Background
Glioblastoma (GBM), the most common primary malignant brain tumor in adults, is genetically and histopathologically highly heterogeneous. A median survival period is 15 months despite the advanced treatment including surgical resection and chemoradiotherapy [
1]. Increasing evidence suggests that cancer stem cells (CSCs) are crucial for tumorigenesis, therapeutic resistance and recurrence in GBM [
2,
3]. Given that subventricular zone (SVZ) consists of enriched neural stem cells that possess the capacity to generate neurons and glia throughout adulthood [
4,
5]. Gliomas are thus often presumed being initiated by neural stem cells in SVZ [
6‐
8]. Indeed, expression of multiple CSC markers in GBM is negatively associated with overall survival in GBM patients [
9,
10]. Therefore, targeting CSCs is considered as a promising therapeutic strategy.
Aldehyde dehydrogenases (ALDHs) are a group of enzymes consisting of 19 isoforms. Besides the metabolic functions [
11,
12], high ALDH activity is considered as a hallmark of CSCs in various cancers [
13]. Targeting ALDH inhibits the proliferation of GBM tumor cells and CSCs [
14]. ALDH1A3 is the most upregulated between ALDH high and low subgroups of glioma cells among 19 isoform of ALDH family [
15]. ALDH1A3 prominently emerges as a CSC marker to be targeted in multiple neoplasms [
16‐
18]. Of note, ALDH1A3 is enriched in mesenchymal subtype (MES) of GBM patients, thereby being a sensitive and specific marker for MES GBM subtype [
15]. ALDH1A3 is crucial for transition from proneural-CSCs to MES-CSC and is important for the maintenance of MES subtype [
19]. Besides, ALDH1A3 plays also important roles in regulating self-renewal, differentiation and chemo/radio-resistance [
20].
Most of the studies of ALDH1A3 expression in dataset were performed at transcriptional level by microarray in GBM. However, the comprehensive association of ALDH1A3 protein expression with clinical outcome remains elusive. The present study focused on the investigation of ALDH1A3 protein expression in a cohort of GBM patients with emphasis on its regional expression pattern and cellular localization and on its correlation with clinical parameters. We also explored the STAT3 and Akt/PTEN signaling cascades which might be involved in the regulation of ALDH1A3 expression and in its functions in GBM as studied by other entities. Through this study, we anticipate providing a broader perspective on this molecule as a prognostic biomarker as well as a potential therapeutic target for GBM.
Discussion
Increasing evidence indicates ALDH1A3 as an important molecule that influences a diverse range of biological processes in CSCs and in tumor cells, thereby being associated with the initiation, progression, and prognosis of various cancers including GBM [
27]. The present study investigated the expression pattern of ALDH1A3 at both mRNA and protein levels in human GBM specimens with emphasis on its association with clinical parameters of the patients.
Analysis of the microarray based TCGA-GBM dataset revealed that the mean level of
ALDH1A3 mRNA in the total cohort was lower than that in control. However, when considered in genomic subtypes of GBM [
28,
29], its mRNA level appeared significantly higher in the MES than in the classical and proneural subtypes; even in the MES of GBM, the mean level of
ALDH1A3 mRNA was comparable with that in control (Supplementary Fig. S
2). On the other hand, the TCGA-GBM dataset showed an inverse association of
ALDH1A3 mRNA expression with OS of GBM
. These findings from the TCGA data encouraged us to further investigate implication of ALDH1A3 in GBM. In the present study, we demonstrated a significant downregulation of
ALDH1A3 mRNA expression in our GBM cohort in comparison to control, consistent with the findings derived from the TCGA dataset. As the cellular, biological functions of a molecule are ultimately determined by its translational/post-translational levels, not by its transcriptional level, we next focused on ALDH1A3 expression at protein levels in individual GBM tissues of our cohort. As shown in Fig.
2l, the immunoreactivity of ALDH1A3 was not detected in the control (normal) brain tissue. Furthermore, by western blot we also demonstrated a 4.66-fold higher protein expression of ALDH1A3 in the subgroup of GBMs with IRS > 2 than that in the control group (Fig.
