Identification of small molecules targeting CtBP1/BARS
In order to identify novel molecules that control the conformational arrangement and the cellular functions of CtBP1/BARS, we performed a virtual screening campaign targeting the NADH-binding Rossmann fold region as binding pocket for the docking simulations. This region was further extended to include a sub-pocket nearby, where inhibitors such as MTOB, PPy and HIPP [22, 64, 65, 72] are known to locate and exert their activity, in order to explore both competitive and non-competitive binding scenarios [73].
The resolved structure of human CtBP1/BARS (PDB: 4LCE) was selected since co-crystallized with both NADH and the inhibitor MTOB, thus offering clear indications about the complete binding region and the crucial residues involved in ligand-protein interaction. Therefore, a single monomer of CtBP1/BARS was targeted during the simulations (Fig. 1A).
The protein structure was prepared for the docking calculations by i) refining and optimizing the overall backbone and residue side chains, ii) removing NADH, MTOB and solvent molecules, iii) relaxing the residues included into a box of 30 Å3, centered on the MTOB center of mass, (so to include both MTOB and NADH environments), via energetic minimization.
A subset of the KEGG Library, accounting for 9.000 small molecules, natural products and metabolites [74] long with a selection of 50.000 commercially available compounds was selected as compound database for the virtual screening campaign. A two-step docking strategy was conducted. In the first run, performed at a standard level of details, the best 2.000 compounds were selected according to the docking score; in the second run, carried out at higher precision with more stringent and time-demanding settings to optimize binding mode and predicted affinity scores, a definitive list of candidates was selected. All in silico calculations, from structure and ligand refinement to docking simulations were performed using the Schrödinger Drug Discovery suite [75] while the entire flow of activities was concerted and monitored with Biovia Pipeline Pilot [76].
Among the top ranked predicted hits, 27 commercially available molecules were acquired and firstly tested for their capability to affect the intracellular localization of CtBP1/BARS in the human A375MM melanoma cell line (a cancer type in which the CtBP1/BARS expression levels are more pronounced than in other tumors; Supplementary Fig. 1A) by immunofluorescence microscopy.
The endogenous CtBP1/BARS localizes both in the nucleus and in the Golgi complex where it acts as transcriptional corepressor and membrane fission inducer, respectively [25, 26, 57]. Three molecules are the most active (at low micromolar concentrations) in promoting the translocation of CtBP1/BARS from the nucleus into the cytoplasm and Golgi membranes, namely; (-)-Epigallocatechin gallate (EGCG), ethinyl estradiol (EE) and N-(3,4-dichlorophenyl)-4-{[(4-nitrophenyl)carbamoyl]amino}benzene-sulfonamide (referred as Comp.11 from here on) (Fig. 1B-D). Similar cytoplasmic re-localization of CtBP1/BARS has been observed in murine melanoma B16F10 cells (Supplementary Fig. 1B) and in other tumors: cervical (HeLa cells) and breast (MCF7 cells) (Supplementary Fig. 2), both tumors transcriptionally regulated by CtBP1/BARS as well. Surprisingly, while EGCG and EE promoted Golgi fission fragmentation (presumably inducing the monomeric fission-prone conformation of CtBP1/BARS) (Fig. 1C), Comp.11 induces Golgi membrane tubulation in all the cancer cell lines (Fig. 1C and 1E, Supplementary Fig. 1B and Supplementary Fig. 2) that resembles a membrane fission defect seen upon inhibition of CtBP1/BARS monomerization [25, 57, 61].
Thus, Comp.11 seems to inhibit both CtBP1/BARS functions: the nuclear transcriptional activity (through protein redistribution into the cytoplasm and in the Golgi Complex) as well as its fission activity (Golgi membrane fission defect, see Fig. 1C and 1E, Supplementary Fig. 1B and Supplementary Fig. 2). This might be due to conformational changes on CtBP1/BARS induced upon Comp.11 binding, and this combination might be very effective in both the transcriptional pro-survival effects and cell cycle control of CtBP1/BARS. Hence, we focus our further investigation on Comp.11.
