved by the Yale Institutional Animal Care and Use Committee.
Transgenic mice expressing the human Kell glycoprotein consisting of the K1 variant (K1 mice) were generated as previously described (55). Briefly, K1 mice were produced by amplifying K2 cDNA from a human bone marrow cDNA library. K2 was mutated to K1 using a QuickChange mutagenesis kit (Stratagene, Santa Clara, CA) and inserted into a previously described RBC expression vector, containing the murine β-globin promoter, the β-globin locus control region, and the β-globin intron 2 and 3′ enhancers (56, 57). Following sequencing to rule out the introduction of other mutations, the construct was injected into fertilized embryos, which were implanted into pseudopregnant C57BL/6 mice. The same approach was described to generate K2 and K1 transgenic mice previously. As the initially described K1 transgenic mouse, previously published as KEL1A, had a low expression of the transgene, additional resources were invested to generate new transgenic founders, including the K1 mouse used in this study (55).
Inflammation-induced alloimmunization
Recipient mice were injected i.p. with 100 μg poly(I:C) (InvivoGen, San Diego, CA) at indicated time points relative to transfusion on day 0. Peripheral blood of K1 mice was collected in 12% citrate phosphate dextrose adenine (Jorgensen Labs, Melville, NY), leuko-reduced with a Pall syringe filter (East Hills, NY), and washed with PBS. Recipient WT, chimeric, or gene-deficient mice were transfused in the lateral tail vein with 75 μl of packed RBCs, the approximate mouse equivalent of 1 U of human RBCs. For CD4+ T cell depletion, 4 and 2 d prior to transfusion, recipient mice were injected i.p. with 200 μg GK1.5 Ab (Bio X Cell, West Lebanon, NH) or an isotype matched control. For in vivo IFN-α treatment, WT mice were cotransfused with K1 RBCs and recombinant mouse IFN-α (HC1040; Hycult Biotech, the Netherlands) at doses of 2–100 × 103 units in PBS. A total of 20 × 103 units of recombinant IFN-α (rIFN-α) approximates poly(I:C)-induced measured serum IFN-α (∼10 × 103 units per ml in 2 ml blood volume) at the time of infusion.
Detection of alloantibodies
Abs produced against the transgenic Kell glycoprotein (K1 variant) are described as anti-K1 IgG and were measured by flow-cytometric crossmatch 7, 14, 21, and 28 d after transfusion as previously described (24). K1 or WT RBCs were incubated with serum from transfused mice and subsequently stained for RBC-bound IgG (goat anti-mouse IgG APC; Jackson ImmunoResearch, West Grove, PA). The adjusted MFI was calculated by subtracting the reactivity of serum with syngeneic WT RBCs from the reactivity of serum with K1 RBCs. To maximize detection sensitivity, serum was not diluted. Figure data illustrates the peak Ab response, 28 d following transfusion. Flow cytometry of RBCs was performed using a BD FACSCalibur (San Jose, CA) and analyzed using FlowJo software (Tree Star, Ashland, OR).
Flow cytometry
For detection of K1 expression, splenocytes, peripheral blood, and platelet-rich plasma were stained with a monoclonal anti-Kell Ab (Mima-8) (58) followed by anti-mouse IgG and Abs against cell type–specific markers. Mima-8 recognizes the Jsb epitope on the human Kell glycoprotein, which is expressed by the K1 transgene. A combination of Mima-8 and Mima-9, which recognizes Kpb epitopes of the K1 transgene, was used to compare K1 expression by KEL1A and K1 mice. Platelet-rich plasma was generated by centrifuging peripheral blood at 8000 × g for 10 min. For analysis of dendritic cells (DCs), spleens were minced with a razor blade and filtered through 100 μm Nylon mesh prior to RBC lysis. Single-cell suspensions were stained with fluorescently conjugated Abs specific for cell-surface proteins, including CD19 (Clone: 6D5), TCRβ (H57-597), and I-A/I-E [MHC class II (MHCII), M5/114.15.2], CD86 (GL-1), Ly6C (HK1.4) and F4/80 (BM8) from BioLegend (San Diego, CA); CD45.1 (A20), CD11c (N418), CD11b (M1/70), CD8α (53-6.7), Ter-119, and Siglec H (eBio440c) from eBioscience (San Diego, CA), and CD41 (MWReg30) from BD Biosciences (San Jose, CA). Zombie-NIR (BioLegend) was used to exclude dead cells. Cells were acquired with a Miltenyi MACSQuant flow cytometer and analyzed using FlowJo.
