NG2/CSPG4 in glioblastoma: about flexibility

See the article by Al-Mayhani and Heywood et al in this issue, pp. 719–729.

The transmembrane proteoglycan neuron-glial antigen 2 (NG2)/chondroitin sulfate proteoglycan 4 (CSPG4) in humans was initially found overexpressed in melanomas and subsequently in other solid tumors. Its subcellular localization and increased expression in multiple tumors favored its consideration as a prototype oncoantigen.¹

In this issue of Neuro-Oncology, Al-Mayhani et al have focused on the role of NG2/CSPG4 in glioblastoma (GBM) “social” organization.² They find that its expression, present in about one third of patient-derived GBM cell cultures, can be switched on and off and that there is a cross-talk between NG2+ and NG2− GBM cells in vitro and in vivo, a finding conflicting with a hierarchical organization of GBM that is, on the contrary, present in NG2+ progenitors from human fetal cortex (see Supplementary Figure 4 in Al-Mayhani et al). Thus their data emphasize the flexibility of GBM expression programs, an exquisite feature of cancer stem cells in GBM and other malignancies underlying their capacity to escape and survive therapeutic or environmental challenges.³

The authors’ experiments underline the important role of NG2+ cells in GBM proliferation. NG2+ cells, obtained by sorting, proliferate faster and are more aggressive in vivo than NG2− cells. Notably, lethal tumors also arose from NG2− cells; however, when examined postmortem, tumors derived from NG2− cells showed positive NG2 expression, possibly due to overgrowth of a small minority escaping short hairpin (sh)RNA inhibition.

One intriguing finding in the paper derives from the examination of expression programs in NG2− cells. The gene most overexpressed in this setting is matrix Gla protein, which according to preclinical data plays a positive role in GBM migration and invasion. Also, the second gene most expressed in NG2− cells is involved in migration/invasion through regulation of MiR-338-5p: epidermal growth factor‒containing fibulin-like extracellular matrix protein (EFEMP1). These observations raise the intriguing possibility that NG2/CSPG4 acts as a sort of “traffic light,” modulating GBM cell behavior according to environmental stimuli that may privilege proliferation (NG2+ cells) or migration/invasion (NG2− cells) (Fig. 1). In line with this, previous data proposed that NG2 phosphorylated at threonine 2314 (Thr2314) colocalizes with beta-1 integrin on microprotrusions from the apical cell surface, favoring cell proliferation.⁴ Thr2256 phosphorylation, on the other hand, leads to NG2 colocalization with beta-1 integrin on lamellipodia, at the leading edge of cells, and is associated with enhanced cell motility.

Fig. 1.

The NG2/CSPG4 “traffic light.” (A) NG2 is constitutively expressed by GBM cells, tumor-associated pericytes, and microglia/macrophages. These GBM can be targeted by NG2 CAR-T cells. (B) Microglia-produced TNF-α can induce NG2 expression in the NG2− fraction of GBM cells. (C) This induction can reduce the risk of tumor escape observed when targeted antigens are heterogeneously expressed, maintaining the tumor susceptible to targeted treatments, including CAR-T cell therapy or alternative approaches such as monoclonal antibodies (mAb9.2.27). (D) NG2− cells showing a different expression program privilege a migration/invasion program, including expression of matrix Gla protein. An increased understanding of the molecular mechanism underlying the switch from proliferative to invasive programs can provide novel therapeutic targets.

The NG2-dependent activation of beta-1 integrin signaling is also involved in proliferation and migration of pericytes.⁵ NG2/CSPG4 is expressed during vasculature formation and by vascular pericytes during neovascularization in malignant brain tumor.⁶ Inhibiting NG2 expression in xenograft GBM by intratumoral delivery of lentivirally encoded shRNA resulted in a significant decrease in NG2+ microvessels. A significant reduction of tumor volume and peritumoral edema was related to the presence of smaller and less hemorrhagic vessels.⁷ Thus, NG2 expression on both tumoral and vascular compartments further contributes to make NG2 an attractive therapeutic target. Specificity, in this setting, may be determined by different expression patterns of NG2 isoforms present in GBM but not in normal tissue.

This information reinforces the importance of NG2/CSPG4 as a therapeutic target in GBM. Expression of NG2/CSPG4 is subject to multiple layers of regulation. Inflammation and hypoxia are important triggers. Inflammatory cytokines, including interleukin-1β and tumor necrosis factor alpha (TNF-α), induced NG2/CSPG4 expression in microglial cells.⁸ NG2 upregulation on the surface of macrophages, induced by inflammatory signals, was involved in the ability of macrophages in extravasation into the tumor mass. The NG2-dependent activation of β1 and β2 integrins seems involved in this mechanism, as suggested by the observation that extravasation is impaired by NG2 ablation.⁹

Recent observations have shown that microglia-released TNF-α upregulates CSPG4 expression in vivo in GBM models.¹⁰ Thus, even GBM cells that in vivo have limited expression of CSPG4 seem to depend on its expression for in vivo development, in keeping with the data of Al-Mayhani et al.

All these observations suggest that the expression of NG2/CSPG4 is differentially regulated in different tissues on the basis of specific cell functions. Overall they emphasize the potential of therapeutic targeting of CSPG4. Anti-CSPG4 chimeric antigen receptors (CAR) have been developed, and their use in the clinical setting is getting close. Understanding CSPG4 biology will be of increasing importance to delineate subsequent targets if in vivo “deletion” can be effectively obtained by CAR-T cells. If the role of CSPG4 in proliferation is prevalent in vivo, a subsequent switch to invasion/infiltration could be expected and targeted specifically (Fig. 1).