Thermodynamic characterization of the Comp.11 – CtBP1/BARS binding
To examine the binding of Comp.11 in the NADH-binding Rossmann fold of CtBP1/BARS we performed fluorescence measurements taking advantages of the peculiar intra-molecular FRET between the residue W307 and the NADH moiety bound to the Rossmann fold of CtBP1/BARS [77] (Fig. 2A). Here the distance between W307 and the nicotinamide moiety of NADH for energy transfer is within the Förster radius of 25 Å [77]; thus, upon CtBP1/BARS excitation at 285 nm, the W307 acts as fluorescence donor to NADH that absorbs at 340 nm (the maximum emission of tryptophan) to emit fluorescence signal at ≅ 425 nm (Fig. 2A, black line). This spectrum agrees with the structural information that NADH binds the Rossmann fold of CtBP1/BARS [30], which, in turn, is in its NADH-bound dimeric conformation [30, 60].
Titration with increasing concentration of Comp.11 (0–15 µM) with fixed CtBP1/BARS (1 µM) was used to monitor the capability of this compound to displace NADH from the Rossmann fold. The addition of Comp.11 reduces both the tryptophan fluorescence peak at 344 nm and the NADH fluorescence peak at 425 nm, in a dose-dependent manner (Fig. 2A, color lines). These data indicate that Comp.11 does not release/displace NADH from the Rossmann fold (which should instead result in loss of FRET signal and accompanied by decreased NADH peak and increased tryptophan peak). These data confirm Comp.11 predicted binding mode.
As schematized in Fig. 2B, Comp.11 is expected to locate near the NADH molecule, the way MTOB, PPy and HIPP do [22, 64, 65, 72], projecting the nitrobenzene moiety towards the center of the binding pocket and the nicotinamide moiety of NADH. Here, the nitro group is stabilized by a salt bridge with the guanidine group of R255 and an H-bond with another arginine, R86. Comp.11 orientation is further strengthened by the π-π stacking and the π-cation stacking that the nitrobenzene ring forms with the aromatic side chain of W307 and positive charge of R86, respectively. At the other extremity of Comp.11, its benzene sulfonamide moiety stacks with the aromatic ring of Y65 and establishes an H-bond with carbonyl backbone of D23, stabilizing the complex with CtBP1/BARS. The binding constant (Kb) of the resulting Comp.11–CtBP1/BARS complex quantified based on static quenching methodologies [78] is Kb = 2.1±0.6×105 M− 1 (Fig. 2C) corresponding to a Kd of 4.7 ± 1.3 µM. Further, we employed isothermal titration calorimetry (ITC) to determine the thermodynamic parameters associated to the binding of Comp.11 to CtBP1/BARS (Fig. 2D). We found that one molecule of Comp 11 binds to the protein, with a dissociation constant of Kd = 0.66 ± 0.20 µM (three orders of magnitude tighter than MTOB, Kd = 1.26 mM) [72]. The measured negative value of the binding enthalpy, ∆H = -8.2 kJ/mol and the positive entropy value, T∆S = 27.1 kJ/mol, suggest that the interaction of Comp.11 is both enthalpically and entropically driven. These results agree with the above computational data, as the negative value of the enthalpy take into account the stabilizing interaction of the ligand and protein; while the water release from the binding site induced by the mostly hydrophobic Comp.11 explain the positive value of the binding entropy. A selection of single point mutants of CtBP1/BARS was also included in the panel in order to explore the contribution of critical (known and presumed) residues to the binding. Some of these residues were selected since they are crucial for NADH binding [G172E, C226A; [30, 60, 79]] and stability [R255A, [79]]; other being predicted to interact with Comp.11 [H66A, R86A; [79]]. Finally, H340L mutant was chosen because H340 plays an important role in the catalytic activity [79]. Related binding isotherms are shown in Fig. 2D and in Supplementary Fig. 3, and summarized with the thermodynamic parameters in Fig. 2F.
As predicted in silico and confirmed via ITC, Comp.11 binding strongly relies on R86, and substitution of this residue completely compromises the establishment of the complex of Comp.11 with CtBP1/BARS. Besides, the substitution of the two histidines, H66 and H304, with hydrophobic residues, alanine and leucine, respectively, would probably improve the binding (lower measured Kd) by enriching the apolarity of the region surrounding the nitrobenzene moiety. Moreover, despite the predicted salt bridge between the nitro group and the side chain of R255, its substitution with alanine does not affect the binding affinity, probably because Comp.11 overall orientation within the pocket is guaranteed by R86. Finally, the G172 mutation leads to the dramatic decrease in the affinity of Comp.11–CtBP1/BARS interaction and we think that this effect does not correlate directly with a loss of direct interaction. We can assume that the presence of NADH itself (as found for MTOB binding), compromised in the G172E mutant [60], is a crucial step for Comp.11 binding.
