Temozolomide (TMZ) is an alkylating chemotherapeutic agent [
28] and established as adjuvant standard-of care treatment for GBM patients [
9]. With the emergence of novel immune-checkpoint inhibitors for cancer therapy, an ongoing discussion addresses the role of TMZ as first-line treatment for GBM patients [
10]. Especially MGMT non-methylated patients may not benefit from TMZ therapy, but will mainly experience adverse side effects such as myelo- and immunosuppression [
29]. At the worst, these effects may even disrupt active antiglioma immune response and lead to poor results of immune checkpoint inhibitors like the PD-1/PD-L1 checkpoint inhibitor Nivolumab. In a previous study, we showed that an extended therapy with TMZ (above 6 adjuvant cycles) significantly lowers the PD-L1 expression in recurrent tumors independently from the MGMT promoter methylation [
16]. We hypothesized that TMZ, in addition to its chemotherapeutic effects, might also induce changes in intracellular signaling cascades or other pathway activations, which leads to decreased PD-L1 signaling. Following this hypothesis, we aimed to investigate TMZ effects on
PD-L1 expression and its associated regulatory pathways in a glioma cell model. First, we used a publicly available database and analyzed differences between de-novo and recurrent GBM specimens. The analysis revealed a relatively constant expression of the typical subtype expression of proneural, mesenchymal and classical expression signatures Fig.
1b, bottom, which is in line with other reports [
30]. The only outstandingly down-regulated pathways were inflammatory response and JAK/STAT activation, which has not been reported to far. In a next step, we aimed to confirm our findings in cell culture models from one commercial (LN229) and two primary, patient-derived cell lines. To respect any possible effects of GBM subgroup properties, both primary cell lines were classified beforehand by RNA-sequencing analysis [
22]. We noticed, that a baseline
PD-L1 expression was only observable in the mesenchymal cell line. This is in line with previous reports that show a significant association between the mesenchymal subgroup and
PD-L1 expression [
27]. The baseline
PD-L1 expression of the cell lines was most likely low due to the lack of inflammatory signaling within the cell culture model. Therefore, we used IFNγ stimulation to simulate a pro-inflammatory environment. In literature, this mechanism of IFNγ stimulation has previously been described to lead to an increased expression of
PD-L1 by activation of the JAK/STAT pathway [
31,
32]. Among other pathways like the MAPK/ERK [
33] and PI3K/AKT [
17] pathway, several studies have identified the JAK/STAT pathway as the main pathway for
PD-L1 regulation [
18,
31,
34]. TMZ was administered to the cells to mimic standard of care treatment and observe effects on the
PD-L1 expression and regulation. We showed that TMZ significantly reduces the
PD-L1 expression compared to the IFNγ stimulation. In line with previous reports that suggest an immune-modulating effect of TMZ [
29], we hypothesized that TMZ interacts with PD-L1 upstream targets such as the JAK/STAT pathway.
We confirmed a reduction of the JAK/STAT pathway by combined TMZ and IFNγ. Different results were observed in the LN229 cell line, TMZ treatment did not induce a significant decrease of
PD-L1 levels, which could be explained by the character of the LN229 cell line. In line with many previous reports, commercial cell lines such as LN229 or U87 drift away from the “real” conditions in GBM tumors passage after passage and lose the properties of a naïve, patient-derived cell line [
35]. Respectively, this may be a possible explanation for false positive or false negative results. As most studies on
PD-L1 are performed using commercial cell lines, this decrease of
PD-L1 expression after TMZ exposure has not been described before. Our findings may highlight the need for a more “personalized” approach to
PD-L1 analysis and the consecutive treatment choices. In a second step, we tried to verify the cell culture results by examining patient-derived brain tumor tissues of de-novo and recurrent samples. Even though all samples showed individual levels of
PD-L1 in the de-novo state, the expression was downregulated in all recurrent samples compared to the de-novo state. All patients had received standard-of-care treatment with TMZ after the first surgery. The extended qRT-PCR panel revealed a significant downregulation of the IFNγ- associated JAK/STAT pathway.