CD5 expression by dendritic cells directs T cell immunity and sustains immunotherapy responses

Abstract

The induction of proinflammatory T cells by dendritic cell (DC) subtypes is critical for antitumor responses and effective immune checkpoint blockade (ICB) therapy. Here, we show that human CD1c⁺CD5⁺ DCs are reduced in melanoma-affected lymph nodes, with CD5 expression on DCs correlating with patient survival. Activating CD5 on DCs enhanced T cell priming and improved survival after ICB therapy. CD5⁺ DC numbers increased during ICB therapy, and low interleukin-6 (IL-6) concentrations promoted their de novo differentiation. Mechanistically, CD5 expression by DCs was required to generate optimally protective CD5^hi T helper and CD8⁺ T cells; further, deletion of CD5 from T cells dampened tumor elimination in response to ICB therapy in vivo. Thus, CD5⁺ DCs are an essential component of optimal ICB therapy.

Fig. 1.. Altered distribution of CD1c⁺CD5⁺ DCs in melanoma-affected human LNs and correlation with patient survival.

(A) Workflow for single-cell analysis of myeloid cells from human melanoma that were isolated from five iLNs and three matched uLNs from patients with melanoma. (B) UMAP displaying nine clusters of myeloid CD45⁺lineage⁻HLA-DR⁺ cells sorted from five iLNs and three matched uLNs from patients with melanoma and processed for scRNA-seq using 10× Genomics technology (see also fig. S1, A and B). MDSC, myeloid-derived suppressor cell; TAM, tumor-associated macrophage. (C) UMAP displaying CD45⁺lineage⁻ HLA-DR⁺ cell distribution based on tissue type. (D) UMAP plots of six marker genes associated with various myeloid cell types in human tumor iLNs and uLNs as defined in Fig. 1, B and C. The color scale represents the normalized expression of the gene. (E) Scatter plot showing the z-score based on the total expression of CD5 for each cluster separated by iLN (red) and uLN (blue). The left y axis corresponds to the dots on the plot, which represent the expression of CD5 in individual cells. The right y axis corresponds to the bars. The height of each bar represents the z-score of the gene in the group of cells of each cluster. (F) Bar plots showing the number of cells expressing CD5 in each cluster (left) and the fraction of cells in each cluster and sample that expresses CD5 (right). Colors represent the samples from which the cells were derived. Shades of blue or red indicate uLNs or iLNs from different donors, respectively. (G) Expression of CD1c⁺CD5⁺ (bottom) and CD14⁻CD5⁺ (top) in iLNs and uLNs of representative patient Mel002. Cells were gated on CD45⁺CD3⁻CD19⁻MHCII⁺CD11c⁺. One representative plot of eight patients is shown. (H) CD1c⁺CD5⁺ (left) and CD14⁺CD5⁻ (right) in eight iLNs and matched uLNs, plus an additional two iLNs from patients with melanoma. Symbols indicate values from individual patients. (I) UMAP displaying CyTOF analysis of CD45⁺ cells from four iLNs and matched uLNs of patients with melanoma. (J) Relative expression of CD5 in each myeloid cell–specific cluster (top) and distribution among the iLN and uLN cells (bottom). For the box-and-whisker plots, the bars are the default setting using interquartile ranges (IQR), where the lower whisker is quartile 1 minus 1.5 × IQR and the upper whisker is quartile 3 plus 1.5 × IQR. (K) Kaplan-Meier curve (top) showing the proportion of overall survival across the TCGA melanoma cohort by median CD5 mRNA level, as generated by RNA-seq, comparing CD5-high (blue) versus CD5-low (red). Strata (bottom) refers to grouping, high versus low. (L) Kaplan-Meier curve (top) showing the proportion of overall survival across the TCGA melanoma cohort by median CD5⁺ DC signature mRNA level comparing CD5⁺DC-high (blue) versus CD5⁺DC-low (red). DC signature (DC Sign) (bottom) was developed using the product of CD5, CD1C, LAMP3, and CLEC10A divided by the mRNA level of CD3G. (M) Volcano plot of the hazard ratios versus −log10(P value) across TCGA cohorts for the CD5⁺ DC signature. Points highlighted in red are significantly (P < 0.05) associated with worse overall survival. SKCM, skin cutaneous melanoma; LUAD, lung adenocarcinoma; SARC, sarcoma; BRCA, breast invasive carcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma.

Fig. 2.. CD5 expression on DCs correlates with CD4⁺ and CD8⁺ T cell activation.

