Functional redundancy between thymic CD8α⁺ and Sirpα⁺ conventional dendritic cells in presentation of blood-derived lysozyme by class II-MHC proteins

Mice

Mice were maintained under pathogen-free conditions in accordance with institutional animal care guidelines. B10.BR mice were purchased from Jackson Laboratory (Bar Harbor, Maine) or bred in our facility. 3A9 TCR transgenic mice against the major HEL48–62 peptide were a kind gift from Dr. Mark Davis (Stanford University, Stanford, CA). mHEL mice expressing HEL under the I-Eα promoter (26), LB11.3 mice expressing a TCR against the minor HEL: 20–35 peptide (17), and Batf3^−/− mice (27) were generated at our institution. LB11.3 and 3A9 mice were mated to Foxp3.GFP mice generated in the laboratory of Dr. Alexander Rudensky (University of Washington, Seattle, WA).

Flow Cytometry

Flow cytometry data were acquired using a FACSCalibur or LSRII flow cytometer (BD Biosciences) and analyzed using FlowJo software (TreeStar). Cell sorting was performed using a BD FACSAria II cell sorter (BD). In all DC sorting experiments, thymic DC subsets were identified using the following mAbs: anti-CD45RA (14.8)biotin (BD Bioscience), anti-Sirpα (P84) PE (BDBiosciences), anti-CD8α (53-6.7) Alexa 700 (Biolegend), anti-CD11c (N418) APC-Alexa 750 (Biolegend) or anti-CD11c (N418) APC-efluor® 780 (eBioscience). In T-cell sorting experiments, thymocytes were stained with anti-CD4 (GK1.5) APC (eBioscience), anti-CD25 (PC61) PE (BD Pharmingen) anti-CD8α (53-6.7) Alexa 700 (Biolegend). In the in vitro thymocyte antigen assays, thymocytes were stained with anti-CD4 (RM4-5) efluor® 450 (eBioscience), anti-CD8α (53.6.7) Alexa 700 (eBioscience), anti-CD69 (H1-2F3) FITC (eBioscience). mTEC and cTEC were identified using anti-CD45 (30-F11) (eBioscience), anti-EpCAM (G8.8) (BD Bioscience or Biolegend), anti-BP-1 (6C3) (eBioscience), anti-UEA-1 (Biomeda). Anti-I-A_α^k (11-5.2) Alexa 488 or FITC (Biolegend) was used to detect MHC class II. 3A9 TCR transgenic T-cells were identified either using anti-Vβ8.1/8.2 (MR5-2) (BD Pharmingen) or 1G12 mAb (26). LB11.3 TCR transgenic T-cells were identified using anti-Vα2 (B20.1) (eBioscience). Intracellular staining with anti-Foxp3 (FKJ-16S) (eBioscience) was performed using the Foxp3 staining buffer set according to the manufacturer's instructions (eBioscience).

Isolation and Testing of Thymic APC

Thymic DC were isolated from groups of 5–20 mice (5–8 weeks of age) injected with HEL and sacrificed at the relevant time-point for the experiment. DC from un-injected mice were isolated side by side. DC were purified as previously described (28) by Nycodenz density gradient fractionation using Nycoprep™ Universal (Accurate Chemical). Briefly, thymic fragments were digested for 30–45 min at room temperature in media supplemented with collagenase D and DNAse (Roche). DC–T cell complexes were further disrupted by the addition of 0.1 M EDTA. Cells were overlaid on a 14% Nycodenz solution (1.077 g/ml). The low density layer enriched for DC was collected after centrifugation at 4C. Low density cells were further segregated by cell sorting into the following subsets: CD45RA^hiCD11c^int (pDC), CD45RA^loCD11c^int (immature DC), CD45RA^loCD8α⁻ Sirpα⁺CD11c^hi (Sirpα⁺ cDC), CD45RA^loCD8α⁺Sirpα⁻CD11c^hi (CD8α⁺ cDC). In some instances, CD45⁻ TECs were also enriched from the thymi of 5–8 week B10.BR mice injected with or without HEL (29, 30).

Presentation Assays

In experiments evaluating the ability of thymic APC to present HEL, 5×10⁴ 3A9 T-cell hybridoma were cultured with titrating numbers of sorted DC isolated from mice injected with 50µg HEL 30min. Thymic DC subsets isolated from uninjected mice were tested for presentation of HEL or 10µM peptide added continuously to the culture, or HEL added for only one hour. After 24 hours, IL-2 production was measured in the supernatant using the CTLL-2 T-cell line.

