Novel Acylguanidine Derivatives Targeting Smoothened Induce Antiproliferative and Pro-Apoptotic Effects in Chronic Myeloid Leukemia Cells

The most relevant therapeutic approaches to treat CML rely on the administration of tyrosine kinase inhibitors (TKIs) like Imatinib, which are able to counteract the activity of Bcr-Abl protein increasing patient’s life expectancy and survival. Unfortunately, there are some issues TKIs are not able to address; first of all TKIs are not so effective in increasing survival of patients in blast crisis, second they are not able to eradicate leukemic stem cells (LSC) which represent the major cause of disease relapse, and third patients often develop resistance to TKIs due to mutations in the drug binding site. For all these reasons it’s of primary interest to find alternative strategies to treat CML. Literature shows that Hedgehog signaling pathway is involved in LSC maintenance, and pharmacological inhibition of Smoothened (SMO), one of the key molecules of the pathway, has been demonstrated to reduce Bcr-Abl positive bone marrow cells and LSC. Consequently, targeting SMO could be a promising way to develop a new treatment strategy for CML overcoming the limitations of current therapies. In our work we have tested some compounds able to inhibit SMO, and among them MRT92 appears to be a very potent SMO antagonist. We found that almost all our compounds were able to reduce Gli1 protein levels in K-562 and in KU-812 CML cell lines. Furthermore, they were also able to increase Gli1 and SMO RNA levels, and to reduce cell proliferation and induce apoptosis/autophagy in both the tested cell lines. Finally, we demonstrated that our compounds were able to modulate the expression of some miRNAs related to Hedgehog pathway such as miR-324-5p and miR-326. Being Hedgehog pathway deeply implicated in the mechanisms of CML we may conclude that it could be a good therapeutic target for CML and our compounds seem to be promising antagonists of such pathway.


Introduction
in an intracellular vescicle [29]. Upon binding of Hh ligands to Ptch receptor, such inhibition is released and the activation of SMO, followed by its migration on primary cilium cell membrane, leads to pathway activation which culminates in a signal transduction cascade. These events cause the nuclear translocation of the Gli family of transcription factors (Gli1-3), and the subsequent activation or inhibition of various cell cycle, proliferation and survival regulating genes such as the D-type cyclins, c-Myc and Bcl-2 [30][31][32][33][34]. As part of a feedback mechanism, Gli target genes also comprise members of the Hh pathway, such as Gli1 and Ptch1 [35,36]. Among the inhibitors of the Hh pathway that interact directly with SMO, vismodegib and sonidegib are the two compounds approved by FDA in 2012 and 2015, respectively, for the treatment of basal cell carcinoma (BCC). Other hedgehog inhibitor compounds have been studied for BCC, like CUR61414 [37], but though they showed a good activity both in vitro and on mice, they failed the clinical phase I studies [38].
On the other hand, preclinical studies are checking the possibility to induce apoptosis in BC cells by treatment with various drugs, alone or in combination. As an example, inhibition of Mek and farnesyl transferase [39], or treatment with a dual Bcr-Abl/Jak2 inhibitor [40], as well as p53 stabilization [18] induce apoptosis and death in human BC CML K-562 cells. Importantly, this cell line expresses all the Hh signaling molecules, including sonic Hh (Shh), Ptch, SMO and Gli1 [41].
Taken together, these results suggest that small molecules able to inactivate the Hh pathway by blocking SMO could be in principle useful either to inhibit BC cell proliferation or, at the same time, to deplete population of LSC whose survival, self-renewal, and expansion is strongly dependent on the Hh pathway [27]. Moreover, a combination therapy comprised of the currently available Bcr-Abl inhibitors and small molecules able to block SMO could represent a very promising and effective tool to deplete CML cells, overcome chemotherapy resistance, eradicate LSC, and thus potentially cure CML.
MicroRNAs (miRNAs) are a class of small non-coding cellular RNAs that are responsible for messenger RNA translational inhibition or degradation. In several human cancers, a downregulated miRNA signature with high Hh signaling does exist. Downregulation of these miRNAs allows high levels of expression of Hh-dependent genes leading to tumour cell proliferation. As an example, miR-324-5p was shown to target the activator components of the Hh pathway, SMO and Gli1, thereby suppressing progenitor and tumour cell growth [42]. Moreover, it has been demonstrated that upregulation of SMO is associated with reduced expression of miRNA-326 [43].
Within a drug design project aimed at identifying new small molecules acting as Hh inhibitors, we have recently found a class of compounds with a very impressive ability to inhibit SMO by direct interaction [44,45]. In particular, in previous studies, compound MRT-83, belonging to the chemical class of acylguanidines substituted with a phenyl group, appeared to be one of the most potent SMO antagonists known so far [46]. On this basis, we planned to design new MRT-83 analogues and to check their ability to block growth and proliferation of two human BC CML cell lines (namely, K-562 and KU-812) and among the tested compounds we found that the one which possessed the phenylethyl terminal group (MRT92) was particularly active toward Hh pathway.

