CD169 + macrophages are sufficient for priming of CTLs with specificities left out by cross-priming dendritic cells Caroline A. Bernhard a , Christine Ried a , Stefan Kochanek b , and Thomas Brocker a,1 a Institute for Immunology, Ludwig-Maximilian University of Munich, 80336 Munich, Germany and b Department of Gene Therapy, University of Ulm, 89081 Ulm, Germany Edited by Emil R. Unanue, Washington University, St. Louis, MO, and approved March 27, 2015 (received for review December 11, 2014) Dendritic cells (DCs) are considered the most potent antigen-presenting cells (APCs), which directly prime or cross-prime MHC I-restricted cytotoxic T cells (CTLs). However, recent evidence suggests the existence of other, as-yet unidentified APCs also able to prime T cells. To identify those APCs, we used adenoviral (rAd) vectors, which do not infect DCs but selectively accumulate in CD169 + macrophages (MPs). In mice that lack DCs, infection of CD169 + MPs was sufficient to prime CTLs specific for all epitopes tested. In contrast, CTL responses relying exclusively on cross-presenting DCs were biased to selected strong MHC I-binding peptides only. When both DCs and MPs were absent, no CTL responses could be elicited. Therefore, CD169 + MPs can be considered APCs that significantly contribute to CTL responses. CD169 + macrophage | dendritic cell | cross-presentation | CD8 T cell | adenovirus D endritic cells (DCs) are known to have superior T-cell ac- tivation ability compared with other cell types and thus are considered the most potent antigen-presenting cells (APCs) (1). Experiments with mice lacking DCs have supported this view in many infection and immunization studies (2, 3). In addition, DCs are specialized to efficiently cross-present exogenous antigen (Ag) via MHC I to CD8 + T cells (4). In contrast, MPs are considered cells of the innate immune system that limit pathogen spread by phagocytosis and degra- dation. They are positioned at strategically important entry points, such as the subcapsular sinus of lymph nodes and the marginal zone and red pulp areas of the spleen, where they capture and filter pathogens (5). Along with producing type I IFN (6), CD169 + marginal zone metallophilic MPs allow re- stricted replication of captured virus to increase the local avail- ability of viral Ag (7) for transfer to B cells for humoral responses (8) and to DCs for cross-priming of CD8 + T cells (9). Participation of MPs in T-cell priming is thought to be restricted to Ag acqui- sition, amplification, and redistribution. Analyzing MP functions in vivo is difficult; genetic models, such as the CD11c-diphteria toxin (DT) receptor (DTR) trans- genic mice (2), cannot make the distinction, because both DCs and monocytes/MPs are equally deleted on DT treatment (10, 11). Recent results with mice lacking CD169 + MPs suggest the role of these MPs in cross-priming of CD8 + T cells specific for the cell-associated model Ag ovalbumin (OVA) in lymph nodes (12). These findings are in contrast with another study in which MPs were unable to cross-present soluble OVA protein in mice lacking DCs (11). Although cross-presentation by DCs is now thought to allow priming of cytotoxic T-cell responses to exog- enous tumor or viral Ag (4), the importance of cross-presentation in vivo has been a matter of intense debate over the last decade (13). Although this debate remains unresolved, cross-presentation has been considered then and now to be an exception rather than a rule (14, 15). To determine the roles of MPs and DCs, we used a mouse model that specifically lacks DCs but conserves all subpopulations of MPs. We show that the presence of directly infected CD169 + MPs is sufficient to prime CTLs, including those recognizing epi- topes not covered by cross-presenting DCs. In contrast, when only DCs cross-prime CTLs without the participation of MPs, the CTL repertoire is incomplete and biased toward the few peptides that interact strongly with MHCI. Our data demonstrate a division of labor between CD169 + MPs and DCs in situations where CD169 + MPs sequester the majority of Ag in the marginal zone. In these situations, bystander DCs have to rely on cross-presentation, which elicits CTLs of limited specificity. Results Adenoviral Targeting of CD169 + MPs. To dissect the relative con- tributions of DCs and MPs to CTL priming, we compared a mouse line that lacks DCs only in secondary lymphoid organs with a mouse line that lacks both DCs and MPs in those organs. As described previously, Δ-DC mice constitutively express diphtheria toxin α-chain (DTA) under control of a loxP-flanked stop cassette in the ROSA26 locus, which is removed on breeding to CD11c-Cre mice, leading to DTA-mediated ablation of con- ventional DCs and plasmacytoid DCs (3). As a consequence, Δ-DC mice lack DCs, but harbor regular MPs, such as marginal zone SIGN-R1 + and CD169 + MPs, and red pulp F4/80 + MPs of the spleen (3) (Fig. 1A). In contrast, CD11c-DTR mice express the receptor for DTA constitutively on all CD11c + cells and, on DTA treatment, lack CD11c + DCs and all types of MPs in the spleen, as described previously (10, 11) (Fig. 1A). Subcapsular CD169 + MPs were present in the lymph nodes of Δ-DC mice, whereas CD11c + DCs and CD11c + CD169 + DCs were not, but all cell types were Significance Although we know much about the molecular mechanisms of cross-presentation, its actual contribution to cytotoxic T cell (CTL) immunity under physiological conditions in vivo is still unclear. Cross-presentation is based on the idea that dendritic cells (DCs) are the only professional antigen-presenting cells able to prime naïve T cells. If DCs are not directly infected, they must take up antigen and present it indirectly. However, recent evi- dence suggests that other cells also may be involved in T cell priming, which probably makes cross-presentation less central. This study shows that cross-priming DCs generate highly restricted CTL repertoires, biased to strong MHC I binding epitopes only. Furthermore, the presence of antigen in CD169 + macrophages is sufficient for generation of CTLs with broader repertoires. Author contributions: C.A.B. and T.B. designed research; C.A.B. and C.R. performed re- search; S.K. contributed new reagents/analytic tools; C.A.B. and T.B. analyzed data; and T.B. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Freely available online through the PNAS open access option. 1 To whom correspondence should be addressed. Email: [email protected]. This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1423356112/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1423356112 PNAS | April 28, 2015 | vol. 112 | no. 17 | 5461–5466 IMMUNOLOGY AND INFLAMMATION Downloaded by guest on December 21, 2020
Transcript
CD169+ macrophages are sufficient for priming of CTLswith specificities left out by cross-primingdendritic cellsCaroline A. Bernharda, Christine Rieda, Stefan Kochanekb, and Thomas Brockera,1
aInstitute for Immunology, Ludwig-Maximilian University of Munich, 80336 Munich, Germany and bDepartment of Gene Therapy, University of Ulm, 89081Ulm, Germany
Edited by Emil R. Unanue, Washington University, St. Louis, MO, and approved March 27, 2015 (received for review December 11, 2014)
Dendritic cells (DCs) are considered themost potent antigen-presentingcells (APCs), which directly prime or cross-prime MHC I-restrictedcytotoxic T cells (CTLs). However, recent evidence suggests theexistence of other, as-yet unidentified APCs also able to prime T cells.To identify those APCs, we used adenoviral (rAd) vectors, which donot infect DCs but selectively accumulate in CD169+ macrophages(MPs). In mice that lack DCs, infection of CD169+ MPs was sufficientto prime CTLs specific for all epitopes tested. In contrast, CTL responsesrelying exclusively on cross-presenting DCs were biased to selectedstrong MHC I-binding peptides only. When both DCs and MPs wereabsent, no CTL responses could be elicited. Therefore, CD169+ MPs canbe considered APCs that significantly contribute to CTL responses.
