Resident CD8+ and Migratory CD103+ Dendritic Cells Control CD8 T Cell Immunity during Acute Influenza Infection

The identification of the specific DC subsets providing a critical role in presenting influenza antigens to naïve T cell precursors remains contentious and under considerable debate. Here we show that CD8+ T lymphocyte (TCD8+) responses are severely hampered in C57BL/6 mice deficient in the transcription factor Batf3 after intranasal challenge with influenza A virus (IAV). This transcription factor is required for the development of lymph node resident CD8+ and migratory CD103+CD11b− DCs and we found both of these subtypes could efficiently stimulate anti-IAV TCD8+. Using a similar ex vivo approach, many publications on this subject matter excluded a role for resident, non-migratory CD8+ DC. We postulate the differences reported can partially be explained by how DC are phenotyped, namely the use of MHC class II to segregate subtypes. Our results show that resident CD8+ DC upregulate this marker during IAV infection and we advise against its use when isolating DC subtypes.


Introduction
Dendritic cells (DC) are a heterogeneous bone marrow-derived cell population widely accepted as the key initiators of T cell immunity [1]. Considerable complexity surrounds understanding the requirement for such heterogeneity along with pinpointing the underlying role each subset has evolved to provide. DC heterogeneity is most apparent in lymph nodes that drain peripheral sites such as the lung, skin and gastrointestinal tract. These lymph nodes are comprised of two major subsets: plasmacytoid DC and myeloid DC, with the latter commonly referred to as conventional or classical DC (cDC) [2]. cDC isolated from the lymph node that drains the lung represent a mixture of lymphoid resident DC, which do not traffic from the periphery, and tissue-derived, ''migratory'' DC. Lymphoid resident DC subsets can be subdivided into CD8aexpressing DC (CD8 + DC) and CD8 2 DC [3,4]. Two defined migratory DC subsets exist and are subcategorized based on their differential expression of the mucosal a E integrin marker CD103 and myeloid marker CD11b (CD103 + CD11b 2 DC and CD103 2 CD11b + DC) [5].
Currently, considerable contention surrounds immune control by these subsets during acute influenza infection [6,7,8,9,10,11,12,13]. Evidence exists for either anti-IAV T CD8+mediated immunity elicited by a combination of lymph node resident and migratory DCs or alternatively sole control by both or only one of the migratory DC populations. The vast majority of these studies utilize ex vivo DC analysis, which is reliant on astute segregation of distinct subsets based on lineage markers defining end-stage subpopulations. In addition, pan-DC markers are widely used; these include surface expression of CD11c and Major Histocompatibility Class II molecules (MHC II). The expression level of MHC II on cDCs is variable and dependent on maturation status. In the absence of infection, cDCs in the lung parenchyma and lymphoid-resident DC in lymphoid organs are present in a socalled ''immature state'' dedicated to surveillance, not T cell priming. In such a state, they possess a high endocytic/phagocytic capacity and intermediate levels of surface MHC class II (MHC II int ). Tissue-derived DC constitutively traffic to the lymph nodes and terminally differentiate into mature cells dedicated to modulating T cell immunity. The coordinated phenotypic changes associated with this maturation event include downregulation of endocytic/phagocytic activity, upregulation of costimulatory molecules and increased surface expression of antigen-MHC II complexes (MHC II high ) [14].
These basic life-cycle events are often extrapolated to conditions when inflammatory stimuli, such as influenza infection, are present. Multiple groups segregate migratory DC and lymphoid resident DC into MHC II high and MHC II int populations, respectively, from the draining mediastinal lymph nodes of influenza infected animals. While DC maturation during acute influenza infection is reported, notably upregulation of the costimulatory molecule CD86 [7,9], altered expression of MHC II in DC subsets has not been thoroughly characterized.
