Notch Signaling Facilitates In Vitro Generation of Cross-Presenting Classical Dendritic Cells

Abstract

The IRF8-dependent subset of classical dendritic cells (cDCs), termed cDC1, is important for cross-priming cytotoxic T cell responses against pathogens and tumors. Culture of hematopoietic progenitors with DC growth factor FLT3 ligand (FLT3L) yields very few cDC1s (in humans) or only immature "cDC1-like" cells (in the mouse). We report that OP9 stromal cells expressing the Notch ligand Delta-like 1 (OP9-DL1) optimize FLT3L-driven development of cDC1s from murine immortalized progenitors and primary bone marrow cells. Co-culture with OP9-DL1 induced IRF8-dependent cDC1s with a phenotype (CD103⁺ Dec205⁺ CD8α⁺) and expression profile resembling primary splenic cDC1s. OP9-DL1-induced cDC1s showed preferential migration toward CCR7 ligands in vitro and superior T cell cross-priming and antitumor vaccination in vivo. Co-culture with OP9-DL1 also greatly increased the yield of IRF8-dependent CD141⁺ cDC1s from human bone marrow progenitors cultured with FLT3L. Thus, Notch signaling optimizes cDC generation in vitro and yields authentic cDC1s for functional studies and translational applications.

Graphical abstract

Figure 1

DL1-NOTCH2 Signaling Induces Differentiation of cDC1s from a DC Progenitor Cell Line The HoxB8-FL cell line was induced to differentiate in vitro by estrogen withdrawal in the presence of FLT3L alone (FL), FLT3L with control OP9 cells (FL+OP9), or FLT3L with OP9 cells expressing the Notch ligand DL1 (FL+Notch). OP9 cells were added on day 3, and HoxB8-FL cells were analyzed on day 7 of differentiation. (A) Representative staining plots of differentiated HoxB8-FL cells. The top row shows total live cells with B220⁺ MHC class II^lo pDCs and B220⁻ MHC class II^hi cDCs highlighted; the other rows show gated cDCs with CD11b^hi cDC2s and CD11b^lo/− cDC1s highlighted. (B) The subset composition of differentiated HoxB8-FL cells. Shown are fractions of pDCs (of total live cells) and cDC subsets (of gated cDCs) and the absolute number of these subsets per 10⁵ initial undifferentiated cells. Data represent mean ± SD of 6 parallel cultures, representative of 3 experiments. (C) The effect of NOTCH2 blockade on HoxB8-FL cell differentiation. HoxB8-FL cells were differentiated in FL+Notch cultures in the presence of anti-NOTCH2 blocking Ab (anti-N2) at 50 ng/ml (gray open bars) or 500 ng/ml (gray textured bars); the fractions of DC subsets are shown as above. Data represent mean ± SD of 5 parallel cultures. (D) The effect of Notch ligands on HoxB8-FL cell differentiation. HoxB8-FL cells were differentiated in co-cultures with OP9 cells expressing the Notch ligand DL1 or DL4; the fractions of DC subsets are shown as above. Data represent mean ± SD of 6 parallel cultures. Statistical significance: ^∗∗∗∗p < 0.0001; ^∗∗∗p < 0.001; ^∗∗p < 0.01; ns, not significant.

Figure 2

DL1-NOTCH2 Optimizes the Differentiation of cDC1s from the BM Total murine BM cells were cultured in the presence of FLT3L alone (FL), FLT3L with control OP9 cells (FL+OP9) or FLT3L with OP9 cells expressing Notch ligand DL1 (FL+Notch). OP9 cells were added on day 3, and BM cells were analyzed on day 7 of differentiation. (A) Representative staining plots of differentiated BM cells. The top row shows total live cells with B220⁺ CD11c^lo pDCs and B220⁻ CD11c^hi cDCs highlighted; the other rows show gated cDCs with CD11b^hi cDC2s and/or CD11b⁻ cDC1s highlighted. (B) The expression of CD8α on gated cDCs in BM cultures differentiated as in (A). (C) The subset composition of differentiated BM cells. Shown are fractions of pDCs (of total live cells) and cDC subsets (of gated B220⁻ CD11c⁺ MHC class II⁺ cDCs) and the absolute number of these subsets per 10⁵ initial BM cells. cDC1s were defined either as CD24⁺ or CD8α⁺. Data points represent values from BM cultures of individual mice pooled from 2 experiments; bars represent mean. (D) Representative expression of the indicated surface markers on gated CD24⁺ cDC1s from FL or FL+Notch cultures. The expression of cDC1 markers on cDC2s is included as a control. (E) Representative expression of CD11c and MHC class II on DC subsets from FL or FL+Notch cultures. Subsets were gated, omitting the marker that is shown in each case. (F) Representative expression of Esam on cDC2s from FL or FL+Notch cultures. The expression on cDC1s is included as a control; the dotted line represents negative staining. (G) The effect of NOTCH2 blockade on the expression of CD8α on cDC1s and of Esam on cDC2s. BM cells were differentiated in FL or FL+Notch cultures in the presence of control immunoglobulin G (IgG) or anti-N2; the fluorescence intensity of marker expression in the indicated subsets is shown. Data represent mean ± SD of 5 parallel cultures for anti-N2 and IgG and 9 cultures pooled from 2 experiments for controls. (H) The effect of transcription factor deletion on DC differentiation. BM from control wild-type (WT) mice or mice with DC-specific deletion of Irf8 (Irf8 conditional knockout, cKO) or with germline deletion of Irf8 or Batf3 were cultured in FL or FL-Notch cultures. Shown is the fraction of the indicated DC subsets among total live cells; data points represent values from individual mice pooled from three (for Irf8) or two (for Batf3) experiments; bars represent mean. Statistical significance: ^∗∗∗p < 0.001, ^∗∗p < 0.01, ^∗p < 0.05.

