CD28 Deficiency Enhances Type I IFN Production by Murine Plasmacytoid Dendritic Cells Free

Type I IFNs (IFN-I) play a crucial role in orchestrating the immune response to multiple disease settings, including viral infections, cancers, tissue injury, and autoimmune disease (1). IFN-I is a pleiotropic cytokine family found among mammalian species that includes several IFN-α and one IFN-β isoforms that signal through a common ubiquitously expressed receptor (IFNαβ-R), promoting both autocrine and paracrine activation and leading to phosphorylation of STAT 1 and 2. The result of these interactions is a positive feedback loop that drives further IFN-I production as well as the induction of hundreds of IFN-I–stimulated genes (ISGs) (2). These ISGs act in concert to create a potent antiviral state and orchestrate the activation of almost all innate and adaptive immune cells. Although almost all cell types can produce IFN-I, plasmacytoid dendritic cells (pDCs) are highly specialized to rapidly secrete copious amounts of these cytokines. Not only do pDCs produce up to 1000 times more IFN-I than other cell types but they also synthesize a broader range of IFN-I isoforms (3). pDCs express endosomal TLR7 and TLR9, which recognize ssRNA and unmethylated CpG-containing motifs (from microbial or self-origin), respectively (4). Engagement of TLR7 or TLR9 in pDCs leads to production of IFN-I (both IFN-α and IFN-β isoforms) as well as proinflammatory cytokines and upregulation of costimulatory molecules such as CD80, CD86, and MHC class II (MHC-II) (5–10).

As such, pDCs play an important role during several in vivo viral infections such as those caused by murine CMV (MCMV) (11, 12), respiratory syncytial virus (13, 14), and mouse hepatitis virus (15), among others (15–17). Furthermore, persistent viruses such as HIV and hepatitis C virus induce substantial IFN-I production upon incubation with pDCs (17, 18), and similar effects are observed early after in vivo infection with persistent strains of lymphocytic choriomeningitis virus (LCMV WE or clone 13; Cl13) (19, 20). However, pDC IFN-I production becomes exhausted during later stages of chronic viral infection, an event accompanied by enhanced susceptibility to opportunistic pathogens (18, 21–23). Similarly, pDC IFN-I production is also attenuated in tumor microenvironments, correlating with cancer progression (24). In contrast, uncontrolled IFN-I production by pDCs is associated with autoimmune diseases such as psoriasis (25), type I diabetes (26), and experimental autoimmune encephalomyelitis (27). In particular, in systemic lupus erythematosus patients, pDCs accumulate in target tissues and exhibit sustained IFN-I production, and pDCs were shown to be critical for promoting systemic lupus erythematosus pathogenesis (28–30). Finally, pDC IFN-I production also promotes innate defenses following tissue injury, playing a critical role in regulating cutaneous wound healing (31). Taken together, these studies demonstrate the importance of fine-tuning the magnitude of pDC IFN-I response and highlight the significant implications of pDC IFN-I regulation for numerous human illnesses.

In the current study, we compared the gene expression profiles of pDCs and conventional DCs (cDCs) to gain insight on putative pDC IFN-I regulators. Unexpectedly, we found that CD28, a cell surface stimulatory receptor constitutively expressed in T cells (32), was highly and selectively expressed in pDCs but not cDCs. Remarkably, CD28 expression negatively regulated pDC IFN-I production in response to in vitro TLR stimulation and in vivo viral infections or tissue injury. Moreover, bone marrow (BM) chimeras revealed a cell-intrinsic effect of CD28 expression in suppressing pDC functions. Thus, our study identified a novel role for the prototypic T cell stimulatory molecule CD28 as a negative regulator of pDC function both in vitro and in vivo. Considering that CD28 is fundamental for T cell priming (32), our work raises the possibility that CD28 may be part of a previously unrecognized molecular pathway that inhibits innate responses while promoting adaptive immunity.

C57BL/6 (wild-type [WT]), C57BL/6 CD45.1⁺, CD80/86 double knockout (dko), and CD28ko mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice (6–12 wk old) were infected i.p. with 1 × 10⁴ PFU MCMV Smith or 2 × 10⁶ LCMV Cl13 i.v. Viruses were propagated and quantified as described previously (33, 34).

