NOD/LtJ (NOD) mice were maintained at the UMS006 animal facility (Toulouse). C57Bl6/JRj (B6) mice were purchased from Janvier Labs (Le Genest Saint Isle, France).

For the in vivo activation of DC subsets, mice were i.v. injected with 5 μg of lipopolysaccharide (LPS) from Escherichia coli 0111:B4), 10 μg of CpG-B ODN1826, and 10 μg of Poly I:C (InvivoGen, Toulouse, France).

For in vivo FLT3L treatment, mice were injected subcutaneously (s.c.) with 5 × 10⁶ B16-Flt3L melanoma cells (21).

All experiments involving animals were performed in accordance with national and European regulations and the Institut National de la Santé et de la Recherche Médicale institutional guidelines. Protocols were approved by the “Midi Pyrénées” ethical committee.

For ex vivo DC isolation, the thymuses or spleens were pooled and cut with blunt scissors before digestion in Roswell Park Memorial Institute (RPMI) 1640 with 2% fetal bovine serum (FCS) supplemented with Liberase (200 μg/ml, Roche) and DNase I (40 μg/ml, Sigma-Aldrich) for 10–15 min. After incubation with 2.4G2 antibody, DCs were isolated using CD11c MicroBeads (Miltenyi Biotec), according to manufacturer's instruction. Each of the cDC subsets and pDCs was subsequently examined by fluorescence-activated cell sorting using a FACSAria™ II high-speed cell sorter (BD Biosciences) following staining with F4-80-, CD45.1- or CD45.2-, CD11c-, CD11b-, B220-, CD24-, and Sirpα-specific antibodies.

For pre-cDC and pDC isolation, BM was extracted from the femurs and tibias, and erythrocytes removed using Gey's treatment. Lineage cells were first removed using fluorescein isothiocyanate (FITC) conjugate anti-CD3, anti-CD19, anti-NKP46, and, for pre-cDC isolation only, anti-B220, and anti-FITC magnetic MicroBeads (Miltenyi Biotec). Pre-cDCs and pDCs were subsequently FACS-sorted following staining with anti-MHC II-, CD135-, Sirpα- and CD11c-specific antibodies or CD19-, B220-, and CD11c-specific antibodies, respectively.

BM cells were extracted, and erythrocytes were removed using Gey's buffer. Cells were cultured for 10 days at a density of 1.5 × 10⁶ cells in a 24-well plate in complete RPMI 1640 medium with 10% FCS containing recombinant mouse 100 ng/ml FLT3L (Miltenyi Biotec), at 37°C in 5% CO₂. Half media were exchanged at day 3 with fresh media containing 50 ng/ml of FLT3L (22).

Cells were stained with a combination of biotinylated, FITC-, PE-, PE-Cy7, allophycocyanin-, allophycocyanin-eFluor780-, allophycocyanin-Cy7-, Brilliant Violet 421-, Brilliant Violet 786-, Pacific Blue-, Brilliant Violet 605-, Pacific Orange-, PerCPVio700, or PerCP-Cy5.5-conjugated monoclonal antibodies. The anti-CD45.1 (clone A20), CD45.2 (104), F4/80 (BM8), CD11c (HL3 or N418), CD11b (M1/70), B220 (RA3-6B2), CD24 (M1/69), CD172α (P84), CD80 (16-10A1), CD86 (GL1), MHC II (OX-6 or m5/114), CD3 (1145-2C11 or 500A2), CD19 (1D3), NKP46 (29A1.4), CD135 (A2F10.1), CCR7 (4B12), XCR1 (ZET), Ly6G (1A8), CD19 (1D3), CD90.2 (53-2.1), CD64 (X54-5/7.1), CCR9 (REA943), SiglecH (eBio440c), PDCA-1 (eBio927), and ESAM (REA722) antibodies were from eBioscience, BD Biosciences, or Miltenyi Biotec. Events were collected within a lymphoid gate based on forward-scatter (FSC) and side-scatter (SSC) profiles, and dead cells were excluded using propidium iodide or Fixable Viability Dye staining. Data were acquired on a BD LSRFortessa™ flow cytometer (BD Biosciences). Data were analyzed using BD FACSDiva™ software (BD Biosciences) or FlowJo software (Tree-star).

