Abstract
Liver Type 1 innate lymphoid cells (ILCs) comprise conventional natural killer (cNK) cells and ILC1s. The main functions of Type 1 ILCs not only include directly killing target cells but also regulating the local immune microenvironment of the liver through the secretion of cytokines. Uncovering the intricate mechanisms by which transcriptional factors regulate and influence the functions of Type 1 ILCs, particularly within the context of liver tumors, presents a significant opportunity to amplify the effectiveness of immunotherapies against liver malignancies. Using Ncr1-drived conditional knockout mouse model, our study reveals the regulatory role of Prdm1 in shaping the composition and maturation of liver Type 1 ILCs. Notably, Prdm1 regulates the ratio between NK cells and ILC1s, promoting a shift in the balance towards the direction of NK cells. Although Prdm1 did not affect the killing function of cNK cells in an in vivo cytotoxicity model, a significant increase in cancer metastasis was observed in Prdm1 knockout mice. IFN-γ, granzyme B, and perforin secretion decreased significantly in Prdm1 deficient Type 1 ILCs. scRNA sequencing data also provided evidence that Prdm1 sustains functional subsets of liver type 1 ILCs and facilitates communications between Type 1 ILCs and macrophages. The present study unveiled a novel regulatory mechanism of Prdm1 in liver Type 1 ILCs, showing promising potential for developing innovative immune therapy strategies against liver cancer.
Graphical abstract
Introduction
Liver Type 1 ILCs consist of cNK cells and ILC1s (1), with distinct developmental trajectories and effect molecules (2). Both NK cells and ILC1s play indispensable roles in combatting viral infections (3), maintaining local immune homeostasis (4), eradicating malignant transformed cells (5), and fostering cross-talk with adaptive immunity (6). NK cells demonstrate potent cellular cytotoxicity, facilitating direct elimination of target cells (7). On the contrary, a defining attribute of ILC1s is their predominant cytokine-mediated functions, with limited cellular killing capacity (8). In a state of homeostasis, liver Type 1 ILCs (CD45+CD3-NK1.1+NKp46+) can be discriminated into cNK cells and ILC1s by the differential expression of CD49a and CD49b (1): liver cNK cells are marked by the expression of CD49b, while liver ILC1s exhibit a distinctive positivity for CD49a. Tumor Necrosis Factor Related Apoptosis Inducing Ligand (TRAIL) is also expressed on liver ILC1s, but not on liver NK cells (9, 10). Transcriptional factors (TFs) such as T-bet, Nfil3, PLZF, and ID2 are also required for Type 1 ILCs development or generation of their progenitors (11–14). Eomes is considered to be necessary for NK cell maturation, but not for ILC1s (15). On the contrary, deficiency of Hobit results in the depletion of ILC1s, while only leaving little impact on liver NK cells (16). These studies suggest that TFs may, or usually, play different roles in NK cells and ILC1s.
During the development of liver cancers, the functions of immune cells are often inhibited, resulting in the formation of an immunosuppressive tumor microenvironment, and sometimes even systemic immune suppression (17). Transforming a cold tumor into a hot tumor with a stronger immune response is one of the goals of many cancer immunotherapies (18). Research targeting the immune system is showing increasing clinical promise in liver cancer treatment (19). Our recent study also found that Toll-like receptor agonists can significantly enhance the anti-tumor effect of Sorafenib by reconstructing the tumor immune microenvironment and reshaping the vascular system (20).
The anti-tumor activity of NK cells in the liver is relatively well-established, whereas the relationship between ILC1 and tumors is still a topic of controversy. Clinical research findings have shown that liver tumors with a higher infiltration of NK cells are associated with a better prognosis (21). Decreased NK cell activity is closely correlated with the malignancy of liver tumors and represents a significant risk factor for recurrence (22). The tumor microenvironment orchestrates the transformation of NK cells into ILC1s in a TGF-β-dependent manner, resulting in their diminished capacity to control tumor growth and metastasis. This process ultimately promotes tumor immunoevasion (23). However, research has also revealed that, distinct from directly inhibiting tumor growth, the primary function of ILC1 is to suppress the seeding of metastatic tumor cells in liver tissue (5). Comprehensive research in this field is essential to harness the precision of Type 1 ILCs targeted immunotherapy for liver cancers.
The transcription factor network governs the function of Type 1 ILCs and the balance between NK and ILC1. Our research team, along with others, has observed that Smad4 promotes the shift in balance from NK cells towards ILC1 in a TGF-beta-independent pathway. This study also revealed that Smad4 positively regulates the expression of another transcription factor, Prdm1(24, 25). PR domain 1 (Prdm1/Blimp1) plays a crucial role in the differentiation of B cells into plasma cells and the homeostasis of T cells (26–28). The function and regulatory network of Prdm1 in NK cells is distinct from B cells and T cells, for its expression is independent of B-cell lymphoma 6 (Bcl-6) and Interferon Regulatory actor 4 (IRF4) but relies on T-bet. In the study that identified Prdm1 as an essential transcriptional factor for NK cell maturation (29), no significant differences were observed in IFN-γ production or cytotoxicity between Prdm1-deficient and wild-type NK cells. Although Prdm1 expression is dependent on IL-15 in immature NK cells and can be further upregulated by IL-12 or IL-21, it plays a role in downregulating the expression of certain cytokine receptors, such as CD25, consequently diminishing the responsiveness of NK cells to IL-2 (30). Notably, some studies have even suggested that Prdm1 suppresses the secretion of IFN-γ by NK cells (31). These findings imply that, although Prdm1 promotes NK cell development, it may function as a negative regulator of their activity. However, direct evidence supporting the impact of Prdm1 on NK cell anti-tumor capabilities is still lacking. Furthermore, there has been no investigation into its influence on the homeostasis of NK cells and ILC1s in the liver, nor has there been an exploration of the underlying mechanisms.
In the current study, we found that the deletion of Prdm1 in Ncr1+ cells resulted in an imbalance in the homeostasis of liver Type 1 ILCs, with a shift towards cNK cells. The data also support the essential role of Prdm1 in the cancer surveillance mediated by NK cells. While Prdm1 positively regulates genes associated with cellular cytotoxicity, it concurrently exerts inhibitory control over certain positive regulators of NK cell development and functionality, such as JunB. Using single-cell RNA sequencing, we have identified a subset of NK cells within the liver characterized by elevated JunB expression. These cells exhibit decreased expression of genes associated with cellular cytotoxicity, and their abundance significantly increases following Prdm1 knockout.
Results
Prdm1 promotes Type 1 ILCs homeostasis and terminal maturation
Examination of 363 liver hepatocellular carcinoma (LIHC) patient samples from The Cancer Genome Atlas (TCGA) revealed a positive correlation between the expression of NK cell-associated genes (NCR1, NCR3, KLRB1, CD160, and PRF1) (32) and PRDM1 expression (Figure 1A). Patients with top and bottom quartiles of NK-PRDM1 signature expression were chosen for survival analysis (Figure 1B). Notably, patients with the NK-PRDM1hi signature had better overall survival compared to the these with NK-PRDM1lo signature (Figure 1C). Similar results were also found in skin cutaneous melanoma (SKCM, n=454) and lung adenocarcinoma (LUAD, n=497) patients (Supplemental Figure 1, A-F). These data suggested that PRDM1 in NK cells might be essential for immune surveillance in some solid tumors, including liver cancer. These findings prompted us to investigate the impact and mechanism of PRDM1 in NK cells and ILC1 within the context of liver cancer.
