Introduction

Glioblastoma (GBM, WHO grade IV) is a highly aggressive and invasive tumor. The ability of GBM tumor cells to invade the surrounding neuroparenchyma critically limits effective treatment leading to tumor recurrence mainly in the peritumoral brain zone (PBZ) of the resection margins [1]. These invading tumor cells are resistant to the standard treatment modalities [2]. Deciphering the molecular underpinnings of such invaded tumor cells and the underlying microenvironment would throw light on understanding the pathobiology of tumor recurrence.

Studies have shown that PBZ differs histologically and molecularly from tumor core and non-neoplastic brain tissue [1, 3,4,5,6]. In our previous study, using a whole genome gene expression microarray approach, we identified 4176 differentially regulated genes in the tumor core and the PBZ of GBM compared to control tissues [6]. In the present study, we selected eight genes that were among the top upregulated genes having almost equal expressions in tumor core and PBZ and with a role in other systemic cancers but rarely studied in GBM, namely; Serine protease inhibitor, clade A, member 3 (SERPINA3), CDC28 kinase regulatory subunit 2 (CKS2), Nucleolar and spindle associated protein 1 (NUSAP1), Fc fragment of gamma binding protein (FCGBP), Cell division cycle 45 like (CDC45L), GTPase activating protein (SH3) domain binding protein 1 (G3BP1), Eukaryotic translation initiation factor 3 subunit M (EIF3M) and Prothymosin α (PTMA). We validated mRNA expression of these genes. Among them, we selected SERPINA3 for detailed evaluation, since it was equally upregulated in the tumor core and PBZ although the latter has limited infiltrating tumor cells. Moreover, association of SERPINA3 expression with poorer prognosis of the patients was already reported [7, 8]. However, the mechanism of action of SERPINA3 has not been elucidated. Hence, we functionally characterized its role in cell proliferation, invasion, migration, promotion of epithelial to mesenchymal transition (EMT), stemness and radioresistance in vitro, in glioma cells. Further, we studied the protein expression of SERPINA3 in tumor core and PBZ and its association with tumor recurrence and patient prognosis.

Materials and methods

Collection and characterization of tissue samples

The study included three sample sets collected from different GBM patient cohorts. The first sample set consisted of tumor core and PBZ tissue samples collected prospectively, from 37 surgically fit adult (> 18 years) patients operated for cerebral hemispheric lobar GBM which were amenable for safe differential biopsy. Pediatric patients as well as patients with tumor located in deep nuclear areas of cerebrum, in the cerebellum and brain stem and cases not amenable for safe differential biopsy were excluded. 22 control samples were collected from patients undergoing surgery for epilepsy at our institute. All these samples were collected in a similar manner as mentioned in our previous study [6], following institutional ethics clearance and informed patient consent. Real time quantitative polymerase chain reaction (qPCR) and immunohistochemistry (IHC) was performed on this cohort. The second and third sample sets included tumor tissues collected from adult patients with cerebral hemispheric lobar GBM. The second sample set (retrospective cohort) comprised 20 paired samples consisting of primary (newly diagnosed) and the corresponding recurrent GBM tissue of individual patients on which IHC was performed. The third sample set (retrospective cohort) included tumor tissues from a cohort of 109 GBM patients who underwent surgery at our institute from the years 2014 to 2016, had received adjuvant radiotherapy and were followed up for survival data. Many of these patients had also completed temozolomide chemotherapy along with radiotherapy. Patients having optimal tissue for molecular characterization were included. IHC and molecular tests for other relevant GBM markers were carried out on this sample set. Sample sets two and three were retrieved as formalin fixed paraffin embedded (FFPE) blocks from the archives of the department of neuropathology of our institute.