Mice backs were shaved and depilated (Veet; Reckitt Benckiser, Slough, U.K.) immediately before injury. Mechanical injury was then applied by tape stripping, using 20 strokes of transparent tape (3M; Scotch) across the back. Injured skin was excised, digested with 1 mg/ml dispase (Sigma-Aldrich, St. Louis, MO), 200 U/ml collagenase type I (Invitrogen, Carlsbad, CA), and 200 U/ml hyaluronidase (SERVA Electrophoresis, Heidelberg, Germany) for 30 min at 37°C and then mechanically dissected to generate a single-cell suspension. Cells were stained with 10 μg/ml anti–PDCA-1 allophycocyanin (JF05-1C2.4.1; Miltenyi Biotec, Auburn, CA), anti–CD11c-PE (HL3), anti–B220-FITC (RA3-6B2; BD Biosciences, San Diego, CA), and anti–CD45-PerCP (30-F11; eBioscience, San Diego, CA). Cells were acquired on a FACSCalibur (BD Biosciences, San Jose, CA) and analyzed using FlowJo software (Tree Star, Ashland, OR).

Residual axillary lymph node samples (n = 5), following diagnostic studies, were used for these studies where samples were incubated with RPMI 1640 medium + collagenase (1 mg/ml; Roche, Indianapolis, IN) for 20 min at 37°C and passed through a 100-μm strainer to achieve a single-cell suspension. Cal-1 cells were provided by Dr. S. Kamihira (Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan). PBLs (n = 7 donors) were isolated using standard Ficoll gradient separation as described previously (35). Cells were stained with CD16 (3G8), CD56 (MEM-188), CD14 (HCD14), CD19 (HIB19)-Pacific Blue, CD3 allophycocyanin Cy7 (UCHT1), B220 Percp Cy5.5 (RA3-6B2), HLADR PE Cy7 (L243), CD11c Alexa 488 (3.9), and CD123 allophycocyanin (6H6) (BioLegend, San Diego, CA). pDCs were identified as Lineage (CD16, CD56, CD14, CD19, B220, CD3)⁻HLADR⁺CD11c⁻CD123⁺. T cells were identified as CD3⁺. T cells and pDCs (lymph node biopsies) or Cal-1 pDCs were stained for CD28 expression with CD28 PE (CD28.2) or isotype control (mouse IgG1) (BioLegend) and acquired on a BD LSRII and analyzed using FlowJo software.

BM cells were isolated from femurs and tibias, and a single-cell suspension was prepared and cultured 7–8 d in the presence of 100 ng/ml Flt3 ligand (Flt3L) (Amgen, Thousand Oaks, CA; Cell Dex Therapeutics, Needham, MA) as described previously (36). Spleens were incubated with 1 mg/ml collagenase D for 20 min at 37°C and passed through a 100-μm strainer to achieve a single-cell suspension. Splenocytes were enriched with PanDC microbeads using an Automacs system (Miltenyi Biotec). PanDC⁺ fractions were stained with propidium iodide (PI) and FACS-purified using a BD ARIA II (BD Biosciences) for pDCs (PI⁻CD11c^{intermediate/dim}CD11b⁻B220⁺PDCA⁺) and CD11b⁺ cDCs (PI⁻CD11c⁺B220⁻CD11b⁺) after B (CD19), T (Thy1.2), and NK (Nk1.1) cell exclusion. BM-pDCs and BM-cDCs were stained and sorted as PI⁻CD11c⁺CD11b⁻B220⁺PDCA⁺ and PI⁻CD11c⁺B220⁻CD11b⁺, respectively. Purity of the cells was >92%. Cells were stimulated with CpG B 1668 (Integrated DNA Technologies, San Diego, CA) at 0.1 μM (BM-derived DC) or 1 μM (splenic DC), 10 μM CpG A 2336 (InvivoGen, San Diego, CA), and 100 μM loxoribine (InvivoGen).