Although we demonstrated that the binding of Comp.11 to CtBP1/BARS results in the fluorescence quenching of CtBP1/BARS (Fig. 2A), it is still unclear whether the binding affects the structure and the microenvironment of the protein. Therefore, we performed circular dichroism (CD) experiments to further investigate the conformational changes of CtBP1/BARS. This protein (at 2 µM concentration, 20°C) in buffer solution shows a well-structured conformation as indicated by the negative bands at 222 nm and 208 nm and the positive band below of 200 nm (Fig. 2E). In the presence of Comp.11 at ratio 1/1 of protein/ligand, the band at 208 nm is slightly reduced, (Fig. 2E) indicating that the protein secondary structure is preserved upon ligand binding.
Comp.11 affects the oligomerization of CtBP1/BARS and inhibits the binding of CtBP1/BARS to partners involved in both transcription and membrane fission.
To investigate the effects of Comp.11 on the oligomerization state, size-exclusion chromatography was performed using purified CtBP1/BARS protein. Full-length CtBP1/BARS in the absence of NAD+ elutes primarily as a dimer mediated by the C-terminal domain [as shown in [30, 60]] near its predicted Mr 96 kDa (Fig. 3A; see DMSO purple line and Ctr/DMSO Western blot panel). The addition of NAD+ shifts the elution patterns so that the majority of CtBP1/BARS elutes as a tetramer, as expected, near its predicted Mr 192 kDa (Fig. 3A; see NAD+ blue line and NAD+ Western blot panel). After the addition of Comp.11, in the absence and in the presence of NAD+, CtBP1/BARS dimerizes and tetramerizes, respectively, in a slightly less packed conformation (Fig. 3A; see Comp.11 green and Comp.11 + NAD+ orange lines and Western blot panels). This indicates that the binding of Comp.11 to CtBP1/BARS alters its oligomerization by favoring an “open oligomeric conformation” probably as a result of a partial hindrance due to the interaction of dichlorophenyl-benzenesulfonamide moiety of Comp.11 with the second CtBP1/BARS monomer during the dimerization/tetramerization process. As reported in Supplementary Fig. 4 this portion of Comp.11 would present at the dimerization interface, in an area normally occupied by a stretch of CtBP1/BARS spanning from residues 153 to 168, somehow impeding the closest approaching of the two monomers.
To define the mechanism through which the binding of Comp.11 to CtBP1/BARS affects the cellular functions of this protein we first investigated whether Comp.11 has an impact on the dynamic equilibrium between monomeric and dimeric conformations of CtBP1/BARS by in vitro pull-down assay with purified recombinant proteins. The preincubation with Comp.11 impaired, in a concentration dependent manner, the ability of GST-CtBP1/BARS to bind to, and dimerizes to, His-CtBP1/BARS immobilized on Ni-NTA–agarose beads (Fig. 3B). NAD+ and acyl-CoA were used as internal pull-down controls: NAD+ to promote the GST-CtBP1/BARS–His-CtBP1/BARS self-association/dimerization and acyl-CoA to inhibit it (Fig. 3B).
After that, we further performed in vitro pull-down experiments, to directly assess whether Comp.11 can alter CtBP1/BARS interactions with the well-known molecular partners involved in transcriptional corepressor activity such as E1A (Fig. 3C) [80, 81] or in membrane fission (Fig. 3D,E) [56, 57, 61]. Figure 3C shows that the preincubation with Comp.11 (as well as with Acyl-CoA) inhibits the ability of His-CtBP1/BARS to bind to GST-tagged E1A, while the addition of NAD+ stabilizes their interaction as previously reported [25, 30].
We demonstrated that CtBP1/BARS, at the trans-Golgi network, is incorporated into a protein complex where the binding to 14-3-3γ adaptor protein stabilizes the monomeric fission-prone conformation of CtBP1/BARS [56, 57], which, in turn, binds to and activates a lysophosphatidic acid (LPA) acyltransferase type δ (LPAATδ), and this LPAATδ−mediated production of phosphatidic acid (PA) is required for fission of post-Golgi carriers [57, 61, 82].