Stable knockdown of SERPINA3 in glioma cell lines

Expression of SERPINA3 in various glioma cell lines was determined in five well-established glioma cell lines (A172, LN229, U251, U343 and U373) and all of them had high expression of SERPINA3 (Fig. S1). Among these cell lines, LN229 and U251 cell lines were selected for functional characterization. LN229 and U251 human glioma cell lines were a kind gift from Professor K. Somasundaram, Indian Institute of Science, Bangalore. These cells were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM, Cat. No. D5648 Sigma Aldrich, USA) supplemented with 10% fetal bovine serum (FBS, Cat. No.10270106, Gibco) and Penstrep (Gibco, Cat. No. 15140122) in a humidified chamber with 5% CO2 at 37 °C. Two shRNAs for SERPINA3 were procured from pLKO.1 shRNA library (Sigma Aldrich). Sequences for shRNA1 and shRNA2 are CCGGACTATAACCTGAACGACATACCTCGAGGTATGTCGTTCAGGTTATAGTTTTTTG and CCGGGCATCACCTGACTATACCTTACTCGAGTAAGGTATAGTCAGGTGATGCTTTTT respectively. Polyethylenimine reagent (PEI, branched, Cat. No. 408727, Sigma Aldrich) was used for transfection of plasmid in selected cell lines according to manufacturer’s protocol. Cells stably expressing the shRNA were selected using puromycin (2 µg/ml DMEM) as a selection marker and were maintained in the low concentration of puromycin (100 ng/ml DMEM). Cells selected following transfection of scrambled shRNA served as control for knockdown cells.

The details about publicly available databases and methods for qPCR, western blotting, in vitro assays, IHC, Sanger sequencing for IDH1 and IDH2 non-canonical mutations, assessment of MGMTp methylation status, patient follow up and statistical analysis are mentioned in Supplementary File 1.

Results

Identification of novel genes from previous microarray study, validation by qPCR and in publicly available databases

From our previous microarray data [6], we selected eight genes among the top upregulated genes, namely, SERPINA3, CKS2, NUSAP1, CDC45L, FCGBP, G3BP1, EIF3M and PTMA which had almost equal fold changes in tumor core and PBZ compared to control tissue although the percentage of tumor cells in PBZ was far less (3–5-fold) compared to tumor core. Gene expression profile for these selected genes across 17 tumor core and PBZ and 8 control samples is represented in the heat map (Fig. 1a). qPCR validation showed that five genes; SERPINA3, CKS2, NUSAP1, FCGBP and CDC45L have higher mRNA expression in tumor core and PBZ compared to control. G3BP1 and EIF3M mRNA expressions were significantly higher in tumor core but not in PBZ compared to control. PTMA did not validate at mRNA level. The log 2-fold changes (FC) of these selected genes are represented in Fig. 1b. The expression of these selected genes in the present study cohort was similar to publicly available databases with respect to their higher expression in tumor core and PBZ in Ivy GAP database (Fig. 1c) and in tumor core in TCGA and REMBRANDT databases compared to their respective controls (Fig. 1d).

Fig. 1
figure 1

a Heat map depicting expression pattern of selected eight genes in tumor core (n = 17) and PBZ (n = 17) compared to control (n = 8). Red and blue colors represent high and low expression, respectively. Rows represent genes and columns represent samples. TC tumor core, PBZ peritumoral brain zone. Note the almost similar pattern of expression in tumor core and PBZ despite PBZ having less number of tumor cells. b Scatter plots depicting mRNA expression by qPCR showing log 2-FC (fold changes) of selected genes in tumor core (n = 37) and PBZ (n = 37) compared to control (n = 22) (for analysis, SERPINA3 had 21 controls, CDC45L had 19 controls and rest of the genes had 22 controls). (p < 0.05, bars represent mean ± SD). c Heat map depicting expression of selected eight genes in cellular tumor (tumor core) and infiltrating tumor (PBZ) compared to leading edge (control) in Ivy GAP database. d Bar diagrams depicting expression of selected genes in tumor core in the present study and in the TCGA (Affymetrix and Agilent platforms) and REMBRANDT databases, indicating similar expression pattern, bars represent mean ± SD

Full size image

SERPINA3 plays an important role in tumor cell proliferation, invasion, migration, EMT, stemness and radioresistance