Generation of bone marrow chimeras
Recipient WT C57BL/6 (CD45.2+) and Ifnar1−/− (CD45.2+) mice were exposed twice to x-ray irradiation (6.35 Gy, 3 h apart) using an X-RAD 320 irradiator (Precision X-ray, North Branford, CT). Recipients were injected i.v. with 3 × 106 bone marrow cells from WT C57BL/6-Ly5.1 (CD45.1+) or Ifnar1−/− mice 2–4 h after irradiation. Peripheral blood was analyzed for lymphocyte reconstitution 6 wk after bone marrow transfer. CD45.1 and CD45.2 congenic markers were used to identify the source of reconstituted cells. Mice were transfused 8–9 wk following bone marrow reconstitution.
Measurement of IFN-α/β
Serum IFN-α was measured by ELISA as previously described (59). For mRNA measurement, splenocytes were enriched for DCs using a mouse pan-DC enrichment kit (19763; Stemcell Technologies, Vancouver, BC). Enrichment was examined by flow cytometry. mRNA was isolated with a Qiagen RNEasy MiniKit (Valencia, CA), treated with DNAse, and reverse-transcribed with a Roche Applied Sciences kit (Indianapolis, IN). cDNA was quantitated with a KAPA SYBR FAST qPCR kit (KAPA Biosystems, Wilmington, MA), using a Stratagene Mx3000P instrument. Primers for Ifnα4 and Ifnβ are: Ifnα4 forward, 5′-CTG CTA CTT GGA ATG CAA CTC-3′; Ifnα4 reverse, 5′-CAG TCT TGC CAG CAA GTT GG-3′; Ifnβ forward, 5′-GCA CTG GGT GGA ATG AGA CTA TTG-3′; Ifnβ reverse, 5′-TTC TGA GGC ATC AAC TGA CAG GTC-3′.
Western blot analysis
Peripheral blood cells from K1 and C57BL/6 WT mice were lysed using hypotonic sodium phosphate. Samples were reduced, electrophoresed on a polyacrylamide gel, and blotted to nitrocellulose membranes. The KEL glycoprotein was detected using the mouse mAb, MM0435-12 × 3 (Novus Biologicals, Littleton, CO) followed by goat anti-mouse IgG1 HRP (Bethyl Laboratories, Montgomery, TX). Detection of β-actin was used as a loading control. Bands were detected with Immobilon Western Chemiluminescent HRP Substrate (Millipore, Darmstadt, Germany).
Statistics
Statistical analyses were performed using Graph Pad Prism software (San Diego, CA). Statistical significance between two groups was determined using a Mann–Whitney U test or an unpaired two-sided Student t test for nonparametric and parametric data, respectively. Significance between three or more groups was determined using a Kruskal–Wallis test with a Dunn posttest. Alloantibody data were analyzed with nonparametric tests. For all bar graphs, bars indicate the mean of data from individual mice, which are indicated by circles.
Results
KEL transgenic mice express the human KEL glycoprotein specifically on RBCs
To examine inflammation-induced alloantibody responses to the K1 Ag from the Kell system, we generated donor transgenic mice that express the Kell glycoprotein (K1 variant) on RBCs (K1 mice). K1 cDNA was inserted into a vector containing the murine β-globin promoter and enhancer regions to promote erythroid-specific transgene expression (56, 57). Multiple founder lines were generated, including K1 mice and the previously described KEL1A mice (55). Using monoclonal Abs against the Jsb epitopes (MIMA-8) and Kpb epitopes (MIMA-9) expressed by the K1 transgene, we observed higher expression on K1 RBCs, compared with KEL1A RBCs (Supplemental Fig. 1A). Given this result, we used K1 mice for subsequent experiments.
Western blot analysis demonstrated expression of K1 on peripheral blood cells of K1 mice (Supplemental Fig. 1B). We then examined K1 cell surface expression on RBCs, platelets, and leukocytes. As shown in Fig. 1A, K1 is expressed by Ter119+ RBCs in the peripheral blood of K1 mice. However, K1 mice do not express K1 on CD41+ platelets in peripheral blood or CD45+ leukocytes in the spleen (Fig. 1B, 1C). Thus, hematopoietic cells of K1 mice express the human KEL glycoprotein in an erythroid-specific manner.
KEL transgenic mice express the human KEL glycoprotein specifically on RBCs. (A–C) Flow cytometric analysis of K1 expression on peripheral blood (A) Ter119+ RBCs and (B) CD41+ platelets, and (C) spleen CD45+ leukocytes and RBCs of K1 and WT mice. Numbers on representative dot plots (left) from K1 mice indicate the percent of cells within the drawn gate. (B) Platelet-rich plasma was gated on CD45− Ter119− cells. (A–C) Histograms are gated on dot plot gates. Representative of three independent experiments.