Total IFN-I bioactivity was measured by luciferase bioassay with reference to a recombinant mouse IFN-β standard (Research Diagnostics, Concord, MA) using a L-929 cell line transfected with an IFN-sensitive luciferase as described previously (37). TNF-α and IL-6 were measured by ELISA (eBioscience). Supernatant cytokine levels were measured 15 h poststimulation.

The following Abs were used to stain murine BM or spleen cells: anti–CD3-PerCP-Cy5.5 (145-2C11; BD Biosciences), Thy 1.2-PE (53-2.1), CD19-PE (eBio1D3), NK 1.1-PE (PK136), CD11c-PE or allophycocyanin (N418), CD11b-PE-Cy7 (M1/70), B220-allophycocyanin eFluor 780 or efluor 450 (RA3-6B2), PDCA-1-FITC (eBio927), CD45.2-allophycocyanin-Cy7 (104), CD45.1-PerCy5.5 (A20), MHC-II (I-A/I-E) efluor 450 (M5/114.15.2), CD28-allophycocyanin (E18) and mouse IgG2b-allophycocyanin isotype control for anti–CD28-allophycocyanin (eBioscience), and CD86 PE Cy7 or PE (IT2.2; BioLegend). PI or Ghost dye (Tonbo Biosciences, San Diego, CA) was used to exclude dead cells and to measure cell viability where indicated. Cells were acquired with an LSRII flow cytometer (BD Biosciences). Data were analyzed with FlowJo software.

WT CD45.1⁺ C57BL/6 recipient mice were sublethally irradiated with 1000 rad and reconstituted with a 50:50 mix of BM cells from CD45.1⁺ WT mice and CD45.2⁺ CD28ko mice. 10 × 10⁶ BM cells were transferred i.v. into the irradiated recipient mice, which were treated with antibiotics (8 mg/ml trimethoprim and 40 mg/ml sulfamethoxazole supplied in drinking water) for 3 wk to prevent infection and allow immune reconstitution. Reconstitution was analyzed 8 wk after BM transfer, and the ratio of WT:CD28ko cells was determined by flow cytometry. Mice were then infected with MCMV or LCMV Cl13 as indicated.

Total RNA was extracted using RNeasy kits (Qiagen, Redwood City, CA), digested with DNase I (RNase-free DNase set; Qiagen), and reverse-transcribed into cDNA using Superscript III RT (Invitrogen). cDNA quantification was performed using SYBR Green PCR kits and a Real-Time PCR Detection System (Applied Biosystems, Carlsbad, CA). The relative transcript levels were normalized against Gapdh as described previously (22). The following primers were used: Ifnα primers recognizing Ifnα 4 and 6, 5′-TATGTCCTCACAGCCAGCAG-3′ (forward) and 5′-TTCTGCAATGACCTCCATCA-3′ (reverse); Ifnβ1, 5′-CTGGCTTCCATCATGAACAA-3′ (forward) and 5′-GAGGGCTGTGGTGGAGAA-3′ (reverse); Cd28, 5′-ACAGTTGGGCCACTTGTTGTCCTTT-3′ (forward) and 5′-GCTCCCAATGGTGCCTTCTGGA-3′ (reverse); Mx1, 5′-CAACTGGAATCCTCCTGGAA-3′ (forward) and 5′-GGCTCTCCTCAGAGGTATCA-3′ (reverse); Rig-I recognizing Ddx58, 5′-CGGGACCCACTGCCTCAGGT-3′ (forward) and 5′-GCATCCAGGGCGGCACAGAG-3′ (reverse); Tnfα, 5′-CCCTCACACTCAGATCATCTTCT-3′ (forward) and 5′-GCTACGACGTGGGCTACAG-3′ (reverse); and MCMV eI, 5′-GAGTCTGGAACCGAAACCGT-3′ (forward) and 5′-GTCGCTGTTATCATTCCCCAC-3′ (reverse). Transcript levels of Il6 were determined relative to Gapdh using primer and probe sets from the Universal Probe Library (Roche). Cytokine transcript levels were measured 6 h poststimulation.

RNA extracted from FACS-purified splenic pDCs, CD8⁺ DCs, and CD11b⁺ DCs from uninfected WT mice were used for DNA microarray using GeneChip mouse genome 430 2.0 arrays (Affymetrix, Santa Clara, CA). Differential gene expression was determined as fold of change of indicated genes over background intensity. Microarray data have been deposited in the National Center for Biotechnology Information’s Gene Expression Omnibus (38) under the accession number GSE75834 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE75834).