SERPINA3 was selected for further functional characterization due to its reported involvement with GBM patient prognosis [7, 8]. shRNA mediated knockdown (KD) approach was used in two glioma cell lines, LN229 and U251. Stable knockdown of SERPINA3 using two different shRNAs showed significant reduction in SERPINA3 protein expression in both cell lines (Fig. 2a). MTT assay was performed to assess the role of SERPINA3 in cell proliferation. There was a significant reduction in cell proliferation in SERPINA3-KD cells as compared to control cells in both cell lines at 24, 48 and 72 h time points (p < 0.05) (Fig. 2b). Cell cycle analysis was carried out to understand the lag in proliferation which showed an increased number of cells in G0/G1 phase in SERPINA3-KD cells compared to control cells (Fig. S2).

Fig. 2
figure 2

SERPINA3 knockdown (KD) and its effect on cell proliferation, invasion and migration and radioresistance. a Western blots representing reduction in SERPINA3 protein expression in knockdown cells compared to control cells (scrambled) in both LN229 and U251 cells. b MTT assay results depicting decreased cell proliferation in SERPINA3-KD cells compared to control cells at 24, 48 and 72 h in LN229 and U251 cells (n = 3, p ≤ 0.05). c Representative images of Matrigel invasion assay in LN229 and U251 cells along with quantitative representation depicting decreased tumor cell invasion in SERPINA3-KD cells compared to control cells. Each experiment was performed in duplicate. Analysis was performed using one-way ANOVA with repeated measures followed by Bonferroni post hoc test (n = 3, p ≤ 0.05). d Representative images of the wound healing assay in LN229 and U251 cells along with quantitative representation depicting decreased migration in SERPINA3-KD cells compared to control cells. Each experiment was performed in duplicate. Analysis was performed using two-way ANOVA with repeated measures followed by Bonferroni post hoc test (n = 3, p ≤ 0.05). e Graphs showing percent survival in LN229 and U251 cell lines assessed by clonogenic assay. In U251, a radioresistant cell line, there is significant reduction in percent survival in SERPINA3-KD cells at 2 and 4 Gy radiation doses. Each experiment was performed in duplicate. Analysis was performed using two-way ANOVA with repeated measures followed by Bonferroni post hoc test (n = 3, p ≤ 0.05). All the data were represented as mean ± SD of three independent experiments. *, ** and *** represents p ≤ 0.05, 0.001 and 0.0001 respectively

Full size image

Matrigel invasion assay was performed to determine the effect of SERPINA3-KD on cell invasion. SERPINA3-KD cells showed significant reduction in the number of cells that invaded through the matrigel chamber compared to control at 16 and 20 h in LN229 (p < 0.05 ) and U251 (p < 0.0001) cells respectively (Fig. 2c). Role of SERPINA3 in cell migration was evaluated by wound healing assay. At 24 h, we observed significant reduction in area of the wound in control cells compared to SERPINA3-KD cells (LN229, p < 0.001; U251, p < 0.001) (Fig. 2d). To further understand role of SERPINA3 in the regulation of epithelial to mesenchymal transition (EMT) phenotype of glioma cells, we assessed mRNA expression of three molecules involved in EMT; N-cadherin, SNAIL and TWIST. In LN229 cells, there was significant reduction in expression of these three markers and in U251 cells, in the expression of SNAIL in SERPINA3-KD clones (Fig. S3). These results suggest the association of SERPINA3 in regulating key factors involved in EMT. Since mesenchymal transition is associated with stemness, we assessed the mRNA expression of three stem cell markers; SOX2, OCT4 and NANOG in both the cell lines and we noted significant reduction in the expression of these markers in SERPINA3-KD cells compared to control cells (Fig. S4).