In vitro Thymocyte Assay

To assay different DC subsets for their capacity to induce negative selection, an in vitro thymocyte antigen assay was adapted from previously described methods (31–35). Briefly, 5×10⁴ whole 3A9 thymocytes were cultured for 14–16 hours in 96-well round bottom-plates in DMEM supplemented with 10% fetal calf serum in the presence of either, i) sorted DC (starting with 5×10⁴ DC/well) isolated from B10.BR mice 30 min after injection of 150 µg HEL i.v., or, ii) 2×10⁴ sorted DC from un-injected mice in the presence of different amounts of HEL, starting with 30µM HEL. The DC subsets were: Sirpα⁺ cDC, CD8α⁺ cDC, immature DC, and pDC. Cells were washed and stained with anti-CD4, anti-CD8 and anti-CD69 mAbs. To determine the number of cells per well, a known concentration of countbright™ absolute counting beads (Molecular Probes) was added to each well prior to analysis by flow-cytometry. The number of CD4^hiCD8^hi (DP^hi) and CD69⁺CD4SP was determined by multiplying the percentage of these subsets by the total number of lymphocytes in each well.

In vitro Treg Differentiation Assay

The assay testing different thymic APC for induction of Tregs was adapted from a previously described method (25). Briefly, 5×10⁴ sorted Foxp3⁻CD25⁻CD4SP from LB11.3 Foxp3GFP were cultured with 1.5×10⁴ APC in the presence of 250U/ml IL-2 in a 96-well round bottom plate for 5 days. APC were isolated and sorted, either from B10.BR mice 30 min after injection of 50µg HEL or from untreated B10.BR mice. In experiments comparing the effectiveness of TECs relative to DC at inducing Tregs in response to in vivo captured Ag, DC and TECs were isolated by sequential magnetic bead isolation using anti-CD11c beads, followed by anti-CD45 beads to deplete CD45⁺ cells (Miltenyi Biotec). In experiments comparing the effectiveness of different DC subsets at inducing Tregs, DC were enriched first using Nycodenz gradient fractionation followed by cell sorting to further segregate the DC populations. In some experiments DC subsets from untreated mice were cultured with T-cells in the presence of titrating levels of HEL, starting with 30µM HEL.

HEL Uptake

Endotoxin-free HEL was labeled using an AnaTag™ HilytePlus™ 647 protein labeling kit according to the manufacturer's instructions (Anaspec). 5–8 week B10.BR mice were injected intravenously with 500µg HEL HilytePlus 647™ or saline. One group was bled and sacrificed from 1 to 30 min later: their blood leukocytes were analyzed by flow-cytometry to determine uptake. A second group was sacrificed at 5 and 30 min post-injection and thymic DC subsets were isolated by Nycodenz gradient fractionation and stained with DC-specific markers to determine extent of HEL uptake. 1×10⁶ total events were acquired by flow-cytometry to identify HEL positive cells.

Liposome Depletion of Blood Sirpα⁺ cDC

PBS liposomes (PBS-LIP) and dichloromethylene-bisphosphonate (Clodronate) liposomes (CLOD-LIP) were encapsulated as previously described (36). B10.BR mice were injected i.v. with 250 µl of either PBS-LIP or CLOD-LIP. At 12 and 24 hours, mice were bled to evaluate depletion of phagocytic cells in the blood by flow-cytometry. At 24 hours, thymi and spleen were also examined. To analyze the effect of blood monocytes including Sirpα⁺ cDC depletion on thymic selection, 3A9 mice were injected i.v. with 250ul PBS or CLOD liposomes. At 16 hours, mice were injected with either 50µg HEL or pyrogen-free saline (PFS). Twenty hours later, thymi were harvested, single-cell suspensions were made, and the effect on thymic cellularity, CD4SP, and Foxp3⁺ CD4SP was evaluated by flow-cytometry.

DC Transfer

DC were isolated from B10.BR mice injected i.p. for three consecutive days with 10µg/ml FLT3L (37). Eight days after the first injection, spleens were harvested and diced into small fragments which were digested with 0.14U/ml Liberase Blendzyme 3 (Roche). After ACK lysis to remove red blood cells, DC were positively selected by magnetic bead isolation using anti-CD11c beads (Miltenyi Biotec). DC were recovered at ≥ 95% purity as determined by flow cytometry. 2×10⁷ DC were transferred i.v. into 3A9 TCR transgenic mice. Thymi were evaluated 24 hours later for effect on total cellularity, CD4SP, and Foxp3⁺CD4SP.