(A) CFSE dilution of allogeneic naïve T cells primed by either CD5⁺ or CD5⁻ DCs isolated from an uLN from a patient with melanoma or from healthy human dermis. Representative results of three melanoma LN donors and at least 10 dermal DC donors are shown. (B) Proliferation index of allogeneic naïve CD4⁺ T cells (left) and CD8⁺ T cells (right) primed by CD5⁺ or CD5⁻ LN DCs or CD5⁺ and CD5⁻ dermal (der) DCs. Representative results of three melanoma LN donors and at least 10 dermal DC donors are shown. The color scale represents the fraction of cells that have diluted CFSE in each of the indicated cell divisions. (C) Number of IFN-γ⁺TNF-α⁺ CD4⁺ T cells (left) and CD8⁺ T cells (right) primed by CD5⁺ or CD5⁻ LN DCs. Composite data of three experiments with three LN donors are shown. (D) TNF-α production measured in the cultures of T cells stimulated by CD5⁺ or CD5⁻ LN DCs (left) or CD5⁺ or CD5⁻ dermal DCs (right) or by control CD14⁺ DCs for 48 hours. Composite data of two experiments with two LN donors and two different dermal donors are shown. Data represent the means ± SEM. (E) Flow cytometric gating scheme showing CD5^hi, CD5^int, and CD5^neg DCs (live, lineage⁻HLA-DR⁺CD11c⁺CD14⁻CD1a^dim) (left). See also fig. S3A. Quantification of surface CD5 expression (right) on CD5^hi, CD5^int, and CD5^neg dermal DCs using BD QuantiBrite beads. One representative quantification of three performed with DCs from three different donors is shown. PE, phycoerythrin. (F) Sorted dermal CD5^hi (blue), CD5^int (yellow), and CD5^neg (red) DCs or control CD14⁺ DCs (black) were cocultured with naïve CD8⁺ T cells at different DC:T cell ratios for 6 days before CD8⁺ T cell numbers were determined using BD TruCOUNT beads. Composite data of three experiments performed with three different donors are shown; data represent means ± SEM. (G) IFN-γ expression by the proliferating CD8⁺ T cells primed by dermal CD1a^(dim) CD5^hi, CD5^int, or CD5^neg DCs. One experiment of six performed with six different donors is shown. (H) Composite data of six experiments performed with six different donors are shown; data represent means ± SEM. (I) Granzyme B and perforin expression by proliferating (CFSE^lo) CD8⁺ T cells primed by CD5^hi, CD5^int, or CD5^neg DCs. One experiment of three performed with three different donors is shown. (J) Proportions of T-CTLs, D-CTLs, and M-CTLs induced by dermal CD1a^(dim) CD5^hi, CD5^int, or CD5^neg DCs. Composite data of three experiments with three different donors are shown. (K) Activated CD4⁺ T cell proliferation in response to different numbers of dermal CD1a^(dim) CD5^hi (blue), CD5^int (yellow), and CD5^neg (red) DCs and CD14⁺ (black) DCs. CD4⁺ T cells were determined on day 6 using BD Trucount beads. Composite data of four experiments with four different donors are shown; data represent means ± SEM. (L) Expression of IFN-γ by proliferating CD4⁺ T cells primed by dermal CD1a^(dim) CD5^hi, CD5^int, or CD5^neg DCs. One experiment of three performed with three different donors is shown. (M) Proliferation of CD8⁺ (left) and CD4⁺ T cells (right) after stimulation with CD5⁺ and CD5⁻ DCs and anti-CD5 blocking mAb. Composite data of two experiments performed with two different donors in triplicates are shown. CTV, CellTrace Violet proliferation dye. (N) IL-2 production measured in the culture supernatant of T cells after stimulation with CD5⁺ or CD5⁻ DCs in the presence or absence of anti-CD5 blocking mAb (LT-1). Composite data of three experiments performed with three different donors in duplicates or triplicates is shown; data represent means ± SEM. In (C), (D), (H), (M), and (N), the numbers over the brackets are P values.

Fig. 3.. Up-regulation of CD5 on human DCs promotes effector T cell activation.