Dendritic cells (DCs) are known to have superior T-cell ac-tivation ability compared with other cell types and thus are
considered the most potent antigen-presenting cells (APCs) (1).Experiments with mice lacking DCs have supported this view inmany infection and immunization studies (2, 3). In addition, DCsare specialized to efficiently cross-present exogenous antigen (Ag)via MHC I to CD8+ T cells (4).In contrast, MPs are considered cells of the innate immune
system that limit pathogen spread by phagocytosis and degra-dation. They are positioned at strategically important entrypoints, such as the subcapsular sinus of lymph nodes and themarginal zone and red pulp areas of the spleen, where theycapture and filter pathogens (5). Along with producing type IIFN (6), CD169+ marginal zone metallophilic MPs allow re-stricted replication of captured virus to increase the local avail-ability of viral Ag (7) for transfer to B cells for humoral responses(8) and to DCs for cross-priming of CD8+ T cells (9). Participationof MPs in T-cell priming is thought to be restricted to Ag acqui-sition, amplification, and redistribution.Analyzing MP functions in vivo is difficult; genetic models,
such as the CD11c-diphteria toxin (DT) receptor (DTR) trans-genic mice (2), cannot make the distinction, because both DCsand monocytes/MPs are equally deleted on DT treatment (10,11). Recent results with mice lacking CD169+ MPs suggest therole of these MPs in cross-priming of CD8+ T cells specific forthe cell-associated model Ag ovalbumin (OVA) in lymph nodes(12). These findings are in contrast with another study in whichMPs were unable to cross-present soluble OVA protein in micelacking DCs (11). Although cross-presentation by DCs is nowthought to allow priming of cytotoxic T-cell responses to exog-enous tumor or viral Ag (4), the importance of cross-presentationin vivo has been a matter of intense debate over the last decade(13). Although this debate remains unresolved, cross-presentationhas been considered then and now to be an exception rather thana rule (14, 15).To determine the roles of MPs and DCs, we used a mouse
model that specifically lacks DCs but conserves all subpopulationsof MPs. We show that the presence of directly infected CD169+
MPs is sufficient to prime CTLs, including those recognizing epi-topes not covered by cross-presenting DCs. In contrast, when onlyDCs cross-prime CTLs without the participation of MPs, the CTLrepertoire is incomplete and biased toward the few peptides thatinteract strongly with MHCI. Our data demonstrate a division oflabor between CD169+ MPs and DCs in situations where CD169+
MPs sequester the majority of Ag in the marginal zone. In thesesituations, bystander DCs have to rely on cross-presentation, whichelicits CTLs of limited specificity.
ResultsAdenoviral Targeting of CD169+ MPs. To dissect the relative con-tributions of DCs and MPs to CTL priming, we compared amouse line that lacks DCs only in secondary lymphoid organswith a mouse line that lacks both DCs and MPs in those organs.As described previously, Δ-DC mice constitutively expressdiphtheria toxin α-chain (DTA) under control of a loxP-flankedstop cassette in the ROSA26 locus, which is removed on breedingto CD11c-Cre mice, leading to DTA-mediated ablation of con-ventional DCs and plasmacytoid DCs (3). As a consequence,Δ-DC mice lack DCs, but harbor regular MPs, such as marginalzone SIGN-R1+ and CD169+ MPs, and red pulp F4/80+ MPs ofthe spleen (3) (Fig. 1A). In contrast, CD11c-DTRmice express thereceptor for DTA constitutively on all CD11c+ cells and, on DTAtreatment, lack CD11c+ DCs and all types of MPs in the spleen, asdescribed previously (10, 11) (Fig. 1A). Subcapsular CD169+ MPswere present in the lymph nodes of Δ-DC mice, whereas CD11c+
DCs and CD11c+CD169+ DCs were not, but all cell types were
Significance
Although we know much about the molecular mechanisms ofcross-presentation, its actual contribution to cytotoxic T cell(CTL) immunity under physiological conditions in vivo is stillunclear. Cross-presentation is based on the idea that dendriticcells (DCs) are the only professional antigen-presenting cells ableto prime naïve T cells. If DCs are not directly infected, they musttake up antigen and present it indirectly. However, recent evi-dence suggests that other cells also may be involved in T cellpriming, which probably makes cross-presentation less central.This study shows that cross-priming DCs generate highly restrictedCTL repertoires, biased to strong MHC I binding epitopes only.Furthermore, the presence of antigen in CD169+ macrophages issufficient for generation of CTLs with broader repertoires.