Figure 3

Notch Signaling Optimizes the Expression Profile of BM-Derived DCs Duplicate samples of sorted DC subsets from FL and FL-Notch cultures of primary BM were analyzed by RNA-seq. (A) Multidimensionality scaling (MDS) analysis of RNA-seq profiles of DC subsets derived from FL cultures (DC1^FL, DC2^FL), FL+Notch cultures (DC1^Notch, DC2^Notch), and primary splenic DC subsets from WT mice (DC1^WT, Esam_high^WT, and Esam_low^WT DC2s). All samples are plotted on the second and third dimension of MDS. (B) Pairwise comparison of RNA-seq profiles of cDC1 versus cDC2 subsets from the indicated culture conditions. Shown are volcano plots of individual genes, with select subset-specific marker genes highlighted. (C) Pairwise comparison of RNA-seq profiles of the indicated DC subsets generated in FL versus FL-Notch cultures. Shown are volcano plots of individual genes, with select highly differentially expressed genes highlighted. (D) Heatmap of Notch-dependent gene expression in cultured and primary DC subsets. Samples (labeled as in A) were hierarchically clustered by the expression of genes that were downregulated preferentially in Notch2-deficient cDC1s, cDC2s, or both cDC1s and cDC2s. Select genes are highlighted; the color scale represents the row Z score.

Figure 4

Notch Facilitates CCR7-Dependent Migration of DCs DCs were generated by culturing murine BM cells in the presence of FLT3L alone (FL) or FLT3L with OP9 cells expressing the Notch ligand DL1 (FL+Notch). (A and B) Cross-priming of CD8⁺ T cells in vitro. FL or FL-Notch cultures on day 7 were incubated with OVA, and either total cultures or enriched cDC1s were incubated with CFSE-labeled OT-I cells at a 1:10 ratio for 3 days. (A) The levels of CFSE versus the activation marker CD44 in gated CD8⁺ T cells from cultures with the indicated DCs, with the individual peaks of CFSE dilution highlighted. No CFSE dilution was observed in the absence of OVA (data not shown). (B) The fraction of T cells in each peak (mean of 4 parallel cultures ± S.D., representative of 2 experiments). (C) Unsupervised clustering of cultured and primary splenic DCs by the expression of chemokine receptors (Table S4). Shown is the clustering dendrogram with individual replicates of the indicated samples. (D) Heatmap of select chemokine receptor expression in cultured and primary splenic DCs as determined by RNA-seq. The color scale represents the row Z score. (E) The expression of CCR7 on the surface of culture-derived DCs. Shown is a representative staining profile and averaged mean fluorescence intensity (MFI) of CCR7 on gated CD11b⁺ cDC2s or CD24⁺ cDC1s. Data represent mean ± range of 2 cultures. (F) DC migration in vitro in a transwell assay. Total DCs from FL or FL+Notch cultures were seeded in top chambers with the indicated recombinant chemokine in the bottom chamber and allowed to migrate for 3 hr. Shown is the fraction of each DC subset that migrated into the bottom chamber. Data represent mean ± SD of 4 parallel transwell cultures, representative of 3 experiments. Statistical significance: ^∗p < 0.05.