Specificity for Hh pathway
Western blotting analysis was applied to evaluate the ability of the compounds to affect Gli1 and suppressor of fused (SuFu) protein expression. Compounds effects were compared with the activity of CUR61414 used as reference compound and AT43 (a known SMO agonist) [37].  (Fig 1b and 1d) cells with the compounds MRTX, MRT94, MRT92, MRT83, at 20 μM for 24 h, significantly reduced Gli1 protein concentration in comparison with non-treated control. A similar decrease of Gli1 protein was found by treatment with CUR61414. We were not able to determine any significant effect of MRTY on both cell lines. While on the contrary, as expected, AT43 was able to significantly increase Gli1 expression. These results showed a negative modulation of Hh pathway by our SMO inhibitors through the reduction of Gli1 protein expression. We also investigated the effect of our compounds on SuFu which is known to be a regulator of Gli proteins counteracting Spop activity, thus preventing Gli factors degradation [47]. On K-562 cells our compounds were not able to induce any significant modification of SuFu and even AT43, despite inducing a significant increase of Gli1, did not produce an increase in SuFu. On the contrary, on KU-812 cells the compounds MRT92 and CUR61414 were able to significantly reduce the amount of expressed protein, while AT43 induced a significant increase that correlated with Gli1 increase. The reduction of SuFu, even if not always significant, is a further confirmation of the ability of tested compounds to interact with Hh pathway. Given the specificity of these compounds to target Hh pathway, their effects on RNA expression were evaluated for the most important pathway components (i.e., Gli1 and SMO). As expected, no significant changes of the Gli1 and SMO RNA expression levels were found in both cell lines after a 3 h incubation with the studied compounds (data not shown). On the contrary, Gli1 RNA expression was increased in K-562 cells by a 24 h treatment with 10 μM inhibitor (Fig 2a), with the sole exception of MRT83. This enhanced expression could be considered as a compensatory effect that balances Gli1 protein level reduction consequent to pathway blockade. Only MRTX and MRTY showed a residual effect after 72 h (Fig 2b). SMO RNA levels were unchanged by treatment of K-562 cells for 24 (Fig 2c) and 72 h (Fig 2d), probably because the compounds act only by binding the protein and preventing its ciliar translocation, without effectively reducing the cellular amount of SMO. In this way, no compensatory mechanisms are required to be activated by the cell.
On the other hand, KU-812 cells responded to the studied compounds in a different way. In fact, only 10 μM MRTY induced an increase of Gli1 RNA expression after 24 h treatment ( Fig  3a) and its effect was maintained at 72 h (Fig 3b). Increasing compound concentrations to 20 and 50 μM led MRTX and MRT92 (in addition to MRTY) to produce an effect after 24 h (Fig  3c), not maintained at 72 h (Fig 3d). Compounds MRT94 and MRT83 were tested only at 50 μM since they did not show any significant activity even at this concentration. These results suggest that KU-812 cells are less sensitive to the studied compounds in comparison to K-562 cells. Consequently, higher compound concentrations were required to determine a biological effect.
A compound concentration of 10 μM did not affect SMO RNA expression after a 24 h treatment (Fig 4a), while increased SMO RNA levels were found after a 72 h treatment with MRT94 and MRT83 (Fig 4b). Increased concentrations (50 μM) of MRT94-Y led to higher SMO RNA expression levels after 24 h (Fig 4c), while all tested compounds were able to increase SMO RNA expression after a 72 h treatment (Fig 4d). In summary, differently from what found in K-562 cells, an increase of SMO RNA expression was found in KU-812 cells after a 10 μM treatment for 72 h, and after a 50 μM treatment after 24 and 72 h. These results suggest that a 10 μM concentration is too low to induce a rapid effect (within 24 h), while higher concentrations (50 μM) exhibit their effects already after 24 h and maintain them for 72 h at least.

