3.1 ALOX5AP expression in AML

We examined the expression of ALOX5AP in TCGA using the TIMER database. ALOX5AP expression was higher in tumor tissues than in normal tissues from glioblastoma multiforme (GBM), lymphoid neoplasm diffuse large B-cell lymphoma (DLBC) and AML (Fig. 1a). Further analysis on GSE24006 and GSE63270 datasets has confirmed the results (P < 0.001 and P < 0.001, respectively, Fig. 1b, c). For further verification, the results of RQ-PCR reveal that ALOX5AP expression was significantly increased in AML patients (3.55–17.16, median 8.404) compared with controls (1.21–9.82, median 2.627) (P = 0.0017, Fig. 1d). Moreover, the increased expression of ALOX5AP was also found in cytogenetically normal acute myeloid leukemia (CN-AML) patients (P = 0.0144, Fig. 1d). ROC curve analysis revealed that the AUC value was 0.8770 (95% CI 0.7738–0.9803, P < 0.0001) in our cohort (Fig. 1e), which suggested the ALOX5AP expression level might be a potential biomarker to discriminate AML from controls. Moreover, ALOX5AP expression was significantly higher in monocytic leukemias (M4 and M5 subtypes according to FAB classification, Fig. 2).

Fig. 1

A study of ALOX5AP expression levels in AML and other types of cancers among the TCGA and GEO dataset. a The differential expression between the tumor and adjacent normal tissues for ALOX5AP across all TCGA tumors via TIMER. b, c The expression changes of ALOX5AP in AML and normal bone marrow cells from GSE24006 (n = 31 normal and 20 AML) and GSE63270 (n = 42 normal and 62 AML), **p < 0.01. c Relative expression levels of ALOX5AP in AML patients and controls. The transcript levels of ALOX5AP in controls, whole-AML patients, and CN-AML patients were evaluated by RQ-PCR. Horizontal lines represent the median level of ALOX5AP expression in each group. n = 102; one-way ANOVA. The data are expressed as the mean ± SEM; P < 0.05 among all groups. d Roc curve analysis of ALOX5AP for distinguishing AML patients from controls. n = 102; Mann–Whitney u-test, ***p < 0.001

Fig. 2

Expression patterns of ALOX5AP in different FAB AML subtypes from the TCGA cohort. n = 171, ns: not significant, *p < 0.05, **p < 0.01, ***p < 0.001

3.2 Differences in clinical characteristics between the low and high ALOX5AP expression groups

We investigated the clinical characteristics of ALOX5AP in AML patients from public TCGA database. Based on the median level of ALOX5AP transcript, we divided the AML patients into high and low expression groups (Table 1). No significant differences were observed in sex and BM blast percentage (P > 0.05). However, patients with high ALOX5AP expression were more likely to be older (P = 0.002), with higher white blood cell (WBC) counts (P = 0.005). We also found that patients with low expression of ALOX5AP carried more favorable karyotype (P = 0.03). With respect to known prognostic markers, patients with low ALOX5AP expression more often had IDH1 mutations (P < 0.05). In order to explore the changes of ALOX5AP expression patterns in the process of hematopoiesis, we analyzed the BloodSpot and HemaExplorer databases, we found that ALOX5AP transcripts was low in hematopoietic stem cells from bone marrow (BM HSCs), with a sharp growth in committed progenitors including common myeloid progenitor (CMP) and granulo-monocyte progenitor (GMP), then remained at a high and stable level during myeloid maturation (Fig. 3).

Fig. 3

The expression of ALOX5AP in cell types. a Hierarchical difference tree about cell types with ALOX5AP expression. b The expression of ALOX5AP in a microarray in cell types. Horizontal lines represent the median expression for each class of cells

3.3 Comparison of ALOX5AP methylation levels in AML and normal samples

To assess the methylation levels of ALOX5AP in primary AML samples, we used Targeted bisulfite sequencing to measure the methylation levels of 2 CpG sites in the first exon of ALOX5AP (chr13: 30735647–30735662) (P < 0.01, Fig. 4a). Detailed methylation data for this sequenced region was provided in Additional file 2. The β values of the 2 CpG sites were then averaged as the DNA methylation level of ALOX5AP gene, which was significantly lower in AML patients compared with healthy controls (P < 0.001, Fig. 4b). Furthermore, the methylation results were also confirmed in the Diseasemeth database (P = 8.8e−06, Fig. 4c) and the cohort GSE63409 (P = 0.01, Fig. 4d). In addition, ALOX5AP methylation levels were negatively correlated with mRNA levels (R = -0.1152, P = 0.0103, Fig. 4e).

