Research on plasmacytoid dendritic cells (pDCs) and interferon (IFN) is almost inextricably interconnected. The first description of IFNs as factors that are rapidly produced by virus-infected cells and that confer viral resistance to neighboring cells dates back to 1957 . Only 1 year later, in 1958, Lennert and Remmele reported the presence of cells with plasmacytoid morphology in human LNs and the spleen . It took more than 40 years, until Siegal et al.  and Cella et al.  identified these pDCs as the so far uncharacterized natural IFN-producing cells, thereby establishing the tight link between pDCs and IFN.
pDCs have the remarkable capacity to produce large quantities of type I IFN upon viral infection [3, 4]. In fact, one cell can secrete up to 10 pg IFN-α . Furthermore, the spectrum of IFN-α subtypes released by pDCs is particularly broad [6, 7]. In addition to the various IFN-α subtypes, activated pDCs produce IFN-β, type III IFNs (IFN-λ1, IFN-λ2, IFN-λ3), and a number of pro-inflammatory cytokines and chemokines . pDCs act in the defense against viruses and bacteria by recognizing pathogenic single-stranded RNA and single-stranded DNA via the endosomal toll-like receptors (TLRs), TLR7 and TLR9, respectively . pDCs are thereby able to sense viruses that enter the cells via endocytosis and do not require direct infection. This is in contrast to other cells such as fibroblasts and conventional DCs, which mainly depend on cytosolic pathogen recognition . For certain single-stranded RNA viruses, such as vesicular stomatitis virus and respiratory syncytial virus, pDCs use autophagy to transport cytosolic viral replication intermediates into endolysosomal compartments, where the nucleic acids are recognized by endosomal TLRs . Hence, the endosomal TLRs play a key role in the recognition of nucleic acids present in various subcellular localizations in pDCs.
There is significant clinical interest in pDCs as central IFN-producing cells. Type I IFNs are clinically used for the treatment of chronic viral hepatitis, certain cancers, and multiple sclerosis. However, high serum levels of type I IFN are thought to be detrimental in different autoimmune diseases, suggesting that pDCs are involved in the erroneous recognition of excessive self-nucleic acids during autoimmunity . Thus, a better molecular understanding of how pDCs sense nucleic acids and generate pro-inflammatory cytokines and IFNs is of immediate relevance for the development of therapeutics.
An impressive amount of knowledge about pDC function and the regulation of type I IFN has been gained since the studies by Siegal et al.  and Cella et al. . However, most studies — in particular those focusing on pDC signaling — were performed with murine cells. Mouse pDCs can be obtained in reasonable numbers from WT and gene-deficient mice, while human pDCs can only be obtained from peripheral blood, where they represent less than 0.5% of the peripheral blood mononuclear cells [3, 4]. Moreover, pDCs are easily activated during the purification process , which complicates their use in signaling studies.
The human pDC cell line CAL-1 is derived from patients with a primary blastic NK-cell lymphoma  and the cytokine response toward CpG DNA of this cell line closely mirrors their human blood-derived counterpart cells, as shown in this issue of the European Journal of Immunology by Steinhagen et al. . Since CAL-1 cells can be cultured to obtain large numbers of cells and manipulated by RNA interference, they can be used for signaling studies as a human pDC surrogate cell line . In a previous study in 2012, Steinhagen et al. comprehensively characterized the transcriptional response of CAL-1 pDCs toward two types of CpG-containing DNAs : CpG type D (also called A), which is characterized by the presence of poly G tails, internal palindromic sequences, a mixed phosphothioate/phosphodiester backbone, and the formation of higher order molecular structures, and CpG type K (also called B), a linear molecule with a complete phosphothioate backbone. The two different types of CpG oligodeoxynucleotides that were used are known to elicit different immune responses [14, 15]: CpG type D is retained in early endosomes for a longer amount of time and acts as a strong inducer of type I IFNs ; on the other hand, CpG type K traffics faster to late endosomes, induces only small amounts of type I IFN in pDCs  but stimulates pDCs to produce high levels of pro-inflammatory cytokines such as TNF-α or IL-6 and is a potent stimulator of B cells [14, 15]. In 2012, Steinhagen et al. showed that, in CAL-1 pDCs, CpG type K induces an early boost of type I IFNs (both IFN-α and IFN-β) . In their current manuscript, the authors analyze the signaling components necessary for this early IFN induction and investigate which key transcription factors contribute to pro-inflammatory cytokine and IFN production in human pDCs. As noted above, our knowledge about TLR9 signaling and the molecular requirements for IFN induction has come mostly from studies with mice. As there are considerable differences between humans and mice, both in the cellular expression patterns of endosomal TLRs as well as in the utilization of transcription factors for the induction of IFNs, this work has important implications.