(A) Flow cytometric analysis of CD5 expression in CD5⁻ dermal DCs 48 hours after nucleoporation with or without dCas9-VPR mRNA and sgRNAs. One representative plot of eight experiments is shown. (B) Flow cytometric analysis of CD5 expression on dermal CD5⁻ DCs 1 to 6 days after nucleoporation with dCas9-VPR mRNA and CD5a sgRNAs or control dCas9-VPR. Symbols indicate values from individual donors. Composite data of CD5 expression analyzed 1 day (n = 7 donors), 2 days (n = 8 donors), 3 days (n = 2 donors), 4 days (n = 4 donors), and 6 days (n = 2 donors) after nucleofection are shown. (C) Volcano plot showing differentially expressed genes between control dCas9-VPR mRNA–electroporated CD5⁻ DCs and CD5_CRISPRa DCs. Genes up-regulated and down-regulated in CD5_CRISPRa DCs are shown in blue and red, respectively. (D) Flow cytometric analysis of the expression of CD72, CD40, CD70, CD80, and CD155 by control dCas9-VPR mRNA–electroporated CD5⁻ DCs and CD5_CRISPRa DCs 24 hours after nucleofection. Composite data of at least four donors analyzed for each marker are shown. (E) Workflow used in (F) to (I). CT, control. (F) Numbers of T cells with diluted CFSE (one division or more) as analyzed on day 6 of the coculture. Composite data of four performed experiments with five donors are shown. (G) CD5_CRISPRa or dCas9-VPR mRNA CT dermal DCs were cultured for 6 days with allogeneic naïve CD4⁺ and CD8⁺ T cells. Percentages of IFN-γ– and TNF-α–expressing T cells were determined by flow cytometry 6 hours after activation. One representative experiment out of three performed is shown (left). The graph shows composite data of three independent experiments; data represent means ± SEM (right). (H) IFN-γ production by T cells after activation for 48 hours with CD5_CRISPRa Langerhans cells or dCas9-VPR mRNA. One representative experiment of five performed with five different donors is shown. Data represent means ± SEM (n = 3) (left). Composite data of four additional experiments performed in the same way using CD5_CRISPRa dermal DCs or dCas9-VPR mRNA CT DCs (right). (I) T helper cytokine (IL-13) production after activation of T cells for 48 hours with CD5_CRISPRa or dCas9-VPR mRNA. The graph shows composite data of four independent experiments. (J) CD5 expression in moDCs 2 days after nucleoporation with or without dCas9-VPR mRNA and sgRNAs. One representative experiment of five performed with five donors is shown. (K) Composite data of CD5 expression analyzed 1 day (n = 5 donors), 2 days (n = 4 donors), 3 days (n = 3 donors), and 5 days (n = 3 donors) after nucleofection are shown. (L) Frequency of IFN-γ– and TNF-α–expressing T cells after 6 days of activation with CD5_CRISPRa or dCas9-VPR mRNA moDCs. One representative experiment of three performed is shown. Data represent means ± SEM (left). Composite data of three experiments performed with three different donors are shown (right). (M) IFN-γ production that was measured in the culture supernatant of naïve T cells after 2 days of activation with CD5_CRISPRa or dCas9-VPR mRNA moDCs at a T:DC ratio of 33:1 or 67:1. One representative experiment is shown. Data represent means ± SEM (left). Composite data of three additional experiments performed at a T:DC ratio of 40:1 in duplicates or triplicates with three different donors are shown (right). In (B), (F) to (I), and (K) to (M), the numbers over the brackets are P values.

Fig. 4.. Deletion of CD5 on DCs compromises the response to ICB therapy and modulates T cell immunity in vivo.

