Chondroitin Sulphate Proteoglycan 4 (NG2/CSPG4) Localization in Low- and High-Grade Gliomas

Background: Neuron glial antigen 2 or chondroitin sulphate proteoglycan 4 (NG2/CSPG4) is expressed by immature precursors/progenitor cells and is possibly involved in malignant cell transformation. The aim of this study was to investigate its role on the progression and survival of sixty-one adult gliomas and nine glioblastoma (GB)-derived cell lines. Methods: NG2/CSPG4 protein expression was assessed by immunohistochemistry and immunofluorescence. Genetic and epigenetic alterations were detected by molecular genetic techniques. Results: NG2/CSPG4 was frequently expressed in IDH-mutant/1p19q-codel oligodendrogliomas (59.1%) and IDH-wild type GBs (40%) and rarely expressed in IDH-mutant or IDH-wild type astrocytomas (14.3%). Besides tumor cells, NG2/CSPG4 immunoreactivity was found in the cytoplasm and/or cell membranes of reactive astrocytes and vascular pericytes/endothelial cells. In GB-derived neurospheres, it was variably detected according to the number of passages of the in vitro culture. In GB-derived adherent cells, a diffuse positivity was found in most cells. NG2/CSPG4 expression was significantly associated with EGFR gene amplification (p = 0.0005) and poor prognosis (p = 0.016) in astrocytic tumors. Conclusion: The immunoreactivity of NG2/CSPG4 provides information on the timing of the neoplastic transformation and could have prognostic and therapeutic relevance as a promising tumor-associated antigen for antibody-based immunotherapy in patients with malignant gliomas.

The aim of the present study was to analyze the role of NG2/CSPG4 in relation to malignant progression and prognosis in a series of 61 adult human gliomas of different molecular subtypes and in nine GB-derived cell lines At the protein level, NG2/CSPG4 expression was evaluated by immunohistochemistry (IHC) in tumor tissues and by immunofluorescence (IF) in cell lines.
Based on these findings, we wanted to evaluate the biological rationale for considering NG2/CSPG4 as a druggable therapeutic target using CAR-Ts in patients with GB.

Brain Tumor Specimens
This study included 61 adult gliomas (seven grade II and two grade III astrocytomas, 14 grade II and eight grade III oligodendrogliomas, and 30 GBs) which were collected at the Units of Pathology of Città della Salute e della Scienza/University of Turin (Turin, Italy) and Maggiore della Carità Hospital/University of Eastern Piedmont (Novara, Italy) ( Table 1).  Patients underwent partial or total resection. Surgical tumor specimens were formalin fixed and paraffin embedded (FFPE), and cut into 5 µm-thick serial sections. The histologic diagnosis was in agreement with the current World Health Organization (WHO) guidelines [71].
Healthy nervous tissue was obtained from the brain autopsies of patients who had died from vascular encephalopathies.
Surgical tumor specimens from five IDH-wild type GBs were split into three consecutive fragments. The first one was FFPE and cut into 5 µm-thick sections. The second one was minced and enzymatically dissociated for expansion in culture, whereas the third fragment was stabilized in RNAlater ® solution (Thermo Fisher Scientific Inc., Waltham, MA, USA) and then stored at −80 • C for molecular genetics.

Ethics Statement
Human brain specimens were obtained and used in compliance with the local institutional review board and the Committee on Human Research and with the ethical human subject principles of the World Medical Association Declaration of Helsinki Research. Written informed consent was obtained from patients after Institutional Ethics Committee approval.

In Vitro Cultures
Fresh surgical tissue from five IDH-wild type GB patients was processed for in vitro culture, as described [72]. Five cell lines were isolated as neurospheres (NS) (TO-02 NS, TO-03 NS, TO-04 NS, TO-07 NS, and TO-09 NS) in Dulbecco's modified Eagle's medium (DMEM)/F-12, supplemented with 20 ng/mL epidermal growth factor (EGF) and 10 ng/mL bFGF for the NeuroSphere Assay (NSA). Four cell lines were developed as adherent cells (TO-02 AC, TO-04 AC, TO-07 AC, and TO-09 AC) in DMEM with 10% fetal bovine serum (FBS). Both cultures were maintained in a 5% O 2 and 5% CO 2 humidified atmosphere. The NS abilities of self-renewal, multipotency, clonogenicity, and tumorigenicity were assessed, as already reported [72].
Cell line authentication from the matched primary tumor was verified by short tandem repeat profiling. Experiments were carried out on cell lines from passages 10-40. In vitro cultures were monthly checked for mycoplasma contamination before experimental use (e-Myco TM Mycoplasma PCR Detection kit, iNtRON Biotechnology, Seongnam-si, Korea).