5b). These findings allow us to postulate that a higher mRNA level of ALDH1A3 does not necessarily refer to its higher protein level. Indeed, several potential mechanisms regulating the translational and post-translation of ALDH1A3 have been identified: the hypermethylation status of ALDH1A3 promoter predicts a low expression of ALDH1A3 protein accompanied by a better prognosis of GBM patients [
30]; USP9X-mediated deubiquitinase plays an important role in ALDH1A3 protein stabilization [
31]; temozolomide (TMZ) treatment at high concentrations does not alter ALDH1A3 mRNA levels, but protein levels through autophagy [
32]. Thus, it is more reliable to analyze the association of clinical parameters with ALDH1A3 protein levels.
GBM tumor is composed of heterogeneous cell populations including a small population, glioma CSCs [
2], which is related to therapeutic resistance of GBM [
3,
33]. As the subventricular zone (SVZ) is often proposed to be the source of CSCs [
6,
8], we evaluated mRNA expression levels of
ALDH1A3 as well as other well-known CSC markers including
Nestin, CD133, CD44, YKL40, OLIG2, SOX2 in GBM. Despite a 2.5-fold higher
ALDH1A3 mRNA detected in the SVZ+ group, none of these investigated CSC markers showed a statistically significant alteration in mRNA levels (Fig.
1d). These results are consistent with previous reports [
34,
35], the mechanism under which need to be further investigated.
The ALDH superfamily is the most important aldehyde metabolic enzyme family in human cells and has been linked to metabolism reprogramming in the initiation, metastasis, and recurrence of cancer [
36]. Among the 19 members, the ALDH1A3 isoform has been served as the major source of total ALDH activity in GBM [
15]. Thus, it is important to check not only
ALDH1A3 mRNA levels but also its protein levels that is more relevant to its enzymatic activity. Our immunohistochemistry study revealed a distinct expression of ALDH1A3 in individual GBM patients and a high inter-tumoral heterogeneity. A clear positive immunoreactivity of ALDH1A3 was detected only in 15 of 30 (50%) GBMs. Among these positive cases, 4 and 11 cases showed medium (IRS > 2) and high (IRS ≥ 6) expression of ALDH1A3, respectively. ALDH1A3 positive cells were mostly located at the tumor infiltrative area, suggesting that ALDH1A3 may participate in tumor cell invasiveness and metastasis. In fact, knockdown of ALDH1A3 expression in vitro models also suppressed cancer cell invasion in different entities [
37,
38]. We demonstrated that a higher ALDH1A3 IRS was significantly associated with a short OS. As supporting, the expression of ALDH1A3 was positively associated with the grade of peri-tumoral edema that is also a prognostic parameter for GBM patients [
39,
40]. Regardless of treatment, MGMT promoter methylation is an independent and favorable prognostic factor in GBM [
41]. MGMT encodes a DNA-repair protein that inhibits the effect of treatment through removing alkyl groups from guanine, a target site for alkylating chemotherapy agents such as TMZ. MGMT promoter methylation is associated with higher therapeutic effect of TMZ and longer OS in GBM patients [
42]. However, controversial results are also observed in several studies [
43,
44]. In the present study, MGMT promoter methylation status was not associated with OS in GBM. Interestingly, when combining MGMT and ALDH1A3 expression together, low expression of ALDH1A3 may sensitively predict a better prognosis than those with higher expression of ALDH1A3 in the subgroup of MGMT+ patients. Hence, evaluation of ALDH1A3 expression might be a powerful prognostic tool in combination with MGMT promoter methylation status. IDH1 mutation is another prognostic maker for GBM. In TCGA database ALDH1A3 mRNA expression is negatively associated with IDH1 mutation [
15]. In our cohort only 3 of 30 patients were identified with IDH1 mutation, and nevertheless, we did analysis of the association of ALDH1A3 expression with IDH1 mutation status. As predicted, no association was found between the level of ALDH1A3 protein or mRNA with IDH1 mutation. Therefore, a larger patient cohort is needed to verify the association of ALDH1A3 protein with IDH1 mutation in the future. Finally, we also observed the expression of ALDH1A3 frequently in the outer layer of different stage of glomeruloid tufts and in some endothelial cells of tumor vessels. In line with the pro-angiogenic functions of ALDH1A3 in vitro [
45,
46], our findings by tissue staining also implicate ALDH1A3 in microvascular proliferation and in neo-angiogenesis in a subgroup of GBM. These clinical associations highlight ALDH1A3 as a potential prognostic biomarker and as a therapeutic target preventing tumor invasion and angiogenesis.