Based on these data, we have investigated the effect of Comp.11 binding to CtBP1/BARS on the assembly and function of the above complex. We first found that Comp.11 inhibits the ability of His-CtBP1/BARS to bind to GST-tagged 14-3-3γ (Fig. 3D) and to Flag-LPAATδ (immunopurified from lysates of HeLa cells transiently-transfected with Flag-tagged LPAATδ, Fig. 3E). Then, on the same line of evidence, CtBP1/BARS bound to Comp.11 was not able to activate the enzymatic activity of LPAATδ and hence the production of PA (Fig. 3F). To this end we performed an in vitro LPAATδ acyltransferase assay [as described in [61]] where extract from Flag-LPAATδ expressing A375MM melanoma cells were incubated with the acyl donor [1-14C]-oleoyl-CoA and the acyl acceptor oleoyl-LPA, with [1-14C]-PA measured as the reaction product (Fig. 3F). The 45% increase in LPAATδ activity in extract from LPAATδ overexpressing cells over the empty Flag-vector transfected cells was completely hindered by Comp.11 (Fig. 3F) which confirms its inhibitory effects on assembly and function of the CtBP1/BARS-mediated fission machinery. Extracts from LPAATδ- and CtBP1/BARS-depleted cells, where LPAATδ is inactive [as described in [61]] were used as internal controls of the LPAATδ acyltransferase assay. As a control of specificity, Comp.11 treatment did not affect the cellular levels of LPAATδ (Supplementary Fig. 5).
Collectively, these results indicate that Comp.11 after binding to CtBP1/BARS induces a oligomerization change that inhibits the ability of CtBP1/BARS to bind both its transcription and membrane fission partners.
Comp. 11 impairs the CtBP1/BARS-controlled protein transport and cell entry into mitosis.
We have previously reported that CtBP1/BARS localizes at the Golgi complex where it controls membrane fission required to support: i) the export of specific class of basolateral cargoes [e.g., human growth hormone (hGH) and vesicular stomatitis virus G protein (VSVG)]; and ii) the Golgi ribbon unlinking during G2/M transition and hence the cell entry into mitosis [56, 61].
To understand whether Comp.11 has an impact on these CtBP1/BARS-controlled cellular functions, we firstly investigated whether Comp.11 affects the export of the basolateral cargo VSVG from the trans-Golgi Network (TGN) to the plasma membrane (PM) using a well characterized transport assay which relies on the thermosensitive mutant protein ts045 from VSV [57]. Briefly, melanoma A375MM cells were infected with VSV and incubated at 40°C to first accumulate the protein in the endoplasmic reticulum (ER) and then shifted to 20°C, a temperature at which cargo proteins exit the ER and reach, but cannot exit, the TGN. The temperature was finally shifted to 32°C, and the formation of VSVG-containing carriers from the TGN was visualized by immunofluorescence and quantified by preventing the fusion of these carriers with the PM with tannic acid [61], resulting in accumulation of carriers close to the cell surface. Comp.11 induced a strong reduction of the formation of the VSVG-positive post-Golgi carriers (Fig. 4A and 4C). Similar results were observed in CtBP1/BARS depleted cells; conversely CtBP2 knockdown does not affect VSVG carrier formation (Fig. 4A and 4C). Of note, Comp.11-treated cells, as well as CtBP1/BARS depleted cells, showed several long (> 10 nm) tubular extensions that contained VSVG. These tubules represent carrier precursors that elongate out of the Golgi but are unable to undergo fission and detach to form mature transport carrier intermediates (Fig. 4A). This fission-defect phenotype resembles that induced by depletion and inhibition of LPAATδ, 14-3-3γ and other components of the CtBP1/BARS-fission machinery [25, 57, 61].
Along this line, we analyzed a soluble basolateral cargo, the stably expressed constitutively secreted GFP-tagged variant of the human growth hormone (hGH) that is retained in the ER and synchronously released in a temperature-independent fashion [83]. The treatment with Comp.11, as well as the depletion of CtBP1/BARS (but not of CtBP2) strongly inhibited export of hGH-FM–GFP from the Golgi to the PM with a similar fission-defect phenotype seen for VSVG cargo (Fig. 4B and 4D).