To achieve this, we developed a conditional knockout mouse model to specifically deplete Prdm1 in Type 1 ILCs. Ncr1-Cre mice were crossed with Prdm1fl/fl mice to generate Ncr1-cre Prdm1fl/fl mice, which specifically knockout exons 6-8 of Prdm1 (33) in NKp46 positive cells (Figure 1D). The mice carrying Ncr1Cre/+Prdm1fl/fl were referred to as Prdm1ΔNcr1 mice, and the mice carrying Ncr1+/+Prdm1fl/flwere referred to as Prdm1+/+ mice or wild-type (WT) mice. To further validate the deletion of Prdm1 in NKp46+ cells of Prdm1ΔNcr1 mice, CD3-NK1.1+NKp46+ cells were sorted and the expression of Prdm1 was quantified by real-time RT-PCR. The expressio n of Prdm1 was almost undetectable in Prdm1ΔNcr1 mice (Figure 1E), indicating successful knockout of Prdm1 in NKp46+ cells.
Proportion and absolute number of cNK cells in blood, bone marrow, lung, liver, spleen, and lymph nodes were analyzed by flow cytometry. Compared with Prdm1+/+mice, the percentage of cNK cells (CD3-NK1.1+NKp46+) among lymphocytes was decreased in all of these tissues except bone marrow and lymph nodes (Figure 1F; Supplemental Figure 2A). However, no significant difference was observed in the percentage of cNK cells among bone marrow-derived lymphocytes between Prdm1ΔNcr1 and Prdm1+/+mice. The absolute number of cNK cells in blood, lung, liver, and spleen also decreased in Prdm1ΔNcr1 mice (Figure 1F; Supplemental Figure 2A). Only a slight decrease in the number of cNK cells was observed in the lymph nodes of Prdm1ΔNcr1 mice, which did not reach statistical significance either (Supplemental Figure 2A). In contrast, the absolute number of cNK cells in Prdm1fl/fl mice bone marrow is moderately higher than Prdm1ΔNcr1 mice (Figure 1F).
NK cell terminal maturation can be divided into four stages according to the expression of CD11b and CD27 (34). Stage IV (CD11b+CD27-) NK cells show higher level of effector molecules than any other stages and are considered as the most mature NK cells (35). Based on the expression of CD11b and CD27, the maturation of cNK cells from blood, bone marrow, lung, liver, spleen, and lymph nodes were assessed. Compared with Prdm1+/+ mice, the proportion of most mature CD11b+CD27- NK cells were significantly decreased in all of the analyzed tissues in Prdm1ΔNcr1mice, without any exceptions (Figure 1G; Supplemental Figure 2B). Remarkably, an increase in immature CD11b-CD27+ NK cells was also observed in these tissues. Killer cell lectin-like receptor subfamily G member 1 (KLRG1) is a lectin-like receptor which was considered as another marker of NK cell maturation (36). In both Prdm1+/+ and Prdm1ΔNcr1 mice, cNK cells from the liver and lung had the highest expression of KLRG1, following by blood and spleen (Supplemental Figure 2, C and D). The lowest KLRG1 expression was observed in cNK cells derived from lymph nodes and bone marrow, indicating the presence of the most immature cNK cells in these tissues. Consistent with CD11b/CD27 based maturation analysis, a significant loss of KLRG1+ NK cells in Prdm1ΔNcr1 mice was observed compared to Prdm1+/+ mice (Supplemental Figure 2, C and D). Together, these data showed that Prdm1 is required for the terminal maturation of cNK cells among various tissues.
Prdm1 is required for Type 1 ILCs to control tumor metastasis
Two subpopulations of liver Type 1 ILCs (gated by CD45+CD3-NK1.1+NKp46+) were further analyzed based on the expression of CD49a and CD49b (Figure 2A). Compared with Prdm1+/+ mice, the Prdm1ΔNcr1 mice exhibited an increased percentage of cNK cells (CD49a-CD49b+) and reduced proportion of ILC1s (CD49a+CD49b-) (Figure 2A). Of note, the absolute number of both cNK cells and ILC1s were decreased in Prdm1ΔNcr1 mice, with a more robustly reduction in ILC1s (Figure 2A). This underscored the crucial role of Prdm1 in maintaining the homeostasis of both liver cNK cells and ILC1s.
The disruption of NK cell homeostasis and maturation due to Prdm1 loss motivated us to further explore whether deficiency of Prdm1 impaired NK cell cytotoxicity. A B2M-deficient cell-based in vivo cytotoxicity assay was used to evaluate the effect of Prdm1 on the cytotoxicity of NK cells (37, 38). B2M-deficient cells do not have detectable Major Histocompatibility Complex I (MHC-Ⅰ) on the cell surface, making them the target of NK cells (39). Health NK cells will reject B2M-deficient donor cells efficiently and the elimination was used to quantify the cytotoxicity of NK cells. Although significant impaired homeostasis and maturation of NK cells were observed in Prdm1ΔNcr1mice, no significant difference in the in vivo cytotoxicity assay were observed between Prdm1+/+ and Prdm1ΔNcr1 mice (Figure 2, B and C).
While our B2M-deficient cell-based assay did not reveal a significant impact of Prdm1 on NK cell cytotoxicity, both our results and those of other researchers have demonstrated Prdm1’s role in positively regulating NK cell development and homeostasis. Besides direct cytotoxicity, other factors influence the anti-tumor capability of NK cells, such as the ability to counteract tumor-induced immune suppression and exhaustion, the secretion of cytokines to activate other immune cells for tumor elimination. Moreover, the cytotoxicity assay, due to its relatively short duration, might not fully represent the anti-tumor activity of NK cells when continuously exposed to immune inhibitory signals in the tumor microenvironment. Therefore, we initiated an in vivo tumor model to further investigate the impact of Prdm1 on NK cell anti-tumor capabilities. B16F10 is a melanoma cell line with low expression of MHC-Ⅰ, which was susceptible to NK cell killing and usually used to evaluate NK cell anti-tumor capacity (40, 41). The B16F10 cells were intravenously (for lung metastasis) or intrasplenic (for liver metastasis) administrated in the mice. The melanoma nodes were quantified three (intravenous injection) or two (intrasplenic injection) weeks after tumor inoculation. Compared with Prdm1+/+ mice, deficiency of Prdm1 resulted in more metastasis nodules in both lung (∼ 2-fold) and liver (∼ 4-fold) (Figure 2, D and F). Histological analysis further confirmed the increased frequency of metastasis tumor foci in Prdm1ΔNcr1mice (Figure 2, E and G).
In agreement with our in vivo data, we also observed decreased IFN-γ secretion in Prdm1ΔNcr1 mice-derived splenic cNK cells, liver cNK cells, and liver ILC1s when stimulated by IL-18 alone or IL-12/IL-18 (Figure 2H; Supplemental Figure3), which indicated that Prdm1 is required for full activation of cNK cells and ILC1s in the context of IFN-γ production. These data implying that although similar cellular killing capacity was observed between Prdm1+/+ and Prdm1ΔNcr1mice, Prdm1 is indispensable for NK cell mediated tumor surveillance.
Bulk RNA-seq depicts Prdm1-mediated functions in peripheral cNK cells
Bulk RNA sequencing of splenic cNK cells (CD3-NK1.1+NKp46+) was conducted to uncover the molecular mechanisms by which Prdm1 regulates NK cell anti-tumor immunity (Figure 3A). Differentially expressed genes (DEGs) between Prdm1+/+ and Prdm1ΔNcr1 mice were determined using a criterion of log2 (fold change) > 0.5 and P < 0.05. 445 DEGs were identified out of 17434 protein-coding genes, which consisted of 223 upregulated genes and 222 downregulated genes (Figure 3B).
Gene Ontology (GO) analysis revealed the enrichment of glucuronate metabolism and lymphocyte differentiation in upregulated genes in Prdm1ΔNcr1 mice derived NK cells (Figure 3C), both of which were associated with cellular growth and development. In contrast, leukocyte mediate cytotoxicity, immune receptor activity, and integrin binding was enriched in the genes which decreased their expression level in in Prdm1ΔNcr1 mice (Figure 3C).