KEL transgenic mice express the human KEL glycoprotein specifically on RBCs. (A–C) Flow cytometric analysis of K1 expression on peripheral blood (A) Ter119+ RBCs and (B) CD41+ platelets, and (C) spleen CD45+ leukocytes and RBCs of K1 and WT mice. Numbers on representative dot plots (left) from K1 mice indicate the percent of cells within the drawn gate. (B) Platelet-rich plasma was gated on CD45− Ter119− cells. (A–C) Histograms are gated on dot plot gates. Representative of three independent experiments.
Inflammation induces alloimmune responses to transfused K1-expressing RBCs
To assess the alloimmune response to K1-expressing RBCs (K1 RBCs), RBCs from K1 mice were leuko-reduced and transfused into WT controls in the presence or absence of poly(I:C) treatment. The anti-K1 IgG response was measured by flow cytometric crossmatch. Although no Ab response to K1 was detected in the absence of poly(I:C), recipients treated with poly(I:C) 3 h prior to transfusion produced anti-K1 IgG alloantibodies (Fig. 2A, 2B, Supplemental Fig. 2). To begin examination of mechanisms underlying poly(I:C)-induced alloimmunization, we assessed the alloimmune responses of WT mice treated with poly(I:C) at varying time points before or after transfusion on day 0. As shown in Fig. 2C, poly(I:C) treatment 3 h prior to transfusion induced maximal alloimmunization. In contrast, treatment one or more days before or after transfusion did not induce significant anti-K1 alloantibody production, compared with untreated controls. Thus, poly(I:C) administered in the immediate peri-transfusion period induces anti-K1 alloimmunization.
Poly(I:C) induces anti-K1 alloimmune responses. WT mice were injected with or without poly(I:C) and transfused with K1 RBCs. Serum anti-K1 IgG 28 d following transfusion was measured by flow-cytometric cross-match. (A) Representative histograms of flow cytometric cross-match. WT and K1 RBCs were incubated with serum of transfusion recipients, followed by anti-mouse IgG. (B) Quantified anti-K1 IgG. The adjusted MFI was calculated by subtracting the MFI of WT RBCs from the MFI of K1 RBCs. (A and B) Poly(I:C)-treated mice were injected i.p. 3 h prior to transfusion. (C) Poly(I:C) was administered at varying time points relative to K1 RBC transfusion on day 0. A dash (-) indicates no poly(I:C) treatment. Representative of three independent experiments with four to five mice per group. Bars indicate the mean of anti-K1 IgG levels from individual mice, indicated by circles. *p < 0.05. n.s., not significant by Kruskal–Wallis test with Dunn posttest.
Poly(I:C) induces anti-K1 alloimmune responses. WT mice were injected with or without poly(I:C) and transfused with K1 RBCs. Serum anti-K1 IgG 28 d following transfusion was measured by flow-cytometric cross-match. (A) Representative histograms of flow cytometric cross-match. WT and K1 RBCs were incubated with serum of transfusion recipients, followed by anti-mouse IgG. (B) Quantified anti-K1 IgG. The adjusted MFI was calculated by subtracting the MFI of WT RBCs from the MFI of K1 RBCs. (A and B) Poly(I:C)-treated mice were injected i.p. 3 h prior to transfusion. (C) Poly(I:C) was administered at varying time points relative to K1 RBC transfusion on day 0. A dash (-) indicates no poly(I:C) treatment. Representative of three independent experiments with four to five mice per group. Bars indicate the mean of anti-K1 IgG levels from individual mice, indicated by circles. *p < 0.05. n.s., not significant by Kruskal–Wallis test with Dunn posttest.
IFNAR signaling in hematopoietic cells is required for inflammation-induced RBC alloimmunization
One of the earliest phases of humoral immune responses is the production of innate cytokines that promote activation of APCs, including DCs. Although poly(I:C) induces multiple inflammatory cytokines, IFN-α/β has been shown to regulate DC activation and RBC consumption in other models (60–62). Thus, immediately prior to transfusion, we measured serum IFN-α of mice pretreated with poly(I:C) at varying time points. Compared to untreated mice, serum IFN-α was only significantly increased in mice treated with poly(I:C) 3 h prior to transfusion (Fig. 3A).