IFN and to a lesser extent pro-inflammatory cytokine production are regulated by various IFN regulatory factors (IRFs). IRFs are transcription factors with versatile roles in cellular differentiation of hematopoietic cells, innate and adaptive immunity, cell cycle, and apoptosis (reviewed in [17, 18]). The mammalian IRF family comprises nine members, namely IRF1–9, and the downstream effects of various IRFs in innate immunity are highly dependent upon the nature of the stimulus and the cell type used [17, 18]. IRFs can cooperate with other transcription factors in a cell- and stimulus-dependent manner, with each combination having distinct, sometimes even opposing, functional outcomes. Furthermore, IRFs are differentially expressed in distinct cell types, and for some IRFs, such as human IRF5, multiple splice variants have been reported . In contrast to human IRF5, murine IRF5 is primarily expressed as the full-length transcript, with only one single splice variant detected in very low levels . Given these and other differences between human and mouse cells regarding the expression and function of IRFs, one cannot directly transfer the results obtained in murine cells to the human system.
The canonical TLR NF-κB signaling pathways, however, appear to be well conserved between humans and mice (Fig. 1). TLR activation leads to the formation of a complex containing MyD88, TRAF6, IRAK4, and IRAK1 (reviewed in ). TRAF6 ubiquitinylates TAK1 that leads to the activation of MAP kinases and IKK-β. Activated IKK-β, in turn, phosphorylates I-κBα and thereby activates NF-κB. NF-κB subsequently translocates to the nucleus and induces the transcription of a range of pro-inflammatory cytokines . The induction of type I IFN happens via a distinct pathway that appears to be less conserved between humans and mice (Fig. 1). In murine immune cells, IRF7 was identified as the master regulator of type I IFN responses . In these cells, the MyD88-TRAF6-IRAK4 complex activates IRF7 through TRAF3, IRAK1, IKK-α, osteopontin, and PI3K. Following ubiquitinylation and phosphorylation, IRF7 then translocates to the nucleus and initiates the transcription of type I IFN . Notably, this pathway is shared by murine TLR7 and murine TLR9, but is distinct from the pathways downstream of murine TLR3 and the cytosolic pathogen recognition receptors RIG-I and MDA5, in which IRF3 and IRF7 are activated by TBK1 and IKK-λ . The induction of relatively low amounts of IFN-β results in the activation of the type I IFN receptor and initiates an IFNAR-mediated positive feedback loop that leads to the upregulation of IRF7 and further induction of IFN-α and IFN-β genes .
IRF8 seems to act as a positive regulator of IFN and pro-inflammatory signaling in the murine system. In BM-derived pDCs and conventional DCs, IRF8 was shown to be involved in the late IFN feedback phase, ensuring sustained RNA polymerase II recruitment to type I IFN promoters . Furthermore, in murine macrophages and DCs, the positive regulation of IL-12 production by IRF8 in concert with IRF1 has been described [24, 25]. In addition, a more general defect regarding pro-inflammatory cytokine induction and the upregulation of activation markers such as CD40 or MHCII is seen in splenic IRF8 KO DCs stimulated with TLR9 ligands . Importantly, murine IRF8 is also crucial for the development of certain DC populations, namely pDCs and CD8α+ DCs [27, 28]. Therefore, the composition of splenic DCs and BM-derived FLT3L-induced DCs is different between WT and IRF8KO mice [27, 28].
The study of IRF5's role in mediating IFN production in mouse cells has been complicated by the appearance of a spontaneous mutation in DOCK2 in some IRF5 KO mouse colonies . This mutation has led to reduced expression of DOCK2 in the affected strains. Both DOCK2 and IRF5 play important roles in immune cell functions. As a result, some of the functions of DOCK2 have been wrongly attributed to the deficiency in IRF5. Two recent reports have demonstrated that abnormal B-cell development in the IRF5 KO line was an effect of DOCK2 rather than IRF5 [29, 30]. However, in IRF5 KO mice that are WT for DOCK2, pDCs retain the pronounced defect in the induction of pro-inflammatory cytokines and IFN-β . In contrast, IFN-α responses were only mildly reduced upon stimulation of TLR7 and TLR9 , suggesting a more important role for IRF5 in the induction of IFN-β than IFN-α.
