Glioblastoma (GBM) is the most lethal brain tumor with a median survival of only 15 months [1
]. Despite an aggressive standard of care consisting of surgical resection, adjuvant radiation and chemotherapy with temozolomide (TMZ) [2
] virtually all GBMs recur [4
GBM is characterized by a high degree of heterogeneity on phenotypic, genetic and cellular levels [5
]. As other solid tumors GBM are composed of various brain resident as well as transformed cell types: there are rapidly multiplying tumor cells which make up the bulk of the tumor mass and, on the other hand, there are self-renewing cell types, often regarded as Glioblastoma stem cells (GSCs) [3
]. Whereas the differentiated tumor cells are eradicated by therapy due to their high proliferation rate, the latter are thought to exert increased resistance to adjuvant therapy and tumor initiating capacity as a source of glioma recurrence [6
]. It is well established that GSC driven recurring tumors are resistant to further treatment, but the underlying molecular changes are not fully understood [10
]. There are several proposed resistance mechanisms such as metabolic inactivation of drugs, inhibition of conversion from prodrug to bioactive drug, increased drug efflux, and increased DNA repair [8
]. Importantly, the detailed analysis of these processes and their contribution to GBM resistance promises the discovery of additional targets for combinatorial therapies to overcome resistance to TMZ.
To identify new potential therapeutic targets, we exploited an in vitro approach of GBM recurrence by generating TMZ resistant primary GSCs. Subsequently, TMZ resistant and DMSO control cells were compared by RNA sequencing analyses. Interestingly, only a small number of genes were consistently affected by TMZ treatment when comparing TMZ-resistant GSCs with recurrent GBM patient samples. The most consistently up-regulated gene in TMZ resistant GSCs was the gene encoding Carbonic Anhydrase 2 (CA2). As proof of principle, we identified CA2 overexpression as a characteristic of TMZ resistant GSCs and recurrent TMZ treated GBMs, moreover its inhibition chemosensitized TMZ resistant GSCs.
Formation of tumor recurrence can be considered to be a Darwinian process. While treatment eliminates most of the malignant cells, it simultaneously selects for resistant ones [12
] preventing repeated and efficient treatment with the same drug. In the case of TMZ, a significant prolonged median survival of patients treated with radiotherapy and TMZ (14.6 months) was observed in comparison to patients treated only with radiotherapy (12.1 months) by Stupp et al. in 2005 [2
]. The two-year survival even increased from 10.4% of patients treated only with radiation compared to 26.5% of those who received TMZ in addition [2
], justifying the establishment of chemotherapy with TMZ into the guidelines for GBM treatment. However, many studies have demonstrated that TMZ also promotes malignant transformation and acquisition of resistance by hypermutation [10
]. Addressing the former, Thunijl et al showed the capacity of TMZ to induce progression of low-grade gliomas (LGG) to GBMs by evoking mutagenesis. GBMs arising from TMZ treated LGG exhibited a 39- to 133-fold increase in mutation rate compared to their initial paired LGG [14
]. Similar evidence of hypermutation was obtained when analyzing matched primary and recurrent GBMs. About 17% of recurrent tumors from patients previously treated with TMZ showed hypermutation (defined here as >500 mutations) and harbored about 10-fold as many somatic mutations as untreated tumor samples. In contrast, none of the recurrent tumors of patients who did not receive any TMZ treatment displayed hypermutation [12
As well documented as the surge of mutation rate caused by TMZ is, there is a major lack of reports defining common pathways in vitro and in vivo which might be responsible for resistance. To achieve this, we generated TMZ resistant GSCs cells by continuous long-term exposure to TMZ and subjected them to RNA sequencing analysis. 49 genes exhibited differential expression (adjusted p-value < 0.1 and |log2-fold change| > 0.5) between TMZ resistant and DMSO control cells. However, it is interesting to note that CA2 is the only gene consistently up-regulated in TMZ resistance in vitro and in recurrent GBM samples. Subsequent validation by qPCR qualified CA2 to be investigated in more detail. As a result, of this, CA2 was found to be up-regulated on mRNA as well as on protein level in TMZ resistant compared to DMSO control cells. Importantly, this increase was confirmed in patient matched tissue samples from primary and recurrent GBMs of patients who were treated with TMZ identifying CA2 up-regulation as a potential mechanism of TMZ resistance.
