Retinoic acid (RA), a metabolically active form of vitamin A, acts as a critical signal transduction molecule [1, 2]. In organism, the catalytic process of vitamin A to RA consists of three parts. Vitamin A derived from colored fruits and vegetables was first catabolized to its alcohol form retinol by retinyl ester hydrolases and then oxidized into retinal by widely expressed alcohol dehydrogenases (ADH) [3, 4]. Finally, retinal dehydrogenases (RALDH) binds to retinal and oxidate it into RA [5]. All-trans RA is the majority isoform and predominated in most tissues, which plays functional role in regulating cell growth and controlling differential [6–8]. Unsurprisingly, dysregulation of RA metabolism was reported to be functionally related to cancer by regulating the expression of many tumor suppression genes and oncogenes, such as mucins and retinoic acid receptor beta (RARβ) family [9–11]. For example, previous reports indicated the combination of RA, interferon-gamma (IFN-γ) and transforming growth factor-beta (TGF-β) induce the expression of MUC4, which is aberrantly expressed in numerous cancers [11, 12]. Retinoic-acid receptor responder protein 1 (RARRES2), a RA inducible gene, is a well-documented tumor suppressor in many types of cancer. Downregulation of RARRES2 expression level promotes tumor proliferation and tumorigenicity [13].
RA metabolism and signaling was reported to be functionally related to immunity. In 1992, Carman et al [14]. argued that RA impaired Th2 cell responses to parasitic during vitamin A deficiency by inhibit IFN-γ production from Th1 cells and CD8 + T cells. In the meanwhile, RA regulates cell differentiation in TGF-β-dependent responses [15]. In serum-free cultures, RA activate T cell in an IL-2-dependent manner and promotes TCR-mediated CD4 + T cell proliferation [16]. Preliminary evidence suggests CD4 + T cell activation was significantly correlated with RARα expression [17, 18], and RARγ was reported to be critical for CD8 + T cell effector differentiation [18], indicating RA controls T cell immunity largely depended on RARα and RARγ. In addition to T cell regulation, RA was found to regulate numerous types of immune cells, such as macrophage, DCs and NK cells [19–21]. For instance, CD103 + DCs mediate T and B cells by high expression of RALDH1 and RALDH2 enzymes, which are crucial for the conversion of retinal to RA [20]. In NK cells, RA was shown to be activated by IFNα and suppress the cytotoxicity of human NK cells [21]. These findings support that RA metabolism is functionally correlated with immunity, and dissecting its instinct molecular mechanism will broaden insight to immunotherapy of cancer.
Targeting RA metabolism has been proved to show effectivity in several cancers [22], however, the researches on retinoic acid metabolism abnormality in sarcoma is still lacking, which limits the development and application of related targeted drugs. Sarcoma is a kind of malignant tumor, which can be classified into bone sarcomas and soft tissue sarcomas based on the tumor-of- origin. The overall 5-year survival for bone sarcomas and soft tissue sarcomas are 66.9% and 64.5% [23, 24].
Recent clinical research results indicate that the overall response rate of sarcoma to immunotherapy is less than 20% [25–27]. Our previous research found that the sarcoma tissue presents an immunosuppressive microenvironment [28]. For example, the degree of cell-killing T cell infiltration is very low, with only (1.8 cells/HPF) [29]. Devalaraja et al. confirmed that sarcoma cells can release RA and promote the differentiation of monocytes in the tumor microenvironment into tumor-associated macrophages in animal models, thereby inhibiting the infiltration of antitumor T-cells [30] and also weakening the clinical effect of immune-checkpoint blockade [31]. This finding was highly recommended by Balkwill for its potential implementation on immunotherapy of sarcomas [32]. Therefore, these studies show us the importance of retinoic acid metabolism in sarcoma. However, there are still many unknowns about the clinical value of RA.
In this study, we systematically analyzed the molecular features of 19 retinoic acid metabolism-related enzymes and revealed two subtypes with distinct metabolic status and prognostic value. The overall design of this research was shown at Figure S1. Gene set enrichment analysis indicated a set of immune and oncogenic pathways were enriched in poor-prognosis group. Connective Map [33] screened 56 potential small molecules specific to different sub-groups. We further generated a 6 genes signature to predict retinoic acid metabolism index based on LASSO-penalized Cox regression model. Several immune cells including CD8 + T cells, Treg cells, Monocytes, and Macrophages showed different abundance between these groups, and immune checkpoint blockade therapy response prediction indicated potential immunotherapeutic efficiency of poor-prognosis group. The robust and powerful metabolic index risk model could provide insightful suggestions to explore the molecular functions and mechanisms of retinoic acid metabolism.