RNA was extracted using a RNA XS Isolation Kit (Macherey Nagel) or RNeasy Mini Kit (QIAGEN), and cDNA was synthetized using SuperScript™ II reverse transcriptase (Invitrogen) and random primers (Invitrogen) according to manufacturers' instructions. A LightCycler® 480 Real-Time PCR System (Roche) was used for quantitative PCR. Results were analyzed with LightCycler 480 V1.5 software. The cycling threshold value of the endogenous control gene (HPRT) was subtracted from the cycling threshold value of each target gene to generate the change in cycling threshold (ΔC_T). The sequences of the primers for target genes are listed in Table 1.

Immature and mature thymic and spleen cDC1 and cDC2 were FACS-sorted based on CCR7 and ESAM expression. RNA was extracted using a RNA XS Isolation Kit (Macherey Nagel), and total RNA's quality and quantity were determined using an Agilent 2100 Bioanalyzer (Agilent Technologies) and a Qubit 2.0 Fluorometer (Life Technologies). Libraries were generated using illumina® Truseq Stranded mRNA library following manufacturer's instructions. Each RNAseq library was sequenced in triplicate on Illumina HiSeq 3000 sequencer using 150 bp/sequence paired-end “reads” at Genotoul genomic platform (Castanet-Tolosan, France). Between almost 70 and 90 million “reads” were obtained per sample. “Reads” were trimmed through use of the Cutadapt tool (version 1.3), with removal of low-quality bases (–q value, <10) and clipping of adaptor sequences. High-quality RNAseq “reads” were aligned to the mouse reference genome mm10 (National Center for Biotechnology Information) with STAR software (version 2.6.0). The Python package HTSeq-count was used to count the number of reads overlapping with each gene using ENSEMBL annotation. Differential expression genes between cDC subpopulations were determined with DESeq2 package of Bioconductor software, with an adjusted P-value of <0.1 (P-value adjusted for multiple testing with the Benjamini–Hochberg procedure) as the cutoff for genes with significantly differential expression in one cell population relative to their expression in another cell population. Ingenuity Pathway Analysis (IPA; QIAGEN Inc., Hilden, Germany) was then used to translate the differential expressed genes into canonical pathways using the Ingenuity Knowledge Base. Two statistical indexes (P-value and z-score) are determined for each inference. The P-value indicates significantly enriched pathways, and the z-score represents the statistical measure of the concordance between differentially expressed genes (DEGs) and the associated canonical pathway.

Statistical analyses were performed in GraphPad Prism. P-values were determined using a two-tailed Mann–Whitney test except for Supplemental Figure 2 where a parametric t-test was used. In cases of multiple comparisons, the data were first analyzed by the Kruskal–Wallis test, and then Mann–Whitney-derived P-values were corrected using the Bonferroni adjustment. Statistical significance was defined as P < 0.05.

For RNAseq analysis significance, DESeq2 uses a Wald test P-value, which is adjusted for multiple testing using the procedure of Benjamini–Hochberg.

As a first step to the understanding of the regulation of the expression of different antigen-processing enzymes by the different DC subsets found in spleen and thymus of NOD mice, we compared their representation and maturation in each tissue. Both cDCs and pDCs were characterized by their expression of CD11c and the absence of the macrophage-specific marker F4/80. Thus, the cDC1 and cDC2 subsets were defined as F4/80⁻CD45.1⁺CD11c⁺B220⁻CD24⁺ or F4/80⁻CD45.1⁺CD11c⁺B220⁻Sirpα⁺ cells, respectively (Figure 1A). Consistent with the use of CD24 as a specific marker to discriminate cDC1 from cDC2 subset, we found that 91 ± 1.2% and 88.2 ± 1.7% of thymic and splenic CD11c⁺B220⁻CD24⁺ cells, respectively, expressed the cDC1 conventional marker XCR1, whereas its expression was barely detectable among the CD11c⁺B220⁻CD24⁻Sirpα⁺ cDC2 subset (Supplemental Figure 1A). The pDC subset was defined as F4/80⁻CD45.1⁺CD11c⁺B220⁺ (Figure 1A). Of note, thymic and splenic CD11c⁺B220⁺ cells were positive for PDCA-1 expression, more than 80% of this population was also CCR9⁺/SiglecH⁺ positive, consistent with our strategy using CD11c⁺B220⁺ to define the bona fide pDC populations (Supplemental Figure 1B). We found that the relative representation of the different DC subsets was distinct between spleen and thymus (Figure 1B), in agreement with published data. Indeed, although the cDC2 subset was the most preponderant subset in the spleen (cDC2 = 71.1 ± 1.9% (mean ± SEM) vs. cDC1 = 8.6 ± 0.3%, n = 6), it was under-represented in the thymus (cDC2 = 13.7 ± 2.5% vs. cDC1 = 25.4 ± 3.2%, n = 6). The relative frequency of the pDC subset also diverged drastically between the spleen and thymus (2.3 ± 2.2 in the spleen vs. 61.0 ± 5.6% in the thymus). In addition, on the basis of CD86 and MHC class II expression, we found that thymic cDCs show a more mature phenotype than their splenic counterparts (Figures 1C,D).