(A) Workflow of MCA1956-mOVA tumor growth in WT control (CT) or CD5ΔDC mice. (B) MCA1956-mOVA tumor growth in CT or CD5ΔDC mice. Results depict tumor growth curves of individual mice from three pooled experiments: CT (WT) (n =15) and CD5ΔDC (n = 11). (C) MCA1956-mOVA tumor growth in CT, CD5ΔDC, or CD5^HET heterozygous (HET) mice. Combined tumor growth curves for CT (n = 15), CD5ΔDC (n = 11), and HET (n = 3) mice are shown. Overall group difference average as measured across time for CD5ΔDC versus CT (P = 0.0055) and CD5ΔDC versus HET (P = 0.0087). There was no difference between HET and CT. Data represent means ± SEM. (D) Binding of OVA:I-A^d tetramer to TDLN cells isolated from MCA1956-mOVA–bearing CT and CD5ΔDC mice. Data represent means ± SEM (n = 4). (E) In vitro OT-II cell proliferation in response to cDC2s isolated from CT, CD5ΔDC, and CD5KO mice and cultured with SLP from the OVA protein (OVA-SLP) that contains the MHC II–restricted peptide. One representative experiment of three is shown. The plot shows means ± SEM of three experimental replicates (left). Proliferated OT-II cells in response to cDC2s loaded with 0.7 μg/ml SLP. The mean of three pooled experiments is displayed with normalization to the control (right). (F) CD5 mean expression on OT-II cells proliferating in response to cDC2s isolated from CT, CD5ΔDC, or CD5KO DCs (left). Data represent means ± SEM of three pooled experiments displayed with normalization to the control (right). (G) Workflow of OVA-SLP vaccination and analysis of the immune response. (H) Representative plot showing CD5 expression on proliferating OT-II T cells in CT or CD5ΔDC mice in vivo. (I) Frequency of proliferating CD5⁺OT-II⁺ cells isolated from the spleen of CT (n = 5), CD5ΔDC (n = 4), or CD5KO (n = 2) mice (left). Mean CD5 expression on proliferating OT-II T cells in CT (n = 5), CD5ΔDC (n = 4), or CD5KO (n = 2) mice in response to SLP vaccine in vivo (right). (J) Workflow of the analysis of immune cell infiltration of MCA1956 tumors in mice treated with anti–PD-1 or IgG2a isotype control. Part of the illustration was created with Biorender.com. (K) Tumor growth from CT mice injected with immunoglobulin G2a (IgG2a) isotype control or anti–PD-1 antibodies (18 to 23 mice per group). Data represent the results of four pooled experiments. (L) Tumor growth from CD5ΔDC mice injected with IgG2a isotype control or anti–PD-1 antibodies (18 to 23 mice per group). Data represent the results of four pooled experiments. (M) Mean tumor diameter in CT or CD5ΔDC mice on day 21 after implantation and treatment with IgG2a isotype control or anti–PD-1 antibodies. Data represent the results of four pooled experiments. (N) Mean tumor diameter in CD5ΔDC and CT mice on day 24 after implantation and treatment with isotype control or anti–CTLA-4 antibodies. (O) CD5 expression by CD4⁺ T cells in the lymphoid immune compartment in the TDLNs of CD5ΔDC or CT mice treated with isotype control or anti–PD-1 antibodies. (P) CD5 expression by CD4⁺ T cells in the lymphoid immune compartment in the tumor of CD5ΔDC or CT mice treated with isotype control or anti–PD-1 antibodies. In (D) to (F), (I), and (M) to (P), the numbers over the brackets are P values. [Illustrations in (A), (G), and (J) created with Biorender.com]

Fig. 5.. Deletion of CD5 on T cells compromises effector T cell priming and response to ICB therapy.

(A) Workflow of tumor growth in control (CT) or CD5ΔT mice. MCA1956 tumor growth in CT (WT) or CD5ΔT mice injected with IgG2a isotype control or anti–PD-1 antibodies. [Illustration created with Biorender.com] (B) Results depict tumor growth curves of individual mice from three pooled experiments for CT (n = 12) and CD5ΔT (n = 13). (C) Results depict tumor growth curves of individual mice from three pooled experiments for CT with anti–PD-1 (n = 13) and CD5ΔT with anti–PD-1 (n = 16). Overall group difference as measured across time for CD5ΔT versus CT (P < 0.0001). (D) Average tumor growth on day 21 in CT or CD5ΔT mice treated with anti–PD-1 or isotype control IgG2a. Data represent means ± SEM. (E) MCA1956-mOVA tumor growth in CT or CD5ΔT mice. Results depict tumor growth curves of individual mice from three pooled experiments for CT (n = 15) and CD5DT (n = 8). Overall group difference as measured across time for CD5ΔT versus CT (P < 0.0001). (F) Frequency of OVA-specific CD8⁺ T cells that were detected in the tumor (left) and spleen (right) of CT and CD5ΔT mice 10 days after tumor cell inoculation. Data represent means ± SEM. (G) Expression of CD5 on T cells (gated on live, CD45⁺CD19⁻CD3⁺; plots show CD4 expression versus CD5 expression) in iLNs and uLNs of two representative patients, Mel028 and Mel018. (H) Composite data showing the frequency of CD5⁺ T cells in iLNs and uLNs of 10 patients with melanoma. Data represent means ± SEM. (I) CD5 and CD6 expression on CT or hCD5ΔT cells after coculture with DCs. (J) CD5 geometric mean expression on CT or hCD5ΔT CD4⁺ or CD8⁺ T cells before coculture with DCs. Composite data of four donors are shown. Data represent means ± SEM. (K) Experimental scheme for (L) to (N). (L) Frequency of IFN-γ expression by CD4⁺ and CD8⁺ T cells in CT or hCD5ΔT cultures with CD5⁺ or CD5⁻ dermal DCs. Data represent one experiment of four performed. (M) Flow cytometric frequency analysis of IFN-γ expressed by CT or hCD5ΔT CD4⁺ (left) or CD8⁺ (right) T cells cocultured with either CD5⁺ or CD5⁻ DCs. Data represent one experiment of four performed with four different donors (see also fig. S11E). Each dot indicates a technical replicate. Data represent means ± SEM. (N) IFN-γ production measured in the culture supernatant of either CT T cells or hCD5ΔT cells cocultured with either CD5⁺ or CD5⁻ DCs. Composite data of four experiments performed with four donors are shown. Data represent means ± SEM. In (D), (F), (H), (J), (M), and (N), the numbers over the brackets are P values.