IHC
Besides haematoxylin and eosin (H&E) staining, IHC was performed using a Ventana Full BenchMark ® XT automated immunostainer (Ventana Medical Systems, Inc., Tucson, AZ, USA). The ultraView TM Universal DAB Detection Kit (Ventana) was the detection system. Heat-induced epitope retrieval was obtained with cell conditioning solution (CC1-Tris based EDTA buffer, pH 8.0, Ventana). Negative controls were obtained by the omission of the primary antibody. The primary antibodies are listed in Table 2. Nestin, Sox2, and MSel1 were considered as representative of GB stem cells (GSCs)/progenitors [73,74].
NG2/CSPG4's immunoreactivity was evaluated using a semi-quantitative system for the percentage of positive cells, the staining intensity, and type of distribution (focal or diffuse) in five randomly selected microscopic high-power fields (HPF), at a ×400 magnification, per tumor section.
IHC for ATRX was used as a surrogate marker for the mutation status of the ATRX gene. The quantification methods for ATRX have been already reported [76].

IF
IF was performed on all nine GB-derived cell lines. Cells were fixed for 20 min with 4% paraformaldehyde at room temperature, rinsed three times with phosphate-buffered saline (PBS), and blocked/permeabilized for 30 min with 1X phosphate-buffered saline (PBS), containing 2% of the appropriate serum and 0.1% Triton X-100. Then, they were stained with the primary antibodies which are indicated with • in Table 2. Negative controls were obtained by omitting the primary antibody. Alexa Fluor ® 488-AffiniPure goat anti-rabbit IgG and Alexa Fluor ® 594-AffiniPure rabbit anti-mouse IgG (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) were used as secondary antibodies. Cell nuclei were counterstained with 4 ,6-diamidino-2-phenylindole (DAPI). Images were acquired on a Zeiss Axioskop fluorescence microscope (Carl Zeiss, Oberkochen, Germany) equipped with an AxioCam MRc5 digital camera coupled to an imaging system (AxioVision Release 4.5, Zeiss).
The frequency of NG2/CSPG4+ cells was quantified by calculating the mean number of positive cells in five randomly selected HPF at a ×400 magnification.
Following the same procedure, IF was also assessed on tissue sections from ten IDH-wild type GBs and five IDH-mutant/1p19q-codel oligodendrogliomas.

Molecular Genetics
Genomic DNA (gDNA) from the FFPE tumor samples was isolated using the QIAamp DNA Mini Kit (Qiagen NV, Venlo, The Netherlands).
Allelic imbalances in the chromosomal regions 9p, 10q, and 17p were assessed by loss of heterozygosity (LOH) analysis and fragment analysis [72]. The EGFR gene amplification status (GenBank accession no. NM_005228) was analyzed as described [72].
Quantitative methylation specific-PCR (MS-PCR), followed by fragment analysis and CE, was used to determine the MGMT promoter hypermethylation status (GenBank accession no. NM_002412). The primer sequences and amplification conditions for MS-PCR were previously reported [79].

Statistical Methods
Associations between the categorical variables were evaluated using 2 × 2 contingency tables by the two-tailed Fisher's exact test. Pearson's correlation coefficient was used to examine the relationship between NG2/CSPG4 immunoreactivity and Ki-67/MIB-1 and Sox2 labelling indices (LIs).
Overall survival (OS) was defined as the time between histologic diagnosis and the patient's death or last follow-up (FU). Patients who were alive at their last FU were considered as censored events. Survival curves were estimated using the Kaplan-Meier method and differences were compared by the log-rank test (Mantel-Cox). For the survival analysis, the cases were dichotomized into the following categories: negative-NG2/CSPG4 expression cases (≤ 1% positive tumor cells) and positive-NG2/CSPG4 expression cases (> 1% positive tumor cells with focal or diffuse immunoreactivity).