These findings indicate that Comp.11 after binding to CtBP1/BARS compromises the fission of basolateral-directed tubular carriers exiting the Golgi complex, and, in turn, blocks the transport of VSVG and hGH to the PM. Of note, Comp.11 treatments had no effect on CtBP1/BARS and CtBP2 protein levels (Supplementary Fig. 6A and 6B), as well as on cell viability, growth and morphology for the duration of the traffic-pulse experiments and longer.
Finally, we investigated the effects of Comp.11 in G2-blocked cells and monitored the effect of this treatment on mitotic entry, which depends tightly on Golgi fragmentation [84]. To this purpose we used HeLa and NRK cells, two well-known cell systems, to synchronize cells at G2/M boundary by the double-thymidine S-phase block. The cells, 4 hours after thymidine washout, were incubated with Comp.11 and then fixed at various times to determine the mitotic index (see Methods). As shown in Fig. 4E there is a strong impairment (by 75%), with no delay, of entry into mitosis in both Comp.11-treated cell lines. These data indicate that Comp.11 inhibits the CtBP1/BARS-mediated cleavage of Golgi ribbon that occurs during the G2 phase of the cell cycle.
The conclusion that can be driven from the above data is that Comp.11 interfering with the assembly of the CtBP1/BARS-fission machinery compromises the cellular function of this protein complex, which results: i) in an impaired transport of basolateral cargoes to the PM, and, ii) in a block of mitotic Golgi fragmentation, a step that controls cell entry into mitosis.
Comp.11 inhibits cell proliferation by inducing G0/G1 phase cell cycle arrest in melanoma cells
Having established the more pronounced expression levels of CtBP1/BARS in the two A375MM and B16F10 melanoma cell lines (Supplementary Fig. 1A), we investigated the cytotoxic effects of Comp.11 on these cells. The cells were treated with different concentrations of Comp.11 up to 150 µM for 24 h, 48 h and 72 h and the cell viability was measured by the MTT assay (Supplementary Fig. 7). As shown in Fig. 5A and 5E, the addition of Comp.11 results in a dose-dependent loss of viability in both cell lines, with EC50 upon 24 h treatment of 23.71 µM on A375MM and of 19.36 µM on B16F10 cells (see also Supplementary Fig. 7). Accordingly, all additional in vitro experiments were performed with 15 µM of Comp.11 treatment.
To further investigate whether Comp.11 inhibited cancer cell proliferation by inducing cell cycle arrest, we treated the two melanoma cell lines with this compound and examined the cell cycle distribution by flow cytometry (see Methods). Figure 5B, 5C and Fig. 5F, 5G indicate an increase in the mean percentage of cells in the G0/G1 phase of the cell cycle from 69.1% and 66.5% (vehicle alone) to 84.05% and 82.85% (Comp.11), respectively. G0/G1 phase cell cycle arrest was accompanied by a decrease in the percentage of cells in the G2/M (as found by thymidine cell-cycle synchronization assays, see Fig. 4E) and S phases (Fig. 5B, 5C and Fig. 5F, 5G). Similar results were observed in the cell cycle distribution of CtBP1/BARS-depleted but not in CtBP2-depleted cells (Fig. 5B, 5C and Fig. 5F, 5G; see also Supplementary Fig. 8A) indicating that Comp.11 inhibits cell proliferation through its specific interaction with CtBP1/BARS. Indeed, Comp.11 does not bind CtBP2 as validated via ITC experiments (Supplementary Fig. 9).
The transcriptional activity of CtBP1/BARS controls several genes involved in cell proliferation and tumor growth, including p16INK4a, p14ARF, Ciclin D1 and p21 [19, 21].
To define the mechanism of the anti-proliferative effects of Comp.11, we explored whether this compound does interfere with the expression of the above cell cycle-related genes by real-time PCR. As shown in Fig. 5D, p16INK4a, p14ARF, Ciclin D1 and p21 exhibited a significant increase in their expression levels upon Comp.11 treatment, in A375MM cells. Similar results were obtained in CtBP1/BARS-depleted cells but not in CtBP2-depleted cells (as a demonstration of on-target specificity of Comp.11 action on CtBP1/BARS functions) (Fig. 5D). In parallel, upon Comp.11 treatment, increased expression levels of Ciclin D1 and p21 were evaluated in murine B16F10 cells (Fig. 5H), which express a level of CtBP1/BARS protein comparable to that in A375MM cells (Supplementary Fig. 1A).