Effects on miRNAs expression
To confirm the specificity of our compounds towards Hh pathway, we investigated the effects they exerted on two miRNAs closely related to the Hedgehog pathway: miRNA-324-5p and miRNA-326. They both suppress the pathway activator SMO, and miR-324-5p also regulates the downstream transcription factor Gli1 [42]. For this analysis, the two compounds with the best activity for both cell lines have been selected. Both K-562 and KU-812 cells were treated with 25 μM MRTX and MRT92 for 24 h. Results show a significant increase of miR-324-5p in both cell lines and the greatest effect toward KU-812 cells was found after 3 h of treatment (Fig  5a and 5c). In addition, MRT92 at 25 μM after 24h of treatment was able to increase miR-324-5p but did not reach significativity. These results are in agreement with the hypothesis that compounds negatively regulate SMO and inactivate the Hh pathway by the reduction of Gli1, thus inducing an increase of miRNA324-5p level. According to previous results showing that SMO expression was not affected in K-562 cells, miR-326 expression did not change in the same cell line (Fig 5b). On the contrary, a significant increase of its expression was found in MRT92-treated KU-812 cells (Fig 5d).
It is noteworthy that changes of the miRNAs levels are in agreement with Gli1 and SMO RNA variations in both tested cell lines. In fact, the levels of both miRNA-324-5p and its target (Gli1) increase after a 24 h treatment in the tested cell lines; the amounts of miRNA-326 and its target (SMO) remain both unaltered in K-562 cells and increase in KU-812 cells with the only exception of miRNA-326 in the sample treated with compound MRTX. This parallel level of expression of the miRNAs and the respective target genes is due to the activation of a control mechanism which promotes miRNAs increase to prevent the uncontrolled production of the target gene RNA.

Antiproliferative activity
Ability of the new compounds and CUR61414 to affect viability of K-562 and KU-812 cell lines (Fig 6) was checked by resazurin proliferation assay. The IC 50 values obtained for each compound are listed in Table 1. A concentration-dependent antiproliferative activity of compounds comparable to or better than that of CUR61414 was found toward both cell lines. In particular, MRT92 and MRTY showed IC 50 values lower than 10 μM in K-562 cells (7.7 and 7.8 μM, respectively, versus 14 μM found for CUR61414). IC 50 values for MRTX and MRT92 toward KU-812 cells were respectively 5.5 and 7.2 μM (27 μM for CUR61414). The remaining compounds showed a two-digit micro molar IC 50 .