IFNAR signaling in hematopoietic cells is required for inflammation-induced K1 RBC alloimmunization. (A and C) IFN-α in serum of mice treated with poly(I:C) at the (A) indicated times or (C) 3 h prior to analysis by ELISA. (B and D) Anti-K1 IgG in serum of indicated mice 28 d following transfusion with K1 RBCs. Mice were administered poly(I:C) 3 h prior to transfusion. (D) Bone marrow chimeras were generated by transferring donor bone marrow into irradiated recipient mice 8–9 wk prior to transfusion. Representative of two (A, C, and D) and three (B) independent experiments with four to five mice per group. *p < 0.05. n.s., not significant by Mann–Whitney U test, **p < 0.01 by Kruskal–Wallis test with Dunn posttest.
IFNAR signaling in hematopoietic cells is required for inflammation-induced K1 RBC alloimmunization. (A and C) IFN-α in serum of mice treated with poly(I:C) at the (A) indicated times or (C) 3 h prior to analysis by ELISA. (B and D) Anti-K1 IgG in serum of indicated mice 28 d following transfusion with K1 RBCs. Mice were administered poly(I:C) 3 h prior to transfusion. (D) Bone marrow chimeras were generated by transferring donor bone marrow into irradiated recipient mice 8–9 wk prior to transfusion. Representative of two (A, C, and D) and three (B) independent experiments with four to five mice per group. *p < 0.05. n.s., not significant by Mann–Whitney U test, **p < 0.01 by Kruskal–Wallis test with Dunn posttest.
To determine whether IFN-α/β plays a role in inflammation-induced alloimmunization, we examined RBC alloimmune responses in mice lacking the only known receptor for IFN-α/β (Ifnar1−/−). Following poly(I:C) treatment and transfusion, production of anti-K1 alloantibodies by Ifnar1−/− mice was significantly diminished, compared to WT controls (Fig. 3B). Notably, WT and Ifnar1−/− mice contained comparable levels of serum IFN-α at the time of transfusion (Fig. 3C). Thus, the decreased alloimmune response in Ifnar1−/− mice is not due to differences in poly(I:C)-induced IFN-α production. However, the diminished response could be due to altered lymphoid structure in Ifnar1−/− mice. To address this possibility, we assessed alloimmune responses in bone marrow chimeric mice generated by reconstituting irradiated recipient mice with WT or Ifnar1−/− bone marrow. As shown in Fig. 3D, reconstitution of Ifnar1−/− mice with WT bone marrow rescued the anti-K1 IgG response. However, the alloimmune response of WT mice reconstituted with Ifnar1−/− bone marrow was completely abrogated. Collectively, these results demonstrate that IFNAR signaling in hematopoietic cells is required for inflammation-induced K1 RBC alloimmunization.
IFN-α/β promotes DC activation during T cell–dependent anti-K1 alloimmune responses
Recent studies in other transfusion models have demonstrated that T cell–dependent alloimmune responses to RBC Ags require Ag presentation by activated conventional DCs (cDCs) (17, 63). To determine whether poly(I:C)-mediated inflammation promotes cDC activation during alloimmunization, we measured expression of the activation marker, CD86, by spleen CD11chi MHCII+ cDCs from WT mice pretreated with poly(I:C) at varying time points. Six hours following transfusion with K1 RBCs, cDCs from mice untreated or treated with poly(I:C) 1 or 7 d prior to transfusion expressed comparable levels of CD86. In contrast, cDCs from mice treated with poly(I:C) 3 h prior to transfusion had elevated CD86 expression (Fig. 4A, 4B). To determine whether the increase in CD86 expression was mediated by IFNAR signaling, CD86 expression was also assessed in Ifnar1−/− mice treated with or without poly(I:C) 3 h prior to transfusion. Compared to WT mice, CD86 upregulation by cDCs from poly(I:C)-treated Ifnar1−/− mice was significantly reduced (Fig. 4C–E). Hence, IFNAR signaling regulates cDC activation during inflammation-induced K1 alloimmunization.
IFN-α/β regulates poly(I:C)-induced activation of cDCs. Indicated mice were transfused with K1 RBCs. (A and C) Flow cytometric analysis of spleen CD11chi MHCII+ cDCs from (A) WT or (C) indicated mice transfused 5 h prior to analysis and injected i.p. with or without poly(I:C) 3 h prior to transfusion. Cells are gated on live CD19− TCRβ− splenocytes. Numbers on dot plots indicate the percent of cells within the drawn gate. (B) CD86 expression of spleen cDCs of WT mice untreated or treated with poly(I:C) at the indicated time point. (D and E) CD86 expression of spleen cDCs gated in (C). (F) Serum anti-K1 IgG of isotype control (iso) and GK1.5 (anti-CD4) Ab treated WT mice 28 d following transfusion and poly(I:C) injection. Representative of three (A, B, and F) or two (C–E) independent experiments with three to five mice per group. *p < 0.05 by Mann–Whitney U test. (E) ***p < 0.001 by unpaired two-sided Student t test.