Gene Set Enrichment Analysis (GSEA) showed enriched NF-kappa B signaling pathway and negative regulation of IL-6 production pathway in Prdm1-deficient cNK cells (Figure 3, D and E), suggesting the potential targets by Prdm1 to regulate NK cell function. Increased expression of multiple TFs such as Junb, Batf3, Nfkb1, Tcf7, and Nr4a2 was observed in Prdm1 knockout cNK cells, suggesting they might be suppressed by Prdm1 (Figure 3F). Downregulation of granzyme B (Gzmb), Perfroin (Prf1) were observed in Prdm1 deficient NK cells (Figure 3F), implied decreased anti-tumor ability, which was consistent with increased melanoma metastasis in Prdm1ΔNcr1mice (Figure 2D and F). Cxcr6, and Cx3cr1 were also decreased in the NK cells derived from Prdm1ΔNcr1 mice (Figure 3F), which was considered to play an important role in promoting the egress of NK cells from bone marrow. This might be the reason for the quantity of NK cells increased in bone marrow while decreasing in other peripheral tissues (Figure 1F). As a result of the reduced expression levels of Cxcr6 and Cx3cr1, NK cells were unable to egress from the bone marrow and accumulated therein. Consistent with decreased production of IFN-γ after stimulated by IL-12/IL-18 (Figure 2H), decreased expression of Il18rap and Il12rb2 were observed in Prdm1ΔNcr1 cNK cells (Figure 3F), implying impaired response to cytokine stimulation.
To confirm the result of RNA-sequencing, the expression of granzyme B and perforin were also analyzed by flow cytometry. Lower granzyme B and perforin production was observed in Prdm1-deficient splenic cNK cells, liver cNK cells and ILC1s (Figure 3, G and I; Supplemental Figure 4, A and B). Notably, granzyme B decreased among all of the maturation stages of cNK cells. Probably because of relatively low level of perforin in CD11b-CD27+ and CD11b+CD27+ cNK cells, there is not statistical difference between Prdm1+/+ and Prdm1ΔNcr1 cNK cells. However, significantly decrease in Perforin was observed in Prdm1ΔNcr1mice derived CD11b+CD27- NK cells, which have the highest level of Perforin. (Figure 3, H and J; Supplemental Figure 4C). These data implied that Prdm1 directly regulates NK cells to exert cytotoxicity, not only due to the less mature phenotype in Prdm1-deficient NK cells.
scRNA-seq reveals unique properties of two clusters from liver type I ILCs following Prdm1 knockout
To further investigate the effect of Prdm1 in liver cNK cells and ILC1s, as well as the changes in the haptic immune microenvironment caused by the deficiency of Prdm1 in Type 1 ILCs, single-cell RNA sequencing (scRNA-seq) was performed for CD45+ cells sorted from the liver (Figure 3A; Supplemental Figure 5A). Initial quality control revealed high-quality of cell purity, library assembly, and sequencing (Supplemental Figure 5B). 10978 cells passed the quality criteria and were selected for further analysis (6161 from Prdm1+/+ mice, 4,817 from Prdm1ΔNcr1 mice). Unsupervised clustering of all sequenced cells based on transcript signatures identified twelve distinct clusters, including B cells, Macrophages, CD4+ T cells, CD8+ T cells, dendritic cells (DCs), NKT cells, cNK cells, ILC1s, Neutrophils, Monocytes, Basophils, and a small number of undefined cells (Figure 4A). A higher percentage of Macrophages, CD8+ T cells, Monocytes, DCs, and a lower percentage of NK cells, ILC1s, CD4+ T cells, and NKT cells was observed in Prdm1ΔNcr1 mice (Supplemental Figure 5C). Liver cNK cells and ILC1s were identified based on the expression of surface markers and TFs (Supplemental Figure 5D). For example, liver cNK cells expressed high levels of Itga2 (CD49b) and Eomes, while ILC1s had high levels expression of Itga1 (CD49a) and Tnfsf10 (Supplemental Figure 5E). In liver Type 1 ILCs, increased percentage of cNK cells and decreased percentage of ILC1s were observed in Prdm1ΔNcr1 mice (Figure 4, B and C), which was in line with the flow cytometry data (Figure 2A).
To better understand the specific function of Prdm1 played in liver cNK cells and ILC1s, the two subpopulations of liver Type 1 ILCs were further analyzed separately using unsupervised clustering. Liver cNK cells and ILC1s were grouped into three clusters respectively and visualized by Uniform Manifold Approximation and Projection (UMAP) (Figure 4, D-I). Based on the cluster specific gene expression signature (Supplemental Figure 6, A and B), the subpopulation of liver cNK cells were referred as “Prf1hi”, “Junbhi ”, and “Cxcr3hi” cNK cells (Figure 4, D-F).
The Prf1hi cNK cell cluster was defined by high expression of cytolysis-related genes, including Ncr1, Gzma, Gzmb, Prf1, and Fgl2 (Figure 4J; Supplemental Figure 6C), indicating the strong target-killing ability of this cluster. GO analysis further revealed the enrichment signatures of cytolysis, response to virus, and lymphocyte mediated immunity in the genes upregulated in Prf1hi cNK cell cluster, further confirming the cytotoxic effects of this cluster (Figure 4K). While this cluster is present in both Prdm1ΔNcr1 and Prdm1+/+mice, there is a significant reduction in Prdm1ΔNcr1 mice (Figure 4F). Additionally, both Gzma and Gzmb expression were downregulated in Prdm1ΔNcr1mice derived cNK cells and ILC1s compared to those from Prdm1+/+mice (Figure 5A). These data underscore the crucial role of Prdm1 in maintaining NK cells with immune effector functions.
The Junbhi liver cNK cell cluster distinguished themselves by higher expression of Junb compared to other clusters (Figure 4J). The expression of Junb was also upregulated in Prdm1ΔNcr1derived cNK cells and ILC1s (Figure 5A). The predominant majority (92.98%) of Junbhi liver cNK cells are derived from Prdm1ΔNcr1mice, with less than ten percent (7.02%) originating from Prdm1+/+mice (Figure 4F). Many signal transduction elements, gene expression regulator, and transcriptional factors, such as Nfkbia, Tnfaip3, Nr4a1/2/3, Batf3, Fos, Fosb, Tcf7, and Kit were upregulated in the Junbhi liver cNK cells (Figure 4J; Supplemental Figure 6C). The expression of cytotoxicity related genes, such as Gzmb and Prf1, in Junbhi cluster was also lower than other cNK cells (Figure 4J). GO analysis showed that the genes upregulated in Junbhi liver cNK cells enriched in cell differentiation, cell activation, and transcriptional regulation (Figure 4K). GSEA indicates that the NF-kappa B, IL-17, MAPK, and TNF signaling pathways were upregulated in this clusters (Figure 5, E-H). GSEA also showed that mitochondrial related pathways, such as mitochondrial protein, oxidative phosphorylation, and respiratory electron transport chain were suppressed in Junbhi cNK cell cluster (Supplemental Figure 6, D-F). Increased proportion of Junbhi cluster in Prdm1 deficient cNK cells suggested impaired anti-tumor activity, which was consistent with more melanoma metastasis in Prdm1ΔNcr1 mice and lower expression of cytotoxicity-related genes in splenic cNK cells based on bulk RNA-sequencing.