Statistical Analysis of Data

All data were analyzed with GraphPad Prism software (GraphPad Software). Statistical significance was assessed by the non-parametric Mann-Whitney U test. Differences with P values less than 0.05 were considered significant and are indicated in the figures.

Presentation of HEL in Culture by Thymic Subsets

Thymic APC were isolated either from mice injected with HEL intravenously or not injected. The APC were examined for their presentation of the HEL epitopes 48–62, or 20–35 to the CD4 T cells 3A9 or LB11.3, respectively.

The experiment presented in Figure 1 compared presentation by the various thymic DC subsets. Thymic DC were isolated and sorted into cDC (Sirpα⁺cDC and CD8α⁺cDC) and CD11c^intCD45RA^hi (pDC) subsets (21, 23, 28) (Figure 1A). CD11c^intCD45RA^lo (immature DC), expressing intermediate levels of CD11c and lower levels of MHC Class II and co-stimulatory molecules (Figure S1), were also sorted (Figure 1A). A small but variable percentage of cDC that expressed CD8α also expressed Sirpα (CD172a). These double-positive CD11c^hi cells could represent Sirpα⁺ cDC that passively acquired surface CD8α from other cells, a phenomenon previously described (22). These double-positive cells were excluded from the sorting analyses shown next and in Figures 3 and 4.

In the experiment of panel B, mice were injected with HEL, the thymic DC were isolated 30min later, and tested for presentation of the HEL 48–62 peptide to the 3A9 T-cell hybridoma. Confirming our previous results (17), both Sirpα⁺cDC and CD8α⁺cDC effectively presented in vivo captured HEL with low level presentation by pDC and none by immature DC (Figure 1B).

This result was compared to presentation by DCs isolated from un-injected mice and assayed upon addition of HEL or the HEL 48–62 peptide. Regardless of the DC subsets tested, there was no difference in the presentation of HEL or peptide (Figure 1C, D). However, when the availability of HEL was limited by incubating APCs with HEL for 1 hour, then washing it away before adding T-cells, presentation by pDC and immature DC was about 2 logs less than that with cDC (Figure 1E).

Thymic APC were examined from Batf3^−/− mice which have a significantly reduced number of CD8α⁺cDC in the thymus and peripheral tissues (27). Their cDC were primarily Sirpα⁺ cDC (Figure 2A). There was normal MHC class II expression among the DC subsets (data not shown). Thymic DC were isolated from Batf3^−/− mice and B10.BR mice injected with HEL 30min before, sorted to CD11c^hi MHC II^hi cells and cultured with the 3A9 T-cell hybridoma. There was no difference in both groups in the presentation of HEL acquired in vivo (Figure 2B). Thus, presentation of blood-borne HEL by class II-MHC molecules of thymic APC can take place in the Batf3^−/− mice by the remaining Sirpα⁺ cDC cells.

Negative Selection and Activation of Thymic Lymphocytes by the Various Thymic DC

Negative selection was assayed in an in vitro thymocyte culture system (31–35) using 3A9 thymocytes, from the TCR transgenic mice. Deletion of DP^hi and activation of CD4SP from 3A9 thymocytes was assayed by flow-cytometry after culturing for 16 hrs in the presence of the sorted DC subsets (Figure 3A, B). In the same culture, the absolute number of immature CD4^hiCD8^hi (DP^hi) remaining was examined, while activation was determined by CD69 up-regulation in the CD4^hiCD8⁻ (CD4SP).

Thymic DC were isolated from mice injected with HEL 30–45min before. Both cDC subsets effectively deleted DP^hi cells at all concentrations and were the best thymic APC in inducing negative selection (Figure 3A). Immature DC did not delete and pDC showed a modest effect.

Sorted cDC, immature DC, and pDC isolated from un-injected mice were cultured for 14 hours with 3A9 thymocytes and different HEL concentrations. Figure 3C is an example of thymocytes cultured with a high dose of HEL (30uM): DP^hi cells were deleted by all the DC subsets, albeit to varying extent (Figure 3C, D). Figure 3D shows the response to titrated amounts of HEL represented as absolute numbers of DP^hi cells. With cDC and pDC, DP^hi cells were nearly ablated at the highest dose, while deletion by immature DC was less severe.