Author contributions: C.A.B. and T.B. designed research; C.A.B. and C.R. performed re-search; S.K. contributed new reagents/analytic tools; C.A.B. and T.B. analyzed data; andT.B. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Freely available online through the PNAS open access option.1To whom correspondence should be addressed. Email: [email protected].
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1423356112/-/DCSupplemental.
missing in CD11c-DTR mice (Fig. S1A). Thus, these two mouselines are differentially deprived of APC subsets that are key topathogen handling and Ag presentation.To assess the CTL-priming capacities of MPs and DCs without
the confounding effects of viral spread and pathogenicity, weinjected C57BL/6 (B6) mice with GFP-expressing recombinantadenovirus particles (rAd-GFP) that are unable to replicate. His-tological analyses showed that these particles are selectivelytrapped by marginal zone CD169+ MPs but not by CD11c+ DCsor any other cells of the spleen (Fig. 1B), just as described pre-viously for WT adenovirus (9, 16). Both CD169+ and SIGN-R1+
MPs of the marginal zone expressed rAd-encoded GFP (Fig. 1B,Lower). Importantly, spleens of Δ-DC mice showed the samedistribution of rAd-directed GFP expression as WT mice, andboth CD169+ and SIGN-R1+ MPs, but no other cells, expressedGFP (Fig. 1C). The analysis of GFP expression kinetics showeddecreasing GFP intensity over time, but with localization remainingconfined to MPs at later time points as well (Fig. S1B). These dataindicated that transcription of rAd-encoded genes occurred ex-clusively in MPs in the spleen, irrespective of the presence orabsence of DCs.Given that the metallophilic CD169+ MPs of the marginal
zone are likely the first cells in the secondary lymphoid organsthat get in touch with pathogens and their products, we askedwhether they are susceptible to TLR ligands. On LPS injection,DCs increased the surface expression of CD80, CD86, andMHCII, whereas CD169+ MPs only up-regulated CD86 andMHCII, but not CD80. The overall levels of CD86 were at leastfivefold lower on CD169+ MPs compared with DCs (Fig. 1D).Taken together, these data show that CD169+ MPs can modulatesurface markers on TLR activation, with CD80 peculiarly excepted.This ability to up-regulate MHC and costimulatory molecules,
together with the fact that CD169+ MPs, but not DCs, selectivelyaccumulate rAd vectors, suggests the possibility that CD169+
MPs could act as APCs for T cells independent of DCs.
Infection of Marginal Zone CD169+ MPs Leads to Priming of CD8+ TCells in the Absence of DCs. With the foregoing tools established,we asked whether directly infected CD169+ MPs are able toinduce CTL responses in the presence or absence of DCs. To thisend, we injected mice with rAd encoding for glycoprotein (GP)from lymphocytic choriomeningitis virus (LCMV) (17) andtested for the priming of CTLs with three different MHC mul-timers: Kb/GP33–41 (GP33), Db/GP33, and Db/GP276–286 (GP276).Interestingly, the GP33 peptide can bind to Kb and Db, although ithas a 1,000-fold higher binding affinity to Kb (18–20). The GP276peptide binds Db with an intermediate affinity and triggers weakerresponses, likely owing to low precursor frequencies (20–22). Asexpected (2), CD11c-DTR mice did not mount any measurableCTL responses above background, indicating that DCs, MPs, orboth are central to CTL priming. In contrast, Δ-DC mice gener-ated CTLs specific for all three epitopes tested despite their lackof DCs (Fig. 2A, Upper). In vitro restimulation with the respectivepeptides showed that CTLs primed in Δ-DC and WT mice dis-played comparable cytokine secretion capacities (Fig. 2A, Lower).These data suggest that rAd-encoded gene products are presentedby cells other than DCs, most likely CD169+ MPs.To extend these analyses to other Ags, we used rAd vectors
expressing cytosolic chicken ovalbumin (OVA), which harborsthree peptides with immunodominance patterns correlated di-rectly with their binding affinity to Kb. The immunodominantOVA257–264 (OVA257) binds 12- to 25-fold better to Kb comparedwith the subdominant peptide OVA55–62 (OVA55) (23, 24). Inaddition, a third epitope OVA107–114 (OVA107) with muchweaker binding capacities to Kb is known (25). On rAd-OVAinfection, Δ-DC mice primed CTLs specific for Kb/OVA257 andKb/OVA55 as detected with the respective MHC peptide multi-mers (Fig. 2B, Upper). Intracellular cytokine detection on in vitrorestimulation with the three peptides revealed an immunodomi-nance hierarchy similar to that in WT mice (Fig. 2B, Lower).Moreover, CD11c-DTR mice could not mount any responses.Taken together, our data demonstrate that CTLs specific for various
Fig. 1. rAd vectors infect MPs in the marginal zone, but not DCs or other cell types. (A) Representative serial micrographs of spleens from C57BL/6 control,Δ-DC, or DTA–treated CD11c-DTR mice analyzed for CD11c, B220 and SIGN-R1, or CD169 or F4/80, respectively. (Scale bar: 200 μm.) (B) Immunofluorescenceanalysis of spleens from C57BL/6 mice at 48 h after i.v. injection with rAd-GFP. After signal amplification for GFP, spleen sections were analyzed for CD11c andCD169 (Upper) or for CD169 and SIGN-R1 (Lower). (Scale bar: 200 μm.) The white frame (Left) was enlarged for detailed analysis of infected macrophages(Right; arrow: GFP+CD169+, marginal metallophilic macrophage; arrowhead: GFP+SIGN-R1+CD169low, marginal zone macrophage). (C) Similar analysis asshown for Bwith spleens from Δ-DC mice. (Scale bar: 200 μm.) (D) Surface marker analysis of DCs and CD169+ macrophages from spleens of mice injected withLPS. Numbers in histograms indicate mean fluorescent intensities of the respective stainings (mean, n = 3).
