Chloride intracellular channel 1 activity is not required for glioblastoma development but its inhibition dictates glioma stem cell responsivity to novel biguanide derivatives

Background Chloride intracellular channel-1 (CLIC1) activity controls glioblastoma proliferation. Metformin exerts antitumor effects in glioblastoma stem cells (GSCs) inhibiting CLIC1 activity, but its low potency hampers its translation in clinical settings. Methods We synthesized a small library of novel biguanide-based compounds that were tested as antiproliferative agents for GSCs derived from human glioblastomas, in vitro using 2D and 3D cultures and in vivo in the zebrafish model. Compounds were compared to metformin for both potency and efficacy in the inhibition of GSC proliferation in vitro (MTT, Trypan blue exclusion assays, and EdU labeling) and in vivo (zebrafish model), migration (Boyden chamber assay), invasiveness (Matrigel invasion assay), self-renewal (spherogenesis assay), and CLIC1 activity (electrophysiology recordings), as well as for the absence of off-target toxicity (effects on normal stem cells and toxicity for zebrafish and chick embryos). Results We identified Q48 and Q54 as two novel CLIC1 blockers, characterized by higher antiproliferative potency than metformin in vitro, in both GSC 2D cultures and 3D spheroids. Q48 and Q54 also impaired GSC self-renewal, migration and invasion, and displayed low systemic in vivo toxicity. Q54 reduced in vivo proliferation of GSCs xenotransplanted in zebrafish hindbrain. Target specificity was confirmed by recombinant CLIC1 binding experiments using microscale thermophoresis approach. Finally, we characterized GSCs from GBMs spontaneously expressing low CLIC1 protein, demonstrating their ability to grow in vivo and to retain stem-like phenotype and functional features in vitro. In these GSCs, Q48 and Q54 displayed reduced potency and efficacy as antiproliferative agents as compared to high CLIC1-expressing tumors. However, in 3D cultures, metformin and Q48 (but not Q54) inhibited proliferation, which was dependent on the inhibition dihydrofolate reductase activity. Conclusions These data highlight that, while CLIC1 is dispensable for the development of a subset of glioblastomas, it acts as a booster of proliferation in the majority of these tumors and its functional expression is required for biguanide antitumor class-effects. In particular, the biguanide-based derivatives Q48 and Q54, represent the leads to develop novel compounds endowed with better pharmacological profiles than metformin, to act as CLIC1-blockers for the treatment of CLIC1-expressing glioblastomas, in a precision medicine approach. Supplementary Information The online version contains supplementary material available at 10.1186/s13046-021-02213-0.


SUPPLEMENTARY FIGURES FIGURE S1
Distribution of CLIC1 in GBM (a) GBM solid tumor cells CLIC1 staining is diffuse and uniform.
(b-c) CLIC1 intensely positive cells are visible near vessels (arrows), and the staining is dot-like and localized especially on plasma membrane (arrows-heads).
Scale bar = 40 m

FIGURE S2
A) CLIC1 expression in GSCs isolated from 14 human GBMs, evaluated by RNA-seq, and expressed as counts per million reads mapped (CPM). Only GBM 39 (arrow) displayed low CLIC1 mRNA content.
The Table reports the corresponding codes of the GSC cultures used in this study and the RNA-seq data deposited at NCBI Geo data set. B) Upper panel: CLIC1 expression evaluated by WB, in total cell lysates from selected GSC cultures.
Membranes were re-probed with α-tubulin antibody after stripping and used as a reference for protein loading. The average response is reported in the Figure 1A of the manuscript.

FIGURE S4
Dose-response curves of novel biguanide derivatives and metformin on individual non-stem differentiated GBM cell cultures (GBM D). The average response is reported in the Figure 1B of the manuscript.

FIGURE S5
Dose-response curves of metformin and novel biguanide derivatives on individual ucMSC cultures. The average response is reported in the Figure 1C of the manuscript.

FIGURE S6
Kaplan-Mayer curves of Q48, Q54, and metformin (MET) depicting the effect of supramaximal concentration of these compounds on zebrafish embryos survival. Limited toxicity, not different from controls, was observed for all the compounds up to 5 days of treatment. Experiments were repeated twice, n = 20 per experimental group. Q48: log rank test for trend p=0.62; Q54: log rank test for trend p=0.64; metformin: log rank test for trend p=0.38.

FIGURE S7
Effect of different doses of metformin and Q54 on chick embryo survival after 10 days of incubation.
Experiments were performed by Inovotion (La Tronche, France). No toxicity was observed for these compounds up to 3 mM. Experiments were repeated twice, n = 18 per experimental group.

FIGURE S8
A. Dose-response curves of Q46, Q48, Q54 and metformin on rat astrocyte cultures. Limited toxicity is observed for all the novel compounds. Only metformin reduced astrocyte viability (-50%) at the higher concentration tested (30 mM). Data are expressed as average of experiments preformed in quadruplicate and repeated twice. B. Table reports IC50 values in non-malignant rat astrocytes and GSCs and the calculated selectivity indices for each compound. According to the "selectivity criteria" all biguanides are considered selective compounds against GSCs (selectivity index >10) (see reference 62).

FIGURE S12
Kaplan-Meier analysis of the relationship between overall survival and CLIC1 expression. The statistical difference between the curves is measured by log-rank test.
The prognostic effect of CLIC1 mRNA level in GBM according to The Cancer Genome Atlas (TGCA) databasea nalyzed using the GEPIA (Gene Expression Profiling Interactive Analysis database) software (A) and Chinese Glioma Genome Atlas (CGCA) (B). TPM: transcripts of per million; n(high): samples with expression level higher than the median of TPM; n(low): samples with expression level lower than the median of TPM.

FIGURE S14
A) Dihydrofolate reductase (DHFR) expression in 14 human GSC cultures evaluated by RNA-seq, and expressed as counts per million reads mapped (CPM). Comparable expression was detected in cells derived from all the GBM analyzed. RNA-seq have been data deposited at NCBI Geo data set (see Figure S2 for the relative codes). B) Effect of Q48, Q54, Q46 and metformin (MET) on DHFR activity, incubating the compounds with purified enzyme. Methotrexate (MTX) was used as positive control. Q48, Q54, and MET, but not Q46, inhibited DHFR activity with an efficacy comparable to MTX. * p< 0.05. C) Effect of Q48, Q54, and metformin, on DHFR activity in living cells. Q48 and metformin, but not Q54, caused a moderate inhibition of enzyme activity in a time-dependent manner. * p< 0.05.