Fig. 4

Methylation patterns of ALOX5AP in AML. a, b ALOX5AP methylation levels are determined by targeted bisulfite sequencing. Detailed methylation data of methylation levels at 2 CpG sites in the ALOX5AP first exon (chr13: 30735647–30735662) are provided in Additional file 1. n = 26 normal and 102 AML; Mann–Whitney u-test. The data are expressed as the mean ± SEM; *p < 0.05, **p < 0.01. c, d ALOX5AP methylation levels in normal and AML patients for the Diseasemeth database (n = 10 normal and 271 AML) and GSE63409 (n = 30 normal and 44 AML), **p < 0.01. e Scatterplot showing the correlation between ALOX5AP methylation and ALOX5AP expression in AML patients. R = − 0.1152, *p < 0.05

3.4 The impact of ALOX5AP expression on prognosis in AML patients

The overall survival of patients with high ALOX5AP expression in the TCGA AML cohort was significantly worse (P = 0.006, Fig. 5a). Furthermore, we evaluated four independent datasets (GSE10358, GSE37642, GSE106291 and GSE146173), all of which indicated that elevated ALOX5AP mRNA levels were significantly associated with poor prognosis in AML patients (Fig. 5b–e).

Fig. 5

The impact of ALOX5AP expression on overall survival in AML patients. a Kaplan–Meier survival curves of OS in AML patients from TCGA cohort (n = 173). b–e Kaplan–Meier survival curves of OS in AML patients from GSE10358 (n = 273), GSE37642 (n = 417), GSE106291 (n = 250), and GSE146173 (n = 246) cohort. *p < 0.05, ** p < 0.01, ***p < 0.001

3.5 Distinct gene—and microRNA-expression signatures associated with ALOX5AP expression in AML

The transcriptomes were compared between ALOX5AP low and high expression groups based on the TCGA database. A total of 1149 differential genes were identified (FDR < 0.05, | log2 FC|> 1.5; Fig. 6a; Additional file 3), with 544 genes positively and 605 negatively associated with the ALOX5AP expression levels, using the median as the cut-off value. To extend the observation that clinical features differ between AML samples with low versus high ALOX5AP expression, we first compared the transcriptomes of low ALOX5AP expressed group with those of high expression. Among these positively related genes, a number of oncogenes comprise the main ones. These contain known cancer-testis antigen genes (MEGA5, MEGA1, TEX15, PNMA5 and ROR1) and a number of new candidate genes (PAX2, NTN4, NTNG2 and PDE10A), these genes can be used as tumour markers or potential therapeutic targets with some clinical value. In parallel, we identified a number of epithelial-to-mesenchymal transition (EMT) inducers, including MMP2, TWIST, HIF and PROX1, PROX1 [18] and MMP2 [19] are thought to be targets of the Wnt signalling pathway. Besides, MMPs could be involved in cell displacement and vascular leakiness, recent studies have suggested MMP inhibition could be a promising complementary therapy to reduce AML growth and limit HSPC loss and BM vascular damage caused by MLL-AF9 and possibly other AML subtypes [20]. Oncogenes were significantly enriched in genes co-expressed with ALOX5AP. In contrast, a number of reported tumour suppressors were found in genes negatively associated with ALOX5AP expression, including negative regulators of EMT, such as ARHGAP29, RBM47 and KLF4; There's also the Wnt antagonist gene DKK2 in the mix, it has been proposed that overexpression of KIF4 may inhibit EMT in gastric cancer cells through the Wnt/β-catenin signalling pathway[21].