Pro-apoptotic activity and autophagy
Prompted by their ability to reduce CML cell line proliferation, the new compounds were also checked for pro-apoptotic activity versus poly-ADP-ribose-polymerase (PARP). We used a single representative compound concentration of 10 μM. Immunoblot analysis of uncleaved and cleaved PARP indicated that a significant PARP cleavage did not occur in K-562 cells after 72 h of treatment except for MRTX treated cells (Fig 7a), while tested compounds, with the exception of MRT94, led to an enhancement of the cleaved PARP in KU-812-cells that were potently induced to apoptosis after 72 h of treatment (Fig 7b).
Moreover, the expression of Bax and Bcl-2 RNA levels was also investigated. In fact, the ratio between Bax and Bcl-2 RNA expression is a critical determinant to induce cells toward apoptosis and represents a direct index of the induction of the apoptotic process [48], thus helping in exploring the apoptosis induction. In agreement with results of the PARP assay, no significant pro-apoptotic effect was measured in K-562 cells after 72h of treatment except for samples treated with MRTY (Fig 8a), while MRTX, MRT94, and MRT92 were able to induce apoptosis in KU-812 cells after 72h (Fig 8b).
We further evaluated the expression of BNIP3, a protein that has been shown to be correlated to autophagy [49]. Seventy-two hours exposure to compounds lead to an increase in BNIP3 expression, particularly this was significant in samples treated with MRT94, MRT92,  and MRTY in K-562 cells (Fig 9a). This result is in agreement with previous findings [49] that show an induction of autophagy in Bcr-Abl-positive CML cells by inhibition of Hh pathway. Differently from K-562 cells, BNIP3 expression in KU-812 cells was not significantly increased (Fig 9b), on the contrary the level of BNIP3 was reduced by the same compounds that elicited apoptosis.
Even if it is reported in literature [50] that the Hh pathway blockade produce an apoptotic response in K-562 cell line, this is not accordant with our results. But since we know that an increase of BNIP3 indicate an early cell damage which most likely will lead to an apoptotic response in longer time frames the differences are probably only due to experimental detection timing.
The reported results demonstrated that some of the tested compounds were able to induce autophagy in K-562 cell line mediated, as shown, by an increase of BNIP3 RNA levels. MRTX and MRT92 induced apoptosis on KU-812 cells as shown by the increase of cleaved PARP and the ratio of Bax/Bcl2. MRT94 was able to significantly increase the ratio of Bax/Bcl2 but did not show any increase in PARP cleavage; probably this compound may require a longer time to fully activate the caspase cascade.

Proliferation comparison on K-562 cells
We evaluated the inhibition of proliferation on K-562 cells after either Gli1 gene silencing or treatment with compounds that have proven to be able to reduce Gli1. For this experiment, we  chose MRTX and MRT92. Gli1 siRNA was inserted by electroporation in K-562 cells and this elicited a reduction on Gli1 expression (Fig 10a). Blocking Hh pathway by Gli1 gene silencing led to a significant reduction on cells viability that was comparable with the reduction of viability in non-silenced K-562 cells induced by MRTX and MRT92 (Fig 10b). In both cases inhibition of cell proliferation was about 90%. It is of particular interest that pathway blockade with our compounds showed the same inhibition of proliferation of biological pathway blockade through Gli1 gene silencing. The current treatment for CML is based on TKIs [6,7,11]. Despite their efficacy, TKIs present several limitations as their inability to improve survival in patients in BC [19], the development of resistance [19] and their inability to kill LSC which represent the reservoir of the disease and the major cause of relapse [20,21].
Given the relationship between Hh-SMO pathway activation and CML progression from LSC to BC, combined with the ability of our compounds either to block SMO or to inhibit CML cell growth and proliferation, our compounds seem to be particularly suitable for a promising therapeutic approach toward CML. A combination therapy comprised of the currently available Bcr-Abl inhibitors (such as ponatinib that is also able to target imatinib-resistant cells) and new small molecules that are able to block SMO could represent a very promising and effective tool to deplete CML cells also in blast crisis, overcome chemotherapy resistance, and eradicate LSC, as already reported in some literature for vismodegib and ponatinib [51].
Among our compounds the one with phenylethyl terminal group (MRT92) appears to be very specific towards Hh pathway as it strongly decreases Gli1 protein expression and modulates Gli1 and SMO RNA levels and miRNAs in both tested cell lines. Furthermore it demonstrated an impressive ability to inhibit proliferation in both tested cell lines with IC 50 values far below 10 μM, it induced apoptosis in KU-812 and seems to provoke autophagy in K-562 cell line. In conclusion our study has proven that MRT92 is certainly a promising therapeutic compound, and the best candidate for further experimental investigations.

