Proteomics of immune cells from liver tumors reveals immunotherapy targets

Abstract

Elucidating the mechanisms by which immune cells become dysfunctional in tumors is critical to developing next-generation immunotherapies. We profiled proteomes of cancer tissue as well as monocyte/macrophages, CD4⁺ and CD8⁺ T cells, and NK cells isolated from tumors, liver, and blood of 48 patients with hepatocellular carcinoma. We found that tumor macrophages induce the sphingosine-1-phospate-degrading enzyme SGPL1, which dampened their inflammatory phenotype and anti-tumor function in vivo. We further discovered that the signaling scaffold protein AFAP1L2, typically only found in activated NK cells, is also upregulated in chronically stimulated CD8⁺ T cells in tumors. Ablation of AFAP1L2 in CD8⁺ T cells increased their viability upon repeated stimulation and enhanced their anti-tumor activity synergistically with PD-L1 blockade in mouse models. Our data reveal new targets for immunotherapy and provide a resource on immune cell proteomes in liver cancer.

Keywords: CRISPR in mouse T cells; HCC; NK cells; T cells; cancer immunotherapy; liver cancer; macrophages; mass spectrometry-based proteomics; profiles of tumor-infiltrating immune cells; proteomes.

Graphical abstract

Figure 1

Schematic of workflow and patient cohort (A) Schematic overview of this study. (B) Overview of the patient cohort. Age and sex are indicated. Pie charts show the percentages of CD4⁺ T cells, CD8⁺ T cells, NK cells, and macrophages that were isolated from HCC tumors. In the center of the pie chart the total number of T cells, NK cells, and macrophages that were isolated from tumors is shown. Cases in which more than a million immune cells/g tumor were isolated are marked in red. See also Figure S1 and Table S1.

Figure 2

Characteristics of HCC and immune infiltrates (A) Volcano plot from differential abundance analysis (two-tailed Welch’s t test) between proteomes of non-tumorous (n = 33) and tumorous liver tissue (n = 32). Each dot represents a protein. Signature proteins of HCC tumors identified in a previous proteomics study are indicated in dark blue. See also Figure S2A and Table S2. (B) Heatmap showing the abundance (Z score) of the five most strongly up- and downregulated proteins in HCC tissue in individual patients. (C) Differential analysis of the arginine and proline metabolism between non-tumorous and tumorous tissue. Enzymes are color-coded according to the fold change as determined by the differential abundance analysis in (A). (D) FACS-purified CD4⁺ and CD8⁺ T cells, CD14⁺ monocytes/macrophages, as well as NK cells from blood, liver, and tumor tissue were analyzed by LC-MS. The plots show the number of identified proteins in each sample. (E) Boxplots showing the abundance of CD4, CD8, CD56, and MRC1 protein in different cell types isolated from blood, liver, and tumor tissue. Each dot represents a sample from a different patient. See also Figure S2C.

Figure 3

TAMs upregulate SGPL1, which impedes their inflammatory anti-tumor functions (A) Volcano plot from differential abundance analysis (two-tailed Welch’s t test) between proteomes of macrophages isolated from non-tumorous liver tissue (n = 33) and tumorous liver tissue (n = 38). Proteins that are most strongly upregulated in tumor macrophages are highlighted. (B) Heatmap showing the abundance of proteins that are significantly upregulated in TAMs across all samples from blood monocytes, liver, and tumor macrophages. (C) Estimates of protein copy numbers (mean of n = 38) are plotted against mRNA copy numbers of TAMs (mean of n = 3). The color code represents the fold change of selected proteins between liver macrophages and TAMs. The ratios on top of the graph indicate the protein-per-mRNA ratio. (D) Volcano plot from differential abundance analysis (two-tailed Welch’s t test) between proteomes of in vitro polarized M1 macrophages (LPS + IFN-γ) and M2 macrophages (IL-4 + IL-13) (n = 4). Proteins that are upregulated in TAMs are shown as black dots. THY-1 and AKR1B10 were not identified in in vitro polarized macrophages. See also Figure S3F. (E) Schematic of the procedure to generate knockout BMBMs. (F–I) BMDMs from Rosa-Cas9 mice were lentivirally transduced with a non-targeting control (NTC) sgRNA or with two different sgRNAs targeting Sgpl1. Then, BMDMs were left untreated or were stimulated with LPS + IFN-γ. After 16 h, CD86 (F), intracellular IL-6 (G), and intracellular IL-12 (H) were analyzed by flow cytometry and secreted IL-12 by ELISA (I). (J) 1 week after co-injecting BMDMs and MC38 cells (500,000 cells, each), tumor sizes were measured. (F–J) P values and the number of samples are shown in graphs and were determined using a two-tailed t test. Bars represent means ± SEM; two independent experiments.

