All cell lines were cultured according to a standard protocol in a controlled environment (37°C, 5% CO₂, humidified). Routine assessments for mycoplasma contamination and cell line authenticity were conducted using the short tandem repeat assay, as detailed previously (Kahlert et al., 2015). All cell lines were passaged every other day. In total, five stem cell lines, namely, GBM1 from classical GBM subtype (Aretz et al., 2022); NCH421k (Kološa et al., 2015), JHH520 (Aretz et al., 2022), BTSC233 (Fedele et al., 2017) from mesenchymal subtype and NCH644 from proneural subtype (Podergajs et al., 2013) were used. Commercially available cell lines (U87-GBM and HUVEC) were used as control. HUVECs were purchased from PromoCell GmbH Heidelberg, U87 was kindly provided by A Weyerbrock, formerly at Neurosurgery Department Medical Center Freiburg, Germany. All sources and specific media components are detailed in Supplementary Tables S1–S4.

The CellTiter-Glo^® (CTG) solution (Promega, Madison, WI) was prepared according to manufacturer description. The resulting THP was initially resuspended to a 50 mM concentration in methanol (MeOH) and subsequently diluted to the required concentrations for the experiment. Cell lines were dissociated using TrypLE, washed then adjusted to a density of 2000 cells in 100 µL of complete media per well. Cells were allocated into black 96-well plates in technical triplicates, with THP concentrations set at 10, 20, 30, 40, and 50 μM, alongside a MeOH control, all maintained in biological triplicates. Cells were subsequently incubated with THP for durations of 0, 2, 4, and 6 days under standard culture conditions (humidified atmosphere, 37°C, 5% CO₂). Plates were retrieved at specified time points (0, 2, 4, and 6 days) followed by the luminescence development. Cell viability was quantified using micro plate reader (Tecan, Austria).

Cell proliferation levels were assessed by Ki67 expression allowing quantification of the percentage of proliferating or non-proliferating cells. Proliferation of tumor cells with or without THP treatment was analyzed after 48 and 72 h using the Muse Ki67 Proliferation Kit. Briefly, after routine centrifugation and washing, an osmotic stabilization solution was added (1 × 10⁵ cells to a 1.5 mL tube), followed by the addition of Muse IgG1-PE or Muse Hu Ki67 PE antibodies. The Muse^® cell analyzer was used to perform cell sorting analysis. Results were expressed as the percentage of Ki67⁺ cells and Ki67^- cells.

Cell cycle analysis was conducted on all cell lines following 48 and 72 h of incubation with and without a 10 µM concentration of THP. Each cell line was processed in triplicate. For the analysis, tumor cells were fluorescence stained using the Muse^® Cell Cycle kit (Guava Muse^®, USA) and analyzed with the Muse cell analyzer. In brief, the supernatant was removed by routine centrifugation (1 × 10⁵ cells to a 1.5 mL tube), the cells were mixed and ice-cold 70% ethanol was added. 200 μL of Cell Dispersion Reagent was added and left in the dark for 30 min at room temperature. For each sample, a total of 1 × 10⁴ events were recorded. The data were analyzed using the Muse^® Cell Cycle software module, facilitated the calculation of cell cycle distributions. The number of gated cells in the G0/G1, S, and G2/M phases were quantified and expressed as percentages.

Fluorescence detection of annexin V served to evaluate the apoptotic events induced by THP treatment for 48 and 72 h. Expression in THP-treated or untreated cells was examined using the Muse^® Annexin V & Dead Cell Kit. The supernatant was removed by routine centrifugation (1 × 10⁶ cells to a 1.5 mL tube), treated with 500 µL TrypLE for 3 min, rinsed the cells and 100 µL of detection reagent was added, incubated in the dark at room temperature for 30 min. Thereafter, stained cells were subjected to Muse^® cell analyzer. Muse^® cell software was served to calculate the percentage of early and late apoptotic cells, as well as necrotic and vital cells.

Total RNA was extracted from cultured cells using TRIzol reagent (Invitrogen) (1,000,000 cells), followed by reverse transcription to synthesize cDNA. Quantitative real-time PCR (qRT-PCR) was performed using BioRad CFX Connect Real-Time System qPCR cycler. Primers, synthesized by Sangon, targeted specific genes, and relative mRNA expression levels were calculated using the method. Primer sequences of HIF-2α, CBS and ZEB1 was detailed in Supplementary Table S5.