Experimental Section
Synthesis of SMO antagonists is described in S1 Appendix and illustrated in S1 Fig of Supporting Information.

Cell lines and treatments
Human CML K-562 cells (American Type Culture Collection) in blast crisis and human CML KU-812 cell line (American Type Culture Collection) in myeloid blast crisis, both expressing Hh signaling pathway and carrying Philadelphia chromosome, were employed for biological assays. Cell lines were cultured in RPMI 1640 medium (Euroclone, Devon, UK), supplemented with 10% or 20% FCS, respectively, 1% L-glutamine 2 mM, streptomycin 100 μg/ml, and penicillin 100 U/mL (Euroclone, Devon, UK), and were maintained in a humidified atmosphere at 37°C and 5% CO 2 .

Proliferation assay
Cell proliferation was evaluated by resazurin fluorescent method. Cells were starved overnight with RPMI 1640 culture medium supplemented with 0.5% FCS, 1% L-glutamine, and antibiotics (100 μg/mL streptomycin and 100 U/ml penicillin), and maintained in a humidified atmosphere at 37°C and 5% CO 2 .
Later, the medium was removed and the culture was refreshed with new medium at the usual concentration of FCS. Cells were plated at a concentration of 10 5 cells/well in a 96 multiwell plate. Then, scalar concentrations of each compound ranging from 0.5 μM to 150 μM or a fixed concentration of compound (10 μM), or no compounds were added to the cells and the plate was incubated in a humidified atmosphere at 37°C and 5% CO 2 for 72 h. Six hours before the end of incubation, resazurin was added at a final concentration of 320 μM and fluorescence was evaluated by fluorimetric analysis employing FLUOstar OPTIMA plate reader (BMG LABTECH, Offenburg, Germany) at an excitation wavelength of 530 nm and emission wavelength of 590 nm.

RNA isolation and quantitative real time PCR
To determine Gli1 and SMO expression, K-562 and KU-812 cells were starved overnight with RPMI 1640 culture medium supplemented with 0.5% FCS, 1% L-glutamine, and antibiotics (100 μg/mL streptomycin and 100U/mL penicillin), and maintained in a humidified atmosphere at 37°C and 5% CO 2 . Later, the medium was removed and the culture was refreshed with new medium at the usual concentration of FCS. Cells were plated at a concentration of 3.5 x 10 5 cells/ml and added with each compound (10 or 50 μM) for 24 or 72 h.
Total RNA isolation was performed by cell lysis with TRI-Reagent (Ambion, Foster City, USA) by taking the upper aqueous phase obtained after centrifugation at 1000g for 10 min. RNA was then washed in isopropanol and cool 75% ethanol, resuspended in nuclease-free water and kept at -20°C for further analysis.
cDNA from total RNA extracted in TRI-Reagent was then synthesized using the iScript™ cDNA Sinthesis Kit (Bio-Rad Laboratories, Hercules, USA) and qRT-PCR analysis of Bax, Bcl-2, BNIP3, SMO and Gli1 RNA expression was performed on cDNAs by using iQ™ SYBR Green Supermix (Bio-Rad Laboratories, Hercules, USA).
Primer were designed using Primer3 [52,53] and purchased from Invitrogen (Carlsbad, USA), sequences are reported in Table 2. Data were analyzed with iQ™ 5 Optical System Software, Security Edition (Bio-Rad Laboratories, Hercules, USA). All values were normalized to β-actin endogenous control and RNA relative expression was measured using the 2 -ΔΔCt method.
cDNA from RNA isolated by miRCURY™ RNA Isolation Kit-cell & plant was synthesized using miRCURY™ LNA Universal RT microRNA PCR (EXIQON, Vedbaek, Denmark) according to manufacturer's instruction and qRT-PCR analysis of Gli1, SMO, miR-324-5p, miR-326 RNA expression was performed on cDNAs by using ExiLENT SYBR 1 Green master mix (EXI-QON, Vedbaek, Denmark) and MicroRNA LNA™ PCR primers. Data were analyzed with iQ™ 5 Optical System Software, Security Edition. All values were normalized to non-coding RNA U6, and RNA spike-ins were used as controls for isolation, cDNA synthesis and PCR. RNA relative expression was measured using the 2 -ΔΔCt method.