CD28 cDNA clone (MGC premier cDNA; TransOMIC Technologies, Huntsville, AL) was amplified by PCR, where XhoI and HpaI restriction sites were incorporated using primers: 5′-GACTCGAGGCCGCCACCATGACACTCAGGCTGCTGTTCTTGG-3′ (forward) and 5′-TCGTTAACTCAGGGGCGGTACGCTGCAAAGT-3′ (reverse). The amplicon was gel-purified and digested with XhoI and HpaI restriction enzymes, according to the manufacturer’s instructions (New England Biolabs, Ipswich, MA), and ligated to previously digested pMIGR (39) (GFP-labeled retroviral vector; RVGFP) (Addgene, Cambridge, MA). 293T cells were transfected with RVGFP-CD28 or empty vector for control, LT1 transfection reagent (Mirus Bio, Madison, WI), and the pcl-Eco packaging plasmid required to produce viral vectors. Seventy-two hours later, supernatants were harvested and stored at −80°C until transfection of pDCs.

At day 3 post-Flt3L culture, CD28ko BM cells were transduced with RVGFP-CD28 or empty vector control (RVGFP) and polybrene reagent (Fisher Scientific) and spin-infected room temperature at 1000 × g at 90 min. Cells were incubated 37°C overnight. On day 4 post-Flt3L culture, cells were again transduced with RVGFP-CD28 or RVGFP and incubated 3 h at 37°C. Cells were then washed in PBS and placed in fresh DC medium + Flt3L. On day 8 postculture, cells were harvested and FACS-purified for GFP⁺ pDC fractions. Purified cells were stimulated with 0.1 μM CpG B 1668 6 h and harvested for mRNA detection of IFN-I and proinflammatory cytokine transcripts relative to Gapdh.

Unpaired Student t tests or ANOVA tests were performed using GraphPad Prism software (GraphPad, La Jolla, CA). Error bars represent mean ± SEM. A p value <0.05 was considered statistically significant.

Human axillary lymph node biopsies sent to the University of California San Diego Clinical Flow Cytometry Laboratory were deidentified and used for research, according to University of California San Diego Institutional Review Board–approved protocol number 130973x. Peripheral blood samples were collected from healthy volunteers (written informed consent was obtained prior to study inclusion) at the University of California San Diego CFAR Clinical Investigation Core Antiviral Research Center, according to University of California San Diego Institutional Review Board–approved protocol number 110522. Mice were bred and maintained in a closed breeding facility and mouse handling conformed to the requirements of the National Institutes of Health and the Institutional Animal Care and Use Guidelines of University of California San Diego, according to approved protocol S07315.

Previous analyses have demonstrated that, although pDCs and cDC subsets (including CD8⁺ and CD11b⁺ cDCs) are derived from a distinct branch of the leukocyte family tree and exhibit functional differences, they maintain an evolutionarily conserved transcriptional signature (40, 41). Therefore, we surmised that a comparison between pDCs and cDCs may highlight regulatory molecules that selectively modulate pDC function. We noted that the expression of the gene encoding the prototypic T cell costimulatory molecule CD28 was 59 and 57 times higher in pDCs compared with CD11b⁺ and CD8⁺ cDCs, respectively (Fig. 1A). Other CD28 family receptors (42) were either undetectable in pDCs (i.e., Ctla4, Pdcd1, and Icos) or equally expressed in all DC subsets (i.e., Btla). Notably, DC subset or T cell–specific gene transcripts were selectively expressed or absent, respectively. To confirm CD28 expression in pDCs, we first determined Cd28 transcript levels by quantitative PCR (qPCR) in both murine BM–derived DCs 7 d postculture with Flt3L and splenic DC subsets. We observed that Cd28 transcripts were undetectable in cDCs but were significantly expressed in both BM-derived and splenic pDCs, albeit to a lesser extent than in splenic T cells (Fig. 1B, Supplemental Fig. 1A, 1B). Surface expression of CD28 protein was also present in pDCs freshly obtained from spleen, BM, blood (Fig. 1C), and BM-Flt3L cultures (Fig. 1D), with levels lower than in T cells but contrasting the background expression in cDCs. Of note, basal surface and intracellular CD28 expression were similar, and CD28 expression did not increase in response to TLR stimulation (data not shown). Finally, we examined CD28 protein expression in human pDCs. For this, we first analyzed the Cal-1 human pDC cell line (43) and compared it with non-pDC human cell lines A549 (lung epithelial cells) (44) and Thp-1 (monocytic cells) (45). Similar to murine pDCs, CD28 expression was observed in Cal-1 but not A549 or Thp-1 cells (Fig. 1E). Furthermore, analysis of primary human pDCs revealed that, although CD28 was undetectable in blood pDCs (data not shown), it was consistently expressed above isotype levels in pDCs obtained from lymph node biopsies (Fig. 1F, Supplemental Fig. 1C). Altogether, these data indicated that the prototypic T cell costimulatory molecule CD28 is constitutively expressed in pDCs.