5462 | www.pnas.org/cgi/doi/10.1073/pnas.1423356112 Bernhard et al.
epitopes can be primed in the absence of DCs as well. Consid-ering that CD169+ MPs up-regulate costimulatory markers andare the only cells that also express Ag, they most likely primethese CTLs.A caveat of the system is that Δ-DC mice have only one-half as
many T cells as WT mice, so that numbers of specific CTLs werereduced by one-half. To correct for this, we adoptively trans-ferred identical numbers of allogeneically marked Db/GP33-specific P14 or Kb/OVA257-specific OTI TCR-transgenic T cellsand analyzed their expansion on rAd-GP or rAd-OVA injection(Fig. 2C). P14 and OT1 T cells expanded in Δ-DC mice as theydid in WT recipients, whereas neither were activated in CD11c-DTR-mice (Fig. 2C). These findings suggest that despite theabsence of DCs, the presence of MPs and infection of CD169+
MPs (Fig. 1C) are sufficient for the priming of CTLs. In contrast,when both DCs and MPs were depleted in CD11c-DTR-mice, noCTL responses could be measured (Fig. 2C). One potentialreason for this failure in CD11c-DTR mice could be the lack ofprimary Ag handling by MPs of the marginal zone (Fig. 1B);however, histological analyses revealed that rAd-directed GFPexpression was also present in spleens and lymph nodes of DTA-treated CD11c-DTR mice, but at different localizations than inorgans of MP-sufficient mice (Fig. S1D). GFP expression wasdetected in deeper zones of lymph nodes (Fig. S1D) rather thanbeing confined to few subcapsular CD169+ MPs as observed inWT nodes (Fig. S1 B and C). In the spleens of CD11c-DTR
mice, GFP expression was seen in only a few cells of the marginalzone, with the majority found in the red pulp area (Fig. S1D).Therefore, although localized differently, rAd-encoded Ag waspresent in spleens and lymph nodes of DC- and MP-depletedCD11c-DTR-mice, but could not be presented for CTL priming.The fact that it worked in Δ-DC mice indicated that infection ofMPs alone was sufficient for CTL responses.
DCs Selectively Cross-Prime CTLs Specific for Strong Epitopes Only.Our results presented so far show that presence of infected MPsis sufficient to prime CTL responses in the absence of DCs.However, they do not reveal whether CTL priming by MPs andDCs is just redundant or qualitatively different. To determinewhether the two APC types prime CTLs of different specificities,we used mice in which only DCs and thymic epithelial cells, andno other cell types, express MHCI molecules (26). These animalsgenerate a normal CD8+ T-cell repertoire that can be primedonly by MHCI+ DCs (27). Using this model, we previouslyshowed that DCs are sufficient to prime different virus-specificCTL responses (27–29) and to cross-present the model Ag OVAto OTI T cells (26). Surprisingly, DC-MHCI mice injected withrAd-GP mounted nearly exclusively Kb/GP33-specific CTL re-sponses, with strongly reduced Db/GP33-specific responses andundetectable Db/GP276-specific responses (Fig. 3A). A com-parison of the distribution of CTL specificities as detected byMHC multimers in WT, Δ-DC, and DC-MHCI mice revealed
Fig. 2. Δ-DC mice prime normal CTL responses despite the absence of DCs. (A and B) Ag-specific CTLs were detected from spleen cell suspensions at 8 d afteri.v. injection of rAd-GP (A) or rAd-OVA (B) using the indicated MHC multimers. Data are gated on CD19−CD8+ cells from two independent experiments (n = 5mice) and were analyzed using the unpaired two-tailed Student t test. Frequencies of the respective CD8+ T cells are shown in the bar graphs. For functionalanalyses, spleen cells were cultured in vitro with the indicated peptides in the presence of CD107a-specific mAb and subsequently stained for CD8 and IFN-γ.(C) P14 or OTI T cells (2 × 106) were adoptively transferred into the indicated mouse strains, which were subsequently immunized with rAd-GP or rAd-OVA.The expansion of CD8+CD90.1+ P14 and OTI T cells was analyzed by flow cytometry 3 d later. Total cell numbers are shown in the bar graph. Data arerepresentative of at least two independent experiments with three mice per group and were analyzed using the unpaired two-tailed Student t test.
Bernhard et al. PNAS | April 28, 2015 | vol. 112 | no. 17 | 5463
that CTL responses induced in that absence of DCs (Δ-DC mice)or presence of DCs and MPs (WT mice) had a very similarcomposition (Fig. 3A); however, DCs alone (DC-MHCI) pref-erentially primed CTLs specific for the strongest of the three GPepitopes, Kb/GP33, but were entirely unable to prime Db/GP33-specific CTLs (Fig. 3 A and B). The Kb/GP33-specific CTLs werefully functional, as demonstrated by their capacity to produceIFN-γ. In addition, after an increase in the precursor frequenciesof specific CTLs by adoptive transfer of carboxyfluorescein suc-cinimidyl ester (CFSE)-labeled Db/GP33-specific P14 T cells, nopriming by DCs was detected in DC-MHCI recipients (Fig. 3C).The distribution of rAd-encoded proteins in DC-MHCI mice,
as determined by histological analyses of rAd-GFP injection,matched the Ag distribution in WT mice (Fig. 1B) and was re-stricted to CD169+ MPs of the marginal zone. From this, weconcluded that the GP33 peptide was most likely transferredfrom CD169+ MPs to DCs for cross-presentation. Apparently,via this route the GP33 peptide is presented only by Kb, to whichit has a much higher affinity compared with Db. Our findingssupport the possibility that DCs cross-present only peptides ofthe highest affinity when Ag is acquired from MPs.To put this hypothesis to a test, we analyzed CTL responses to
rAd-OVA with its OVA257 peptide, which binds to Kb withmore than 10-fold higher affinity than OVA55 does (23). Wedetected CTLs specific for both epitopes in WT mice, whereasDC-MHCI mice mounted only CTLs specific for Kb/OVA257,and not those specific for Kb/OVA55 (Fig. 3D and Fig. S2 A andB). These experiments also revealed that rAd-OVA–immunizedWT mice primed functional CTLs to the very weakly bindingOVA107 peptide, which were not induced by cross-priming inDC-MHCI mice (Fig. 3D and Fig. S2B).Our data thus extend the evidence supporting the hypothesis
that DCs present only high-affinity peptides from Ags handedover from MPs. This conclusion is based on the assumption thatin DC-MHCI mice, the directly infected MPs are negative forMHCI and are unable to directly prime CTLs. Indeed, CD169+
MPs of DC-MHCI mice did express very low levels of MHCI,probably too low for CTL priming, whereas DCs of DC-MHCImice expressed WT MHCI surface levels (Fig. 3E). Taken to-gether, our results indicate that the repertoires induced by cross-
presenting DCs are very narrow and biased toward high-affinitypeptides from Ags handed over from MPs; however, this gap inthe repertoire is filled, and those CTLs spared by DCs areprimed, when MPs accumulate Ags and express MHC class I(Fig. 3 A and D).