IHC
A total of 61 grade II-IV gliomas were studied. In the healthy nervous tissue, NG2/CSPG4 was expressed in the cytoplasm of neurons, whereas it was not found in glial cells or ECs. It was absent or not detectable in rarely visible vascular pericytes.
In gliomas, reactive astrocytes, some tumor cells, vascular pericytes, ECs, normal neurons, and macrophages expressed NG2/CSPG4 in the cytoplasm and/or on the cell membrane.
Through the analysis of the semi-serial sections, NG2/CSPG4 immunoreactivity was mutually exclusive from that of GFAP in tumor cells but not in reactive astrocytes. No correlation was found with Nestin expression that was much more diffusely distributed; NG2/CSPG4-positive cells could be either Nestin-positive or Nestin-negative. In GBs, most CD163-positive macrophages showed NG2/CSPG4 immunopositivity. The NG2/CSPG4 immunoreactivity according to the different molecular subtypes is reported in Table 3.

Relationship between NG2/CSPG4 Expression and Tumor Neo-Vasculature
NG2/CSPG4 was detected in the vascular pericytes or ECs of normal blood vessels or quiescent tumor blood vessels neither in low-nor in high-grade gliomas (LGGs and HGGs, respectively). By contrast, it was strongly expressed in the cytoplasm of the vascular pericytes and ECs of newly proliferated tumor vessels (microvascular proliferations (MVPs) and glomeruli) ( Figure 2E,F). Vascular pericytes were revealed by α-SM actin (α-SMA) and PDGFRβ immunoreactivity ( Figure 2G,H).
Compared to NG2/CSPG4, α-SMA was positive in a lower number of pericytes, whereas PDGFRβ was detected in ECs and pericytes in a higher number.

IF
A total of fifteen tissue sections (from IDH-wild type GBs and IDH-mutant/1p19q-codel oligodendrogliomas) and nine GB-derived cell lines were analyzed by IF. In the former, the distribution pattern of NG2/CSPG4 was similar to the immunohistochemical findings.

Molecular Genetics
The frequency of the above reported molecular changes with respect to the NG2/CSPG4 immunoreactivity is described in Table 3.

Relationship between NG2/CSPG4 and Clinical and Molecular Features
The immunoreactivity of NG2/CSPG4 was not associated with patient age, gender, or anatomic location (all p > 0.05, Fisher's exact test). It increased with the histologic malignancy grade in both astrocytic and oligodendroglial tumors but without statistical significance.
The comparison between NG2/CSPG4 and Ki-67/MIB-1 LIs showed a positive correlation neither in LGGs nor in GBs (p > 0.05, Pearson's test). NG2/CSPG4-positive cells could be either mitotically active or not. Accordingly, in GBs, no correlation was found between NG2/CSPG4 and Sox2 LIs, the latter having previously been described as a marker of cell proliferation for this tumor type [80].

Survival Analysis
The relationship between NG2/CSPG4 immunoreactivity and overall survival OS was analyzed in two tumor sub-groups, including 18 oligodendroglial tumors (ten grade II and eight grade III) and 24 astrocytic tumors (seven grade II, two grade III, and 15 GBs). For the survival analysis, the cases were dichotomized into positive (diffuse or focal NG2/CSPG4 immunoreactivity) and negative ones.
For 19 cases (four oligodendrogliomas and 15 GBs), clinical data and FU were not available. Survival analysis of the tumor sub-group of 15 GB patients with recorded FUs was not possible because among the three NG2/CSPG4-positive cases, one died after nine months and two were still alive at 12 months.