The Cxcr3hi cNK cell cluster was characterized with high expression of Cxcr3, Ccr2, and some genes encoding ribosomal subunits such as Rps7 (Supplemental Figure 6C). The enrichment of chemokine receptors in the genes upregulated in the Cxcr3hi cluster implying a greater likelihood of them being tissue-resident compared with other cNK cell clusters (Figure 4K). The significant enrichment of ribosomal subunits and cytoplasmic translation in Cxcr3hi cluster (Figure 4K) revealed their distinct and active metabolic profile and the capability to mount immune responses. The Cxcr3hi cNK cell cluster had a remarkably decreased proportion in Prdm1ΔNcr1 mice (Figure 4F), which emphasized the critical role of Prdm1 in maintaining this cluster of liver cNK cells. Consisting with the flow cytometry result that showed an increase in the number of NK cells in the bone marrow and a decrease in NK cells in peripheral tissues, bulk RNA sequencing data also found that the expression of Cx3cr1 and Cxcr6 decreased in Prdm1ΔNcr1 cNK cells. These findings supported the hypothesis that Prdm1, through regulating chemokine receptor expression levels, influenced the distribution of NK cells in the bone marrow and peripheral tissues, particularly within the liver tissue.
Three clusters of ILC1s were identified from liver ILC1s (Figure 4, G-I), comprising “Il7rhi”, “Klrahi”, and “Gzmahi” ILC1s. The first two clusters of ILC1s were characterized by higher expression of Il7r (CD127) and Klra5 separately, while the Gzmahi ILC1 cluster was identified by elevated expression of both Gzma and Gzmb (Supplemental Figure 7, A-C). The Il7rhi ILC1s cluster was distinguishable from other ILC1 clusters by its high expression of Il7r, IL18RA (Il18r1), and IFN-γ (Ifng) (Figure 5B; Supplemental Figure 7C). The high expression of Il18r1 and Ifng in Il7rhi ILC1s indicated this cluster of cells was highly responsive to IL-18. Module scores, calculated based on the expression of feature genes within the Junbhi cNK cell cluster, revealed a comparable Junbhi signature expression pattern within the Il7rhi ILC1 cluster (Figure 5C). GO analysis and GSEA showed that IL-17, NF-kappa B, TNF, MAPK signaling pathway and T cell differentiation were activated in the Il7rhi ILC1 cluster (Figure 5D, I and J; Supplemental Figure 7D-F). Considering the close relationship between IL-17 mediated immunity response and ILC3 (42), it is plausible that the observed heterogeneity within this cluster may be attributed, at least in part, to potential plasticity between ILC1 and ILC3 subsets. These two clusters appear reminiscent of an intermediate stage in the conversion process from cNK cells to ILC1s, as previously described (23, 43). There was no significant difference in the proportion of this cluster between WT and in Prdm1ΔNcr1mice.
The second liver ILC1 cluster, characterized by high expression of Ly49E (Klra5) and Ly49G (Klra7), was designated as the Klrahi ILC1 cluster (Figure 5B). Notably, there was an elevated proportion of Klrahi ILC1s in Prdm1ΔNcr1 ILC1s (39.7%) compared to Prdm1+/+ ILC1s (28.1%). Liver Ly49E+ ILC1s have been identified as possessing greater cytotoxic potential and a more robust viral response compared to liver Ly49E- ILC1s (44). The Klrahi cluster exhibited notably high expression of Ccl5 (Supplemental Figure 7C). Previous research has underscored the pivotal role of CCL5, produced by both cNK cells and ILC1s, in facilitating the accumulation of DCs within the tumor microenvironment, thereby impeding tumor immune evasion, as highlighted in studies (32, 45). The expression of Ccl5 was reduced in the Klrahi cluster of Prdm1ΔNcr1 ILC1s compared to Prdm1+/+ ILC1s (Supplemental Figure 7C). This decrease in Ccl5 expression could potentially have a detrimental impact on the ability of Klrahi ILC1s to develop a connection between innate and adaptive immune responses.
The Gzmahi ILC1 cluster was identified according to the high expression of Gzma, Gzmb, and Fgl2 (Supplemental Figure 7C), compared to other ILC1 clusters. GO analysis also revealed the enrichment in cytolysis and stimulus-response capacity (Figure 5D) of the Gzmahi ILC1 cluster. Consistent with the Prf1hi cNK cell cluster, the proportion of the Gzmahi cluster among liver ILC1s exhibited a considerable reduction in Prdm1ΔNcr1 mice compared to Prdm1+/+ mice, and the expression of Gzma also downregulated in Prdm1ΔNcr1mice (Figure 5B). Previous reports showed that GzmA+ ILC1 constituted the main population of liver ILC1s at birth, with the potential target-killing ability (46, 47).
Prdm1 facilitates the intercellular communication between liver Type 1 ILCs and macrophages
The reciprocal crosstalk between Type 1 ILCs and macrophages plays a critical role in maintaining liver immune homeostasis and anti-cancer immune surveillance. Based on the scRNA-seq data, four clusters were generated from liver macrophages according to the transcriptional profile. They were labeled as “Ly6c2hi”; “Cxcl2hi”; “Ear2hi”; and “C1qhi” macrophages (Mac) (Figure 6A, Supplemental Figure 8, A-D).
High-resolution interactions among liver cNK cells, ILC1s, and macrophages were established and compared between Prdm1+/+ and Prdm1ΔNcr1 mice using the CellChat program (48). Interactions between ILC1s and macrophages were higher than that between cNK cells and macrophages (Supplemental Figure 9, A and B). There is a significant decrease in the interaction number between liver Type 1 ILCs (including cNK cells and ILC1s) and macrophages (∼2 fold) in Prdm1ΔNcr1 mice (Figure 6, B and C). A reduction in the interaction of ligand-receptor, such as Mif-CD74, Cxcl16-Cxcr6, and Cxcl10-Cxcr3 was observed in Prdm1ΔNcr1 mice compared to Prdm1+/+ mice (Supplemental Figure 10). Liver cNK cells and ILC1s contributed equally to macrophages through most pathways, while CXC chemokine ligand (CXCL), Thy-1 cell surface antigen (THY1), and C-type lectin (CLEC) pathways participated more in ILC1-Mac interaction than cNK-Macrophage interaction (Supplemental Figure 9, C and D). Compared to Prdm1+/+ mice, the information flow of CXCL and macrophage migration inhibitory factor (MIF) pathways significantly decreased in Prdm1ΔNcr1mice (Figure 6, D and E). These pathways play a crucial role in facilitating macrophage migration. The pathway Thrombospondin (THBS) was upregulated in Prdm1ΔNcr1, which was consistent with the higher expression of Thbs1 in Prdm1ΔNcr1macrophages (Supplemental Figure 8C). The CXCL signaling was sent from Cxcl2hi macrophage and C1qhi macrophage, targeting all ILC1 clusters and Cxcr3hi cNK cell clusters (Figure 6F). Of note, although the population of Cxcl2hi macrophage primarily comprised cells from Prdm1ΔNcr1 mice, the interaction within the CXCL pathway between macrophages and Type 1 ILCs was only observed in Prdm1+/+ sample (Figure 6F). These changes could be linked to a decreased population of ILC1s and Cxcr3hi cNK cell cluster in Prdm1ΔNcr1 mice, implying that the homeostasis of Cxcl2hi macrophages required sufficient signals from cNK cells and ILC1s. The impaired CXCL-CXCR interactions might subsequently lead to reduced recruitment and activation of Type 1 ILCs and macrophages within the tumor microenvironment.
Prdm1 safeguards Type 1 ILCs from exhaustion-like phenotypes in the tumor microenvironment
The suppression of mitochondrial related pathways in Junbhi cNK cell cluster, along with a significant increase of this cNK cell cluster in Prdm1ΔNcr1 mice, encouraged us to explore mitochondrial function through flow cytometry. MitoTracker, MitoSOX, and Tetramethylrhodamine methyl ester (TMRM) were used to assess the mitochondrial mass, superoxide production, and mitochondrial membrane potential. A substantial decrease in mean fluorescent intensities (MFIs) of MitoTracker was observed in Prdm1ΔNcr1 splenic cNK cells, liver cNK cells, and liver ILC1s when compared to their Prdm1+/+ counterparts (Figure 6A). This observation aligns with the enrichment of downregulated genes from Prdm1 deficient sample in mitochondrial related pathway, as revealed by RNA sequencing data (Supplemental Figure 6, D-F). There was no significant difference in MitoSOX and TMRM between Prdm1ΔNcr1 and Prdm1+/+ mice (Figure 7, B and C), which suggested that the ATP synthesize capacity was minimally affected by Prdm1.