IFN-α/β regulates poly(I:C)-induced activation of cDCs. Indicated mice were transfused with K1 RBCs. (A and C) Flow cytometric analysis of spleen CD11chi MHCII+ cDCs from (A) WT or (C) indicated mice transfused 5 h prior to analysis and injected i.p. with or without poly(I:C) 3 h prior to transfusion. Cells are gated on live CD19− TCRβ− splenocytes. Numbers on dot plots indicate the percent of cells within the drawn gate. (B) CD86 expression of spleen cDCs of WT mice untreated or treated with poly(I:C) at the indicated time point. (D and E) CD86 expression of spleen cDCs gated in (C). (F) Serum anti-K1 IgG of isotype control (iso) and GK1.5 (anti-CD4) Ab treated WT mice 28 d following transfusion and poly(I:C) injection. Representative of three (A, B, and F) or two (C–E) independent experiments with three to five mice per group. *p < 0.05 by Mann–Whitney U test. (E) ***p < 0.001 by unpaired two-sided Student t test.
Given that cDC activation enhances Ag presentation to T cells, we determined whether production of anti-K1 alloantibodies requires T cell help. Treatment of mice with the anti-CD4 Ab, GK1.5, has been shown to deplete CD4+ T cells, which remained undetectable for at least 21 d (64). WT mice were pretreated with GK1.5 or an isotype control Ab prior to poly(I:C) treatment 3 h before transfusion. Compared to control mice, the alloimmune response of GK1.5-treated mice was fully abrogated (Fig. 4F). Collectively, these results indicate that IFN-α/β may regulate K1 T cell–dependent alloimmune responses, at least in part, by promoting cDC activation.
Poly(I:C)-mediated inflammation induces IFN-α/β production by CD8α+ cDCs
cDCs and plasmacytoid DCs (pDCs) have been shown to produce IFN-α/β in response to inflammatory stimuli (28, 65). Hence, to determine whether DCs produce IFN-α/β during inflammation-induced alloimmune responses, we measured IFN-α/β mRNA in WT splenocytes enriched for DCs by magnetic cell selection (Supplemental Fig. 3A). In contrast to PBS treated mice, DC-enriched splenocytes from poly(I:C)-treated mice produced IFN-α and IFN-β mRNA 8 h following treatment (Supplemental Fig. 3B, 3C).
DCs are comprised of multiple CD11c+ MHCII+ DC-subset populations, including CD11b+ and CD8α+ cDCs and Siglec H+ pDCs. To determine which DCs produce IFN-α/β during poly(I:C)-induced alloimmunization, we used IFN-β reporter mice that produce the yellow fluorescent protein (YFP) transcript linked to endogenous IFN-β mRNA (IFN-βmob/mob mice) (65). Scheu et al. (65) reported that YFP expression was not observed in poly(I:C)-treated IFN-βmob/mob mice until 6 h after treatment. Given that DC activation typically peaks 6–8 h following treatment with inflammatory stimuli (66), IFN-β/YFP expression was analyzed 8 h following treatment. Poly(I:C) treatment of IFN-βmob/mob mice induced IFN-β/YFP expression in a low percentage of CD11c+ and MHCII+ cells, but not CD11b+ Ly6C+ monocytes (Fig. 5A) or lymphocytes (Supplemental Fig. 4A, 4B). Analysis of CD11c+ MHCII+ DC subsets demonstrated that spleen CD11b+ cDCs and Siglec H+ pDCs from poly(I:C) and PBS-injected mice expressed comparable levels of IFN-β/YFP (Fig. 5B–E, Supplemental Fig. 4C, 4D). However, poly(I:C) treatment resulted in a significant increase in the percentage of IFN-β/YFP-expressing CD8α+ cDCs, compared with PBS-injected mice (Fig. 5D, 5E). This result was also represented by an increase in total IFN-β/YFP expression by CD8α+ cDCs (Supplemental Fig. 4C, 4D). These results indicate that a subset of CD8α+ cDCs produce IFN-α/β during inflammation-induced K1 alloimmunization.