To investigate a putative role for CD28 in pDC differentiation, survival, and/or function, we first examined pDCs obtained from WT versus CD28ko BM-Flt3L cultures. Similar percentages and numbers of live pDCs were obtained at day 7–8 postculture, indicating normal pDC and cDC development from BM progenitors (Supplemental Fig. 2A, 2B). In contrast, significantly higher levels of IFN-I were detected upon stimulation of FACS-purified CD28ko BM-derived pDCs with the TLR7 agonist loxoribine or TLR9 agonist CpG (Supplemental Fig. 2C). We next sought to extend these findings to pDCs freshly isolated from spleens of WT and CD28ko mice. Consistent with normal development of DCs in the absence of CD28, we observed comparable percentages and numbers of pDCs and cDCs in WT and CD28ko spleens (Fig. 2A and data not shown, respectively). Importantly, as described for BM-Flt3L cultures, significantly increased levels of IFN-I were detected when FACS-purified splenic pDCs from CD28ko mice were stimulated with loxoribine, CpG A, or CpG B. Moreover, quantification of TNF-α and IL-6 in these same culture conditions revealed a similarly increased production of these proinflammatory cytokines by CD28ko compared with WT pDCs, whereas viability of both WT and CD28ko pDCs was similar following agonist stimulation (Fig. 2B). Despite changes in cytokine response, CD86 (a hallmark of DC maturation (46)) and MHC-II were equally expressed in WT versus CD28ko pDCs before and after TLR stimulation, although a minor increase in MHC-II was observed in loxoribine-treated CD28ko pDCs (Fig. 2C). Furthermore, although incubation of murine WT pDCs and human Cal-1 pDCs with agonistic anti-CD28 Ab or rCTLA4-Ig (47, 48) showed no effect on cytokine production (data not shown), we did observe reduced Ifna in CD28ko pDCs upon reconstitution of CD28 levels (Fig. 2D), ruling out any off-target effects in IFN-I production by CD28ko pDC. The levels of Tnfa and Il6 transcripts were, however, unchanged in CD28ko pDCs with restored CD28 expression, raising the possibility that the effect of CD28 may be more profound and/or selective for IFN-I than for proinflammatory cytokines.

Given that CD80 and CD86 engage CD28 to promote T cell activation and are upregulated in pDCs upon TLR stimulation (Fig. 2C) (8–10), we next evaluated their putative role in pDC IFN-I production. Interestingly, FACS-purified pDCs from CD80/86 dko BM-Flt3L cultures exhibited enhanced IFN-I production in response to loxoribine and CpG stimulation when compared with WT BMpDCs (Fig. 2F), although this effect was not observed with FACS-purified splenic pDCs from CD80/86 dko mice (data not shown), possibly because of other anomalies in CD80/86 dko mice. Altogether, these data demonstrated that, in the absence of CD28, pDC development and maturation were unchanged, but IFN-I production was significantly increased in response to TLR stimulation, revealing a novel regulatory role for CD28 in limiting the magnitude of pDC cytokine responses. Furthermore, our data suggest that CD80/CD86 molecules, which are natural CD28 ligands, may also mediate IFN-I downregulation in pDCs.