Furthermore, stemness is known to induce therapy resistance, therefore we studied the role of SERPINA3 in radioresistance at 0, 2, 4 and 6 Gy radiation doses by using gamma irradiator, which was evaluated by colony formation assay. We observed that U251 cells were more resistant to radiation compared to LN229 cells. In this radioresistant cell line, there was significant reduction in the percent survival of cells in SERPINA3-KD cells compared to control at 2 and 4 Gy radiation doses. These findings suggest a role of SERPINA3 in radioresistance (Fig. 2e). Thus, SERPINA3 plays an important role in tumor cell proliferation, invasion, migration, EMT, stemness and radioresistance in GBM.

Increased SERPINA3 expression in PBZ compared to tumor core and in recurrent GBM tumors

As SERPINA3 mRNA expression was almost similar in tumor core and PBZ, despite PBZ having fewer infiltrating tumor cells, we further studied the immunoreactivity of SERPINA3 in the tumor core and PBZ by IHC. SERPINA3 showed variable immunopositivity within tumor core (Fig. 3a) and predominantly stained tumor cell cytoplasm (Fig. 3a, inset). An increased SERPINA3 expression was noted in the infiltrating tumor cells in PBZ (Fig. 3b) compared to tumor core. The median LI (mean ± SD) in the tumor core and PBZ was 30 (30.44 ± 14.16) and 40 (41.18 ± 14.77), respectively, showing significantly higher (p = 0.0042) expression in PBZ. In addition, SERPINA3 staining was noted in the cells spreading in the subpial region (Fig. 3c). There was no neuronal or glial staining in the control tissues (Fig. 3d).

Fig. 3
figure 3

SERPINA3 protein expression by immunohistochemistry (IHC). Tumor core shows variable staining (a). The staining is predominantly cytoplasmic (a, inset). Several infiltrating tumor cells in the PBZ are stained (b), including cells lodged within subpial region (c). Control tissues are negative for SERPINA3 (d). Immunostaining is higher in recurrent GBM tissue (f) when compared to the paired primary tissue (e). IHC image in inset (a) original image magnification × 400, images a, b, d–f original magnification × 200, image c original magnification × 100

Full size image

Furthermore, SERPINA3 immunoreactivity was studied by IHC in 20 paired GBM tumors (sample set 2) comprising primary GBM and the corresponding recurrent GBM tissue. The median LI (mean ± SD) for the primary GBM was 27.5 (25.75 ± 12.49) and 35 (33.5 ± 11.25) for recurrent GBM, which was significantly higher (p = 0.020). Representative IHC images are depicted in Fig. 3e, f.

Correlation of SERPINA3 protein expression with patient age and clinically relevant GBM markers

Association of SERPINA3 with age, clinically relevant molecular markers (IDH mutations and MGMTp methylation) and prognosis was studied in the survival cohort (n = 109). Median age of the patients was 55 years (range 22–77 years). Among the 109 tumors, 8 had IDH mutations (IDH1 R132H = 7, IDH1 R132G = 1). MGMTp methylation was noted in 60% of tumors. Spearman correlation analysis showed that increased SERPINA3 expression correlated with increased age of the patients (p = 0.015) and IDH wild type tumors (p = 0.008), but not with MGMTp methylation.

SERPINA3 is associated with poor patient prognosis

Survival analysis was carried out only on patients with IDH wild type tumors (n = 101), in order to maintain uniformity of the cohort and adhering to the guidelines of the consortium to inform molecular and practical approaches to CNS tumor taxonomy, not officially WHO (cIMPACT-NOW) update 5 where in, IDH mutant GBMs will now be denoted as IDH mutant astrocytoma, grade 4 [9]. Median overall survival (OS) and progression free survival (PFS) of the cohort were 12 months and 9 months respectively. We studied the prognostic significance of SERPINA3 and clinically relevant markers in this retrospective cohort using Cox regression model. On univariate Cox regression analysis; increased age of the patient and SERPINA3 expression were associated with poor patient prognosis while MGMTp methylation was associated with better prognosis. Multivariate Cox regression analysis showed age of the patient, MGMTp methylation status and SERPINA3 as independent prognostic markers. The details are shown in Table 1.