Figure 4

PD-1⁺ T cells in tumors with dysfunctional properties upregulate AFAP1L2 (A) Volcano plot from differential abundance analysis (two-tailed Welch’s t test) between proteomes of NK cells isolated from liver and tumor tissue. (B) Boxplots showing abundance of selected proteins in NK cells isolated from blood, liver, and tumor tissue. Each dot represents a sample from a different patient. (C) Heatmap showing the percentage of PD-1 in CD4⁺ and CD8⁺ T cells isolated from blood, liver, and tumor tissue of HCC patients as determined by flow cytometry. Contour plots on the right side show an example of the PD-1 staining for patient HCC25. (D) Volcano plot from differential abundance analysis (two-tailed Welch’s t test) between proteomes of CD8⁺ T cells with a high (>60%, n = 10) and low (<20%, n = 15) percentage of PD-1. (E) Heatmap showing the relative abundance (Z score) of selected proteins in CD8⁺ T cell proteome samples of individual patients based on the differential analysis in (D). (F) Number of CD8⁺ T cells that were isolated and FACS-purified from HCC tumors compared in (D). Dots are color-coded according to the percentage of PD-1⁺ T cells as determined in (C). The p value is indicated in the graph and was determined using a two-tailed t test.

Figure 5

AFAP1L2 is upregulated in chronically activated T cells and blunts their survival (A) FACS-purified naive CD8⁺ T cells from four healthy donors were either analyzed directly by LC-MS or after increasing times of activation with antibodies to CD3 and CD28. Heatmap shows the abundance of proteins that were significantly upregulated in tumor-infiltrating CD8⁺ T cells with a high percentage of PD-1⁺ cells (see Figure 4). Each column represents a different donor. (B) Schematic of the activation conditions of naive CD8⁺ T cells and flow cytometry plots showing the expression of PD-1, TIM-3, CD39, and LAG-3. See also Figure S4B for quantifications. (C) Naive CD8⁺ T cells were activated as indicated in the schematic in (B) and then analyzed by LC-MS. Heatmap shows the abundance of indicated proteins, where each column represents a different donor. (D) Plot shows AFAP1L2 protein copy numbers at different time points following activation. n = 4 for resting T cells, and n = 3 for other time points. (E) AFAP1L2 protein copy numbers in CD8⁺ T cells following different activation schemes. n = 3. Two-tailed t test. (F) Same as in (E) but T cells were analyzed by RNA-seq, and AFAP1L2 mRNA counts are shown. (G) NTC and AFAP1L2-edited CD8⁺ T cells were labeled with CTV and then stimulated with ImmunoCult CD3/CD28/CD2 Activator (1:333). Proliferation (CTV intensity) was followed by flow cytometry over the time course of 5 days. n = 9 from three different donors. (H) NTC and AFAP1L2-edited CD8⁺ T cells were stimulated with plate-bound anti-CD3/CD28 antibodies for 8 and 12 days, and the number of surviving T cells was quantified by flow cytometry. n = 5 from five donors. Two-tailed t test.

Figure 6

T cells devoid of Afap1l2 have superior anti-tumor activity (A) Schematic showing the workflow to analyze the impact of gene editing on the anti-tumor activity of T cells. (B and C) 5 x 10⁵ B16.OVA cells were subcutaneously injected into C57BL/6 mice. 5 days later, 10⁶ control or Afap1l2-edited OT-I T cells were adoptively transferred into mice. 5 days later, tumors were disaggregated, and the number of OT-I T cells and endogenous T cells was quantified by flow cytometry. A representative plot is shown in (B), and quantifications from 10 mice are shown in (C). P value is shown on the graph and was determined using a two-tailed t test. Bars represent SEM throughout. (D and E) Same as in (B) and (C) but isolated OT-I T cells were re-stimulated with PMA/Ionomycin, and intracellular cytokines were analyzed by flow cytometry. A representative plot is shown in (D), and quantifications from six mice are shown in (E). (F) 5 x 10⁵ B16.OVA cells were subcutaneously injected into C57BL/6 mice. 6 days later, 10⁶ control or Afap1l2-edited OT-I T cells (sgRNA 1) were adoptively transferred into mice, and the size of tumors was followed. P value was determined by two-way ANOVA and is shown on the graph together with numbers of mice. (G) Same as in (F) but Afap1l2 was edited with a different sgRNA. (H and I) 5 x 10⁵ B16.OVA cells were subcutaneously injected into C57BL/6 mice. 6 days later, 10⁶ control or Afap1l2-edited OT-I T cells were adoptively transferred into mice. Mice received four subcutaneous injections of anti-PD-L1 antibodies indicated by arrows on the graph. Tumor growth curves are shown in (H) and survival curves in (I). P values in (H) were determined by ANOVA and in (I) by Mantel-Cox log rank test. (J) 5 x 10⁵ MC38.OVA cells were subcutaneously injected into Cd3e^−/− mice. 5 days later, 10⁶ control or Afap1l2-edited OT-I T cells were adoptively transferred into mice, and the size of tumors was followed. P value was determined by two-way ANOVA and is shown on the graph together with numbers of mice.