Consistent with this , Steinhagen et al. find that IRF5 and NF-κB p50 are key regulators of IFN-β and IL-6 induction upon stimulation of human pDCs with CpG type B . Using RNA interference, the authors identified MyD88, TRAF6, IRF5, and NF-κB p50 as being essential for the upregulation of both IFN-β and IL-6, whereas NF-κB p65 knockdown only affected IL-6, but not IFN-β mRNA levels. Within 1 h of stimulation, IRF5, NF-κB p50 as well as NF-κB p65 translocated to the nucleus. Furthermore, nuclear co-localization was observed for IRF5 and NF-κB p50, but not for p65, suggesting that IRF5 and NF-κB p50 co-regulate certain genes. Late upon stimulation, IRF1 was also found to accumulate in the nucleus and, moreover, upregulated IRF1 and IRF7 mRNA levels could be detected. These late events were dependent on the type I IFN feedback loop. In contrast to mouse cells, however, knockdown of IRF1 or IRF7 did not have any significant effect on IFN-β or IL-6 induction. IRF8 was identified as a negative regulator, as knockdown of IRF8 resulted in increased mRNA levels of IFN-β and IL-6 .
The current study of Steinhagen et al.  points to a crucial role for human IRF5 in the induction of type I IFN. This is in accordance with a previous study reporting the involvement of IRF5 in type I IFN induction upon TLR7/8 stimulation . TLR7/8 signaling via MyD88, IRAK1, and TRAF6 was found to activate IRF5 and IRF7, but not IRF3 . Moreover, knockdown of IRF5 in THP1 cells resulted in strongly reduced IFN-α levels upon stimulation with R848 . In addition to the synthetic TLR7/8 ligand R848, various viral stimuli including Newcastle disease virus, vesicular stomatitis virus, and HSV1, but not Sendai virus, were identified as activators of human IRF5 . A more complex role for human IRF5 was, however, suggested in an overexpression study carried out in the 2fTGH cell line : depending on the dimerization partner, IRF5 could either act as an activator (i.e. IRF5 homodimers and IRF3/5 heterodimers) or as a repressor (i.e. IRF5/7 and IRF5/1 heterodimers) of IFN gene induction .
In the current study by Steinhagen et al. , IRF8 was identified as a negative regulator of TLR9-stimulated IFN-β and IL-6 induction in human pDCs. Interestingly, Li et al.  have reported the opposite role for IRF8 in human blood monocytes stimulated with LPS, Pam3CSK4, or Sendai virus. In these cells, IRF8 and IRF3 were shown to cooperatively regulate the rapid induction of IFN-β, suggesting cell type-specific functions of IRF8. Future studies will need to elucidate the exact mechanisms by which IRF8 acts as an activator in one cell type, but as a repressor in another.
Surprisingly, Steinhagen et al.  could not identify any significant role for IRF7 in mediating early type I IFN induction. This is in contrast to murine IRF7, which acts as a master regulator of type I IFN production upon TLR7 and TLR9 stimulation in murine cells . Further studies will be needed to clarify the function of human IRF7 in type I IFN production in different human cell populations and how human IRF7 is involved in late type I IFN induction. In this regard, it would also be interesting to test whether IFN-α levels are affected by IRF7 knockdown.
The CAL-1 pDC cell line is shown by Steinhagen et al.  to functionally resemble primary human pDCs and thus this cell line serves as a valuable tool to gain insights into pDC signaling pathways. However, donor-specific effects and abnormalities due to malignant transformation cannot be fully excluded. Notably, special IRF5 transcript variants, for example, have been identified in human malignancies, but not in healthy tissue  and the transcript variants of the different IRFs in CAL-1 pDCs have not yet been studied extensively.
Nevertheless, recent reports on IRF5 and IRF8 in the pathogenesis of lupus underline the high relevance of the current investigation. As mentioned above, murine IRF8 is essential for the generation of pDCs [27, 28]. In lupus-predisposed mice that lack IRF8, pDCs fail to develop, SLE-related autoantibodies are almost absent, and the disease is significantly milder . Notably, the same effect can be achieved by mutating Slc15a4, a solute carrier transporter that is essential for pDC-mediated type I IFN production in response to TLR7 and TLR9 ligands . The importance of murine IRF5 in the development of pristane-induced lupus has been confirmed by two recent studies [36, 37] using mice that lack the DOCK2 mutation mentioned above. In the absence of IRF5, hypergammaglobulinemia and type I IFN levels are reduced  and pristane-induced monocyte trafficking to the peritoneal cavity is impaired .
These studies are of high clinical interest, as genetic variants in IRF5 and IRF8 have been found to be associated with increased serum levels of IFN-α in SLE patients and high SLE susceptibility (as reviewed in [38, 39]). Furthermore, a study by Stone et al.  now provides the first direct evidence of altered IRF5 activation in human monocytes of SLE patients. Despite considerable progress, the exact roles of IRFs in human cells that are relevant for autoimmunity, cancer and infection still require further clarification, with the study by Steinhagen et al.  leading the way.