We next determined whether these phenotypic differences between thymic and splenic DCs were also correlated with difference in the expression levels of proteases of the class II presentation pathway, namely, TSSP, CTSS, and CTSL. We therefore FACS-sorted each individual subset and analyzed the expression of the different proteases by RT-qPCR (Figures 1E,F). We found that Tssp was barely expressed by the different splenic DC subsets and expressed at significantly higher levels by all thymic DC subsets (Figure 1E, upper row and Figure 1F, left panel). It is worth noting that thymic cDC2 and pDCs had the highest levels of Tssp mRNA. By contrast, Ctss was expressed by all splenic and thymic DC subsets; but, as observed for Tssp mRNA, thymic cDC2 and pDCs expressed significantly higher levels of Ctss mRNA than their splenic counterpart (Figure 1E, middle row and Figure 1F, middle panel). Although Ctsl is highly expressed by cortical thymic epithelial cells, its expression was also extended to other APCs including macrophages and pDCs (15, 23, 24). We found that the Ctsl mRNA level was very low in both cDC subsets and similarly high in the pDC subset of the spleen and thymus. Moreover, a significant increase in Ctsl transcripts was observed in thymic cDC2 compared with their splenic counterparts (Figure 1E, lower row and Figure 1F, right panel).

Collectively, these different results showed that TSSP is selectively expressed by thymic DCs and that, overall, thymic DCs expressed higher levels of the different antigen-processing enzymes examined.

Given that NOD mice present several genetic defects and develop autoimmune pathologies linked to defects in central tolerance, we wondered if the protease expression pattern observed in NOD mice was unique to that strain of mice or was also observed in non-autoimmune prone mouse strains. We therefore similarly cell sorted each individual subset and analyzed Tssp, Ctss, and Ctsl expression patterns in the different thymic and splenic DC subsets of B6 mice by RT-qPCR (Figure 2). Of note, we found that the frequency of the different DC subsets differs between NOD and B6 mice (compare Figure 1B with Figure 2A), consistent with previous studies highlighting the unusual DC distribution in the NOD background (25–27). As observed in NOD mice, we found that Tssp mRNA was mainly expressed by thymic cDC1 and cDC2, whereas it was barely detectable in splenic cDC1 and cDC2 (Figure 2B, upper row). Thymic pDCs also showed slightly higher levels of Tssp mRNA, although this did not reach statistical significance. By contrast, Ctss mRNA was expressed by all thymic and splenic DC subsets (Figure 2B, middle row). As observed in NOD mice, the thymic cDC2 subset expressed higher Ctss mRNA level than its splenic counterpart. Finally, as observed in NOD mice, Ctsl was primarily expressed by thymic and splenic pDCs and to a similar level in both tissues (Figure 2B, lower row). Unexpectedly, the level of Tssp transcript was significantly higher in the cDC1 subset in B6 mice as compared with NOD mice; and, conversely, the level of Ctsl transcript was significantly higher in the cDC2 subset in NOD mice as compared with B6 mice.

Collectively, these different analyses showed that in the thymus, cDCs express overall higher levels of the antigen-processing enzymes TSSP and CTSS than in the spleen in both autoimmune NOD mice and B6 mice.