Poly(I:C)-induced IFN-α/β is produced by CD8+ cDCs. IFN-βmob/mob mice were injected with poly(I:C) or PBS 8 h prior to analysis. (A) Representative flow cytometric analysis of IFN-β/YFP expression by spleen CD11chi cDCs, MHCII+ cells, and CD11b+ or Ly6C+ monocytes. Gated on live CD19− TCRβ− nonlymphocytes. (B) Representative flow cytometric analysis of spleen CD8α+ and CD11b+ cDCs (right) from poly(I:C)-treated IFN-βmob/mob mice, gated on total splenic cDCs (left). (C) Representative analysis of spleen CD11c+ Siglec H+ pDCs, gated on live CD19− TCRβ− cells. (D) IFN-β/YFP+ cDCs, and pDCs, gated as in (B) and (C). (E) Summary data of IFN-β/YFP+ cells identified in (D). (A–D) Numbers on plots indicate percent of cells within the drawn gate. Representative of three independent experiments with four to five mice per group. *p < 0.05. n.s., not significant by unpaired two-sided Student t test.
Poly(I:C)-induced IFN-α/β is produced by CD8+ cDCs. IFN-βmob/mob mice were injected with poly(I:C) or PBS 8 h prior to analysis. (A) Representative flow cytometric analysis of IFN-β/YFP expression by spleen CD11chi cDCs, MHCII+ cells, and CD11b+ or Ly6C+ monocytes. Gated on live CD19− TCRβ− nonlymphocytes. (B) Representative flow cytometric analysis of spleen CD8α+ and CD11b+ cDCs (right) from poly(I:C)-treated IFN-βmob/mob mice, gated on total splenic cDCs (left). (C) Representative analysis of spleen CD11c+ Siglec H+ pDCs, gated on live CD19− TCRβ− cells. (D) IFN-β/YFP+ cDCs, and pDCs, gated as in (B) and (C). (E) Summary data of IFN-β/YFP+ cells identified in (D). (A–D) Numbers on plots indicate percent of cells within the drawn gate. Representative of three independent experiments with four to five mice per group. *p < 0.05. n.s., not significant by unpaired two-sided Student t test.
Mitochondrial antiviral signaling protein–dependent IFN-α/β production is required for K1 RBC alloimmunization
The abrogated anti-K1 response in Ifnar1−/− mice indicates that IFN-α/β production may also be required for alloimmunization. As shown in Fig. 6A, poly(I:C) can induce IFN-α/β by binding endosomal TLR3, which utilizes the adaptor protein, TIR-domain–containing adaptor protein inducing IFN-β (TRIF), to mediate downstream signaling (28). MDA5 and RIG-I in the cytosol can also recognize poly(I:C) and use the signaling protein, mitochondrial antiviral signaling protein (MAVS), to induce IFN-α/β (27). Both pathways converge upon the canonical transcription factors in the nucleus, IFN regulatory factor (IRF) 3 and IRF7 (31).
IFN-α/β production is necessary and sufficient for K1 RBC alloimmunization. (A) Schematic of pathways leading to IFN-α/β production. Recognition of poly(I:C) by pattern recognition receptors activates multiple pathways leading to IFN-α/β induction. (B and C) Indicated mice were injected with poly(I:C) 3 h prior to (B) analysis or (C) transfusion with K1 RBCs. (B) IFN-α in serum of indicated mice. (C and D) Anti-K1 IgG in serum of indicated mice 28 d following transfusion. (D) WT mice were cotransfused with K1 RBCs and rIFN-α. Representative of two independent experiments with four to five mice per group. (B–D) *p < 0.05 by Mann–Whitney U test. MyD88, myeloid differentiation primary response gene 88.
IFN-α/β production is necessary and sufficient for K1 RBC alloimmunization. (A) Schematic of pathways leading to IFN-α/β production. Recognition of poly(I:C) by pattern recognition receptors activates multiple pathways leading to IFN-α/β induction. (B and C) Indicated mice were injected with poly(I:C) 3 h prior to (B) analysis or (C) transfusion with K1 RBCs. (B) IFN-α in serum of indicated mice. (C and D) Anti-K1 IgG in serum of indicated mice 28 d following transfusion. (D) WT mice were cotransfused with K1 RBCs and rIFN-α. Representative of two independent experiments with four to five mice per group. (B–D) *p < 0.05 by Mann–Whitney U test. MyD88, myeloid differentiation primary response gene 88.