Spleen DCs Cannot Cross-Present Db-GP33 in Vitro. To further ana-lyze the Ag-presenting capacities of DCs, we isolated DCs fromspleens of rAd-injected mice and tested them in vitro for theircapacities to prime specific T cells. DCs from rAd-OVA-injectedB6 and DC-MHCI mice could prime OVA257-specific CD8+
OTI T cells equally well, as when loaded with synthetic pOVA257directly (Fig. 4A, Left). In marked contrast, DCs from rAd-GP33-immunized mice did not stimulate P14 T cells in vitro even if theDCs originated from either strain, whereas peptide loadingworked for both (Fig. 4A, Right). These results suggest that DCsare able to cross-present Kb/OVA257 following OVA acquisitionfrom CD169+ MPs, but not Db/GP33.A simple explanation for this finding is that DCs are not able
to generate the complex at all or destroy it proteolytically.Therefore, we tested whether DCs can be infected in vitro withrAd-GFP, and detected a low but reproducible infection rate of∼1% (Fig. 4B, Upper). In addition, rAd-GP33–infected DCs couldefficiently trigger proliferation of P14 T cells (Fig. 4B, Lower).These results confirm our in vivo findings that DCs do not cross-present the Db/GP33 epitope on rAd-GP33 injection (Fig. 3 Cand D), although they do so when provided as preprocessedpeptide (Fig. 4A) or when infected directly with rAd-GP33(Fig. 4B).To test whether infection or external loading of DCs would
also lead to the priming of P14 T cells in vivo, we infected micewith LCMV, which infects spleen DCs directly (30). Indeed,adoptively transferred P14 T cells were primed with similarefficiency in both B6 and DC-MHCI mice, corroborating ourprevious results (28) (Fig. 4C). Moreover, when mice were im-munized with GP33 and LPS, CFSE-labeled P14 T cells pro-liferated similarly in the two strains (Fig. 4D). Taken together,these data suggest that DCs are able to prime Db/GP33-specificCTLs when directly infected, but not by cross-presentation fromexternal sources.
A
25755
107
rAd
-OVA
CFSE
cells
0 102 103 104 105
44,5±10,6
0 102 103 104 105
9,5±0,3
C57BL/6 DC-MHCIC
MHCI
cells
0 10 2 10 3 10 4 10 5
CD169+CD11c+
0 10 2 10 3 10 4 10 5
MHCI KODC-MHCIC57BL/6
0 102 103 104 105
0
102
103
104
105
CD
11c
CD169isotype0 102 103 104 105
E
C57BL/6 DC-MHCI -DC
Db-GP33Kb-
GP33
GP276
rAd
-GP
02468
1012141618*
0
1
2
3
4
% o
f CD
8 T-
cells
0
1
2
3
4 *
cel
ls10
5
cont
rol
C57BL/
6
DC-MHCI
0
2
4
6
8
10
12
14
cont
rol
C57BL/
6
DC-MHCI
Db/GP33 Kb/GP33B
D C57BL/6 DC-MHCI -DC
% o
f CD
8 T-
cells
cel
ls10
5
cont
rol
C57BL/
6
DC-MHCI
cont
rol
C57BL/
6
DC-MHCI
Fig. 3. DCs cross-prime CTLs against high-affinity MHC I/peptide complexes only. (A) Spleens of rAd-GP–infected mice were analyzed as described in Fig. 2with respective MHC multimers. Pie charts show the frequency of CTLs with the respective specificities in the different mice. Data are representative of morethan three independent experiments with three mice per group. (B) Analysis of Db/GP33- and Kb/GP33-specific T cells in WT and DC-MHCI mice at 8 d after i.v.injection of rAd-GP33. Bar graphs indicate frequencies of specific CTLs as well as total numbers. (C) Adoptive transfer of 2 × 106 CFSE-labeled P14 T cells intomice and immunization with rAd-GP33. Three days later, the CFSE profiles of CD8+CD90.1+ P14 T cells were monitored; numbers represent SE. Data showpercentages from three mice per group and are representative of two independent experiments. Data were analyzed using the unpaired two-tailed Studentt test. (D) Pie charts displaying the distribution of CTLs with the indicated specificities at 8 d after immunization with rAd-OVA. Data are representative ofmore than three independent experiments with three mice per group. (E) Analysis of MHC I expression on DCs and CD169+ MPs from spleens of untreated B6mice, DC-MHCI mice, and MHCI KO mice. Data are from one of three independent experiments with similar results.
5464 | www.pnas.org/cgi/doi/10.1073/pnas.1423356112 Bernhard et al.
