PD-L1 blockade immunotherapy rewires cancer-induced emergency myelopoiesis

Abstract

Introduction: Immune checkpoint blockade (ICB) immunotherapy has revolutionized cancer treatment, demonstrating exceptional clinical responses in a wide range of cancers. Despite the success, a significant proportion of patients still fail to respond, highlighting the existence of unappreciated mechanisms of immunotherapy resistance. Delineating such mechanisms is paramount to minimize immunotherapy failures and optimize the clinical benefit.

Methods: In this study, we treated tumour-bearing mice with PD-L1 blockage antibody (aPD-L1) immunotherapy, to investigate its effects on cancer-induced emergency myelopoiesis, focusing on bone marrow (BM) hematopoietic stem and progenitor cells (HSPCs). We examined the impact of aPD-L1 treatment on HSPC quiescence, proliferation, transcriptomic profile, and functionality.

Results: Herein, we reveal that aPD-L1 in tumour-bearing mice targets the HSPCs in the BM, mediating their exit from quiescence and promoting their proliferation. Notably, disruption of the PDL1/PD1 axis induces transcriptomic reprogramming in HSPCs, observed in both individuals with Hodgkin lymphoma (HL) and tumour-bearing mice, shifting towards an inflammatory state. Furthermore, HSPCs from aPDL1-treated mice demonstrated resistance to cancer-induced emergency myelopoiesis, evidenced by a lower generation of MDSCs compared to control-treated mice.

Discussion: Our findings shed light on unrecognized mechanisms of action of ICB immunotherapy in cancer, which involves targeting of BM-driven HSPCs and reprogramming of cancer-induced emergency myelopoiesis.

Keywords: bone marrow; cancer; hematopoietic stem and progenitor cell; immunotherapy; inflammation.

Figure 1

αPD-L1 immunotherapy reduces peripheral MDSC frequencies during tumor progression. (A, B) Quantification through flow cytometry of the GMFI of PD-L1 surface expression measurements of intratumoral (A; n = 7 control, n = 6 αPD-L1) and splenic (B; n = 6 control, n = 7 αPD-L1) CD3⁺ cells, CD11c⁺ DCs, and CD11c⁻CD11b⁺Gr1⁺ MDSCs after 8 days in B16.F10 melanoma-bearing C57BL/6 mice treated either with PBS or αPD-L1. Representative data from four independent experiments. (C–E) Representative FACS plots (C; numbers denote the percentages of gated populations) and quantification of the frequencies in total cells (D) of intratumoral CD45⁺ cells, DCs, and MDSCs in PBS- or αPD-L1-treated C57BL/6 mice after 8 days of B16.F10 (C, D; n = 5 control, n = 5 αPD-L1) and 12 days of MB49 (E; n = 5 control, n = 5 αPD-L1) tumor progression. Data from one experiment (D, E). (F–H) Representative FACS plots (F; numbers denote the percentages of gated populations) and frequencies in total cells of splenic DCs and MDSCs during the 8th day of B16.F10 (F, G; n = 6 control, n = 6 αPD-L1) and MB49 (H; n = 7 control, n = 7 αPD-L1) tumor progression in C57BL/6 mice treated with PBS or αPD-L1. Data from two independent experiments (G) and data from two combined independent experiments (H). p < 0.05*, p < 0.01**, p < 0.001***, p < 0.0001****. If not stated otherwise, unpaired two-tailed t-tests were performed. Means and SEM are depicted in all bar plots. n = biologically independent mouse samples.

Figure 2

Murine and human HSPCs express PD-L1. (A, B) Representative histograms (left) and quantification (right) of surface PD-L1 expression in murine BM HSPCs 8 days following B16.F10 (A; n = 5 naive, n = 5 B16.F10) or MB49 (B; n = 5 naive, n = 5 MB49) inoculation in C57BL/6 mice. Representative flow cytometry data from one (B) and two (A) independent experiments. (C–E) Representative gating strategy (C; numbers denote the percentages of gated populations) of human BM stem (CD45^lowCD34⁺CD38⁻) and progenitor (CD45^lowCD34⁺CD38⁺) cells isolated from HL patients at diagnosis. Representative overlays (left) and GMFI quantification (right) of PD-L1 in CD34⁺CD38⁻ (D; n = 5) and CD34⁺CD38⁺ (E; n = 5) HL patients compared to their counterpart FMO (representation in a log10 scale). (F, G) Representative histograms (left) and quantification (right) of PD-L1 surface staining in BM HSPCs during the 8th day of B16.F10 (F; n = 8 control, n = 11 αPD-L1) or MB49 (G; n = 3 control, n = 3 αPD-L1) tumor development in C57BL/6 mice treated with PBS or αPD-L1. Representative flow cytometry data from 2 (G) and 10 (F; 2 of them are displayed) independent experiments. p < 0.05*, p < 0.01**, p < 0.001***, p < 0.0001****. If not stated otherwise, unpaired two-tailed t-tests were performed. Means and SEM are depicted in all bar plots. n = biologically independent mouse or human samples.