Here we report mice deficient for the basic leucin zipper transcription factor 3 (Batf3 o/o ) [15,16], which have a selective deficit in CD8 + and CD103 + CD11b 2 DC within the mediastinal lymph node, lack anti-IAV T CD8+ immunity. This result implicates a crucial role for one or both of these subsets in orchestrating T CD8+ immunity during an acute viral infection. Our study shows that both ex vivo-isolated lymphoid resident CD8 + and migratory CD103 + CD11b 2 DC are capable of MHC class I-restricted presentation after lung infection. Multiple studies incorporate a similar ex vivo approach to interrogate DC subset contribution for generating T CD8+ immunity and do not reach the same conclusion [17]. This prompted us to further scrutinize the phenotypic changes associated with DC behavior during the early stages of IAV infection. Our results show that the lymphoid resident CD8 + DC population rapidly upregulates MHC II complexes early during infection. Multiple studies employ this marker to distinguish between resident and migratory DC. Our results suggest not relying on this marker when isolating DC subsets, especially during an inflammatory setting. This is an important point as it identifies a potential experimental pitfall and we hope it provides the first steps to providing an explanation into the outstanding discrepancies that exist within the current literature. This question is something that has puzzled the field for a long time and it is paramount that a unified view on this matter be reached in order to move the research forward. Fluorescent labeling of DC migration in the lung PKH26 dye was diluted in diluent C (Sigma) 1:100 and 40 mL was administered by intranasal delivery to each mouse after anesthesia.

Generation of T CD8+ line specific to IAV NP 366-374
Splenocytes were isolated from B6.Ly5.1 mice primed with PR8 at least 30 days prior to harvest. 10% of these cells were pulsed with 1 nM NP 366-374 for 60 minutes. The cells were then washed with PBS and co-cultured with the remaining 90% of memory splenocytes in the presence of IL-2 (Chiron). T cell lines specific to IAV NP 366-374 were generated using a recently described protocol [18]. T cell lines showing .90% T cell activation by bonemarrow-derived DC pulsed with NP 366-374 were used.
DC isolation from lymph nodes and co-culture with IAVspecific T cells DCs were isolated as previously described [19]. Briefly, single cell suspensions from the mediastinal lymph node were enriched for DC using an antibody depletion and magnetic enrichment. Purity of enriched cell suspensions usually yielded .40% CD11c + DC. These enriched DC were stained with anti-CD11c, -CD8, -CD11b (BD) and -CD103 (eBiosciences) for flow cytometric sorting. Serial dilutions of each sorted DC population were cocultured in vitro with 5610 4 NP 366-374 -specific T cells. Antigenspecific T cells were enumerated by ICS as described above.

Results
Batf3 o/o mice are unable to mount an efficient antiinfluenza T CD8+ response DCs are well understood to mediate T cell activation and initiate anti-viral immunity. We previously argued that two specific subtypes, the CD8 + and CD103 + CD11b -DC subsets, drive T CD8+ responses during herpes simplex virus skin infection [20]. We sought to determine whether this was also the case during pulmonary IAV infection. We hypothesized that in the absence of the CD8 + and CD103 + CD11b -DC subsets, naïve influenzaspecific T cell priming should be compromised. A recent series of novel studies identified that the transcription factor Batf3 is required for the development of both the splenic CD8 + and dermal CD103 + CD11b -DC subset lineages and mice lacking this gene were ideal for testing our hypothesis [15,16]. However, our experimental models utilize C57BL/6 (B6) mice and mice lacking Batf3 on this background are reported to retain completely normal numbers of functional CD8 + DC within the skin-draining lymph nodes [21,22]. We did observe similar results within the skindraining inguinal lymph nodes of both B6 wildtype and Batf3 o/o mice. Interestingly, a selective loss of the CD8 + DC subset remained in the lung-draining mediastinal lymph node of Batf3 o/o mice of this background (Fig 1B). Furthermore, a complete loss of the migratory CD103 + CD11b 2 DC subset was observed in both skin-and lung-draining lymph nodes ( Fig 1C). Thus, Batf3 o/o mice on a B6 background remain an excellent opportunity to test our hypothesis.
We infected control B6 and Batf3 o/o mice with influenza (strain PR8) and the expansion of various well-characterized influenzaspecific T CD8+ populations were analyzed at the reported peak of the immune response (10 days p.i.) [23,24]. Consistent with previous published results, robust numbers of T CD8+ specific to NP 366-374 , PA 224-233 , PB1-F2 62-70 , and PB1 703-711 were observed in both the spleen ( Fig 1D) and bronchoalveolar lavage (Fig 1E) of control mice. In contrast, the magnitude of the T cell response to each of these epitopes was clearly diminished in both compartments within Batf3 o/o mice (Fig 1D-E). Near identical results were observed following infection with a different virus strain (X-31), further highlighting this impairment in Batf3 o/o mice (data not shown). It is important to note that the absence of these DC subsets did not compromise naïve CD8 + T cell precursors specific to influenza as robust T CD8+ responses against PA 224-233 were generated in both strains primed with peptide pulsed bone marrow derived-DC (Fig 2). These results provide in vivo evidence for a critical role of one or both the CD8 + and/or CD103 + CD11b 2 DC subsets in initiating T CD8+ immunity during acute IAV infection.