The TI consists of various cell types including infiltrating tumor cells, CSCs, reactive astrocyte, microglia, oligodendrocytes, inflammatory and immune cells, endothelial cells, and stromal cells thereby creating a complex microenvironment feasible for tumor growth and invasion as well as survival and therapy-resistance [
47,
48]. To identify a cell type expressing ALDH1A3, we performed double staining of ALDH1A3 with the following cell type specific markers: GFAP, CD68 and CD44. Among ALDH1A3 positive cells, there were some GFAP positive cells likely reactive astrocytes or tumor cells due to their distinct morphology. As the co-expression of ALDH1A3 with the stem cell marker CD44 was also found, we propose that ALDH1A3 positive tumor cells might possess the stem cell-like properties involved in tumor progression and therapy resistance. Tumor-associated macrophages (TAMs) are highlighted in GBM due to the considerable size of their population (30–50%) [
49]. But, the present study did not observe cells co-expressing ALDH1A3 and CD68. The possible implication of ALDH1A3 in immunoreaction mediated by immune cells needs to be further studied.
Along with important roles of ALDH1A3 in cancer and cancer stem cells, its underlying regulatory mechanism has become of interest. Several non-coding microRNAs including miR-600hg [
50], miR-7 [
51], miR-187 and miR-125a/b [
27] have been identified to recognize ALDH1A3 as a target gene and to suppress ALDH1A3 expression. Forkhead Box D1 (FOXD1) regulates the transcriptional activity of ALDH1A3 in glioma stem cells (GSCs) of MES subtype [
52]. The hedgehog pathway can significantly increase the expression of ALDH1 in cancer stem cells of ovary, which can also be a possible mechanism in GBM [
53]. In a recent study, ALDH1A3 is stabilized upon ubiquitin-specific protease 9X (USP9X) in the MES GSCs. In contrast, deletion of USP9X induces degradation of ALDH1A3, and a subsequent decrease in p-STAT3, whereas overexpression of ALDH1A3 can restore p-STAT3 expression in GSCs in which downregulation of ALDH1A3 was induced by USP9X depletion [
31]. On the other hand, in NSCLC CSCs, activation of STAT3 pathway significantly increases ALDH1A3 expression while multiple inhibitors of STAT3 signaling can decreased ALDH1A3 expression [
17]. Taken together, we assume that there might be a feedback loop between ALDH1A3 and STAT3 to promote tumor progression in cancers. Developing STAT3 signaling inhibitors has been a new strategy for anticancer therapy. However, so far none of them is satisfied in the clinic studies [
54]. STAT3 has two mainly isoforms, STAT3α and STAT3β, of which the latter is a suppressive factor of STAT3α due to lacking the transactivation domain [
55]. In the present study, we found that ALDH1A3 protein expression was inversely associated with the levels of total STAT3β and p-STAT3β but no association was seen between the expression of ALDH1A3 and STAT3α. The association and the underlying regulatory mechanism of ALDH1A3 and STAT3 in GBM need to be further elucidated by using gene techniques in the future. Akt/PTEN is an important signaling implicated in numerous malignant tumors including GBM. We demonstrated a significant increase in p-Akt levels concomitantly accompanied by a downregulation of PTEN in GBM. However, the level of PTEN/p-Akt did not appear to be associated with ALDH1A3 expression in GBMs, suggesting that PTEN/Akt may not be a direct downstream signaling cascade trigged by ALDH1A3.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.