Altogether, these results demonstrate that binding of Comp.11 to CtBP1/BARS abrogates its transcriptional activity and inhibits cell proliferation by inducing cell cycle arrest in the G0/G1 and G2/M phases.
Comp.11 induces apoptosis in melanoma cells
Flow cytometry analysis by annexin V/PI staining was performed in order to investigate the induction of apoptosis in melanoma cells by Comp.11. The percentages of viable, early apoptotic, late apoptotic, and necrotic cells after 24 h of treatment with 15 µM of Comp.11 are shown in Fig. 6A. Significant differences were observed between control and treated cells: 7.6% of total percentage of apoptotic cells in vehicle-treated A375MM cells versus 28.7% in Comp.11-treated cells. Similar effects of Comp.11 treatments were observed in CtBP1/BARS-depleted cells but not in CtBP2-depleted cells (Fig. 6A and 6B and Supplementary Fig. 8B).
To understand the mechanism of apoptosis induced by Comp.11, we examined the expression levels of p53 and PTEN tumor suppressors genes, which are both reported to inhibit cell cycle progression and promote apoptosis under the transcriptional control of CtBP1/BARS [20]. The treatment of A375MM with Comp.11, or with CtBP1/BARS siRNAs, resulted in a marked increase in PTEN (7-fold) and in p53 (3-fold) mRNA levels (Fig. 6C) compared with that in vehicle- or non-targeting-treated cells. A negligible effect in p53 mRNA level was observed in CtBP2 depleted cells (Fig. 6C).
CtBP1/BARS is also involved in genome instability through its transcriptional regulation of BRCA1 gene, which dampens DNA-damage repair in melanoma. Indeed, in melanoma patients, the increased expression level of CtBP1/BARS correlates with decreased expression and function of BRCA1, and this contribute to genome instability and melanoma initiation [19]. We found that upon CtBP1/BARS depletion the expression levels of BRCA1 increased (Fig. 6D). In light of this finding, we investigated the possibility that by blocking CtBP1/BARS functions with Comp.11, we could prevent melanoma progression by increasing the BRCA1/BRIP1-controlled DNA-Damage Response Pathway [85]. As shown in Fig. 6D, both the mRNA levels of BRCA1 and BRIP1 increased 4-fold in Comp.11- and in CtBP1/BARS siRNAs-treated cells compared to the vehicle- or non-targeting-treated A375MM cells. Similar data were observed in murine B16F10 melanoma cells upon the above treatments (Fig. 6E-H).
In conclusion, these observations indicate that Comp.11 treatment is able to reverse the CtBP1/BARS-mediated transcriptional repression of PTEN, p53, BRCA1 and BRIP1 genes, activating apoptosis and reducing melanoma initiation.
Comp.11 impairs motility and invasion of melanoma cells by affecting the CtBP1/BARS- mediated transcription of EMT-related genes.
A key pro-oncogenic function in which CtBP1/BARS has been implicated is the ability to promote cancer cell migration and invasion [21, 86], which is related to CtBP1/BARS role in induction of EMT and metastasis [22]. Indeed, CtBP1/BARS expression and activity has been found to be upregulated in metastatic cancer types [87] where this protein repressed epithelial marker genes such as E-cadherin (CDH1), plakoglobin, desmoglein-2, occludin [35, 36, 51], beta-catenin [88] and increased the expression of mesenchymal marker genes including vimentin, N-cadherin, Snail [89] and versican [VCAN [86]].
We have thus evaluated whether Comp.11 can affect melanoma cell migration and invasion by reverting the transcription of these CtBP1/BARS-controlled EMT-related genes. Total mRNA extraction and real-time PCR determination were applied on A375MM and B16F10 cells, both recognized to possess strong migratory and invasive abilities. As shown in Fig. 7A and 7B, the addition of Comp.11 increased by several folds the epithelial JAM-1, E-cadherin, beta-catenin, Zona occludens 1 (ZO-1), occludin, desmoglein-2 (DSG2) and plakoglobin genes, and conversely impaired the induction of mesenchymal N-cadherin, vimentin and VCAN genes compared to vehicle-treated cells. Similar results were observed in both A375MM and B16F10 cells, upon depletion of CtBP1/BARS but not of CtBP2 (see also Fig. 7C and 7D). Western blot analysis of cell lysates treated with Comp.11 (or with CtBP1/BARS siRNAs) showed a reduction of the well-known signaling proteins associated with melanoma development and progression: NRAS and Akt [identified as CtBP repression target, [36]], and NF-KB and STAT3 [90] (Fig. 7C and 7D).