Discussion
In the present study, we analyzed the role of NG2/CSPG4 in relation to the origin, progression, and prognosis of 61 adult gliomas. Compared to low-grade tumors, malignant gliomas show an increased and aberrant NG2/CSPG4 overexpression that correlates with a worse prognosis. The positive immunostaining of reactive astrocytes is one of the most important findings.
The subcellular location of NG2/CSPG4 at the protein level could be related to the different functions of the protein domains. The positive staining of the cytoplasm in neurons could be referred to as the intracellular domain of NG2/CSPG4. Conversely, the membranous distribution in tumor cells and in reactive astrocytes is probably due to the transmembrane and extracellular domains. Its variable expression in GB may also be due to the intra-tumor heterogeneity [83,84].
NG2/CSPG4 marks OPCs in approximately 60% of the oligodendrogliomas. During cytogenesis, NG2/CSPG4 is expressed from a certain stage of development until terminal differentiation in adult oligodendrocytes. NG2/CSPG4-positive or NG2/CSPG4-negative tumors could originate from the transformation of precursors/progenitors at different timepoints (either before expressing NG2/CSPG4, during its expression, or after its loss) ( Figure 5) [85]. They could correspond to tumors with different maturation degrees, proliferation rates, and survival times, according to their early or late derivation. Canonically, NG2-glia (NG2+ glial restricted progenitors) has the capability to proliferate and differentiate into oligodendrocyte precursor cells (OPCs), then giving rise to oligodendrocytes in the immature and mature brain. However, NG2-glia can also differentiate into astrocytes. Additionally, in vitro, NG2-glia can differentiate into type-2 astrocytes through the O2A progenitor cells. (B) NG2+ cells can also be considered as potential cells for the origin of malignant glioma. During tumorigenesis, the glioma stem cell is believed to derive from the transformation of neural stem cells or from the dedifferentiation and transformation of NG2+ glial restricted progenitors, OPCs, or mature cells (astrocytes and oligodendrocytes). The arising tumor cells show potential for self-renewal and express markers associated with both stem and progenitor cell types [85].
Taking into account the low number of cases, we suggest a prevalence of NG2/CSPG4 expression in malignant gliomas and a tendency toward a better prognosis for NG2/CSPG4-negative tumors. As a matter of fact, in the survival analysis, NG2/CSPG4 immunoreactivity was significantly associated with a worse prognosis in astrocytic gliomas after patient stratification for the histologic grade. A similar tendency has been observed in oligodendrogliomas, but the difference in the median survival time between NG2/CSPG4-positive and NG2/CSPG4-negative cases is not statistically significant. The number of censored cases is too high with respect to the temporal values of our collection. This consideration is also valid for the tumor sub-group of GBs.
The most important observation is the NG2/CSPG4 expression in reactive astrocytes. As their frequency varies in the different surgical samples (mainly in GB where they can be found in a very high number), they could be responsible for the variable expression of NG2/CSPG4. In the present study, we confirm an inter-grade NG2/CSPG4 variability in gliomas and its correlation with the malignancy grade, in agreement with previous findings [44][45][46][47]86]. An intra-grade comparison was not possible because of the poor number of cases.
The positive NG2/CSPG4 immunostaining in reactive astrocytes could be responsible for the focal positivity in tumors but does not affect the general interpretation because of the dismal significance of the reactive astrocytes [87]. As a matter of fact, the survival analysis, if limited to cases characterized by diffuse immunoreactivity, did not reveal any significant correlation with prognosis. To explain the widespread NG2/CSPG4 expression, we can speculate that tumors may originate from the transformation of progenitors after acquiring NG2/CSPG4 expression.
Reactive astrocytes can be found either at early onset or at a later stage, both in peritumor tissue and entrapped in the tumor. They can appear as hyperplastic Nestin-positive and GFAP-positive cells with a small and globoid cytoplasm without processes or as hypertrophic GFAP-positive and Nestin-negative cells with a larger cytoplasm and thick and long processes [88,89]. The advancing tumor incorporates reactive astrocytes either as round cells or as GFAP-positive multipolar, spindle, or gemistocytic cells, with long processes on vessels. Their morphological aspect could depend on their age and on the invasion speed of the tumor.