The family of carbonic anhydrases (CAs) consists of 13 zinc containing metalloenzymes which catalyze the reversible hydration of carbon dioxide to a bicarbonate ion and a proton (Figure 7
As such CAs play an important role in several physiological processes including pH homeostasis, regulation of glycolysis and gluconeogenesis, and CO2
]. Furthermore, several members including CA2, 9 and 12 have been associated with neoplastic growth [16
]. As a highly active cytosolic isoform, CA2 is expressed in almost all tissues throughout the body including the brain [16
]. It has been investigated in several tumor entities such as leukemia, melanoma, neuroectodermal tumors, medulloblastomas and gliomas [16
]. In gliomas, CA2 expression not only seems to correlate with malignancy but also with survival, suggesting a link between high CA2 expression levels and a shorter overall survival [19
]. Moreover, several studies show a reduction of invasiveness by inhibition of CAs [16
]. Sulfonamides are the most commonly used inhibitors of CAs which find clinical application in the treatment of glaucoma, epilepsy, congestive heart failure, mountain sickness and gastric as well as duodenal ulcers [20
]. One such sulfonamide is acetazolamide (ACZ) which has previously been reported to have synergistic effects when combined with TMZ [21
]. ACZ in combination with TMZ was shown to decrease cell viability and increase apoptosis in GBM cells in a more distinct manner than TMZ alone [21
]. In established human GBM cell lines an increased apoptosis rate was due to up-regulation of Bax as well as Caspase 9 activity and a concomitant down-regulation of Bcl-2 [21
] indicating that ACZ treatment interferes with anti-apoptotic mechanisms. Indeed, CA2 was identified as a Bcl-3 target gene [23
]. Consequently, an increase in TMZ induced cytotoxicity was observed upon knockdown of CA2, whereas a reduction of TMZ caused an induced cell death when CA2 was overexpressed. Moreover, the authors demonstrate a chemosensitizing effect of ACZ in vitro as well as in vivo using intracranial xenografts [23
]. This favors a combinatorial therapy of ACZ with TMZ, an ongoing phase I clinical trial [24
] highlighting the promising therapeutic potential of ACZ.
While these studies suggest a therapeutic benefit of CA2 inhibition by ACZ for untreated GBM, none of them examined the effect on TMZ resistant GBMs. To our knowledge, this is the first study to substantiate the value of CA2 inhibition for TMZ resistant GBM recurrence. We show here a synergistic effect of ACZ and TMZ in TMZ resistant GBM stem-like cells. It is tempting to speculate that the pH regulating function of CA2 might be involved in the chemosensitizing effect of ACZ.
The pH value is defined as the decimal logarithm of the reciprocal of the proton activity [25
]. Demonstrated by the reaction outlined above the concentration of protons depends on the activity of CA2 among other factors. Maintenance of intracellular pH values is essential for basic cellular functions including enzyme activity, energy metabolism and posttranslational modification of proteins [25
]. Apart from physiological processes, the efficacy of pharmaceutical agents can also depend on pH levels, as is the case for TMZ.
TMZ as prodrug spontaneously converts into the metabolite 5-(3-methyltriazen-1-en-1-yl)-1H-imidazole- 4-carboxamide (MTIC). MTIC is then transformed further into the inactive 4-amino-1H-imidazole-4-carboxamide (AIC) and a methyldiazonium ion which actually mediates the cytotoxic effect by transfer of its methyl group onto the O6
-position of guanine. The two steps of this reaction depend on pH, in particular TMZ bioactive conversion most efficiently at physiological pH (Figure 8
Thus, further studies analyzing pH values as well as hypoxic features influencing TMZ treatment either alone or in combination with ACZ will be necessary to mechanistically explain the ACZ driven chemosensitization.
We demonstrate a causal relationship between TMZ resistance in primary stem-like cells and CA2 up-regulation. In line with previous studies mentioned above, co-treatment of ACZ and TMZ led to a significant re-sensitization of cells to TMZ, revealing a therapeutic benefit of CA2 inhibition. Mechanistically, these therapeutic benefits might be due to a subsequent intracellular acidification [20
]. As mentioned before, the highest TMZ efficacy is achieved at physiological pH values. However, tumor cells often exhibit an intracellular alkaline pH, so that acidification as a result of CA2 inhibition causes a shift towards a physiological pH value. This is due to their altered metabolism as a consequence of enhanced glycolysis, which results in accumulation of protons and acids such as lactate. Contrary to expectations this does not lead to intracellular but rather to extracellular acidification [25
]. It is, therefore, likely that the ACZ dosage of 100 mM causes a “therapeutic window” in which the intra- and extracellular pH values are optimal for increased TMZ efficacy. Moreover, the described alterations of the metabolism in vivo often correlate with hypoxia in a causal manner [22
]. Hypoxia is a characteristic feature for the GBM microenvironment and impacts several processes which leads to progression of tumors such as differentiation, invasion and angiogenesis [28
]. Further studies analyzing pH values as well as hypoxic features influencing TMZ treatment either alone or in combination with ACZ will be required to explain the mechanistic meaning of our results. These revealed hypoxia as one of five hallmark gene sets significantly enriched in TMZ resistant compared to DMSO control cells. In this respect, it is interesting to note that TMZ resistant GSCs selected can recapitulate gene expression changes relevant for GBM recurrence.