We next investigated the effect of CD28 signaling on pDC IFN-I response and host defense to viral infection in vivo. We first infected WT and CD28ko mice with LCMV Cl13, an ssRNA virus in which pDCs, via TLR7, contribute to peak IFN-I response at 24 h postinfection (p.i.) (20, 22). At this time point postinfection, we detected higher IFN-I in serum from CD28ko versus WT mice infected with LCMV, whereas no difference in systemic TNF-α levels was observed (Fig. 3A, 3B). Importantly, splenic pDCs isolated at this same time point from LCMV-infected CD28ko mice showed significantly higher levels of Ifnα and Ifnβ transcripts compared with pDCs from WT infected mice, whereas cDCs from both WT and CD28ko mice demonstrated low to undetectable IFN-I levels, as described previously (Fig. 3C) (20). However, we observed no differences in LCMV Cl13 replication in CD28ko versus WT pDCs or liver homogenates at day 1 p.i. (data not shown), potentially because of abundant and sufficient IFN-I already present at this time point in WT controls.

To further validate the inhibitory role of CD28 on pDC antiviral function in vivo and to examine its biological relevance in a model where the contribution of pDCs to viral control is well established, we next evaluated WT and CD28ko mice infected with MCMV (11, 12). During MCMV infection, pDCs recognize and respond to viral infection in a TLR9-dependent manner and are essential for the systemic IFN-I peak observed at 36 h p.i. (11, 12). We observed that, although systemic IFN-I was undetectable in WT and CD28ko mice at 24 h p.i., they were enhanced at 36 h p.i. and that this elevation was significantly higher in CD28ko compared with WT infected mice (Fig. 3D), whereas serum TNF-α levels were similar between both groups (Fig. 3E). Of note, at 36 h p.i., pDC numbers were similar between WT and CD28ko mice, demonstrating that the difference in systemic IFN-I levels was not the result of increased numbers of pDCs in CD28ko mice (data not shown). Consistently, FACS-purified splenic pDCs isolated from CD28ko MCMV-infected mice 36 h p.i. demonstrated higher levels of Ifnα and Ifnβ transcripts than their WT counterparts (Fig. 3F). As expected, undetectable levels of Ifnα and Ifnβ transcripts were observed in WT and CD28ko cDCs at 36 h p.i. (Fig. 3F) (36). Consistent with the enhanced IFN-I response in CD28ko mice, expression of prototypic ISGs (i.e., Mx1 and RigI) were upregulated in both pDCs and cDCs from CD28ko compared with WT MCMV-infected mice (Fig. 3G). Importantly, in the absence of CD28, we detected lower viral replication, as indicated by decreased numbers of PFU and transcript levels of the MCMV early inducible (eI) gene in the livers of CD28ko compared with WT infected mice (Fig. 3H). Taken together, these results indicate that CD28 inhibited pDC and systemic IFN-I production during in vivo viral infection with both RNA (LCMV) and DNA (MCMV) viruses. Furthermore, this effect was associated with a restricted antiviral state and influenced viral control early after MCMV exposure, supporting the biologically relevant role of CD28 on pDC IFN-I response.

To investigate whether the inhibitory effect of CD28 on pDC-IFN-I production during viral infection was cell-intrinsic or resulted from CD28 deficiency on other cell types, we generated WT:CD28ko mixed chimeras (Fig. 4A). Chimeric mice showed similar proportions of total lymphocytes, pDCs, and cDCs in WT versus CD28ko compartments (Supplemental Fig. 3). Notably, pDCs isolated from LCMV-infected WT:CD28ko mixed chimeric mice 24 h p.i. showed enhanced expression of Ifnα and Ifnβ transcripts when derived from the CD28ko compared with WT origin (Fig. 4B). Moreover, 36 h after MCMV infection, FACS-purified splenic pDCs from the CD28ko compartment showed an even more dramatic increase of Ifnα and Ifnβ expression compared with WT pDCs (Fig. 4C). Similarly, proinflammatory cytokine levels were elevated in CD28ko pDCs compared with their WT counterpart (Fig. 4D). Interestingly, pDC Mx1 expression was also enhanced in CD28ko pDCs during MCMV infection, suggesting autocrine IFN-I signaling in this setting (Fig. 4E). Furthermore, expression of the MCMV eI gene was greatly downregulated in pDCs (but not cDCs) from the CD28ko origin compared with their WT counterparts, indicating that intrinsic CD28 signaling (potentially through suppression of an autocrine IFN-I effect) was promoting MCMV replication in pDCs (Fig. 4F). Altogether these data indicate that cell-intrinsic CD28 signaling on pDCs significantly downregulated their IFN-I and proinflammatory cytokine production in response to in vivo RNA and DNA viral infections. In the case of MCMV infection, cell-intrinsic CD28 signaling also resulted in a restricted antiviral state and increased viral replication, indicating that CD28 signaling directly regulated pDC antiviral defense during in vivo infection.