IFN-γ is a critical cytokine for NK cells mediated cancer surveillance(49, 50) and impaired production of IFN-γ was considered as a key hallmark of exhausted NK cells (51, 52). To evaluate the IFN-γ secreting capacity of liver cNK cells and ILC1s in tumor microenvironment, B16F10 tumor cells were inoculated to the liver via splenic injection and the IFN-γ levels in response to stimulation of IL-12 and/or IL-18 were assessed by flow cytometry (Supplemental Figure 11). Compared with Prdm1+/+, significant deceased of IFN-γ were observed in Prdm1ΔNcr1 liver cNK cells and ILC1s under the combinate stimulation of IL-12/IL-18 (Figure 7D), which was more remarkable in liver ILC1s. Similar trends were observed when IL-12 or IL-18 was used alone, although only liver ILC1s showed a significant decrease in response to IL-18 stimulation (Figure 7D). These findings were consistent with the heavy tumor burden observed in Prdm1ΔNcr1 mice.
Discussion
Using Ncr1-driven conditional knockout transgenic mice, which specifically delete Prdm1 in Type 1 ILCs, we not only validated Prdm1’s positive regulation of NK cell maturation, but also demonstrated its indispensable role in NK cell anti-tumor activity. The compromised mitochondrial function and reduced IFN-γ, granzyme B, and perforin production appear to be potential contributing factors to Prdm1-mediated cancer surveillance. Decreased expression of Cx3cr1 and Cxcr6 was observed in Prdm1ΔNcr1 splenic cNK cells (Figure 3F), both of which are essential for NK cells egressing from bone marrow (53, 54). The quantity of cNK cells increased exclusively in the bone marrow, with reductions observed in all other tissues (Figure 1F). This indicated that Prdm1 might regulate chemokine receptors to facilitate the egression of cNK cells from the bone marrow to peripheral tissues. In addition, higher expression of Cxcr6 compared to cNK cells is also a key factor for liver tissue residence of ILC1s (55). Decreased expression of Cxcr6 in Prdm1 deficient Type 1 ILCs may also contribute to the balance shift towards cNK cells.
scRNA sequencing analysis reveals that both liver cNK cells and ILC1s can be further divided into three subgroups based on their gene expression patterns. Within the cNK cells, the abundance of Junbhi cluster substantially increased in Prdm1ΔNcr1mice (75.8%). Notably, these cells account for less than 5% of WT cNK cells. High Junb expression stood out as a prominent feature of this subgroup. Junb is a crucial transcriptional factor for the cytotoxic function of CD8+ T cells and NK cells. However, excessive Junb expression has been found to promote T cell exhaustion (56). Our previous study showed that as NK cells mature, the expression level of Prdm1 increased while the expression level of Junb gradually decreased (57). Our data demonstrated that in NK cells, the expression level of Junb significantly increases upon the deletion of Prdm1, indicating that Junb expression is suppressed by Prdm1. As Junb expression decreases with NK cell maturation, and it is inhibited by the gradually increasing Prdm1 during maturation. This implies that constraining Junb expression is likely a fundamental prerequisite for NK cell maturation. However, the precise mechanism by which Prdm1 downregulates Junb in NK cells still needs further research. Furthermore, Junbhi NK cells exhibit lower expression levels of cytotoxic genes and reduced mitochondrial-related signaling pathways. Mitochondrial fragmentation has also been confirmed to be closely associated with NK cell exhaustion(58). While our current data is not sufficient to definitively classify these cells as exhausted NK cells, it supports that this particular group of NK cells demonstrates an exhausted phenotype. The significant increase in this cell population following Prdm1 knockout in NK cells may potentially be one of the reasons why Prdm1ΔNcr1mice lose their tumor-killing capacity.
Our scRNA-seq data revealed that Prdm1 plays distinct roles in regulating cNK cells and ILC1s despite being required for both lineages. Specifically, Prdm1 appears to be more involved in promoting the resistance against exhaustion in cNK cells, whereas in ILC1s, it may play a role in the plasticity between ILC1s and ILC3s. In both our previous study and a study by Colonna et al. (57, 59), it was demonstrated that Smad4, a transcriptional factor involved in TGF-β signal pathway, upregulated Prdm1 in NK cells and depletion of Smad4 resulted in a decreased ratio of NK cells to ILC1s in the liver. However, knocking out Prdm1 in Ncr1+ cells increased the ratio of NK cells to ILC1s in liver type 1 ILCs (Figure 2A). These findings suggest the possibility of a Smad4-independent pathway through which Prdm1 promotes the maintenance of ILC1s, or that Prdm1 plays a more significant role in maintaining ILC1s compared to its role in NK cells.
Chronic inflammation is a crucial factor in promoting tumorigenesis, and macrophages play a significant role in this process. Macrophages interact with both cNK cells and ILC1s. However, the TFs that regulate these interactions are poorly understood. Fortunately, recent advances in scRNA-seq technology and the CellChat software tool (48) have allowed us to gain a better understanding of the Prdm1 signaling pathway in Type 1 ILCs and its impact on macrophage function at the transcriptional level. It is worth noting that normal ILC1-macrophage interactions are more prevalent than those with NK cells (Supplemental Figure 9, A and B). Moreover, the finding of increased liver metastasis in Prdm1ΔNcr1 mice indicated the crucial role of Prdm1 in maintaining homeostasis between Type 1 ILCs and macrophages. Specifically, Prdm1 may be critical in preventing the overactivation of macrophages that can lead to cancer development.
Methods
Mice
Prdm1fl/fl mice were purchased from The Jackson Laboratory. Ncr1-iCre and B2m-/- mice were purchased from Shanghai Model Organisms Center, Inc.. Six- to twelve-week-old littermates were used for the experiment.
Experimental metastasis model
For lung metastasis model, 0.3 × 106 B16F10 cells were intravenous injected into mice. Three weeks later, mice were euthanized for analysis. For liver metastasis model, mice were inoculation with 0.5 × 106 B16F10 via intrasplenic injection. Three weeks later, mice were euthanized for analysis. Lung and liver from tumor-bearing mice were fixed in 10% formalin and embedded in paraffin. Sections were stained with H&E.