THP-treated or untreated cells (GBM1, NCH644, JHH520) were subjected to RNA sequencing after 48 h. Total RNA samples designated for transcriptome analyses were quantified using the Qubit RNA HS Assay (Thermo Fisher Scientific) and assessed for quality via capillary electrophoresis with the FragmentAnalyzer and DNF-471 Standard Sensitivity Assay (Agilent Technologies, Inc., Santa Clara, USA). All samples achieved excellent RNA Quality Numbers (RIN >8.8). Library preparation was carried out according to the ‘VAHTS Universal V6 RNA-Seq Library Prep Kit for Illumina V9.1’ protocol. Briefly, 400 ng of total RNA was employed for mRNA capture, fragmentation, cDNA synthesis, adapter ligation, and library amplification. Following bead purification, libraries were normalized and sequenced on the HiSeq 3,000 system (Illumina Inc., San Diego, USA) using a 2 × 150 bp read setup. The bcl2fastq2 software (version 2.17.1.4) was utilized to convert bcl files into fastq files, and to perform adapter trimming and demultiplexing.

All raw data were processed by R software with the screening criteria of logarithmic fold change |log FC|>1.5 and p < 0.05. Afterwards, Venn diagram was used to identify the shared differentially expressed genes in the three datasets. Finally, Gene set enrichment analysis (GSEA) was performed using Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis identified the signaling pathways significantly enriched by the feature (Minoru et al., 2021). The complete KEGG gene set list was downloaded from mSigDB database https://www.gsea-msigdb.org/gsea/msigdb(accessed autumn 2023).

The survival analysis of genes of interest were executed on data from The Cancer Genome Atlas (TCGA) using TCGAbiolinks R package and statistical testing was performed using log rank test.

Western blot was used to evaluate the CBS related protein expression of the THP processed to each tumor cell line (GBM1, NCH644, JHH520) after 72 h. Additionally, the changes in protein expression after 48 and 72 h of treatment were observed in JHH520. Total proteins were isolated using RIPA buffer (1 × 10⁵ cells to a 1.5 mL tube). Tumor cell lysates were resolved on ready-made BAA gels, with the gel concentration selected based on the molecular weight of the target proteins. Electrophoresis was conducted for 60 min at 110 V. Proteins were then transferred to nitrocellulose membranes over a period of 1 hour at the same voltage. The membranes were subsequently blocked with nonfat dry milk for 1 h to prevent nonspecific binding, followed with the primary antibodies and secondary antibodies were applied. Proteins were finally identified by chemiluminescence (Bio-Rad, USA).

Protein oxidation after THP treatment was analyzed using the OxyBlot™ Protein Oxidation Detection Kit (S7150, Merck Millipore). Briefly, protein samples (5 ng/μL for all samples) were resuspended in 6% SDS buffer and subsequently derivatized with or without 2,4-dinitrophenylhydrazine. Equal amounts of derivatized and non-derivatized samples (2.5 μg/sample) were resolved by SDS-PAGE, transferred from a polyacrylamide gel to the membrane. Thereafter, membranes were incubated with the first antibody (90,450) (1:150 in 1% bovine serum albumin/PBS) for 1 h at 25°C, followed by the secondary antibody (90,451) (1: 300 in 1% bovine serum albumin/PBS) for 1 h at 25°C. The oxidatively modified proteins were finally detected by SuperSignal West Pico chemiluminescent substrate (Pierce Biotechnology). Band intensity data were obtained using ImageQuant TL software (GE Healthcare).

In addition, we detected oxidized and persulfidated proteins using the protein persulfide detection protocol (ProPerDP) established by Dóka et al. (2016).

Animal studies were performed in accordance with the German Animal Welfare Act and approved by local authorities (Landesamt für Gesundheit und Soziales, LaGeSo Berlin, Germany) under the permission E0023-23. 6–8 weeks old NMRI nu:nu mice (Janvier Labs) were used. Intracerebral injections were performed as previously described (Alcaniz et al., 2023). Prior to surgery, all mice received 4 mg/kg meloxicam subcutaneously and were anesthetized with a combination of 90 mg/kg ketamine and 12 mg/kg xylazine. Anesthetized mice were fixed in a stereotactic frame, the skin on the skull was opened and 2 μL cell suspension containing 1 × 10⁵ cells were slowly injected through a hole in the skull into the cortex of the right hemisphere. After cell injection, the syringe was removed slowly and the skin was closed using tissue adhesive. The following day mice received 4 mg/kg meloxicam subcutaneously again.