Aside from its role in antiviral defense, pDC IFN-I production also promotes innate defense during tissue injury (31, 49). Indeed, pDCs sense host-derived nucleic acids that are released following common skin injuries, migrate to the site of cutaneous lesion, and secrete IFN-I, promoting tissue re-epithelialization and wound repair (31). To investigate the role of CD28 in pDC IFN-I response triggered by tissue injury, we compared pDC infiltration and IFN-I levels in WT and CD28ko mice following mechanical injury where the upper epidermal layer of the skin is removed (50). No differences were observed in pDC infiltration between WT and CD28ko mice at 24 or 48 h postinjury, the time at which pDC infiltrate peaks (data not shown and Fig. 5A, respectively). Interestingly, despite no change in pDC number, CD28ko mice demonstrated an enhanced IFN-I signature, which in this model, is known to be fully dependent on pDCs (31). In fact, 48 h postinjury, we observed increased transcript levels of Ifnα5 and Ifnα6 (Fig. 5B) and the IFN-I response gene Ifi202b. Irf7 and Isg15 also trended toward increased levels in CD28ko mice, but differences did not reach statistical significance (Fig. 5C). These data indicate that CD28 suppression of pDC IFN-I response was not restricted to viral infection but instead also influenced nonviral innate responses such as those following tissue injuries.

pDCs are highly specialized to produce IFN-I, which dramatically influence antiviral and anticancer defense, autoimmunity, and wound healing, among other human illnesses (1). Therefore, understanding the regulation of pDC IFN-I production at the molecular level is of great importance to ultimately fine-tune the magnitude of IFN-I responses as well as their multiple (and often opposing) biological consequences (1). We found that pDCs constitutively expressed the prototypic T cell stimulatory molecule CD28, which unexpectedly restrained (rather than stimulated) pDC IFN-I production in response to TLR stimulation. Importantly, CD28-mediated pDC suppression was also observed upon infection with RNA and DNA viruses and following skin injury in vivo, indicating that it is a general mechanism that downregulates pDC innate response in different immune settings.

CD28 costimulation in T cells, through interaction with CD80 and/or CD86 molecules on APCs, is one of the best established events that bridge innate and adaptive immunity following pathogen recognition (32, 51). Interestingly, CD28 expression has been previously reported in other immune cells including NK cells, eosinophils, neutrophils, and plasma cells. For example, engagement of CD28 on a subset of NK cells promotes activation and NK-mediated cytotoxicity (52), whereas CD28 activity on eosinophils promotes their activation and secretion of IL-2, IFN-γ, and IL-13 (53, 54). Furthermore, CD28 ligation on neutrophils enhances CXCR-1 expression, promoting their migration (55), and also induces neutrophil IFN-γ secretion during Leishmania major infection (56). More recently, plasma cells were found to express CD28 (57), but the role of CD28 in regulating plasma cell survival and function is unclear as evidence for both promoting and limiting plasma cell survival and Ab production have been reported (58, 59). Importantly, our in vivo experiments with BM-mixed chimeras indicated that CD28 dampened pDC function and its antiviral state, promoting viral replication in a cell-intrinsic manner, rather than through signaling via the aforementioned non-pDC cells that also express CD28. This is consistent with the enhanced cytokine production in CD28ko pDCs differentiated in BM-Flt3L cultures where pDCs are the only CD28-expressing cells. Of note, reconstitution of CD28 in CD28ko pDCs from such BM-Flt3L cultures downregulated their IFN-I production, ruling out any off-target effects in IFN-I production by CD28ko pDCs. Overall, it is striking that although CD28 appears to induce stimulatory signals that promote activation, enhanced function, and/or survival in most cells that express it, it exerts a suppressive role in pDCs. It is tempting to speculate that such opposing roles of CD28 in pDCs versus other immune cells might have evolved to counterbalance the immune response to best fight infections with minimal collateral tissue damage.