In vivo cytotoxicity assay
Donor splenocytes harvested from B2m deficient (B2m-/-) mice were labeled with 5 µM CFDA-SE. Donor splenocytes harvested from B2m-adequate (B2m+/+) mice were labeled with 5 µM eF670. Labeled splenocytes from two mouse strains were mixed in a 1:1 ratio, and 1 × 107 cells in total were injected i.v. into Prdm1+/+ and Prdm1ΔNcr1 mice. One day after administration, spleen and liver cells were isolated from recipient mice, and the population of labeled cells was analyzed by flow cytometry. Rejection % was quantified according to the following formula:
Flow cytometry
Flow cytometry and cell sorting were performed with a Cytoflex S/SRT (Beckman Coulter). Data were analyzed using FlowJo software. The following antibodies were used (all purchased from BioLegend unless otherwise indicated): CD45-PE-Cy7 (catalog 103114, clone 30-F11); CD3ε-PerCP-Cy5.5 (catalog 100327, clone 145-2C11); NK1.1-BV421 (catalog 108741, clone PK136); CD335-AF647 (catalog 560755, clone 29A1.4, BD Bioscience); CD49a-PE (catalog 562115, clone Ha31/8, BD Bioscience); CD49b-FITC (catalog 108906, clone DX5); CD27-BV510 (catalog 124229, clone LG.3A10); CD11b-AF700 (catalog 101222, clone M1/70); KLRG1-APC (catalog 561620, clone 2F1, BD Bioscience); CD49a-BV421 (catalog 740046, clone Ha31/8, BD Bioscience); IFN-γ-PE (catalog 505807, clone XMG1.2); Granzyme B (catalog 372207, clone QA16A02); Perforin (catalog 154305, clone S16009A); CD49b-APC-Cy7 (catalog 108919, clone DX5); IgG-PE (catalog 402203, clone 27-35); CD335-BV510 (catalog 137623, clone 29A1.4); TIGIT-PE (catalog 622205, clone A17200C); CD3ε-APC-Cy7 (catalog 100330, clone 145-2C11). For mitochondrial metabolic assay, fresh cells were incubated in 37℃ media for 30min with 100 nM MitoTracker Green (catalog M7514, Invitrogen), 100 nM TMRM (catalog T668; Invitrogen), and 10 uM MitoSOX Red (catalog M36008; Invitrogen), respectively. Surface-stained after washing with PBS and then detected by flow cytometry. For intracellular IFN-γ staining, Cells freshly obtained from liver and spleen were stimulated 12 hours with or without cytokine. GolgiStop (BD Biosciences) was added 4 hours before intracellular staining.
Bulk RNA sequencing
Total RNA from FACS sorted splenic NK cells of Prdm1+/+ and Prdm1ΔNcr1 mice was extracted by TRIzol reagent (Invitrogen), then reverse transcribed into cDNA. Library construction was prepared using Illumina mRNA Library kit, and sequencing was performed by the BGISEQ-500. Standard methods were used to analyze the RNA-seq data, including aligning the reads to the genome by HISAT2 (v2.1.0) (60), and gene expression values (Counts) were calculated using RSEM (v1.3.1) (61). DEGs were identified using DEseq2 (v1.4.5) (62) with a cutoff of log2(fold change) > 0.5 and P < 0.05. The “clusterProfiler” package (v4.4.4) (63) and gene sets from molecular signatures database (MSigDB) were used for GSEA and GO analysis. The heatmap was plotted using the “Pheatmap” package (v1.0.12).
Single-cell RNA sequencing
FACS-sorted liver CD45+ cells with more than 80% cell viability were used for library preparation. Each sample contained cells from three Prdm1+/+ or Prdm1ΔNcr1 mice. Gel Bead-in-Emulsions (GEMs) were generated using the 10X Genomics Chromium system, which combinates Master Mix, Single Cell 3’ v3.1 Gel Beads, and Partitioning Oil with single cells. GEMs were mixed with cell lysate and reverse transcription reagent to produce full-length cDNA. After incubation, the GEMs were broken, and recovered cDNA were amplified via PCR. Fragmentation, End repair, A-tailing, and Adaptor Ligation were performed to obtain final libraries, which contain P5 and P7 sequences. The 3’ library was sequenced on Novaseq 6000 with approximately 50k read pairs/cell sequencing depth. The “Seurat” R package (v4.2.0) (64) was used for data analysis. Initial quality control was performed to filter out the low-quality cells and cell doublets. Cells with 200-5500 expressed genes and no more than 10% mitochondrial genes were considered for high-quality. Doublets were filtered with the “scDblFinder” Package (v1.10.0) (65). After quality control, we totally recovered 6161 cells and 4817 cells from Prdm1+/+ and Prdm1ΔNcr1 mice, respectively. Principal component analysis (PCA) was used for cluster analysis. The first 15 PCs were picked for clustering and further visualized by UMAP. Clusters-specific marker was defined using the “FindAllMarkers” function, and clusters were manually annotated based on the top 30 or 15 markers. The “clusterProfiler” package and gene set from MSigDB were used for GSEA and GO analysis. “CellChat” package (v1.4.0) (48) was utilized to predict the cell-to-cell communication from scRNA-seq data.
TCGA datasets assay
The normalized gene expression and survival datasets of cancer patients collected in The Cancer Genome Atlas (TCGA) were downloaded from UCSC Xena (http://xena.ucsc.edu/) (66). The mean expression of NK cell-associated genes (NCR1, NCR3, PRF1, KLRB1, CD160) was defined as NK signature according to the previous study (32). NK cell-associated genes, together with PRDM1, constitute the NK-PRDM1 signature in this study. The mean expression of per genes was ordered from high-to-low and plotted by heatmap using the “Pheatmap” package. The overall survival of patients in the high and low quartiles expression of NK-PRDM1 signature was selected for analysis. Kaplan-Meier curves were plotted by GraphPad Prism.
Statistics
For experiment results, two-tailed t tests were used to measure the continuous and normally distributed between the two independent groups. Paired t-tests were used to determine the statistical significance between two paired groups. Log-rank tests were used to compare the overall survival distribution between the two groups of patients. A P value less than 0.05 was considered significant and data were presented as mean±SEM.
Study approval
All animal experiments were approved by The Tianjin University Animal Care and Use Committee (protocol 00000000202010100023). No human subjects were performed in this study.
Acknowledgements
This work was supported by The National Key Research and Development Program of China (2022YFF1202900), and The Zhejiang Provincial Natural Science Foundation of China (LY21H150002).
Declaration of interests
The authors have declared that no conflict of interest exists.
Supplemental Figure 1. High expression of PRDM1-NK signature predicts better overall survival of cancer patients.
(A and D) Correlation between the average expression of NK cell-associated genes(NCR1, NCR3, PRF1, CD160, KLRB1) and PRDM1 in skin cutaneous melanoma (SKCM, n=454) (A) and lung adenocarcinoma (D) (LUAD, n=497) patients from TCGA datasets. (B and E) Heatmap of the ordered, z-score normalized expression values for PRDM1 and NK cell-associated genes in skin cutaneous melanom (B) and lung cancer (E) patients. The top and bottom quartiles of all patient samples are indicated. (C and F) Prognostic value of the NK-PRDM1 signature for overall survival of skin cutaneous melanom (C) and lung cancer (F) patients comparing high and low quartiles. P, P-value; r, pearson correlation coefficient.
Supplemental Figure 2. Prdm1 plays an essential role in type I ILCs homeostasis and maturation.
(A) Representative flow cytometric plots (left) and quantification (right) of the proportion and absolute number of CD3-NKp46+ cells among lymphocytes in blood, spleen, and lymph nodes (n=8). (B) Representative flow cytometric plots (left) of the CD11b and CD27 expression within CD3-NK1.1+NKp46+ cells in blood, spleen, and lymph nodes (n=7). Right panel showed the percentage of distinct stages of NK cells. (C and D) Representative flow cytometry analyses (C) and quantification (D) of KLRG1 expression in CD3-NK1.1+NKp46+ cells in blood, bone marrow, lung, liver, spleen, and lymph nodes (n=5). Data are presented as the mean±SEM and were analyzed by 2-tailed, paired t-test. Differences were evaluated between littermates. Each circle and square on graphs represents an individual mouse; P, P-value; *, P<0.05; **, P<0.01, ns, not significant.
Supplemental Figure 3. Prdm1 affects the IFN-γ secretion ability of type I ILCs.
(A) Representative flow cytometric plot showing the frequency of IFN-γ+ cNK cells and ILC1s in spleen and liver between Prdm1+/+ and Prdm1ΔNcr1 mice (n=5). Liver cells and splenocytes were costimulated in the presence or absence of IL-12 and IL-18 for 12 hours. GolgiStop was added 4 hours before intracellular staining of IFN-γ. Data are presented as the mean±SEM and were analyzed by 2-tailed, paired t-test. Differences were evaluated between littermates. Each circle and square on graphs represents an individual mouse; P, P-value; *, P<0.05; **, P<0.01, ns, not significant.