CD8 + and CD103 + DC are essential for generating optimal anti-influenza T CD8+ immunity We next turned to an ex vivo antigenic presentation assay to determine the individual antigen presentation contributions of the different DC subsets, especially those implicated in the above in vivo experiments. This involves DC isolation from the lung draining lymph node of IAV-infected mice, segregation into defined subpopulations before co-culture with IAV-specific T cells. If the T cells respond, it is indicative that the co-cultured DC subtype is presenting antigen. We prepared DC at the reported peak of antigen presentation, D3 p.i. [6], by first depleting lymph node suspensions of other cell types and then staining for pan-DC marker CD11c as well as the markers CD8, CD103 and CD11b. Then, we sorted CD11c + DC by flow cytometry on the basis of differences in expression of these markers to delineate the subsets. We collected CD11c 2 cells as a negative control as well as CD11c + CD8 + DC before fractionating the CD11c + CD8 2 population into the CD103 + CD11b 2 DC subset and a CD11b + subset ( Fig 3A). It is imperative to isolate CD8 + DC first, as we have previously demonstrated they can express CD103 [20] and could potentially contaminate the migratory CD103 + CD11b 2 population. Following sorting, we examined these populations for their presentation of an influenza-derived epitope by co-culturing independent subsets with a T cell line capable of responding to as little as 10 213 M of the immunodominant influenza-derived nucleoprotein peptide (NP 366-374 ) (Fig 3B). Figure 3C shows that both CD8 + and CD103 + DC could stimulate NP 366-374 -specific T cells to generate the effector molecule IFNc. These findings suggest that both these subpopulations have acquired influenza antigen and are capable of evoking anti-IAV T CD8+ immunity.

DC subset maturation during acute influenza infection
To examine if phenotypic changes occur in migratory DCs trafficking from the infectious site to the lung draining mediastinal lymph node, mice were instilled i.n. with a solution containing the dye PKH26 16 hours prior to influenza infection. Dye-labeled DCs were readily identifiable in the mediastinal lymph node 1-3 days post infection (Fig 4A). CD11c + CD8 + DC did not stain for the dye, consistent with the notion they are a non-migratory, lymphoid resident cell population [25] (Fig 4A GI). Dye-labeled cells (Fig 4A GII) were a mixture of the migratory CD103 + CD11b 2 DC and CD103 2 CD11b + DC populations previously described [5] (Fig 4B). Consistent with previously published results, a higher proportion of migrated pKH26 + CD103 2 CD11b + DCs was observed (data not shown). Day 1 p.i., the expression of MHC II surface molecules on CD8 + DC were at intermediate levels as compared to a high level of expression on their migratory PKH26 + counterparts (Fig 4A  histogram). However, a clear shift in MHC II expression within the resident CD8 + DC subpopulation was readily observed on day 2 (Fig 4C bottom). Upregulation of this marker by the majority of this DC subset was most evident on day 3 (Fig 4D bottom), a time point previously reported as the peak of antigen presentation [6].  In addition, the kinetics of expression of the co-stimulatory marker CD86 in this DC subset mirrored MHC II expression (Fig 4E-F  bottom). No alteration in MHC II expression was observed in the PKH26 + migratory populations (Fig 4C-D top) between D1-3 p.i.. Interestingly, no alteration in CD86 expression was observed on pKH26 + migratory DCs on D2 p.i. (Fig 4E top). Increased expression of this accessory molecule was only observed D3 p.i. (Fig 4F top) in both the pKH26 + CD11b + CD103 2 and pKH26 + CD11b 2 CD103 + subpopulations (data not shown). These results clearly show that DC subpopulations are responsive to a peripheral pathogenic encounter, progressing from an immature phenotype to a more mature state.