Treatment with THP started on day 12 after tumor cell inoculation. Animals received THP as i.p. administered monotherapy five constitutive days per week starting with first week dosage 1 mg/kg body weight, week 2 with 2 mg/kg body weight and week 3 with 3 mg/kg body weight. Bioluminescence imaging was performed once per week with the NightOwl II LB983 in vivo imaging system. The IndiGO 2.0.5.0 software is used for initial analysis, color-coding of the signal intensity and quantification. Animal condition was checked daily and body weight was measured at least twice weekly. Mice were sacrificed for ethical reasons when showing behavioral abnormalities and body weight loss, surrogates for progressive intracranial tumor growth.

All in vitro experiments were performed with a minimum number of three biological repetitions, each in three technical replicates. Statistical significance was calculated with ANOVA or a t-test. All statistical calculations and graphing were done through GraphPad Prism 8.0 software (La Jolla, USA). Significant findings (*) with a p-value <0.05 and highly significant findings (**) with a p-value <0.01 or (***) with a p-value <0.001. All statistical calculations and graphing were done through GraphPad Prism 8.0 software (La Jolla, USA) and R soft (Version 4.2.0).

THP therapeutic potential was replicated and confirmed in all tested models, independently of the molecular subclass of glioblastoma they represent. The efficacy of varying concentrations of THP on tumor stem cell models, as well as on the non-cancerous control cell line HUVEC, was assessed using the CTG assay at intervals of 0, 2, 4, and 6 days. Overall, THP treatment led to a significant reduction in the viability of tumor stem cells in a dose-dependent manner, as illustrated in Figure 1. Of note, classical monolayer cultured U87 grown in serum containing medium did show relative high levels of resistance to the drug, revealing growth reduction on only on day 4 under high dosage therapy. This dose-response relationship became increasingly evident over time. At a concentration of 10 μM, THP’s impact on HUVEC cells was minimal. Interestingly, on day 6, the viability of HUVEC cells treated with 10 µM THP surpassed that of the untreated control group, suggesting that THP exhibits low toxicity towards non-malignant cells at this concentration. Based on these results, we selected the optimal concentration of 10 µM THP for subsequent experiments. These findings indicate that THP could offer a potent therapeutic advantage in glioblastoma treatment, with reduced adverse effects on healthy cells (Figure 1).

Subsequent assays were conducted on cells treated with 10 µM THP concentration for 48 h and 72 h. Analysis of the Ki67 proliferation index revealed significant differences in three out of five GBM cell lines (Figure 2). Specifically, the proliferation rates of JHH520, NCH644, and U87 were significantly reduced after 48 h of THP treatment. In contrast, GBM1 and BTSC233 did not show significant differences in their proliferation rate, and neither did the control cell line HUVEC. At 72 h post-treatment, all cell lines, with the exception of BTSC233 and the non-cancerous HUVEC cells, demonstrated a significant reduction in proliferation. These results suggest that a concentration of 10 µM THP effectively inhibits the proliferation of a majority of glioma stem cells, while exerting negligible effects on non-malignant cells.

To evaluate the impact of THP on the cell cycle, we treated the stem cell models with 10 µM THP for 48 h. The results showed no significant cell cycle alterations in any of the cell lines, although a slight increase in the G0/G1 phase upon THP treatment was detected (Figure 3). These findings indicate that the inhibitory effects of THP on the growth and proliferation of glioma stem cells are most likely not a consequence of cell cycle alterations. Although all our experiments are performed under standardized conditions as defined by our previously described quality control system (Hewera et al., 2020) enforcing highly repetitiveness of genetic and functional properties of our in vitro platform (Nickel et al., 2021) we acknowledge the fluctuate growth of cell model NCH644 in between the different biological repetitions. This results in large error bars when evaluating the cumulative data consequently also impacting the statistical results.