Table 1 Univariate and multivariate Cox regression analysis for overall survival (OS) and progression free survival (PFS)
Full size table

We divided the survival cohort into lower (n = 50) and higher (n = 51) expression groups (based on median LI = 25) to derive the clinical significance of SERPINA3 with respect to prognosis, by Kaplan–Meier survival analysis using log rank test. Higher SERPINA3 expression group had a significantly shorter OS and PFS than the lower expression group (Fig. 4a, b). Our findings on association of SERPINA3 with prognosis correlated with the TCGA and REMBRANDT databases (Fig. 4c, d).

Fig. 4
figure 4

Kaplan–Meier (K–M) survival curve analysis of the present study cohort showing high expression of SERPINA3 is associated with shorter overall survival (OS, a) and progression free survival (PFS, b) in GBM. K–M survival curves analysis showing high expression of SERPINA3 is associated with shorter OS in TCGA (c) and REMBRANDT (d) databases. p < 0.05 was considered statistically significant

Full size image

Discussion

There are limited studies that throw light on the molecular underpinnings of the PBZ in GBM, which is a niche space for tumor recurrence. Few studies have shown that although PBZ may contain limited number or devoid of infiltrating tumor cells, its genetic and molecular make up are comparable to the tumor core, laying emphasis on the role of molecular alterations of the infiltrating tumor cells or the tumor microenvironment towards tumor cell survival and recurrence [3, 6, 10]. Some studies have demonstrated expression of kinases such as; phosphorylated extracellular signal regulated kinases 1/2 (pERK1/2) and phosphorylated C-jun NH2 terminal kinases (pJNK), stem cell markers such as, Nestin, SOX2, Musashi, CD133, GD3, NG2, cMet, angiogenesis related factors such as VEGF, CD105, HIF1α/2α in both tumor core and PBZ [10,11,12,13,14,15]. These studies hint at the phenomenon of PBZ transformation prior to tumor recurrence. On similar lines, we previously reported two novel markers; PDZ binding kinase (PBK) and Myosin light chain 9 (MYL9) which are highly expressed in tumor core as well as in the PBZ and associated with GBM recurrence [6, 16].

Among the genes selected for validation from our microarray data, SERPINA3, CKS2, NUSAP1, FCGBP and CDC45L, mRNA expressions were significantly higher in tumor core and PBZ compared to control. Some of these genes are overexpressed in various malignancies, such as melanoma, endometrial, hepatocellular, breast, colon cancers and others. These genes have also been shown to influence patient prognosis and tumor recurrence. Their role in cancer related processes is also known [17,18,19,20,21,22]. Higher expression of these genes in tumor core and PBZ suggest that they could contribute to invasion of GBM tumor cells into the surrounding neuroparenchyma. On the other hand, G3BP1 and EIF3M are known to be involved in cell division and translation initiation in cancers [23, 24]. Expression of these genes could be mainly involved in tumor cell proliferation in GBM.

We selected SERPINA3 from the above-mentioned validated genes for functional characterization because of its equally high mRNA expression in tumor core and PBZ in our study, in Ivy GAP database and its high expression in GBM tumor tissue in the TCGA and REMBRANT databases. SERPINA3 is a member of the serpin superfamily of protease inhibitors and is also known as α1-antichymotrypsin (ACT). It is approximately 55–66 kDa, secreted, acute phase protein [25]. SERPINA3 overexpression has been reported in various malignancies such as lung, breast, melanoma, endometrial, colon cancers and others [17, 18, 26,27,28].

SERPINA3 expression in the subsisting tumor cells in the milieu of PBZ might contribute to GBM tumor cell invasion, migration and stemness and probably escaping the effects of radio-chemotherapy and resulting in tumor recurrence. In support, SERPINA3 has been shown to promote cell proliferation, invasion and migration in melanoma, endometrial and colon cancers using knockdown approach [18, 29, 30]. We noted that SERPINA3-KD decreased cell proliferation by arresting cells in G0/G1 phase of cell cycle in GBM. Yang et al. demonstrated role of SERPINA3 in proliferation of endometrial cancer cells by regulating cells at G2/M check point and further suggested activation of MAPK/ERK1/2 and PI3K/ AKT signaling pathways [17]. A study on melanoma cancer has reported that STAT3 is involved in cancer cell invasion and migration through its transcriptional target SERPINA3 suggesting that SERPINA3 is the immediate downstream molecule of STAT3 executing its oncogenic role [31]. On the contrary Zhu et al. reported role of SERPINA3 as a tumor suppressor gene in hepatocellular carcinoma using both overexpression and knockdown approaches and demonstrated opposite effects [32].