Reactive gliosis around gliomas is a typical finding [90], but it is also known that reactive astrocytes in gliomas have a different meaning in comparison to gliosis in other pathological conditions [89]. However, in the neocortex after injury, NG2/CSPG4-positive cells are not the major source of reactive astrocytes [91]. In addition, progenitors may originate protoplasmic astrocytes of the cortex [90,92] before expressing NG2/CSPG4, consistently with the capability of NG2/CSPG4-positive precursors to originate reactive astrocytes [93,94]. In our experience, tumor astrocytes do not express NG2/CSPG4 in these areas.
The fate of reactive astrocytes entrapped in the tumor still remains unknown. In our opinion, they remain visible for a long time in the proliferative tumor due to their GFAP positivity and NG2/CSPG4 expression, undergoing the pathoplastic influence from tumor cells, mainly in relation to vessels. The double staining for GFAP/ATRX demonstrates that reactive astrocytes retain ATRX expression in the nucleus (i.e., with ATRX-wild type), whereas in diffuse astrocytomas, tumor astrocytes are ATRX-mutant [76]. Different explanations can be suggested: (i) they are converted into tumor cells, (ii) they remain as vestiges, or (iii) they are destroyed.
NG2/CSPG4 immunoreactivity only partially overlaps with the expression of CNPase and PDGFRβ; vascular pericytes express PDGFRβ, NG2/CSPG4, and α-SMA only in a late stage. Their increase corresponds with the growth of newly formed vessels (MVPs and glomeruli) and, therefore, they can be considered a reliable marker of neo-angiogenesis.
By IF, we could confirm that NG2/CSPG4 positivity marks immature progenitor cells that represent the majority of cells under in vitro culture conditions. NG2/CSPG4-negative cells are possibly those with a lower grade of commitment, i.e., with features closer to stem cells. The increase in NG2/CSPG4-positive cells in late passages may indicate the aging of the in vitro culture, i.e., the increase of cell differentiation. The greater number of NG2/CSPG4-positive cells in NS derived from tumors with the highest cell proliferation index and poor survival time is in line with this interpretation. In AC, the subcellular location around the nucleus, rather than on the cell membrane, might suggest the involvement of the intracellular domain in differentiation [11,14], controlling, at the same time, cell proliferation and pericyte migration.
There is an apparent inconsistency between the predominant NG2/CSPG4 positivity of NS and the negativity of most tumor cells in GB. We analyzed a panel of primary cell lines, isolated by IDH-wild type GBs, that we supposed to be derived directly from the transformation of stem cells or NG2/CSPG4-negative early progenitors. We can hypothesize that: (i) tumors mainly develop from NG2/CSPG4-negative cells, in spite of the co-existence of positive ones under in vitro cultures, (ii) the latter have lost the capability of expressing NG2/CSPG4 during the malignant transformation, or (iii) the microenvironment triggers tumor cells towards stemness [95] and, therefore, NG2/CSPG4 negativity.
In astrocytomas, tumor astrocytes may derive from OPCs and lose NG2/CSPG4 positivity by differentiation; on this point, previous speculations about astrocyte derivation from OPCs become relevant [2][3][4]6,[11][12][13]. Oligodendrogliomas might derive from NG2/CSPG4-expressing precursors or from not yet or no longer expressing it, so that the different progression potentials could compensate for the survival evaluation.
The heterogeneous expression of NG2/CSPG4 in GB-derived NS indicates that they are probably composed of both GSCs and progenitors. Our data suggest that in NS, no more than 10%-15% of the cells are true GSCs that do not yet express NG2/CSPG4; the remaining 85%-90% of the cells would be composed of NG2/CSPG4-expressing progenitors. By contrast, the ACs all express NG2/CSPG4 as they are already committed. The majority of the NG2/CSPG4-positive cells would proliferate as progenitors and not as GSCs [90].
The highest NG2/CSPG4 expression in oligodendrogliomas, if compared to the most frequent astrocytic tumors, would suggest that oligodendroglial cells exhibit the hallmarks of progenitors rather than of neural stem cells (NSCs). This finding supports the origin of the tumor from the transformed (NG2/CSPG4-expressing) progenitors, rather than from the transformed NSCs, that may contribute to the better outcome.
Finally, the significant association between NG2/CSPG4 immunoreactivity and EGFR gene amplification (a well-known negative prognostic marker) in both astrocytic and oligodendroglial tumors is in line with previous findings [45]. It corresponds to the unfavorable prognosis of NG2/CSPG4-positive cases revealed by the survival analysis. NG2/CSPG4 could play a functional role in the EGFR-PI3K-AKT signaling pathway by enhancing the activation of the tyrosine kinase domain of EGFR and the proliferative capability of GB cells [38,95].
All the above findings on NG2/CSPG4 in gliomas may have a translational relevance. NG2/CSPG4 could be regarded as a new biomarker of malignancy and poor prognosis, and a promising TAA for a CAR-Ts therapy in patients with malignant gliomas [96,97].