Fig. 6

Gene/microRNA signatures associated with ALOX5AP expression. a Left: heatmap showing the gene expression signature associated with ALOX5AP expression. Right: volcano plot showing gene expression differences between patients with low and high ALOX5AP expression. (FDR < 0.05, |log2 FC|> 1.5). b Analysis of GO and KEGG pathway associated with ALOX5AP expression. c Enrichment of differentially expressed genes in tissues and cells. d Heatmap showing the microRNA expression signature associated with ALOX5AP expression (FDR < 0.05, |log2 FC|> 1.5). Up regulated and down regulated microRNAs mentioned in the text are indicated

Furthermore, Gene Ontology and KEGG enrichment analysis showed that these genes are involved in several biological processes (Fig. 6b). The most enriched were inflammatory responses, neutrophil degranulation, and immunoregulatory interactions between a Lymphoid and a non-Lymphoid cell, which confirmed that ALOX5AP is required for the synthesis of leukotrienes and is involved in various types of inflammatory responses. Several of these GO entries occurred in biological processes, such as positive regulation of cytokine production and positive regulation of MAPK cascade, all these implied that ALOX5AP-related genes are pro-carcinogenic. In addition, these DEGs were significantly enriched in blood and bone marrow (Fig. 6c), which further suggested a potential role for ALOX5AP in haematological tumours.

We also obtained ALOX5AP-associated microRNA expression profiles containing 75 microRNAs (FDR < 0.05, |log2 FC |> 1.5; Fig. 6d; Additional file 4). Among the 75 miRNAs, 68 miRNAs were up-regulated and 7 were down-regulated in ALOX5APhigh patients. Among the down-regulated microRNAs, miR-582, miR-187, miR-9–2, miR-9–1 and miR-9–3, which could inhibit the development of a variety of tumours, including bladder, gastric and lung cancers [22,23,24]. We also found that miR-582 and miR-187 were involved in inhibiting the EMT process [25, 26]. Most up-regulated miRNAs acted as oncogenes in tumour diseases, while in haematological tumours, miR125b, miR-93 and miR-196 could promote leukemogenesis through increasing the leukemic stem/progenitor cell population, promoting cell proliferation, blocking cell differentiation, and diminishing cell apoptosis [27].

3.6 Correlation between ALOX5AP and immunocytes infiltration

Tumor tissue is composed of various types of cells, including stromal cells, fibroblasts, and immune cells. These cells constitute the tumor’s microenvironment. Therefore, we focused on the correlation between ALOX5AP expression and the level of immune cell infiltration. We evaluated the relationship between ALOX5AP expression and tumor-infiltrating immune cells in the AML microenvironment in the TCGA database using the CIBERSORT algorithm. ALOX5AP expression was significantly positively correlated with monocytes and macrophages M2 (P < 0.001, Fig. 7a). This is consistent with the significantly elevated expression of ALOX5AP in monocytic leukaemia (M4 and M5 subtypes of the FAB classification) mentioned above. In contrary, those cells performing anti-tumor responsiveness (i.e., activated CD4 T cells, effector memory CD4 T cells, and CD8 T cells) were significantly negatively correlated ALOX5AP expression (P < 0.001). Meanwhile, we found that ALOX5AP significantly enhanced the immune score in AML (P < 0.05) (Fig. 7b). All these data suggested that ALOX5AP was closely related to tumor immune infiltration. Furthermore, given that immune checkpoints are promising therapeutic targets for cancer therapy, we also evaluated the relationship between ALOX5AP and the set of checkpoint genes (Fig. 7c). We found that the more widely studied immune checkpoint CD86 was significantly positively associated with ALOX5AP expression, While HAVCR2 (TIM-3), CD274 (PD-L1), and TIGIT were negatively correlated with ALOX5AP expression.

Fig. 7

Tumor-infiltrating immune cells in the AML microenvironment and checkpoint genes associated with ALOX5AP expression. a The relation between ALOX5AP expression and immune cell infiltration. Violin plot showing the differences of immune cell fractions between patients with low and high ALOX5AP expression. The overall immune cell compositions were estimated by CIBERSORT across TCGA microarray. b Correlation of ALOX5AP expression and immune score in AML across TCGA database. c Correlation betweenALOX5AP and other immune checkpoints in AML across TCGA database