Altogether, these data indicate that Comp.11 is capable of reversing the CtBP1/BARS-mediated transcriptional activity and this could be explained by alteration in the oligomerization state of CtBP1/BARS bound to Comp.11, which, in turn, is unable to form the active CtBP1/BARS transcriptional complex [22].
Based on the above results, we analyzed the effects of these gene alteration upon Comp.11 treatment in cell migration and invasion by wound healing and transwell matrigel invasion assays. Both A375MM and B16F10 cells, exposed to 15 µM of Comp.11 or to CtBP1/BARS depletion (see Methods), showed a significantly reduced cell migration ability compared with the controls (Fig. 8A and 8D and Supplementary Fig. 10). Quantification performed at 16 h after scratching showed that the wound closure was reduced by 60% compared to the controls (Fig. 8B and 8E). The transwell invasion assays also showed that Comp.11 or CtBP1/BARS siRNAs treatment inhibited cell invasion (Fig. 8C and 8F and Supplementary Fig. 10), indicating an inhibitory role of CtBP1/BARS bound to Comp.11 on melanoma cell invasion.
Overall, our data indicate that Comp.11 directly binds to CtBP1/BARS and thus inhibits the transcriptional complex function of CtBP1/BARS shifting gene expression patterns from mesenchymal to a more epithelial phenotype, leading to an impairment of melanoma cell migration and invasion.
Comp.11 triggers reduction in colony formation and in vivo primary tumor growth.
We decided to explore further the anti-tumor effects of Comp.11 in anchorage-dependent and anchorage-independent (in soft agar) colony formation assays. The A375MM cell line was particularly sensitive, and exhibited a significant decrease in colony-forming potential (see Methods) in response to treatment with Comp.11. As shown in Fig. 9A and 9B, the inhibition of colony formation was dose-dependent, with higher doses of 15 µM (consistent with the dose that induced apoptosis; see also Fig. 6), and indicating the ability of Comp.11 to interfere with melanoma cell growth.
In order to confirm in vivo the antitumor effect observed in vitro, we evaluated the effect of Comp.11 in a A375MM xenograft model (Fig. 9C-G). Specifically, A375MM cells were injected in the right flank of fifteen mice, when the tumors became palpable, the mice were randomly assigned to three experimental groups (n = 3) to receive Comp. 11 (10 mg/kg daily for 2 weeks or 20 mg/kg three times/week for 2 weeks) or its vehicle as schematized in Fig. 9C (see Methods). Comp 11 20 mg/kg produced a statistically significant tumor growth inhibition (p = 0.02) compared with control group, evaluated as % of means of the fold change of the tumour volume for each group (Fig. 9D) after 2 weeks of treatment, compared with the baseline. The waterfall plot of the fold-change in tumor volume compared with the baseline for each mouse was reported in Supplementary Fig. 11A.
A statistically significant tumor inhibition (p = 0.03) of Comp. 11 20 mg/kg treatment was confirmed by evaluating the weight of tumors collected at the end of study (day 14) as shown in Fig. 9E. Moreover, by calculating the percent change in tumor volume from the time of initial treatment (day 0) to day 7 or day 14 (end of the treatment) of the study, Comp.11 20 mg/kg treatment reduced the tumor burden by 3.3% and 39,39%, respectively, in spite of the other treated group (Comp.11 10 mg/kg) that reduce tumor burden of 0.68% and 18,4% respectively (Fig. 9F). The treatment was well tolerated by xenografted mice, as shown by the maintenance of body weight (Fig. 9G) and by the absence of other signs of acute or delayed toxicity.
Next, we confirmed, a statistically significant inhibition of tumor volume after Comp.11 20 mg/kg treatment respect to vehicle (P = 0.04) by the High Frequency Ultrasound (HFUS) (Supplementary Fig. 11B).
Altogether, these data suggest that Comp.11, with the best results at the dosage of 20 mg/kg, once bound to CtBP1/BARS, inhibits growth of melanoma tumor in xenograft models, indicating its potential effective antitumor activity.