Supplemental Figure 4. Prdm1 deficiency impairs production of Granzyme B and Perforin in type I ILC.
(A and B) Representative flow cytometric plot (left) and summary data (right) showing the relative mean fluorescence intensities (MFIs) of Granzyme B (A) and Perforin (B) in splenic NK cells from Prdm1+/+ and Prdm1ΔNcr1 mice (n=5). (C) Splenic NK cells were divided into different stages according to the expression of maturation markers CD11b and CD27. Representative flow cytometric plot showing the MFIs of Granzyme B and Perforin in evaluated between littermates. Data are presented as the mean±SEM and were analyzed by 2-tailed, paired t-test. Differences were evaluated between littermates. Each circle and square on graphs represents an individual mouse; P, P-value; r, pearson correlation coefficient; *, P<0.05; **, P<0.01, ns, not significant.
Supplemental Figure 5. scRNA-seq identified subsets of liver CD45+ cell from Prdm1+/+ and Prdm1ΔNcr1 mice.
(A) Liver CD45+ cells from Prdm1+/+ and Prdm1ΔNcr1 mice were FACS-sorted for scRNA-seq. (B) Quality control of the scRNA-seq data. Violin plot showing the nFeature_RNA, nCount_RNA, and percentage of mitochondrial genes of sequencing cells. nFeature_RNA represents the number of genes detected in each cell, and nCount_RNA represents the total number of molecules detected within a cell. (C) Population of twelve liver CD45+ cell clusters between Prdm1+/+ and Prdm1ΔNcr1 mice. (D) Dot plot showing the normalized expression of marker genes in liver cNK cells and ILC1s clusters between Prdm1+/+ and Prdm1ΔNcr1. The dot size represents the percentage of cells expressing selected genes, and color intensity represents the average expression. (E) Violin plots showing the expression of marker genes Itga2, Itga1, Eomes, and Tnfsf10 between liver cNK cells and ILC1s between Prdm1+/+ and Prdm1ΔNcr1 mice.
Supplemental Figure 6. Cluster-specific markers of liver cNK cell clusters.
(A) Heatmap showing the expression of top 30 upregulated DEGs by log fold change (computed using Wilcox test in the “FindAllMarkers” function of Seurat, avg_log2FC>0.25; P<0.05), across the three liver cNK cell clusters (Prf1hi cNK, JunbhicNK, Cxcr3hi cNK) within Prdm1+/+ and Prdm1ΔNcr1mice. (B) Feature plots showing the normalized expression of selected markers for liver cNK cell populations. (C) Violin plots showing the normalized expression of DEGs for each cNK cluster within Prdm1+/+and Prdm1ΔNcr1 mice. (D-F) GSEA of the enrichment of Mitochondrial protein (D), Oxidative phosphorylation (E), and Respiratory electron transport chain (F) in Junbhi cNK cluster compared with clusters of Prf1hi and Cxcr3hi cNK cells. NES, normalized enrichment score.
Supplemental Figure 7. Cluster-specific markers of liver ILC1 clusters.
(A) Heatmap showing the expression of top 30 upregulated DEGs by log fold change (computed using Wilcox test in the “FindAllMarkers” function of Seurat, avg_log2FC>0.25; P<0.05), across the three ILC1 clusters (Il7rhi ILC1, Klrahi ILC1, Gzmahi ILC1) within Prdm1+/+ and Prdm1ΔNcr1 mice. (B) Feature plots showing the normalized expression of selected markers for liver ILC1 populations. (C) Normalized expression of DEGs for each ILC1 cluster within Prdm1+/+ and Prdm1ΔNcr1 mice. (D) GSEA plots (left) depicting the enrichment of NF-kappa B (D), TNF (E), and MAPK (F) signaling pathways) in Il7rhi ILC1 cluster compared with clusters of Klrahi and Gzmahi ILC1s. Right panel showed dynamic relative expression of the given gene sets from cluster1 to cluster3 between Prdm1+/+ and Prdm1ΔNcr1. Dots represent the average expression of given gene set in each cell, which was calculated through the sum of normalized expression of each individual gene within the designated gene set in every single cell. NES, normalized enrichment score.
Supplemental Figure 8. Identification of four distinct macrophage clusters.
(A) Heatmap showing the top 15 upregulated DEGs across four macrophage clusters (computed using Wilcox test in the “FindAllMarkers” function of Seurat, avg_log2FC>0.25; P<0.05). (B) Percentage of macrophages from Prdm1+/+ and Prdm1ΔNcr1 mice among total cells (left) and within clusters (right). (C) Violin plots showing the expression of selected markers of each macrophage cluster within Prdm1+/+ and Prdm1ΔNcr1mice. (D) GO analysis of four distinct macrophage clusters. Dot size represents enriched gene number, and color intensity represents significance.
Supplemental Figure 9. Cell-Cell communication between cNK cells-macrophages and ILC1s-macrophages.
(A and B) Circle plots (A) and cumulative data (B) showing the interaction numbers between cNK cells-macrophages and ILC1s-macrophages. The thickness of the line indicates the number of enrich pairs, and the arrow reflects the direction of the interaction. (C) Heatmap of overall signaling pattern recognized from ligand-receptor pairs between cNK cells-macrophages and ILC1s-macrophages, which contained the sum of signaling from the sender and target cells. (D) Bar graphs showing the information flow in selected active signaling patterns between cNK cells-macrophages and ILC1s-macrophages. Relative information flow was calculated as the sum of the communication probability in given signaling patterns.
Supplemental Figure 10. Ligand-receptor interaction between type I ILCs and macrophages.
Bubble plots showing the significant ligand-receptor pairs between cNK cells, ILC1s, and macrophages within Prdm1+/+and Prdm1ΔNcr1 mice. The highlighted Mif-Cd74 and Cxcl-Cxcr signaling was significantly decreased in Prdm1ΔNcr1 mice. Dot size represents the P-value, and color intensity represents the communication probabilities. Empty space indicates a communication probability of zero. P-value were calculated by the one-sided permutation test.
Supplemental Figure 11: Prdm1 maintains the IFN-γ production of liver type 1 ILCs in tumor microenvironment.
Representative flow cytometric plot of the frequency of IFN-γ+ liver cNK cells and ILC1s from Prdm1+/+and Prdm1ΔNcr1 tumor-bearing mice at day 14 after inoculation with B16F10 melanoma cells via intrasplenic injection (n=5). Data are presented as the mean±SEM and were analyzed by 2-tailed, paired t-test. Differences were evaluated between littermates. Each circle and square on graphs represents an individual mouse; P, P-value; *, P<0.05; **, P<0.01, ns, not significant.