To determine which pathway induces IFN-α/β production during alloimmunization, we measured serum IFN-α 3 h following poly(I:C) treatment in mice lacking MAVS (Mavs−/−), IRF3 and IRF7 (Irf3/7−/−), or the entire TLR pathway (MyD88−/− Trif−/−). WT and MyD88−/− Trif−/− mice produced comparable levels of IFN-α. However, levels of IFN-α produced by Mavs−/− and Irf3/7−/− mice were significantly diminished, compared with WT mice (Fig. 6B). To determine whether IFN-α/β production regulates inflammation-induced alloimmunization, we assessed anti-K1 alloimmune responses in these knockout mice. Although WT and MyD88−/− Trif−/− mice produced similar levels of anti-K1 IgG, alloimmune responses by Mavs−/− and Irf3/7−/− were significantly diminished (Fig. 6C). Collectively, these results indicate that IFN-α/β production is required for K1 RBC alloimmunization. Additionally, although TLR3-mediated signaling is not required for poly(I:C)-induced IFN-α/β production or alloimmunization, MAVS-dependent signaling is required.
IFN-α is sufficient to induce alloimmunization to transfused K1 RBCs
Although IFN-α/β production and signaling are required for anti-K1 alloimmune responses, poly(I:C) may promote alloimmunization by inducing multiple cytokines. Thus, to determine whether IFN-α/β is sufficient to induce alloimmunization in the absence of poly(I:C), we assessed alloimmune responses of WT mice cotransfused with K1 RBCs and rIFN-α. As shown in Fig. 6D, rIFN-α treatment induced anti-K1 alloantibody responses in a dose-dependent manner. Thus, IFN-α is sufficient to induce K1 RBC alloimmunization.
Discussion
Multiple studies have established that certain inflammatory disorders and inflammatory stimuli increase the frequency and magnitude of alloimmune responses to RBC Ags (12–16, 19–21). However, the molecular mechanisms underlying these findings have not been previously studied. We demonstrate that IFN-α/β production and IFNAR signaling are required for alloimmune responses to the K1 Ag in a murine model of inflammation-induced alloimmunization. Although poly(I:C)-induced alloimmunization was initially described years ago (20), the receptor-associated pathways that recognize poly(I:C) and induce RBC alloimmune responses have not been understood until now. In this manuscript, we show that MAVS-mediated pathways are required for poly(I:C)-induced IFN-α/β production and alloimmunization. Further, we report that IFN-α, in the absence of an adjuvant, is sufficient to induce RBC alloimmunization.
Assessing the alloimmune response of transfusion recipients exposed to inflammatory stimuli at varying times provides insight into the mechanisms underlying inflammation-induced alloimmunization. Previous studies in other murine transfusion models have shown that immediate pretreatment or cotransfusion of pathogen-associated molecular patterns can enhance RBC alloimmunization (20, 21). The results of the current study further indicate that the recipient’s inflammatory state at the time of transfusion can dictate the immune response to RBC alloantigens. Treatment of recipient mice with poly(I:C), only during the peri-transfusion period, induced cDC activation and T cell–dependent alloimmunization to the human K1 Ag. These findings agree with prior studies demonstrating a critical role for cDC activation in T cell–dependent alloimmune responses to stored RBCs expressing the HOD Ag (18, 63). A recent study reported that the timing of poly(I:C) treatment influences alloantibody responses to transfused HOD RBCs (19). However, in contrast to our findings, the anti-HEL alloimmune response of mice treated with poly(I:C) 7 d prior to HOD RBC transfusion was significantly higher than those treated just 4 h prior to transfusion. These somewhat conflicting findings may be due to qualitative or quantitative differences in the response to different RBC alloantigens. Whereas transient MAVS-mediated IFN-α/β production is critical to induce K1 alloimmunization in part by activating DCs, anti-HEL alloantibody responses may be regulated by alternate mechanisms.
In contrast to alloimmune responses to HOD or K2 RBCs (25, 24), we found that anti-K1 responses only occurred following treatment with inflammatory stimuli. Thus, the K1 RBC model allowed for direct evaluation of inflammatory pathways critical for alloimmunization. Examination of alloimmune responses in Ifnar1−/− mice and bone marrow chimeras revealed that IFNAR signaling in hematopoietic cells was required for K1 alloimmunization. Given that Ifnar1−/− and WT mice were previously reported to produce comparable IgM and IgG responses to immunization with multiple soluble T dependent Ags (49), this result was unlikely due to altered hematopoiesis in Ifnar1−/− mice. Rather, it is likely that IFNAR signaling in multiple hematopoietic cell types promotes K1 alloimmunization. We show that IFNAR signaling regulates inflammation-induced cDC activation following transfusion. Additionally, IFN-α/β has been shown to directly promote activation of other cell types, including lymphocytes, during T-dependent humoral immune responses (31).