Fig. 6.. CD5⁺ DCs are modulated in the TME.

(A) Workflow of tumor growth and immune cell infiltrate analysis in control (CT) or CD5ΔDC mice. [Illustration created with Biorender.com] (B) CD5 expression of tumor-infiltrating cDC1 (TIDC1) (left) or cDC2 (TIDC2) (right) from CT mice treated with anti–PD-1 (n = 8; three data points are outside the axis limits for TIDC1) or IgG2a isotype control (n = 7). Data represent means ± SEM. (C) Frequency of CD5⁺ cDC1 (left) or CD5⁺ cDC2 (right) in TDLNs of CT mice treated with IgG2a isotype control (n = 18) or anti–PD-1 antibodies (n = 19). Data represent means ± SEM. (D) Plots show the expression of Annexin V by CD5⁺ and CD5⁻ splenic DCs after 24 and 48 hours of incubation. One representative experiment is shown. Cells are gated on live, lineage⁻F4/80⁻CD64⁻IE/IA⁺CD11c⁺ cells (left). Composite data of three experiments are shown. Each dot represents a different mouse. Annexin V expression was normalized to CD5⁺ DCs for each indicated time point (right). Data represent means ± SEM. (E) CD5⁺ cDC1s and CD5⁺ cDC2s isolated from TDLNs of MCA1956 tumor–bearing mice treated with IgG2a isotype control or anti–PD-1 antibodies and stained with an apoptosis detection antibody (APO-BrdU TUNEL assay). Composite data of three mice per group from one of two independent experiments are shown. (F) UMAP shows CD5 gene expression by the myeloid DC clusters isolated from scRNA-seq of the peripheral blood mononuclear cells from a patient with inherited PD-1 deficiency and the patient’s brother, as well as control subjects. The cells were analyzed per the authors’ Mendeley data report [(50); see https://data.mendeley.com/datasets/nb26v3mx3x/2]. (G) Frequency of IL-6–producing immune cells within tumor-infiltrating immune cells of mice bearing MCA1956 tumors and treated with IgG2a isotype control or anti–PD-1. Data represent means ± SEM (n = 9 mice per group). (H) DC1 (CD24⁺) and DC2 (Sirp-α⁺) output from BM cells that were cultured with either Fms-like tyrosine kinase receptor 3 ligand (FLT3-L) or FLT3-L and 4 ng/ml IL-6 for 8 days. CD5 median expression is shown in heatmap statistic plots. (I) BM cells were cultured for 8 days with FLT3-L (black line), FLT3-L with 0.5 ng/ml IL-6 (dashed blue line), or FLT3-L with 4 ng/ml IL-6 (solid blue line). CD5 expression on output BM DC1s or DC2s (left) is shown. Composite data showing CD5⁺ DC output from mouse BM DCs that were cultured with FLT3-L and the indicated amounts (0 to 20 ng/ml) of IL-6; each dot represents an independent experiment (right). Data represent means ± SEM. (J) CD5 expression on CD1c⁺ DCs derived from human CD34⁺ HPCs that were differentiated with GM-CSF, FLT3-L, and SCF (GM, FL, and SCF) in the presence or absence of IL-6 (100 ng/ml) for 7 days. One experiment is shown. (K) CD5⁺ DC output within the CD34-derived CD1c⁺ compartments. Cells are gated on live, lineage⁻HLA-DR⁺CD11c⁺CD1c⁺. The graph shows composite data of 10 independent experiments. In (B) to (D), (G), (I), and (K), the numbers over the brackets are P values.