Interestingly, our findings suggest the possibility that CD80/CD86 interaction with CD28 could regulate the ability of pDCs to produce IFN-I. Because we observed enhanced pDC-IFN-I production in FACS-purified CD80/86 double-deficient BM-pDCs, it is possible that CD80/CD86 can act in a pDC-autonomous manner. However, it is also conceivable that other CD80/86-expressing cells (such as cDCs) might also modulate pDC IFN-I production via CD28 signaling. Moreover, although it is still unclear at what stage of the pDC lifespan CD28 suppression becomes effective, failure to alter IFN-I production by ex vivo CD28 stimulation or blockade in fully differentiated WT pDCs (concomitantly or 1–2 h before TLR stimulation) suggests that CD28 may homeostatically regulate pDCs (i.e., before pDCs encounter pathogen-associated molecular patterns). Similarly, a recent study demonstrated homeostatic regulation of pDCs by the microRNA miR-126, which (in steady-state conditions) is required for pDC survival and optimal expression of molecules involved in pDC TLR responses, affects subsequent pDC antiviral responses (60). Interestingly, miR-126 regulates pDCs through the VEGFR2 protein, which, similar to CD28, is undetectable in human pDCs isolated from blood but significantly expressed in pDCs purified from tissues (60). Furthermore, the T cell regulatory molecule lymphocyte activating gene 3 (61, 62) also controls the homeostatic proliferation and expansion of pDCs (63). Taken together with CD28, these and other homeostatic pDC regulators such as E2-2 (64), Flt3 (65), and the PI3k/mTOR pathway (66) highlight the important and intricate nature of pDC control under steady state, where these multiple pathways may partially overlap but may also regulate distinct aspects of pDC biology, preconditioning subsequent pDC responses.

Although CD28 signaling in pDCs may be critical for homeostatic pDC function, it also holds the potential for therapeutic exploitation to limit excessive IFN-I responses that contribute to autoimmunity or immunopathology. In fact, in several autoimmune diseases, pDCs play a pathogenic role via TLR recognition of self nucleic acids, resulting in excessive IFN-I production that promotes activation and survival of autoreactive T and B cells (26, 27, 29, 30, 67). Given that CD28 promotes activation of conventional T cells while also expanding regulatory T cells and dampening pDC innate responses, it is not surprising that the effect of ubiquitous CD28 inhibition or stimulation varies in different autoimmune disease settings. Indeed, although studies have shown that CD28-mediated signaling prevents spontaneous development of autoimmune diabetes (68) and reduces experimental autoimmune neuritis (69), others have shown that CD28 promotes autoimmune diseases such as collagen-induced arthritis, experimental autoimmune encephalomyelitis, and systemic lupus erythematosus (70–72). Although these studies have primarily focused on the effects of CD28 on conventional or regulatory T cells, our results open the possibility that altered IFN-I production by pDCs may also play a role in the abovementioned autoimmune disease models. Future studies with cell-specific and temporally controlled ablation of CD28 may illuminate how to best harness CD28 and its downstream signaling pathways toward therapeutically attenuating autoimmune responses while promoting host defense and wound healing.

In conclusion, our study revealed an unexpected role for the prototypic stimulatory receptor CD28 in suppressing innate immunity through cell-intrinsic inhibition of pDC IFN-I production. These findings broaden our understanding of the molecular networks that coordinate innate responses and may aid future pDC-related therapies.

We thank Dr. Jack Bui for facilitating studies with human biopsies and critically reading the manuscript.

The authors have no financial conflicts of interest.