Regardless of whether HEL was captured in vivo or was provided in the cultures, there was activation of CD4SP by all DC subsets, although the CD8α⁺cDC were particularly effective (Figure 3B, C bottom, E). Both cDC subsets were better relative to immature DC and pDC at inducing CD4SP activation in response to in vivo captured HEL, causing CD69 up-regulation above background, even with the lowest number of APC tested (~800 cells/well) (Figure 3B).

In vitro Treg Induction by Thymic APC Subsets

Injection of low-dose HEL resulted in the induction of Tregs in both 3A9 and LB11.3 TCR transgenic mice (17). Treg induction was best found with the LB11.3 TCR transgenic mice specific for the minor HEL epitope, 20–35. To identify which APCs were capturing and presenting HEL for Treg induction (38, 39), thymic DC were isolated from mice injected 30 min before with HEL. The APCs were co-cultured in the presence of IL-2 with sorted Foxp3⁻CD25⁻CD4SP from LB11.3Foxp3.GFP mice (Figure 4A–C). There was an increase in the percentage and absolute number of Foxp3⁺CD4⁺ cells, whether the APC were Sirpα⁺ cDC or CD8α⁺cDC (Figure 4B, C). In contrast, there was minimal to no induction with pDC, immature DC (Figure 4B, C) or TECs (Figure S2). DC from uninjected mice did not induce Tregs, indicating that induction required cells to have acquired HEL in vivo (Figure 4A–C). We concluded that Treg induction in response to blood-borne HEL was mediated by cDC.

We compared Treg induction by DC isolated from un-injected mice in the presence of HEL (Figure S2). All DC induced a level of LB11.3Foxp3GFP+ cells. The most effective were the cDC. The cDC induced a response at the lowest level of HEL much better than at the higher dose.

Uptake of Blood-borne HEL by Intra-thymic cDC

The uptake of HEL by thymic APC was examined following injection of HEL labeled with HilytePlus™647. 0.28% and 0.41% of all thymic cells were HEL⁺ at 5 or 30 min after injection, respectively (Figure 5A). 21.3% and 27.4% of CD11c⁺ cells in the DC-enriched fraction contained HEL at 5min and 30min, respectively (Figure 5B). DC enriched cells were further analyzed gating on the different DC subsets (Figure 5C). HEL was acquired by all APC, but the proportion with HEL and their intensity varied. pDC had higher levels of HEL compared to the other DC subsets, while immature DC contained the least amount (Figure 5C). Among the cDC, the Sirpα⁺ cDC showed heterogeneity in uptake but higher on average than the uptake by CD8α⁺cDC (Figure 5C). As previously reported with ovalbumin (18), a higher percentage of Sirpα⁺ cDC acquired HEL compared to CD8α⁺cDC at both 5min and 30min (Figure 5C). [Of note, among the gated CD11c^hi cells expressing Sirpα, there was also HEL uptake by a subset of cells that co-expressed CD8α (Figure 5C).]

Together, these results indicate uptake by all APC with some differences in degree. The differences in uptake between the two cDC subsets, about five fold, had little functional consequences: the presentation experiments showed that CD8α⁺ cDC and Sirpα⁺ cDC presented in vivo captured HEL to a similar extent (17) (Figure 1).

Role of Peripheral Sirpα⁺ cDC in the Establishment of Tolerance to Blood-borne Ags

We considered whether Sirpα⁺ cDC in the blood could take up some of the injected HEL, enter the thymus, and contribute to T-cell selection events. In addition to monocytes and granulocytes, mouse blood includes pDC and two types of myeloid DC subsets (40). Among peripheral DC, blood Sirpα⁺ cDC and pDC entered the thymus at low frequency (18, 19, 24, 25), and Sirpα⁺ cDC induced negative selection and Tregs (19, 25). Additionally, blood monocytes were shown not to migrate to the thymus in the steady state (24). To confirm in our experimental system that DC could enter the thymus and induce negative selection (19), FLT3L elicited CD11c from transgenic mice expressing HEL in all APCs (mHEL mice) that were injected into 3A9 TCR transgenic mice and the effect on thymocytes was evaluated. Injected mHEL DC induced negative selection of 3A9 thymocytes as indicated by significant reduction in cellularity (Figure 6A), number of CD4SP (Figure 6B) and number of Foxp3⁺CD4SP (Figure 6C). Participation of peripheral DC in thymic selection events could be due to direct presentation or transfer of antigen to resident DC in the thymus as previously described between migratory DC and lymph node resident DC (41, 42).