DiscussionThe necessity for cross-presentation is based on the idea thatDCs are the only professional APCs able to prime naïve T cells(1). If Ag is not available to DCs directly, then DCs must take itup and present it indirectly. However, recent evidence suggeststhat other, as-yet unidentified cells might be involved in CD8T-cell priming as well (11). Such priming by other cells wouldpossibly make cross-presentation by DCs less important. Ourfindings support this view. We demonstrate that CTL repertoiresinduced by cross-priming DCs have only very restricted speci-ficities only for a few strong epitopes. Normal CTL responses canbe mounted despite a complete lack of DCs. When MHCI+
CD169+ MPs of the marginal zone do capture Ag, CTLs specificfor all epitopes tested can be primed, as well as those spared outby cross-presenting DCs. Therefore, we identified CD169+ MPsas important, nonredundant APCs of similar significance as DCs.In summary, although infection of CD169+ MPs leads to gen-eration of CTLs specific for all epitopes tested for, cross-pre-senting DCs additionally strengthen only a few CTL specificitiesfrom a broad repertoire.Owing to its deficiency for DCs, the Δ-DC mouse model is a
valuable tool for analyzing the contribution of MPs to CTLpriming, given the impossibility of isolating viable marginal zoneMPs for functional T-cell stimulation assays in vitro (12). MPs ingeneral have received much attention recently. Initially consid-ered scavenger cells and pathogen filters, they were thought tohave little or no T-cell priming capability; however, they werefound to modulate immunity via cytokine production and re-cruitment. Their Ag-handling capacities remained restricted toAg redistribution to DCs and B cells. Although it is not possibleto isolate marginal zone MPs for a direct demonstration of theirAPC capacities in vitro, our findings indirectly support the notionthat they also might be able to prime CTLs directly, for severalreasons. First, we found that virus-encoded Ag colocalizes selec-tively in CD169+ MPs. Although Ag detection by microscopy most
likely is not sensitive enough to exclude the presence of low levelsof DC infection, this is insufficient for priming of CTLs specific forweaker epitopes, such as P14 T cells. Second, in theory CD169+
MP-borne Ag also could reach other cells via the exogenous cross-presentation route, as reported previously for CD8+ DCs (9);however, DCs are generally considered the most potent cross-presenting APCs in vivo (4, 31) and a cross-presenting non-DCable to more potently prime CTLs also specific for weak epitopesis currently unknown. Third, we show that exclusive cross-pre-sentation by DCs generates CTL-responses specific for just two ofthe six epitopes studied, whereas in Δ-DC mice the presence ofinfected MPs is sufficient for priming of CTLs to all epitopesderived from OVA and LCMV-GP. Taken together, our resultssupport the interpretation that CD169+ MPs are sufficient toprime CTLs also specific for weaker epitopes.CD169+ MPs are rather sessile in the spleen marginal zone or
subcapsular sinus of lymph nodes; therefore, it is conceivable thatCD8+ T-cell priming by MPs requires T-cell migration to themarginal zone or lymph node sinus areas. Although such in-teractions have been observed during memory responses in vivo(32), whether interactions with naïve CD8+ T cells may also occurthere is unclear. Blood-borne Ag is first trapped by CD169+
marginal zone MPs, making a role for CD169+ MPs in early CTL-priming events rather likely. Thereafter, additional cross-primingmay occur once Ag reaches CD8+ DCs in T-cell areas.Previous studies showed that APC functions of CD169+ MPs
are limited to the immunodominant epitope Kb/OVA257 (12,32), which also can be cross-presented efficiently, but redundantly,by DCs. In contrast, we demonstrate that CD169+ MPs also primeCTLs specific for weaker epitopes that cannot be cross-primed byDCs. Such a division of labor might occur selectively in situationswhere DCs either are not directly infected or do not receive suf-ficient amounts of Ag necessary for cross-presentation (33). This isthe case with the rAd vectors used in the present study, but mightbe different for other infectious agents that preferentially target DCs.
0 102 103 104 105
53.6 1.21
0.6244.8
0 102 103 104 105
0
102
103
104
10553.3 0.25
046.8CD
11c
GFP
rAd-GFPcontrol
rAd-GP33
CFSE
cells
0 102 103 104 105
77.2
1:20 1:40 1:80
DC:T cell ratio
0
1020
30
4050
60
7080
90
% d
ivid
ed c
ells
rAd-GP33 (MOI 2)
rAd-GP33 (MOI 5)
control
0
2.5
5
7.5
10P14
0
1
2
3OTI C57BL/6
DC-MHCI
cont
rol
pOVA25
7-26
4
rAd-
OVA
cont
rol
pGP33
-41
rAd-
GP33
104
cells
per
wel
l B
C D
A
CFSE
cells
pGP33-41/LPS
0 102 103 104 105
57.1±1.3
0 102 103 104 105
49.9±5.3
C57BL/6 DC-MHCI
0
20
40
60
80
0
10
20
30
40
P14
T-c
ells
(%C
D8+
CD
90.1
+)
106
cells
cont
rol
C57BL/
6
DC-MHCI
cont
rol
C57BL/
6
DC-MHCI
LCMV
Fig. 4. Only directly infected DCs can prime Db/GP33-specific T cells. (A) DCs isolated from spleens of B6 or DC-MHCI mice that had been immunized 24 hearlier with i.v. injection of 1 × 1010 particles of rAd-GP33 or rAd-OVA were cultured with naïve P14 or OT-I T cells. As a positive control, DCs isolated fromnaïve mice were loaded with pGP33 or pOVA257. Peptide-free DCs from naïve mice served as negative controls. Priming of P14 and OT-I T cells was analyzedon day 4 of culture. Data are representative of two independent experiments with similar results. (B) (Upper) rAd-GFP–infected CD11c+ bone marrow DCswere analyzed for GFP expression by flow cytometry. (Lower) Proliferation of CFSE-labeled CD8+CD90.1+ P14 T cells cocultured with rAd-GP33-transduced DCswas analyzed by flow cytometry. (C) Mice of the indicated strains were infected with 100 pfu of LCMV at 1 d after adoptive transfer of 5 × 105 P14 T cells. CD8+
CD90.1+ P14 T cells were identified by flow cytometry, Bar graphs show frequencies of P14 T cells as well as total cell numbers (n = 3). (D) CFSE dilution of P14 Tcells adoptively transferred into WT or DC-MHCI mice at 4 d after immunization with GP33-41 peptide plus LPS (thin line) or LPS alone (thick line). Data arerepresentative of two independent experiments with similar results (n = 3 mice per group) and were analyzed using the unpaired two-tailed Student t test.