Given that thymic cDCs are more mature than their splenic cDC counterparts, we considered the possibility that the increased Tssp and Ctss mRNA levels in thymic cDCs may simply reflect their maturation. To address this issue, we FACS-sorted immature and mature thymic cDCs and immature splenic cDCs on the basis of CCR7 and ESAM expression, and we performed RNAseq analysis (Supplemental Figure 2A and Figure 3). Consistent with the highest expression levels of the co-stimulation markers CD80 and CD86, we found that thymic CCR7⁺/ESAM⁺ cDC1 and cDC2 exhibited a more mature phenotype than splenic and thymic CCR7⁻/ESAM⁻ cDC1 and cDC2 subsets (Supplemental Figures 2B,C). We first made a global analysis of the genes that were differentially expressed by splenic and thymic cDCs (Supplemental Figure 3). As expected, principal component analysis (PCA) of the RNAseq data showed that the cDC1 and cDC2 subsets have distinct transcriptional programs. Although splenic and thymic immature cDC subsets cluster together, mature CCR7⁺ESAM⁺ thymic cDCs were distant. Indeed, using a greater than two-fold change, we identified 4,855 and 3,263 genes that were differentially expressed between immature and mature thymic cDC1 and cDC2, respectively. Nonetheless, splenic and thymic cDCs show some differences, with 877 and 1,173 DEGs between splenic and thymic cDC1 and cDC2, respectively. IPA indicates that these transcriptional signatures correspond to multiple biological pathways (Supplemental Figures 3B–E). We next focused on Tssp, Ctss, and Ctsl, and we found that overall immature and mature thymic cDC expressed similar levels of Tssp, Ctss, and Ctsl transcripts (Figure 3A). The only exception was Tssp transcripts that were increased in mature thymic cDC2 as compared with immature thymic cDC2, although this did not reach statistical significance. Here again, the level of Tssp transcripts was higher in thymic cDCs as compared with splenic cDCs. We did not find, however, in this analysis a significant increase in the number of Ctss transcripts in thymic cDC2 as compared with splenic cDC2, although there was a trend (spleen cDC2: 36,205 ± 1,969; CCR7– thymic cDC2: 54,387 ± 708.5; CCR7+ thymic cDC2: 42,909 ± 3,283). Other cathepsins such as Ctsb or Ctsd, in which their role in the class II presentation pathway is less clear, were expressed similarly by immature and mature thymic and splenic cDCs (Supplemental Figure 3F). Similarly, AEP (Lgmn) expression was only increased in mature thymic cDCs. Furthermore, the expression of genes coding for class I or class II molecules was very similar between spleen and thymic cDCs and immature and mature thymic cDCs, coherent with the expression of Ciita or Rfx5 in these different populations. cDCs also express several serpin encoding cysteine/serine inhibitors, some of which have been shown to inhibit CTSS, CTSL, or papain in vitro (28). The expression of some of these serpins is down-regulated in thymic vs. splenic cDCs, suggesting an additional mechanism for increased antigen-presentation by thymic cDCs.

Overall, these analyses show that the overall increased expression of Tssp or Ctss by thymic cDC as compared with splenic cDC is not simply resulting from the increased maturation of thymic cDC.

We next determined whether, conversely, maturation of splenic cDCs was associated with an increased expression level of Tssp, Ctss, and Ctsl mRNA. For this experiment, we i.v. injected B6 mice with a combination of LPS, CpG-B ODN1826, and poly I:C; FACS-sorted the cDC1 and cDC2 subsets 20 h later; and analyzed Tssp, Ctss, and Ctsl mRNA expression by each subset. According to such treatment, cell surface expression of CD80, CD86, and MHC class II was up-regulated, consistent with a successful activation of both splenic cDC1 and in a lower extent cDC2 subsets (Figures 3B,C). We found that in vivo TLR stimulation had no effect on the Tssp mRNA expression level by splenic cDC1 and induced only a modest increase in its expression in the cDC2 subset that remained, however, significantly lower than that of thymic cDCs (P < 0.004; compare Figure 1E and Figure 3D). Similarly, in vivo TLR stimulation induced a significant increase in Ctss mRNA levels in both splenic cDC1 and cDC2 that reach levels comparable (cDC2) or even higher (cDC1, P < 0.012) than their thymic counterparts (compare Figure 1E and Figure 3D). Finally, in vivo TLR stimulation increased Ctsl mRNA levels in both splenic cDC1 and cDC2 subsets, reaching levels that were significantly higher than those found in thymic cDC1 and cDC2 (P < 0.024 and P < 0.006, respectively, compare Figure 1E and Figure 3D).

Collectively, these experiments highlight two distinct mechanisms of regulation of Tssp and Ctss expression by individual cDC subsets. First, the thymic environment was associated with enhanced expression of each protease as compared to the splenic environment, independently of the maturation status of the individual cDC subsets. Second, cDC maturation upon TLR stimulation is mainly associated with an increased expression of Ctss mRNA.