EMT is known to induce migration and invasion in tumor cells by acquiring mesenchymal phenotype [33]. Previously Li et al. identified SERPINA3 as a gene associated with glioma cell invasion using knockdown approach and also reported that it was highly upregulated in astrocyte/microglia co-cultured glioma stem cells, GSCs [8]. Furthermore, upregulation of STAT3 and SERPINA3 along with activation of FAK/AKT signaling was demonstrated in melanoma cancer stem cell sphere culture [34]. In line with this, we report reduced expression of key molecules that regulate EMT and stem cell markers in SERPINA3-KD cells, reiterating the role of SERPINA3 in mesenchymal transition and stemness in GBM cells. Furthermore, increased stemness of glioma cells is shown to be associated with resistance to radiotherapy in vitro [35]. We assessed role of SERPINA3 in radioresistance in both LN229 and U251 cell lines. Previously it has been reported that LN229 as radiosensitive and U251 as radioresistant glioma cell lines [36]. Our observations were in concordance with this study. We observed there was significant reduction in the number of colonies formed at 2 and 4 Gy radiation doses in SERPINA3-KD cells in radioresistant U251 cell line. This suggests role of SERPINA3 in conferring radioresistance to glioma cells by regulating DNA damage response.

In our study, SERPINA3 immunoreactivity was seen mainly in the tumor cell cytoplasm. A similar immunostaining pattern has been documented in glioma, melanoma and endometrial cancer cells [7, 8, 17, 18]. We observed significantly higher immunoreactivity in the infiltrating tumor cells of PBZ compared to tumor core. SERPINA3 immunoreactivity was also evident in the tumor cells spreading in the subpial region. Our study is the first to report this unique immunoreactivity pattern of SERPINA3 in the PBZ. Our observations point to the fact that SERPINA3 could be involved in the remodeling of extracellular matrix, leading to tumor recurrence. This was supported by the observation that there was a significantly increased expression of SERPINA3 in recurrent GBM tissues compared to their primary counterparts. Our findings are supported by the observation that SERPINA3 expression is significantly increased in recurrent ovarian cancer and invasive breast cancer suggesting its role in chemoresistance [27, 37]. In addition, we showed that SERPINA3 correlated significantly with increasing patient age as well as IDH wild type status, once again suggesting its association with an aggressive phenotype of GBM.

Further, to understand the clinical importance of SERPINA3 expression, we studied the association of SERPINA3 with patient survival on a uniform cohort with IDH wild type GBM cases and showed that its high expression was associated with poor patient prognosis. Our findings are in concordance with the survival data of TCGA and REMBRANDT databases and with two other studies that reported association of SERPINA3 with adverse clinical outcome in gliomas [7, 8]. Also, SERPINA3 expression was shown to correlate with worse prognosis in gastric, lung and breast cancers, melanoma, and in acute leukemia [26, 29, 38,39,40]. However a study on hepatocellular carcinoma reported that increased expression of SERPINA3 was associated with better patient prognosis [32]. These findings suggest a context dependent role of SERPINA3 in human cancers.

In conclusion, our study identified SERPINA3 as a novel GBM biomarker, highly expressed in the PBZ of GBM, having a role in proliferation, invasion, migration, EMT, stemness and radioresistance. Its increased expression in recurrent GBM tissues and association with poor patient prognosis suggests that it contributes to the aggressive phenotype of GBM. Validating these protumorigenic actions of SERPINA3 through in vivo experiments will further strengthen our findings.