Discussion
While the above results are in agreement with several of the aforementioned studies [6,7], it conflicts with multiple reports where one attributes the CD103 2 CD11b + DC as the major antigen presenting cell [8] and another study that deems the CD103 + CD11b 2 DC as the responsible subset [11]. Interestingly, both of these studies employed the same TCR-like monoclonal antibody to detect recombinant IAV-derived H-2K b /SIINFEKL [26]. Additional recent studies describe influenza antigen presentation restricted to both CD103 + CD11b 2 and CD103 2 CD11b + migratory DC subsets [9,10]. In one case, the CD103 + CD11b 2 DC subset preferentially stimulated naïve T CD8+ differentiation [9], while in the second study this was achieved by CD103 2 CD11b + DC [10]. The simple fact so many contradictory results are prominent within the literature clearly identifies a problem. The studies comprehensively rely on ex vivo presentation assays, which in turn rely on segregation of DC via lineage markers. It is plausible that the discrepancies may partly be attributed to differing gating strategies prompting an analysis of the various DC populations for phenotypic changes during IAV infection. Our results demonstrate that lymphoid-resident DCs increase their expression of MHC II during a peripheral viral infection, indicating they are transitioning from an immature to a mature phenotype.
These findings draw awareness to gating strategies based on separating lymphoid resident and migratory DC subpopulations using MHC II expression levels in an inflammatory setting. This strategy was utilized in several publications that concluded lymphoid resident DC were not involved in generating antiinfluenza T CD8+ responses [9,10]. It is possible that the gating sequence these groups employed place matured lymphoid resident DC, particularly the CD8 + subset, into the MHC II high gate. In fact, we readily observe many of the CD8 + DC subpopulation undergoing a phenotypic change from CD11c high MHC II int in naïve mice to CD11c int MHC II high expression in IAV infected mice ( Fig. 4G red dots). This latter phenotype is a classical migratory DC phenotype and we believe a segregation strategy based on MHC II and CD11c expression would ultimately misrepresent both lymphoid resident and migratory DC subsets by excluding a large proportion of the matured resident DC and placing them into the migratory subsets as a contaminating population. Such an experimental design possibly provides misinformation on the identity of the DC groups capable of providing T cell stimulation. In fact, a large proportion of the influenza-induced matured CD8 + DC (Fig 4) would be completely overlooked and a false negative result could be observed when the remaining MHC high -depleted CD8 + subpopulation is utilized in ex vivo experiments.
While our results reiterate a crucial role for CD8 + and CD103 + CD11b 2 DC subsets during IAV infection, it is difficult to ascribe overall individual contributions by either subset from these experiments. Our experiments displayed a trend of less efficient antigen presentation by the CD8 + DC subpopulation as compared to CD103 + CD11b 2 DC subpopulation ( Fig 3C). From our data, it is difficult to interpret if this is on a per cell basis or if far fewer CD8 + DC are involved in antigen presentation. However, clearly both populations possess influenza antigens; therefore, retaining the capacity to stimulate naïve influenzaspecific precursors recruited to the draining lymph node. Each may further provide unique independent roles, such as the CD103 + CD11b 2 migratory DC acting as a transporter to donate antigenic cargo to the CD8 + DC in order to amplify T CD8+ immunity [25]. It is well documented that the CD8 + DC subset excels at cross-priming T cells and in a recent study Langlois et al. imply cross-presentation is a significant pathway used to generate influenza specific T CD8+ immunity [27]. Furthermore, we recently published both subsets possess the capacity to prime herpes simplex virus specific T cells and that the dermal CD103 + CD11b 2 DC subset specializes in cross-presentation [20]. Thus, we believe a central role for these DC in providing anti-viral immunity, with significant contribution from the cross-presentation pathway, is becoming increasingly apparent. Multiple groups report CD11b + DC play a key role in promoting influenza-specific T CD8+ immunity. If this is the case, they evolved to rely on factors supplied by the CD8 + and CD103 + CD11b 2 DC subsets.
In summary, the role of DC subsets following influenza infection of the lung continues to be an active and intense area of debate. Our results directly challenge recent proposals and provide strong support to the original descriptions of DC presentation of antigen after influenza virus infection. Considering the complex network of DC subtypes require careful and elegant techniques for their identification and isolation, we suggest caution by DC biologists when incorporating MHC II as a segregation marker for ex vivo analysis of DC under inflammatory settings such as influenza infection. Finally, we hope this study provides the first steps in unifying the view in the field on specific DC subpopulation control of T CD8+ responses during IAV infection and that it provides an explanation for some of the discrepancies. Knowledge of the specific DCs orchestrating anti-viral immunity is critical for the development of targeted vaccination strategies that will provide broad-spectrum T CD8+ immunity against IAV.