Bernhard et al. PNAS | April 28, 2015 | vol. 112 | no. 17 | 5465
IMMUNOLO
GYAND
INFLAMMATION
Dow
nloa
ded
by g
uest
on
Dec
embe
r 21
, 202
0
Accordingly, we did not observe direct priming of Kb/OVA257-specific CTLs in Δ-DC mice immunized with MVA-OVA (3), avaccine that depends on infection of DCs for CTL immunity (34).Cross-presentation depends on high doses of Ag (33). In par-
ticular, LCMV-GP is a relatively instable protein with a short half-life (30). Furthermore, GP33-41 is located in the signal sequenceof the protein, which might further reduce epitope stability (35),explaining why this peptide is notoriously inefficiently cross-presented. Initial studies using cells expressing high doses of GPor stable variants of peptides as a source for cross-presentingAg were able to trigger GP33 cross-presentation (36, 37). Mostof those studies did not differentiate between Db/GP33- andKb/GP33-specific CTLs (reviewed in refs. 14, 15), however, and thusKb/GP33-specific CTLs might have been overlooked in some cases.The observed difference in cross-presentation efficiency of GP33
compared with OVA257 stimulated a debate more than a decadeago on the relative importance vs. the very restricted contribution ofcross-presentation to CTL priming (13–15). Determining a precisecross-presentation efficiency is difficult, because it is impossible toquantify the amount of peptide available to the Ag-presentingmachinery of cross-presenting DCs. The most likely explanation forthe cross-presentation preferences of GP33 via Kb but not Db is thehigher affinity under conditions where the Ag quantity is limiting.This also might explain why Kb/OVA257 is cross-presented effi-ciently compared with Kb/OVA55 and Kb/OVA107. In general,weakly binding peptides are poorly cross-presented (38).
Our findings also raise the question of whether similar rulesapply for peripheral tolerance induction by cross-presentation.We previously showed that DCs are sufficient to cross-presentself-Ag for tolerance induction of CD8+ T cells (26, 39). It ispossible that such a bias for strong epitopes also might restrictperipheral cross-tolerance induction.Although we were able to show that CTLs primed by DCs and
CD169+ MPs have different specificities, it remains to be clari-fied whether those CTLs induced by cross-priming vs. directpriming or by DCs vs. MPs are qualitatively equal with respect tolongevity and memory formation. Thus, the precise analysis ofCTL responses initiated by different APC types might greatlyadvance our knowledge for vaccine improvement.
Materials and MethodsAll mice were bred and maintained at the animal facility of the Institute forImmunology, Ludwig-Maximilian University of Munich. Unless stated oth-erwise, mice were immunized i.v. with the respective rAd vectors (1 × 109
particles). Flow cytometry was performed with a FACSCanto flow cytometer(BD Biosciences), and data analysis was conducted using FlowJo software(Tree Star).
More detailed information is provided in SI Materials and Methods.
ACKNOWLEDGMENTS. We thank Reinhard Obst for critical discussions andFlorian Kreppel for providing adenoviral vectors. This work was supportedby the Deutsche Forschungsgemeinschaft (Grants SFB1054 TPB03, BR1889/5,and BR1889/6, to T.B.). Initial support for this work came from FP6-COMPUVACto T.B. and S.K.
1. Banchereau J, Steinman RM (1998) Dendritic cells and the control of immunity. Nature392(6673):245–252.
2. Bar-On L, Jung S (2010) Defining dendritic cells by conditional and constitutive cellablation. Immunol Rev 234(1):76–89.
3. Ohnmacht C, et al. (2009) Constitutive ablation of dendritic cells breaks self-tolerance ofCD4 T cells and results in spontaneous fatal autoimmunity. J Exp Med 206(3):549–559.
4. Joffre OP, Segura E, Savina A, Amigorena S (2012) Cross-presentation by dendriticcells. Nat Rev Immunol 12(8):557–569.
5. Oehen S, et al. (2002) Marginal zone macrophages and immune responses againstviruses. J Immunol 169(3):1453–1458.
6. Iannacone M, et al. (2010) Subcapsular sinus macrophages prevent CNS invasion onperipheral infection with a neurotropic virus. Nature 465(7301):1079–1083.
7. Honke N, et al. (2012) Enforced viral replication activates adaptive immunity and isessential for the control of a cytopathic virus. Nat Immunol 13(1):51–57.
8. Junt T, Scandella E, Ludewig B (2008) Form follows function: Lymphoid tissue mi-croarchitecture in antimicrobial immune defence. Nat Rev Immunol 8(10):764–775.
9. Backer R, et al. (2010) Effective collaboration between marginal metallophilic mac-rophages and CD8+ dendritic cells in the generation of cytotoxic T cells. Proc NatlAcad Sci USA 107(1):216–221.
10. Probst HC, et al. (2005) Histological analysis of CD11c-DTR/GFP mice after in vivodepletion of dendritic cells. Clin Exp Immunol 141(3):398–404.
11. Meredith MM, et al. (2012) Expression of the zinc finger transcription factor zDC(Zbtb46, Btbd4) defines the classical dendritic cell lineage. J Exp Med 209(6):1153–1165.
12. Asano K, et al. (2011) CD169-positive macrophages dominate antitumor immunity bycrosspresenting dead cell-associated antigens. Immunity 34(1):85–95.
