Research ArticleCANCER

Tumor immunological phenotype signature-based high-throughput screening for the discovery of combination immunotherapy compounds

See allHide authors and affiliations

Science Advances  22 Jan 2021:
Vol. 7, no. 4, eabd7851
DOI: 10.1126/sciadv.abd7851
  • Fig. 1 The expression pattern of TIP signature genes is associated with TIPs.

    (A) TIP signature that contains 15 immune genes. Red column, highly expressed genes. NK, natural killer. (B) Expression pattern of TIP signature genes within the hot and cold melanoma cohorts. (C) The TIP signature could distinguish hot and cold tumors with 89.7% accuracy in a melanoma dataset. (D) The receiver operating characteristic curve analysis was applied to evaluate the accuracy of the signature genes for tumor classification in a melanoma dataset (23). AUC, area under curve. (E) The correlation analysis of the expression level of six TIP signature genes with CD8+ T cell infiltration level in breast invasive carcinoma (METABRIC n = 1980); RSEM, RNA sequencing (RNA-seq) by expectation-maximization.

  • Fig. 2 The expression pattern of TIP signature genes predicts the response of immunotherapy.

    (A) Averaged expression level of TIP-associated gene in pretreatment tumors of ICB responders and nonresponders over four cohorts (2629), represented separately (left) and merged (right). Within each group, the scattered dots represent the averaged z score of hot tumor–related genes of each sample, and the thick line represents the median value; the bottom and top of the boxes are the 25th and 75th percentiles (interquartile range). Whiskers encompass 1.5 times the interquartile range. (B) Heatmap showing the expression of TIP-associated gene that discriminates between anti–PD-1 responders and nonresponders. (C) Kaplan-Meier plots of overall survival (OS) for 42 patients with melanoma who accepted ICB therapy (28) and were grouped with hot tumor–related genes highly or lowly expressed. (D and E) AUC for signatures in predicting ICB response in seven datasets (27, 28, 3135), represented merged (D) and separated (E). AUC in (D) is averaged AUC among seven datasets; the performance of a random predictor (AUC, 0.5) is represented by the dashed line. NA, the data are not available. TMB, tumor mutation burden; MSI, microsatellite instability signature.

  • Fig. 3 The TIP gene signature is correlated with the clinical outcome of patients with cancer.

    (A and B) The impact of TIP signature genes on OS in patients over 28 cancer types, dotplot with log-rank P value (A) and forest plot of Cox proportional hazard ratio and 95% confidential interval (CI) of OS (B) were showed. (C and D) The genomic alterations (C) and the percentage of patients with each genetic alteration type (D) for TIP signature genes in the TCGA breast cancer dataset (58). (E) Fisher’s exact test of genetic deletion (or mRNA down-regulation or missense mutation) and amplification (or up-regulation) for hot tumor– and cold tumor–related genes. (F and G) Impact of TIP signature genes on OS in patients with breast cancer; log-rank test was used. (H) Top: Ranked risk score of 1980 patients with breast cancer evaluated by the expression and the risk coefficient of TIP signature genes. Middle: OS distribution of the 1980 patients ranked according to the risk score (from the top). Bottom: Heatmap of the expression pattern of each of the TIP signature gene in the 1980 patients.

  • Fig. 4 Aurora kinase inhibitors up-regulate expression of hot tumor–associated genes and down-regulate expression of cold tumor–associated genes.

    (A) Potential hot tumor–boosting compounds ranked by the scores of reversing activities for the TIP signature genes. (B) The highest scored compounds are ENMD-2076, JQ1, ICG001, epinephrine bitartrate, and flunixin meglumine. (C) Analysis of chemokines mRNA expression by qPCR. Data are representative of three replicates. (D) GSEA of RNA-seq results of ENMD-2076– and TAK-901–treated MDA-MB-231 cells. NES, normalized enrichment score. FDR, false discovery rate. (E) Selected pathways and corresponding representative genes perturbed by ENMD-2076 and TAK-901 in MDA-MB-231 and 4T1 cell lines.

  • Fig. 5 AKIs activate TH1-type chemokine expression through inhibiting AURKA-STAT3 signaling pathway.

    (A and B) Effects of ENMD-2076 (A) and TAK-901 (B) on the mRNA expression of CXCL10 and CXCL11 in breast cancer cells. (C) Protein level of CXCL10 and CXCL11 measured in MDA-MB-231 cells by enzyme-linked immunosorbent assay after ENMD-2076 treatment. DMSO, dimethyl sulfoxide. (D) The knockdown efficiency of AURKA (left) and the expression of CXCL10 and CXCL11 were quantified by qPCR (right two panels). (E) The overexpression of AURKA was confirmed by Western blot (left), and the expression of CXCL10 and CXCL11 was quantified by qPCR (right two panels). GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (F) The knockdown efficiency of STAT3 (left) and the expression of CXCL10 and CXCL11 (right two panels) were quantified by qPCR. (G to I) MDA-MB-231 cells were treated with ENMD-2076 (5 μM) or interferon-γ (IFN-γ) for 24 hours, and STAT protein levels were monitored by Western blot with the indicated antibodies. (J and K) MDA-MB-231 cells were transfected with the shRNAs (J) or overexpression lentiviral vectors (K) of AURKA, and STAT3 protein levels were monitored. shNC, negative control. (l) Illustration of AKIs enhancing CXCL10/CXCL11 expression through inhibition of STAT3 phosphorylation. *P < 0.05,**P < 0.01, and ***P < 0.001 (Mann-Whitney U test). Error bars depict SEM.

  • Fig. 6 AKIs improve the infiltration of CD8+ T cell and the efficacy of anti–PD-1 immunotherapy in TNBC in vivo.

    (A) Representative flow cytometric analysis of CD8+ T cells migration in vitro. (B) The percentage of CD8+ T cell migration described in (A) from independent sample (Mann-Whitney U test). (C) Illustration of AKIs enhancing CD8+ T cell chemotaxis. (D) Illustration of animal models. 4T1 cells were injected into the mammary fat pads of mice. Animals were administrated when the volumes of tumors were about 100 mm3. (E) Representative images of Immunohistochemistry (IHC) staining showing the distribution of CD8+ T cells. Scale bar, 50 μm. P, E, and T represent PD-1–, ENMD-2076–, and TAK-901–treated groups, respectively; P+E and P+T represent particular combination therapies, respectively. (F) Representative IHC analysis of CD8+ T cells in independent samples (Mann-Whitney U test). (G) The expression level of Cxcl10 and Cxcl11 in 4T1 tumors was analyzed by qPCR; each square represents data from an independent mouse. (H) The growth of 4T1 tumors was measured by tumor volume; volume (mm3) = [width2 (mm2) × length (mm)]/2. (I) Tumor sizes at day 25 (sample-paired Student’s t test). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Error bars depict SEM.

  • Fig. 7 Schematic illustration of the novel approach for the discovery of combination immunotherapy agents and the mechanism of AKIs enhancing immunotherapy efficacy.

Supplementary Materials

  • Supplementary Materials

    Tumor immunological phenotype signature-based high-throughput screening for the discovery of combination immunotherapy compounds

    Haiyan Wang, Shasha Li, Qianyu Wang, Zhenshuo Jin, Wei Shao, Yan Gao, Lu Li, Kequan Lin, Lin Zhu, Huili Wang, Xuebin Liao, Dong Wang

    Download Supplement

    This PDF file includes:

    • Figs. S1 to S6
    • Table S1

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article