As discussed above, cDCs derive from a BM precursor, the pre-cDC, that recirculate via the bloodstream to lymphoid tissues where the pre-cDCs complete their differentiation into immature cDC1 and cDC2 (6–8). By contrast, pDC differentiation occurs entirely within the BM, and the differentiated pDCs recirculate via the bloodstream to seed lymphoid tissues (4). The expression pattern of Tssp and Ctss mRNA by thymic and splenic cDC suggested that tissue-specific factors might control the expression of these proteases during their final differentiation in the thymus or in the spleen. To further address this issue, we first examined the expression level of Tssp, Ctss, and Ctsl mRNA by BM pre-cDCs and differentiated BM pDCs. The two BM subsets were FACS-sorted as described in Figure 4A and subsequently subjected to RT-qPCR analysis. We found that both pre-cDCs and BM pDCs expressed relatively high levels of Tssp and Ctss mRNA (Figures 4C,D). In agreement with their commitment to the cDC lineage, pre-cDCs did not express significant levels of Ctsl mRNA, whereas BM pDC did (Figures 4C,D right row). Tssp and Ctsl expression patterns in pre-cDC and cDC subsets suggested that their expression may be enhanced/sustained by thymic environmental factors, whereas it may be down-regulated by splenic environmental factors. If this hypothesis was correct, one would expect that when generated in the “neutral environment” of in vitro cultures of BM precursors in the presence of FLT3L, cDC1, cDC2, and pDCs should maintain a profile similar to that of their respective BM precursors. We therefore cultured BM cells in the presence of FLT3L for 10 days; FACS-sorted the cDC1, cDC2, and pDCs as described in Figure 4B; and subsequently performed RT-qPCR on each subset. We found that in vitro-differentiated cDC1 and cDC2 maintained high levels of Tssp and Ctss mRNA expression and low levels of Ctsl mRNA expression (Figures 4C,D). By contrast and similarly to the pDC isolated from the BM, in vitro-generated pDCs expressed relatively high levels of the three transcripts (Figures 4C,D, right row).

To ensure that the transcript levels detected in in vitro-generated DCs was not due to overt stimulation with FLT3L, we examine Tssp, Ctss, and Ctsl mRNA levels in the different DC subsets isolated from the spleen and thymus of mice inoculated with B16-Flt3L melanomas. At day 9 post B16-Flt3L injection, the number of thymic and splenic cDCs was increased as was the frequency of the cDC1 subset relative to the cDC2 subset, reflecting enhanced provision of systemic FLT3L (not shown). Tssp mRNA expression levels by spleen cDC1 and cDC2 were moderately increased in treated mice as compared with untreated mice (Supplemental Figure 4A, upper row). The expression levels in treated mice remained, however, lower than those detected in the corresponding thymic cDC subset (P < 0.004). Similarly, pDCs in the spleen of FLT3L treated mice expressed significantly higher Tssp mRNA levels than those from untreated mice, reaching levels found in thymic pDCs. In contrast, FLT3L treatment did not modify the levels of Ctss or Ctsl mRNA in the different DC subsets, nor did it alter the expression of the different proteases in thymic cDC (Supplemental Figures 2B, 4A,B). Hence, FLT3L signaling has no major effect on the expression of Tssp, Ctss, or Ctsl mRNA.

Collectively, these different results indicated that the tissue environment in which pre-cDCs differentiate modulate the level of expression of Tssp, Ctss, and Ctsl transcripts. To better appreciate this tissue imprinting on the expression of the different proteases, we expressed the mRNA levels of Tssp, Ctss, and Ctsl detected in the spleen or thymus cDC subsets as fold of its BM precursor (Figure 5A). Tssp mRNA levels were reduced following differentiation of pre-cDC into cDC1 or cDC2 but significantly more when differentiation occurs in the spleen than in the thymus (Figure 5A, upper row). Ctss mRNA levels were not significantly changed during cDC1 differentiation but were significantly increased during cDC2 differentiation in the spleen and thymus, although this increase was far more important in the thymus as compared with the spleen. Regardless of the tissue considered, Ctsl mRNA expression was reduced in differentiated cDCs except for thymic cDC2 that maintain the low pre-cDC level. pDCs, which differentiate in the BM, show a different profile with a marked increase in Ctsl transcripts in thymic and splenic pDCs (Figure 5B). However, as observed for cDCs, expression of both Tssp and Ctss mRNA was higher in thymic pDCs as compared with splenic pDCs.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.