References
- 1.Liver-resident NK cells confer adaptive immunity in skin-contact inflammationJ Clin Invest 123:1444–56
- 2.Liver type 1 innate lymphoid cells develop locally via an interferon-gamma-dependent loopScience
- 3.ILC1 Confer Early Host Protection at Initial Sites of Viral InfectionCell 171:795–808
- 4.TRAIL+ NK cells control CD4+ T cell responses during chronic viral infection to limit autoimmunityImmunity 41:646–56
- 5.Conventional NK cells and tissue-resident ILC1s join forces to control liver metastasisProc Natl Acad Sci U S A 118
- 6.Group 1 ILCs regulate T cell-mediated liver immunopathology by controlling local IL-2 availabilityScience immunology 7
- 7.“Natural” killer cells in the mouse. I. Cytotoxic cells with specificity for mouse Moloney leukemia cells. Specificity and distribution according to genotypeEuropean journal of immunology 5:112–7
- 8.Type 1 Innate Lymphoid Cells Protect Mice from Acute Liver Injury via Interferon-gamma Secretion for Upregulating Bcl-xL Expression in HepatocytesImmunity 52:96–108
- 9.TRAIL identifies immature natural killer cells in newborn mice and adult mouse liverBlood 105:2082–9
- 10.Involvement of tumor necrosis factor-related apoptosis-inducing ligand in surveillance of tumor metastasis by liver natural killer cellsNat Med 7:94–100
- 11.T-bet regulates the terminal maturation and homeostasis of NK and Valpha14i NKT cellsImmunity 20:477–94
- 12.Development of peripheral lymphoid organs and natural killer cells depends on the helix-loop-helix inhibitor Id2Nature 397:702–6
- 13.The basic leucine zipper transcription factor E4BP4 is essential for natural killer cell developmentNature immunology 10:1118–24
- 14.A committed precursor to innate lymphoid cellsNature 508:397–401
- 15.The transcription factors T-bet and Eomes control key checkpoints of natural killer cell maturationImmunity 36:55–67
- 16.Hobit and Blimp1 instruct a universal transcriptional program of tissue residency in lymphocytesScience 352:459–63
- 17.The Liver-Immunity Nexus and Cancer ImmunotherapyClin Cancer Res 28:5–12
- 18.Approaches to treat immune hot, altered and cold tumours with combination immunotherapiesNature reviews Drug discovery 18:197–218
- 19.Highlighting novel targets in immunotherapy for liver cancerExpert review of gastroenterology & hepatology 16:1029–41
- 20.The Combination of R848 with Sorafenib Enhances Antitumor Effects by Reprogramming the Tumor Immune Microenvironment and Facilitating Vascular Normalization in Hepatocellular CarcinomaAdv Sci (Weinh
- 21.NK cell infiltration is associated with improved overall survival in solid cancers: A systematic review and meta-analysisTransl Oncol 14
- 22.Natural killer cell activity is a risk factor for the recurrence risk after curative treatment of hepatocellular carcinomaBMC Gastroenterol 21
- 23.Tumor immunoevasion by the conversion of effector NK cells into type 1 innate lymphoid cellsNat Immunol 18:1004–15
- 24.SMAD4 impedes the conversion of NK cells into ILC1-like cells by curtailing non-canonical TGF-beta signalingNature immunology 18:995–1003
- 25.SMAD4 promotes TGF-beta-independent NK cell homeostasis and maturation and antitumor immunityThe Journal of clinical investigation
- 26.Identification and characterization of a novel repressor of beta-interferon gene expressionGenes Dev 5:868–79
- 27.Transcriptional repressor Blimp-1 is essential for T cell homeostasis and self-toleranceNat Immunol 7:466–74
- 28.Transcriptional repressor Blimp-1 regulates T cell homeostasis and functionNat Immunol 7:457–65
- 29.A role for Blimp1 in the transcriptional network controlling natural killer cell maturationBlood 117:1869–79
- 30.PRDM1 decreases sensitivity of human NK cells to IL2-induced cell expansion by directly repressing CD25 (IL2RA)Journal of leukocyte biology 109:901–14
- 31.PRDM1/Blimp-1 controls effector cytokine production in human NK cellsJ Immunol 185:6058–67
- 32.NK Cells Stimulate Recruitment of cDC1 into the Tumor Microenvironment Promoting Cancer Immune ControlCell 172:1022–37
- 33.Blimp-1 Is Required for the Formation of Immunoglobulin Secreting Plasma Cells and Pre-Plasma Memory B CellsImmunity 19:607–20
- 34.CD27 dissects mature NK cells into two subsets with distinct responsiveness and migratory capacityJ Immunol 176:1517–24
- 35.Maturation of mouse NK cells is a 4-stage developmental programBlood 113:5488–96
- 36.NK cell maturation and peripheral homeostasis is associated with KLRG1 up-regulationJ Immunol 178:4764–70
- 37.Rejection of class I MHC-deficient haemopoietic cells by irradiated MHC-matched miceNature 349:329–31
- 38.Inducible down-regulation of MHC class I results in natural killer cell toleranceThe Journal of experimental medicine 216:99–116
- 39.Selective rejection of H-2-deficient lymphoma variants suggests alternative immune defence strategyNature 319:675–8
- 40.SHP-1-mediated inhibitory signals promote responsiveness and anti-tumour functions of natural killer cellsNat Commun 5
- 41.Generation of drug-resistant variants in metastatic B16 mouse melanoma cell linesCancer Res 47:2604–8
- 42.Innate lymphoid cells--a proposal for uniform nomenclatureNat Rev Immunol 13:145–9
- 43.Toxoplasma gondii infection drives conversion of NK cells into ILC1-like cellsElife 8
- 44.Ly49E separates liver ILC1s into embryo-derived and postnatal subsets with different functionsJ Exp Med 219
- 45.NK cells with tissue-resident traits shape response to immunotherapy by inducing adaptive antitumor immunitySci Transl Med 14
- 46.Granzyme A and CD160 expression delineates ILC1 with graded functions in the mouse liverEuropean journal of immunology 51:2568–75
- 47.Effector differentiation downstream of lineage commitment in ILC1s is driven by Hobit across tissuesNat Immunol 22:1256–67
- 48.Inference and analysis of cell-cell communication using CellChatNat Commun 12
- 49.IFN-gamma production by lung NK cells is critical for the natural resistance to pulmonary metastasis of B16 melanoma in miceJournal of leukocyte biology 90:777–85
- 50.IFN-gamma-dependent NK cell activation is essential to metastasis suppression by engineered SalmonellaNature communications 12
- 51.NK-cell exhaustion, B-cell exhaustion and T-cell exhaustion-the differences and similaritiesImmunology 166:155–68
- 52.Blockade of the checkpoint receptor TIGIT prevents NK cell exhaustion and elicits potent anti-tumor immunityNature Immunology 19:723–32
- 53.CX3CR1 expression defines 2 KLRG1+ mouse NK-cell subsets with distinct functional properties and positioning in the bone marrowBlood 117:4467–75
- 54.CXCR6 Expression Is Important for Retention and Circulation of ILC PrecursorsMediators Inflamm 368427
- 55.Human liver-resident CD56(bright)/CD16(neg) NK cells are retained within hepatic sinusoids via the engagement of CCR5 and CXCR6 pathwaysJ Autoimmun 66:40–50
- 56.c-Jun overexpression in CAR T cells induces exhaustion resistanceNature 576:293–300
- 57.SMAD4 promotes TGF-beta-independent NK cell homeostasis and maturation and antitumor immunityJ Clin Invest 128:5123–36
- 58.Mitochondrial fragmentation limits NK cell-based tumor immunosurveillanceNature immunology 20:1656–67
- 59.MicroRNA-362 negatively and positively regulates SMAD4 expression in TGF-beta/SMAD signaling to suppress cell migration and invasionInt J Med Sci 18:1798–809
- 60.HISAT: a fast spliced aligner with low memory requirementsNat Methods 12:357–60
- 61.RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genomeBMC Bioinformatics 12
- 62.Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2Genome Biol 15
- 63.clusterProfiler 4.0: A universal enrichment tool for interpreting omics dataInnovation (Camb) 2
- 64.Integrated analysis of multimodal single-cell data. Cell 184:3573–87
- 65.Doublet identification in single-cell sequencing data using scDblFinderF1000Res 10
- 66.Overcoming barriers to early disease interventionNat Biotechnol 38:669–73
Article and author information
Author information
Version history
- Sent for peer review:
- Preprint posted:
- Reviewed Preprint version 1:
- Reviewed Preprint version 2:
- Version of Record published:
Copyright
© 2024, He et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
- views
- 713
- downloads
- 82
- citations
- 0
Views, downloads and citations are aggregated across all versions of this paper published by eLife.