Figure 3

Administration of αPD-L1 drives the expansion of the HSPC compartment and promotes their activation. (A–D) Representative FACS plots (A, C; numbers denote the percentages of gated populations) and frequencies in total cells of BM HSPCs (LSK: (Lin)⁻Sca1⁺cKit⁺), in PBS- or αPD-L1-treated C57BL/6 mice inoculated with B16.F10 (B; C; n = 5 control, n = 5 αPD-L1) and MB49 (D; 5 control, n = 5 αPD-L1) and sacrificed after 8 days. (E, F) Representative FACS plots (E; numbers denote the percentages of gated populations) of BM HSPCs isolated from PBS- or αPD-L1-treated C57BL/6 mice inoculated with B16.F10 (E, F; n = 5 control, αPD-L1 n = 5) and MB49 (G; n = 4 control, n = 5 αPD-L1). After 8 days, mice were sacrificed and stained with the proliferation marker Ki-67 for cell cycle analysis. Frequencies of HSPCs (F, G) in the G0 and G1/S/G2/M cell cycle phases. Two-way ANOVA was performed. Representative data from three (D; HSPCs) and nine independent experiments (B). Data from one experiment (F, G). p < 0.05*, p < 0.01**, p < 0.001***, p < 0.0001****. If not stated otherwise, unpaired two-tailed t-tests were performed. Means and SEM are depicted in all bar plots. n = biologically independent mouse samples.

Figure 4

Tumor immunogenicity dictates the differentiation potential of αPD-L1-targeted HSPCs. (A) Gating strategy of the BM LK pool (Lin⁻Sca1⁻cKit⁺) and the subclusters CMPs (LK CD34⁺CD16/32⁻), GMPs (LK D34⁺CD16/32⁺), and MEP (LK CD34⁻CD16/32⁻) isolated from PBS- or αPD-L1-treated C57BL/6 mice inoculated with B16.F10 and sacrificed after 8 days. Numbers denote the percentages of gated populations. (B, C) Frequencies in total cells of the BM LK compartment, CMPs, GMPs, and MEPs isolated from PBS- or αPD-L1-treated C57BL/6 mice inoculated with melanoma (B; n = 4 control, n = 4 αPD-L1) or MB49 (C; n = 4 control, n = 4 αPD-L1) and sacrificed after 8 days. Representative data from two (C) and four (B) independent experiments. (D–F) Gating strategy of common lymphoid progenitors (CLPs) in the BM (Lin⁻Sca1^lowcKit^low IL-7Rα^hiCD135^hi) (D; numbers denote the percentages of gated populations) and their frequencies in total cells in C57BL/6 mice treated with PBS or αPD-L1 and inoculated with either B16.F10 (E; n = 4 control, n = 4 αPD-L1) or MB49 (F; n = 4 n = 4 control, n = 4 αPD-L1) and sacrificed on the 8th day of tumor development. Representative data from one (F) and two (E) independent experiments. p < 0.05*, p < 0.01**, p < 0.001***, p < 0.0001****. Means and SEM are depicted in all bar plots. If not stated otherwise, unpaired two-tailed t-tests were performed. n = biologically independent mouse samples.