In conclusion, we show here for the first time to our knowledge that re-sensitization of TMZ resistant GSCs by inhibition of CA2 bears a promising therapeutic potential to overcome chemotherapeutic resistance and potential marker to assess TMZ sensitivity.
4. Materials and Methods
4.1. Isolation of Primary GBM Stem-Like Cells
We obtained approval from the Ethics committee of the Faculty of Medicine, Philipps University Marburg (institutional review board number 185/11), to collect tumor tissue samples from patients who underwent surgical resection of GBM after giving written informed consent. The tumor tissue was mechanically minced and enzymatically digested with accutase for 30 min at 37 °C. The supernatant was then passed through a 40 µm cell strainer to remove tissue fragments. Erythrocytes were lysed by incubating in Red Cell Lysis Buffer for 10 min. Cells were then seeded on a 6 well coated with 20 µg/mL laminin in PBS.
4.2. Cultivation of Primary GBM Stem-Like Cells
The established human GBM cell line U87 was cultured in DMEM medium containing 10% FCS, Penicillin/Streptomycin (0.1 mg/mL), Non-essential amino acids (1×), Sodium Pyruvate (1 mM). The patient-derived GBM stem-like cells were grown in DMEM/F12 (GlutaMAX) containing 2% B27 Supplement, 1% Amphothericin, 0.5% HEPES and 0.1% Gentamycin with addition of EGF and bFGF in a final concentration of 0.02 ng/µL in non-cell-culture-treated petri dishes, where they formed spheres. All cells were grown in a humidified atmosphere at 37 °C under 5% CO2.
4.3. Differentiation Assay
300,000 cells were seeded in 6-wells and cultured in 3 different conditions for 7 days:
After 7 days RNA was extracted and cells were analyzed for the mRNA expression of stem cell marker CD133 and differentiation marker GFAP.
4.4. Invasion Assay
The invasion assay was executed as described previously [30
]. Briefly, the percentage of the 25,000 seeded cells which invaded along an FCS gradient through the 8 µm pores of the transwell membrane into the matrigel was quantified.
4.5. Generation of TMZ Resistant Cells
To generate TMZ resistant cells, they were exposed to a daily dose of 10 µM TMZ on 5 out of 7 weekdays for a period of at least 20 weeks. For comparison an untreated control as well as a solvent (DMSO) control were also done. Resistance was validated and quantified by Dose-Response-Curves based on viability assays.
4.6. Viability Assay
96 well plated were coated with 50 µg/mL collagen (Sigma-Aldrich, Munich, Germany) in 0.01 M HCl and 500 cells/well were seeded. After 24 h the indicated concentrations of TMZ and/or ACZ were administered to wells in triplicates. Cells were cultured for 2 weeks before viability was measured. For this, 50 µL CellTiter-Glo® 3D Cell Viability Assay (Promega, Fitchburg, MA, USA) was added directly to the well, followed by 15 min of vigorous shaking and 15 min incubation at room temperature in the dark. Luminescence was determined using a 96 well plate reader (FLUOstar OPTIMA Microplate Reader, BMG LABTECH, Ortenberg, Germany). Data was normalized to untreated cells. The IC50 values were determined by graphical analysis of the plotted data.
4.7. Side Population Analysis
5 µg Hoechst 33342 dye was added to 1 million cells in 1 mL medium and incubated for 90 min at 37 °C with or without 50 µM Verapamil (Sigma-Aldrich, Munich, Germany). After this time cells were constantly kept on ice to inhibit further efflux of the dye. After centrifugation and washing, 2 µg/mL PI was added and cells were analyzed at the FACS Analyse MoFloTM Astrios High End Sorter (Beckman Coulter GmbH, Krefeld, Germany). By gating properly debris, doublets and dead cells were excluded before measuring Hoechst intensity at two wavelengths (355–488 and 355–620).
4.8. RNA Extraction and qPCR
To extract RNA, 50 mg tumor tissue was mechanically homogenized in 1 mL Qiazol (Qiagen, Hilden, Germany), whereas cells were lysed in 1 mL Qiazol by resuspending. The following steps were performed as previously described [30
], including transcription to cDNA and qPCR.