To ascertain whether THP’s cell-killing effect was mediated through apoptosis, all cell lines were treated with 10 µM THP for 48 and 72 h, followed by Annexin-V staining to detect apoptotic cells. The results showed that NCH644, GBM1, and HUVEC cell lines did not undergo apoptosis at neither time point, with only minimal differences in cell viability observed. In contrast, U87 cells exhibited a significant increase in apoptosis after 48 h (p < 0.01), a trend that persisted at 72 h. Interestingly, the most notable apoptotic response to THP treatment was observed in the BTSC233 cell line, where the percentage of apoptotic cells significantly increased at 72 h post-treatment (Figure 4). Together with the previous observations, these results suggest that THP’s inhibitory action on certain glioma stem cell types could be attributed to its ability to induce apoptosis, although our data focused on assessing late and early apoptosis only.

To investigate the molecular impact of THP on GSCs, total mRNA sequencing was conducted on three GBM stem cell lines: GBM1, JHH520, and NCH644. The cells were treated with 10 µM THP for 48 h, followed by RNA extraction and RNA sequencing. The results showed that that only five genes were differentially regulated across all 3 cell lines upon THP treatment. Four of these were pseudogenes, which are coding gene copies lacking normal function in protein coding (HS6ST1P1, RN7SL4P, RNU6-132P, STAG3L1). The fifth gene, Cystathionine beta-synthase-like (CBS-L), showed significant upregulation in all 3 cell lines (Figures 5A, B). CBS-L is involved in the transsulfuration pathway, crucial for metabolizing L-methionine and detoxifying L-homocysteine.

In further studies, the mRNA sequencing data from THP treated versus non-treated cells was subjected to KEGG pathway analyses. The results showed that the pathways associated with amino acid biosynthesis and the metabolism of cysteine, methionine, glycine, serine, and threonine were upregulated in the THP-treated samples in all cell lines (Figure 5C). These findings suggest that the effect of THP on GBM cells may occur via alterations of the molecular pathways associated with cell metabolism.

Western blot analyses were conducted to confirm the impact of THP on CBS-L protein levels in three tumor cell lines (GBM1, JHH520, and NCH644) treated with 10 µM THP for 72 h. However, CBS-L is not detected by anti-CBS antibodies, or the effect on mRNA is not visible on protein level. Despite observing various dysregulation in all 3 cell lines, the changes were not statistically significant (Supplementary Figure S1A). Treatment of cell line NCH644 for 48 h showed again no significant differences compared to the control group (Supplementary Figure S1B). To further explore the assumed role of CBSL in H₂S formation during THP treatment, we detected protein persulfidation, the oxidative posttranslational modification of cysteine residues initiated by H₂S. Unfortunately, were not able to see persulfidated proteins. However, the ratio of oxidized to reduced proteins nearly doubled following THP treatment whereas the number of protein thiols (reduced cysteine residues) decreased, indicating significant changes in the oxidative state of proteins (Figure 6A). Further analysis was conducted to assess protein carbonylation in three GBM cell lines at 48 h and 72 h post-THP treatment (Figures 6B, C). In line with previous findings demonstrating that H₂S diminished protein carbonylation (Koike et al., 2016; Mao et al., 2022), protein carbonylation levels decreased in NCH644 and JHH520 cells at both time points. However, the GBM1 cell line exhibited a slight increase in carbonylated proteins after 72 h of THP treatment, suggesting variable responses to THP across different GBM cell lines. These findings underscore the complex effects of THP on protein chemistry within tumor cells.

Three PDX models of MES-GBM were established and treated with THP as described. JHH520 showed the highest in vivo tumor formation capacity and comparably homogeneous intracranial growth, Models BTSC233 and NCH421k grew approximately half as fast (Figure 7). In models JHH520 and NCH421k we did not observe tumor growth inhibition, with signal intensities under treatment with THP increasing comparably to control groups, albeit being slightly higher on average. Furthermore, THP had no effect on surrogate survival in these two models. First animals in THP groups and their respective control groups had to be terminated for ethical reasons within 2 days. In model BTSC233 mean signal intensities increased comparably in control animals and animals treated with THP for the first 2 weeks, but eventually decreased in individual animals in the THP group in treatment week three. Furthermore, first two animals in the control group had to be terminated on day 43, while first animals treated with THP had to be terminated on day 54. While monitored differences were not significant, this overall indicates some degree of tumor growth inhibition of THP in the orthotopic BTSC233 in vivo model. THP did not cause any side effects in the performed studies.