Figure 5

Notch Facilitates cDC1-Mediated T Cell Cross-Priming In Vivo DCs from FL or FL+Notch cultures of primary murine BM were incubated with OVA, and total DCs or enriched cDC1s were injected i.v. into naive WT syngeneic recipient mice. The priming of endogenous OVA peptide-specific CD8⁺ T cells was determined 7 days later by staining of PB leukocytes with H-2K^b-OVA peptide tetramer. (A) Representative staining profiles of gated CD44⁺ TCRβ⁺-activated T cells in the PB, with the CD8⁺ tetramer⁺ cells highlighted. (B) The fraction of OVA-specific T cells among total CD44⁺ TCRβ⁺ CD8⁺ cells after vaccination with the indicated numbers of total DCs. Mice that received no injection (control) or a mock PBS injection are shown as well. Data points represent values from individual mice pooled from 2 experiments; bars represent mean. (C) The fraction of OVA-specific T cells after vaccination with 3 × 10⁵ enriched cDC1s. Data points represent values from individual mice pooled from 3 experiments; bars represent mean. (D) Kaplan-Meier survival plot of animals that were vaccinated with OVA-pulsed total DCs and subsequently challenged with the OVA-expressing melanoma cell line. Statistical significance: ^∗∗∗∗p < 0.0001, ^∗∗∗p < 0.001.

Figure 6

Notch Facilitates cDC1 Development from Human BM Progenitors (A and B) Sorted CD34⁺ stem and progenitor cells purified from human BM were cultured for 2 weeks in the presence of a FLT3L-containing cytokine mix (FSGM) or on monolayers of control OP9, OP9-DL1, or OP9-DL4 cells. (A) Representative staining profiles highlighting the indicated DC subsets in the resulting cultures; primary DCs from the PB are included for comparison. Numbers represent the percentage of gated cells. (B) Absolute numbers of DC subsets generated in a 0.2-mL culture standardized to 3,000 CD34⁺ progenitor cell input. Data points represent values in BM cultures from different donors (n = 8 for OP9, 7 for OP9-DL1, and 2 for OP9-DL4); bars represent mean with SEM. The indicated p values were derived by unpaired two-tailed t test. The data represent the absolute number of DC subsets per input progenitor cell. (C) Comparison of expression profiles of culture-derived and primary DC subsets based on the Nanostring nCounter analysis. Shown is the principal-component analysis of the indicated triplicate samples after removal of the “culture signature” derived by pairwise comparison of all culture-generated versus ex vivo cells. (D and E) Representative staining profiles (D) and cell yields (E) of the cultures of BM from the two patients with biallelic IRF8 mutations (IRF8^108E/108E or IRF8^83C/291Q). The experiments were done as in (A) and (B).

Figure 7

Notch-Driven Differentiation Yields Functional Human cDC1s DCs generated from CD34⁺ progenitors in cultures with OP9 or OP9-DL1 were analyzed in parallel to primary DCs from PB. (A and B) Cytokine production by DCs stimulated for 14 hr with a cocktail of TLR agonists (poly-I:C, LPS, CL075, and CpG). (A) Representative flow cytometric analysis of intracellular cytokine production (TNF, IL-12, and IFN-α) in the indicated gated DC subsets from OP9-DL1 cultures. Grey contours represent unstimulated cells; numbers represent the cytokine-positive fraction. (B) Proportion of cytokine-positive cDC1s (purple), cDC2s (red), or pDCs (blue) generated from CD34⁺ progenitors in culture with OP9 (n = 4) or OP9-DL1 (n = 3) cells compared with PB primary (n = 7) cells following TLR stimulation. Circles, histograms, and bars represent individual experiments, mean, and SEM, respectively; p values are indicated. (C and D) T cell stimulation by DCs cultured with sorted allogeneic blood CD3⁺ T cells. (C) Representative flow cytometric analysis of T cell proliferation in response to culture with DCs. The data show output of a T cell and cDC1 (generated with OP9-DL1) culture. CD11c⁺ CD141⁺ cDC1s could be identified (purple gate) and gated out. CD3⁺ T cells were subdivided by CD8 and CD4 expression. Cell division was indicated by CFSE dilution (turquoise gate). (D) Proportion of CD4⁺ or CD8⁺ T cells that underwent division (CFSE dilution) in culture with cDC1 (purple) or cDC2 (red) isolated from PB or generated in culture with OP9 or OP9-DL1 (DL1) cells. T cells cultured alone or with beads coated with anti-CD3 plus anti-CD28 were used as negative (Neg) and positive (Pos) controls, respectively. Responses to blood DCs were generated from 2–3 DC donors and 3 T cell donors (2–6 independent experiments). Responses to cultured DCs were generated from 2 BM donors combined with 3 T cell donors (4–6 independent experiments). Each circle represents an independent experiment (mean of 1–3 technical replicates). Histograms and bars represent mean and SEM, respectively. The p values were derived from unpaired two-tailed Student’s t test.