4.9. RNA Sequencing, Subsequent Data Analysis, Differential Expression Analysis, and Gene Set Enrichment Analysis (GSEA)
RNA for Sequencing was isolated using the Macherey-Nagel NucleoSpin RNA Plus Kit according to manufactures’ instructions. Integrity of total RNA was assessed on the Bio-Rad Experion. Sequencing libraries were prepared with the TruSeq Stranded mRNA Kit (Illumina, San Diego, CA, USA). On-board cluster generation using the TruSeq Rapid SR Cluster Kit-HS (Illumina) and single read 50 nucleotide sequencing was performed on a HiSeq Rapid SR Flow Cell (Illumina) on the Illumina 1500 platform.
Transcript quantification was carried using the Salmon
approach and the transcript reference from the GRCH38 annotation [31
]. Data import into R and gene expression quantification was realized with tximport
R package [32
]. Differential gene expression analysis between TMZ resistant cell group (all three cell lines) and DMSO control cell group (all three cell lines) was performed with a paired design (TMZ resistant vs. DMSO control), using the methods implemented in the DESeq2
(Version 1.14.1, default settings) R package, wherein genes with an absolute log2-fold change > 0.5 and an adjusted p
-value < 0.1 were considered differentially expressed [33
]. Gene Set Enrichment Analysis (GSEA) was conducted in pre-ranked mode, using the log2-fold change values obtained from DESeq2
for the ranking of the gene list and the fgsea
R package [34
]. The Hallmark, (GO biological process, and Reactome gene sets) were downloaded from Molecular Signature Database [36
] and used for enrichment testing. Gene sets with an adjusted p
-value <0.05 were considered significantly enriched.
4.10. Protein Extraction and Western Blot
Protein extraction, sample preparation and western blotting were conducted as previously described [30
]. The following primary antibodies were used: anti-CA2, Abcam ab124687, 1:1000 in 5% milk in TBST and anti-β-Tubulin, Novus Biologicals, NB600-936, 1:2000 in 5% milk in TBST.
Formalin fixed and paraffin embedded tissue sections (3 µm) were stained using the VECTA Stain Elite Kit (Vecta Laboratories, Burlingame, CA, USA) according to manufacturer’s instructions. After deparaffination at 60 °C for 45 min, sections were hydrated using descending alcohol concentrations. Demasking of epitops was achieved by boiling in citratbuffer (10 mM Trisodium citrate dihydrate, pH = 6). Endogenous peroxidase was blocked by incubating in 3% H2
in methanol for 30 min before incubation in 1.5% goat serum for blocking of unspecific binding. Followed by incubation with primary antibody (anti-CA2, Abcam ab124687, 1:250 in PBS) at 4 °C overnight. Incubation with the respective secondary biotinylated antibody and ABC reagent were followed by DAB staining with the ImmPACTT DAB Kit (Vecta Laboratories, Burlingame, CA, USA). For counterstaining hematoxylin (Carl-Roth, Karlsruhe, Germany) was used. Finally, sections were dehydrated by ascending alcohol concentrations and covered with mounting medium. Images of whole sections were acquired with Axio Scan.Z1 (Zeiss, Jena, Germany) and processed using QuPath [37
] and Fiji ImageJ [38
4.12. Stereotactic Injection and In Vivo Analysis of Tumor Growth
For in vivo experiments 10 to 12-week-old athymic nude mice were obtained from Harlan, Indianapolis, USA. Mice were anesthetized using 1–3% isoflurane and injected with 100,000 cells in 10 µL PBS into the corpus striatum using a stereotactic device. Tumor growth was monitored for 2 to 8 weeks post injection using a 7T MRI (Clinscan 70/30 USR Bruker). When abortion criteria were reached, animals were euthanized in accordance with the local guidelines and the whole brain was immediately collected for histological analysis. Mouse brains were fixed overnight in 4% formalin and embedded in paraffin. Paraffin sections were stained as described above.
Animal facilities and experiments were authorized by the Regierungspräsidium Gießen Germany, according to the German and Hessen animal welfare regulations (file number G60-2016). Animals were housed in the special pathogen-free facility, where a constant temperature of 26 °C, a 12 h light–12 h dark electric cycle, water ad libitum and a commercial laboratory animal diet were provided.
4.13. Statistical Analyses
Data were analyzed using the statistical software R. If not indicated otherwise, data are presented as mean values ± SD. Student’s t-tests were employed to determine significance, which was indicated with one, two or three stars for p-values of <0.05; <0.01 and <0.001 respectively.