13. Melief CJ (2003) Mini-review. Regulation of cytotoxic T lymphocyte responses bydendritic cells: Peaceful coexistence of cross-priming and direct priming? Eur J Im-munol 33(10):2645–2654.
14. Zinkernagel RM (2002) On cross-priming of MHC class I-specific CTL: Rule or excep-tion? Eur J Immunol 32(9):2385–2392.
15. Zinkernagel RM (2014) On the role of dendritic cells versus other cells in inducingprotective CD8+ T cell responses. Front Immunol 5:30.
16. Di Paolo NC, et al. (2009) Virus binding to a plasma membrane receptor triggers in-terleukin-1 alpha–mediated proinflammatory macrophage response in vivo. Immu-nity 31(1):110–121.
17. Flatz L, et al. (2010) Development of replication-defective lymphocytic choriomeningitisvirus vectors for the induction of potent CD8+ T cell immunity. Nat Med 16(3):339–345.
18. Hudrisier D, Oldstone MB, Gairin JE (1997) The signal sequence of lymphocytic cho-riomeningitis virus contains an immunodominant cytotoxic T cell epitope that is re-stricted by both H-2D(b) and H-2K(b) molecules. Virology 234(1):62–73.
19. Achour A, et al. (2002) A structural basis for LCMV immune evasion: Subversion ofH-2D(b) and H-2K(b) presentation of gp33 revealed by comparative crystal structureanalyses. Immunity 17(6):757–768.
20. Kotturi MF, et al. (2007) The CD8+ T-cell response to lymphocytic choriomeningitis virusinvolves the L antigen: Uncovering new tricks for an old virus. J Virol 81(10):4928–4940.
21. Gallimore A, et al. (1998) Induction and exhaustion of lymphocytic choriomeningitisvirus-specific cytotoxic T lymphocytes visualized using soluble tetrameric major his-tocompatibility complex class I-peptide complexes. J Exp Med 187(9):1383–1393.
22. Basler M, Youhnovski N, Van Den Broek M, Przybylski M, Groettrup M (2004) Im-munoproteasomes down-regulate presentation of a subdominant T cell epitope fromlymphocytic choriomeningitis virus. J Immunol 173(6):3925–3934.
23. Chen W, Khilko S, Fecondo J, Margulies DH, McCluskey J (1994) Determinant selectionof major histocompatibility complex class I-restricted antigenic peptides is explainedby class I-peptide affinity and is strongly influenced by nondominant anchor residues.J Exp Med 180(4):1471–1483.
24. Skolnick J, Milik M, Kolinski A (1999) Prediction of relative binding motifs of bi-ologically active peptides and peptide mimetics. US Patent 5933819.
25. Lipford GB, Hoffman M, Wagner H, Heeg K (1993) Primary in vivo responses to ov-albumin: Probing the predictive value of the Kb binding motif. J Immunol 150(4):1212–1222.
26. Kurts C, Cannarile M, Klebba I, Brocker T (2001) Dendritic cells are sufficient to cross-present self-antigens to CD8 T cells in vivo. J Immunol 166(3):1439–1442.
27. Gruber A, et al. (2010) Parenchymal cells critically curtail cytotoxic T-cell responses byinducing Bim-mediated apoptosis. Eur J Immunol 40(4):966–975.
28. Richter K, Brocker T, Oxenius A (2012) Antigen amount dictates CD8+ T-cell exhaus-tion during chronic viral infection irrespective of the type of antigen presenting cell.Eur J Immunol 42(9):2290–2304.
29. Torti N, Walton SM, Brocker T, Rülicke T, Oxenius A (2011) Non-hematopoietic cells inlymph nodes drive memory CD8 T cell inflation during murine cytomegalovirus in-fection. PLoS Pathog 7(10):e1002313.
30. Honke N, et al. (2013) Usp18 driven enforced viral replication in dendritic cells con-tributes to break of immunological tolerance in autoimmune diabetes. PLoS Pathog9(10):e1003650.
31. Jung S, et al. (2002) In vivo depletion of CD11c+ dendritic cells abrogates priming ofCD8+ T cells by exogenous cell-associated antigens. Immunity 17(2):211–220.
32. Chtanova T, et al. (2009) Dynamics of T cell, antigen-presenting cell, and pathogeninteractions during recall responses in the lymph node. Immunity 31(2):342–355.
33. Kurts C, Miller JF, Subramaniam RM, Carbone FR, Heath WR (1998) Major histocom-patibility complex class I–restricted cross-presentation is biased towards high doseantigens and those released during cellular destruction. J Exp Med 188(2):409–414.
34. Liu L, Chavan R, Feinberg MB (2008) Dendritic cells are preferentially targeted amonghematolymphocytes by Modified Vaccinia Virus Ankara and play a key role in theinduction of virus-specific T cell responses in vivo. BMC Immunol 9:15.
35. Wolkers MC, Brouwenstijn N, Bakker AH, Toebes M, Schumacher TN (2004) Antigenbias in T cell cross-priming. Science 304(5675):1314–1317.
36. Ochsenbein AF, et al. (2001) Roles of tumour localization, second signals and crosspriming in cytotoxic T-cell induction. Nature 411(6841):1058–1064.
37. Freigang S, et al. (2007) A lymphocytic choriomeningitis virus glycoprotein variantthat is retained in the endoplasmic reticulum efficiently cross-primes CD8(+) T cellresponses. Proc Natl Acad Sci USA 104(33):13426–13431.
38. Pavelic V, Matter MS, Mumprecht S, Breyer I, Ochsenbein AF (2009) CTL induction bycross-priming is restricted to immunodominant epitopes. Eur J Immunol 39(3):704–716.
39. Luckashenak N, et al. (2008) Constitutive crosspresentation of tissue antigens bydendritic cells controls CD8+ T cell tolerance in vivo. Immunity 28(4):521–532.
5466 | www.pnas.org/cgi/doi/10.1073/pnas.1423356112 Bernhard et al.