Whether APCs in the blood could take up HEL and enter the thymus was next examined. Mice were injected either with saline or HEL HilytePlus™647, bled at different time-points and the cellular uptake assessed by flow-cytometry. As early as 1min after injection, both CD11c⁺ and CD11c⁻ cells were positive for HEL (Figure 7A–C). HEL⁺ CD11c⁻ subsets were cleared from the blood within five minutes post-injection (Figure 7B), but some HEL⁺CD11c⁺ cells were detected over time, albeit with decreasing frequency (Figure 7C). Most of the HEL was found in both CD11b^hi and CD11b^int/lo subsets within the CD11c negative population (Figure 7D). This population included cells that were CD11c⁻ Sirpα⁺CD11b^hi, a phenotype consistent with the expression pattern for blood monocytes (21, 43, 44). Among CD11c positive cells, there was HEL uptake by both Sirpα⁺ and Sirpα⁻ subsets: the CD11c⁺Sirpα⁺ cells were CD11b^hi, whereas the CD11c⁺Sirpα⁻ subset expressed intermediate levels of CD11b. Thus, while blood HEL is captured pre-dominantly by CD11c negative cells (CD11c⁻Sirpα⁺, and CD11c⁻Sirpα⁻), CD11c⁺ cells, including CD11c⁺Sirpα⁺ cDC, the dominant cell that brings antigens into the thymus, also acquired HEL (Figure 7D). The expression of CD11b by blood Sirpα⁺ cDC is consistent with published reports (18, 24).

The effect on negative selection and Tregs when phagocytic cells were depleted from the blood prior to HEL administration was examined in the experiment shown in Figure 8. Phagocytes were depleted using CLOD-LIP (36, 44). Mice were injected with liposomes, either PBS-LIP or CLOD-LIP, and bled at 12 hours and 24 hours. At 12 and 24 hours there was major depletion of Sirpα⁺cDC (Figure 8A). In addition to Sirpα⁺cDC, other APCs were reduced (44), including cells that were CD11c⁻Sirpα⁺CD11b^hiB220^lo cells, consistent with the expression pattern for blood monocytes (40, 43) (Figure 8A and data not shown). At 24 hours these CD11c⁻Sirpα⁺CD11b^hiB220^lo cells decreased from 48.1% in PBS-LIP mice to 27.5% in CLOD-LIP treated mice (Figure 8A). Moreover, cells with the phenotype of pDC (CD11c⁺Sirpα^intCD11b^loB220⁺) (45, 46) were reduced at 12 hours, but by 24 hours the percentages were comparable to the PBS-LIP control (Figure 8A and S3). Myeloid DC (CD11c⁺Sirpα⁻CD11b⁺)(40, 47) were variably depleted at 12h with normal levels by 24h (Figure 8A and S3). Examination of the thymus 24 hours post-injection showed that the proportion of thymic Sirpα⁺ cDC in CLOD-LIP injected mice was comparable to mice that received the PBS-LIP control (Figure 8A). This was in contrast to the spleen where there was depletion of not only Sirpα⁺ CD11c⁺ cells, but other populations as well in mice that were injected with CLOD-LIP (36) (data not shown). In brief, CLOD-LIP depleted only the circulating APCs.

Based on these findings, we chose 16 hours as an appropriate time-point to examine effects of HEL in the absence of blood Sirpα⁺ cDC. 3A9 TCR transgenic mice were injected first with CLOD-LIP or PBS-LIP, and at 16 hours with 50µg HEL or PFS. Twenty hours later, T cells were examined by flow-cytometry. The HEL dose chosen resulted in near deletion of CD4SP and significant reduction in the number of Tregs in wild-type 3A9 TCR transgenic mice (17). 3A9 CD4SP thymocytes were negatively selected regardless of whether mice had received CLOD-LIP or PBS-LIP as indicated by reduced thymic cellularity (Figure 8B) and number of CD4SP thymocytes (Figure 8C). Further, there was also deletion of Tregs (Figure 8D) that occurred whether mice had been treated with PBS or CLOD LIP.

These results indicate that blood Sirpα⁺ cDC are dispensable for the induction of thymic tolerance by class II-MHC molecules to blood-borne HEL. While CLOD-LIP treatment was most effective at depleting Sirpα⁺ cDC (Figure 8A), other circulating APC (monocytes, pDC) were not completely depleted (Figure 8A and S3). However, given the poor presentation of HEL by intra-thymic pDC (Figure 1) and the lack of entry of monocytes into the thymus (24), it is doubtful that these cells contributed to the negative selection observed here.