Figure 5

Targeting of the PD-L1/PD-1 axis expands the HSPC compartment in the BM. (A, B) Representative FACS plots (A; numbers denote the percentages of gated populations) and Lin⁻ cells and HSPC frequencies in total cells (B; n = 7 control, n = 7 αPD-1, n = 7 αCTLA-4) in C57BL/6 mice inoculated with B16.F10 and treated with either αPD-1, αCTLA-4, or PBS with BM analysis performed after 8 days. Data from two combined independent experiments. One-way ANOVA was performed. (C) Frequencies in total cells of BM Lin⁻ cells and HSPCs, isolated from B16.F10-inoculated WT or PD-1^−/− C57BL/6 mice (n = 6 WT, n = 5 PD-1^−/− ), sacrificed during the 8th day of tumor development. (D–F) Representative FACS plots (D; numbers denote the percentages of gated populations) and frequencies in total cells (F; n = 9 control, n = 8 αPD-L1) of Lin⁻ cells and HSPC subpopulations in RAG1^−/− mice inoculated with B16.F10 and treated with either αPD-L1 or PBS with BM analysis performed after 8 days. Representative histograms (up) and GMFI quantification (down) (E; n = 9 control, n = 8 αPD-L1) of the PD-L1 surface expression of the aforementioned HSPCs. Data from one experiment (C) and two combined independent experiments (B, E, F) assessed using flow cytometry. p < 0.05*, p < 0.01**, p < 0.001***, p < 0.0001****. If not stated otherwise, unpaired two-tailed t-tests were performed. Means and SEM are depicted in all bar plots. n = biologically independent mouse samples.

Figure 6

Immunotherapy induces transcriptomic reprogramming of BM HSPCs. (A) RNA-seq heatmap of the representative DEGs categorized by their involvement in hematopoiesis and inflammation (|FC| ≥ 1.5, FDR < 0.05) of BM HSPCs isolated from B16.F10 tumor-bearing C57BL/6 mice treated with PBS (control; n = 2) or αPD-L1 (n = 2). (B) Pathway analysis of DEGs from BM HSPCs isolated from B16.F10 tumor-bearing C57BL/6 mice treated with PBS (n = 2) or αPD-L1 (n = 2). (C) GSEA plot showing the positively enriched pathways “negative regulation of myeloid cell differentiation” (NES 1.44, FDR 0.18), “hematopoietic stem cell differentiation” (NES 0.11, FDR 0.18), and “TNF-α signaling via NF-kB” (NES 1.72, FDR 0.03) of the αPD-L1 group compared to control [FDR (q-value) < 25%]. (D) BM-specific IPA of signaling pathways in BM-HSPCs from melanoma αPD-L1-treated mice, as compared to PBS-treated mice (control). The bar color reflects the IPA activation z-score of an enriched pathway which indicates the direction of effect associated from gene to pathway, with orange representing a direct association and blue representing an indirect association between pathway activation/inhibition and gene expression. (E) RNA-seq heatmap of representative DEGs involved in hematopoiesis of CD34⁺ cells from the BM of HL patients isolated at diagnosis (n = 2) and αPD-1-treated (nivolumab; n = 2) HL patients (p-value < 0.05). (F) Pathway analysis of DEGs of CD34⁺ BM cells isolated from untreated and αPD-1-treated HL patients. n = biologically independent mouse and human samples. Heatmaps are normalized log2(CPM); counts per million.

Figure 7

αPD-L1 immunotherapy rewires cancer emergency myelopoiesis. (A) Experimental scheme: HSPCs (Lin⁻Sca1⁺cKit⁺) isolated 8 days from B16.F10 tumor-bearing C57BL/6 (CD45.2⁺CD45.1⁻) mice treated with PBS (control-HSPC) or αPD-L1 (αPD-L1-HSPC) and then adoptively transferred to NBSGW mice (CD45.2⁻CD45.1⁺). Following 7 weeks of HSPC engraftment, the recipient mice were either sacrificed and analyzed [A(1)] or inoculated with B16.F10 and sacrificed after 17 days of tumor development [(A(2)]. (B) Numbers denote the percentages of gated populations. Representative FACS plots of splenic DC cells, G-MDSCs, and M-MDSCs isolated from melanoma-bearing NBSGW [as in (A(2)]. (C, D) Frequencies in total cells of splenic CD45.1⁻ cells, DCs (C; n = 6 control-HSPC, n = 5 αPD-L1-HSPC), G-MDSCs, and M-MDSCs (D; n = 6 control-HSPC, n = 5 αPD-L1-HSPC) in NBSGW mice inoculated with B16.F10 and sacrificed after 17 days [as in A(2)]. Data from two combined independent experiments. p < 0.05*, p < 0.01**, p < 0.001***, p < 0.0001****. If not stated otherwise, unpaired two-tailed t-tests were performed. Means and SEM are depicted in all bar plots. n = biologically independent mouse samples.