Results of experiments in Ifnar1−/− mice indicated that IFN-α/β production may also be necessary for K1 alloimmunization. Indeed, the only WT mice that produced anti-K1 alloantibodies were transfused in the presence of elevated serum IFN-α. Moreover, following poly(I:C) treatment, mice lacking the canonical IFN-α/β transcription factors, IRF3 and IRF7, were unable to produce IFN-α or anti-K1 alloantibodies. In agreement with prior studies in other models (60, 65), we observed that a fraction of CD8α+ cDCs are the primary hematopoietic source of poly(I:C)-induced IFN-β. Although TLR3 has been shown to mediate poly(I:C)-induced IFN-α/β production (28), TLR3 signaling was dispensable for IFN-α production and K1 alloimmunization. Given that TLR3 is preferentially expressed in murine spleen CD8α+ cDCs (67), this result was unlikely due to lack of TLR3 expression in IFN-α/β–producing cells. The finding that poly(I:C)-induced IFN-α/β production is MAVS-dependent indicates that poly(I:C) gains access to the cytosol in CD8α+ cDCs and stimulates RLRs. Thus, we conclude that cytosolic RLRs of spleen CD8a+ cDCs recognize poly(I:C) and induce MAVS-dependent IFN-α/β production that is required for alloimmunization.
Although not addressed in this study, the dependence of K1 alloimmunization on IFN-α/β production and IFNAR signaling does not rule out a contributory role for NFҡB cytokines induced by poly(I:C). It is plausible that a lack of IFNAR signaling in Ifnar1−/− mice may inhibit production of other critical cytokines. However, Salem et al. (29) reported that poly(I:C) treatment of WT and Ifnar1−/− mice induced comparable levels of TNF-α, macrophage chemoattractant protein, IL-6, and IFN-γ. Treatment with rIFN-α was sufficient to induce K1 alloimmunization, yet other cytokines may augment the quality or magnitude of the response. IL-6 was recently shown to enhance alloimmunization to stored HOD RBCs by promoting T follicular helper cell differentiation (17). The role of other inflammation-induced cytokines in alloimmunization will be the subject of future study.
Clinically significant alloantibodies are formed in only 3–10% of transfusion recipients, and the majority of transfusion recipients never form detectable RBC alloantibodies (22). Given the role of inflammation in promoting alloimmunization, these nonresponders may have been initially transfused in the absence of inflammatory stimuli. Consistent with this notion, prior studies in some murine models have shown that transfusion of allogeneic RBCs in the absence of an inflammatory stimulus induces long-term nonresponsiveness to the Ag (68). Although not directly tested, transfusion of K1 RBCs in the absence of inflammation may have a similar result. In this study, demonstration that cotransfusion of rIFN-α is sufficient to prevent nonresponsiveness indicates that exposure to IFN-α/β–inducing stimuli during transfusion may be one factor that dictates responder versus nonresponder status.
In summary, we report that MAVS-mediated IFN-α/β production and IFNAR signaling are required for alloimmune responses to the human K1 Ag in a murine model of inflammation-induced alloimmunization. These findings provide a potential mechanistic basis for past observations of inflammation-induced alloimmunization. If they extend to human studies, patients with IFN-α/β–associated conditions and patients receiving IFN-α/β therapy may have an elevated risk of alloimmunization and may benefit from personalized transfusion protocols, including extended Ag matching.
Footnotes
This work was supported by grants from the National Blood Foundation (R13672) (to D.R.G.) and the National Institutes of Health/National Heart, Lung, and Blood Institute (R01 HL126076) (to J.E.H.) and (T32 HL007974-14) (to Brian Smith, Chair of the Department of Laboratory Medicine, Yale University School of Medicine).
This work was presented in abstract form at the annual meeting of the American Society of Hematology, December 3, 2016, San Diego, CA.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- cDC
conventional DC
- DC
dendritic cell
- HEL
hen egg lysozyme
- IFNAR1
IFN-α and -β receptor 1
- IRF
IFN regulatory factor
- MAVS
mitochondrial antiviral signaling protein
- MDA5
melanoma differentiation–associated gene 5
- MHCII
MHC class II
- pDC
plasmacytoid DC
- poly(I:C)
polyinosinic-polycytidylic acid
- rIFN-α
recombinant IFN-α
- RIG-I
retinoic acid inducible gene-I
- RLR
retinoic acid inducible gene-I–like receptor
- SLE
systemic lupus erythematosus
- TRIF
TIR-domain–containing adaptor protein inducing IFN-β
- WT
wild type
- YFP
yellow fluorescent protein.
References
Disclosures
The authors have no financial conflicts of interest.