Protein expression
To assess Gli1, SuFu and ß-actin protein production and PARP cleavage, K-562 and KU-812 cells were starved overnight with RPMI 1640 culture medium supplemented with 0.5% FCS, 1% L-glutamine and antibiotics (100 μg/ml streptomycin and 100 U/ml penicillin), and maintained in a humidified atmosphere at 37°C and 5% CO 2 . Later, the medium was removed and the culture was refreshed with new medium with 10% FCS. Cells were plated at a concentration of 3.5 x 10 5 cells/mL and added with each compound (10 or 20 μM) for 24 or 72 h. Later, cells were harvested and lysed in an appropriate buffer containing 1% Triton X-100 and protease inhibitors. Proteins were quantitated by the BCA method (Pierce, Rockford, USA). Equal amounts of total cellular protein were resolved by SDS-polyacrylamide gel electrophoresis, with 10% acrylamide for PARP and 8% for Gli1 and SuFu. Blotted proteins were transferred by electroblotting to a PVDF membrane (Hoefer Pharmacia Biotech, San Francisco, USA) for 1 h at 100 v and 4°C. After a saturation step of 1 h with a solution of 5% nonfat dry milk and 0.1% TBST 10X in agitation at room temperature, anti-PARP, anti-β-actin, anti-Gli1 or anti-SuFu (Cell Signaling Technology, Boston, USA) antibodies were added to the PVDF membrane according to manufacturer's instruction. On the day after, incubation with HRP-linked secondary antibodies was carried out for 1 h in agitation at room temperature and then HRP substrate was added (Bio-Rad Laboratories, Hercules, USA). Nonsaturated, immunoreactive bands were detected with a CCD camera gel documentation system (ChemiDocXRS, Bio-Rad Laboratories, Hercules, USA) and then quantitated with Image Lab ver.5.1 analysis software (Bio-Rad Laboratories, Hercules, USA). β-actin was used as loading control.

Gene silencing
Gli1 gene silencing on K-562 cells was performed by inserting into cells a siRNA (AUAUCUU GCCCGAAGCAGGUAGUGC) towards Gli1 or a control scrambled siRNA owning the same CG ratio, at a final concentration of 30 nM by means of electroporation. Briefly, cells were centrifuged at 200g for 10 min. at room temperature, washed with sterile PBS and centrifuged again. Then, cells were resuspended in resuspension buffer R (Invitrogen, Carlsbad, USA) at a final density of 1 x 10 7 cells/ml and siRNA towards Gli1 (or scrambled siRNA) was added at a final concentration of 30 nM. Then cells were electroporated using Neon™ Trasfection System (Invitrogen, Carlsbad, USA) and according to the following parameters: Pulse Voltage: 1350 v; Pulse Width: 10 ms; Pulse Number: 4; Cell Density: 3 x 10 7 . After electroporation, cells were plated on a 24-well plate at a concentration of 3 x 10 5 cells/well in a final volume of 500 μL RPMI supplemented with 1% glutamine without antibiotics and with 10% FCS and incubated for 24 h in a humidified atmosphere at 37°C, 5% CO 2 for further analysis.

Statistical analysis
Reported data are Mean ± SEM of at least three independent experiment performed in triplicate. The statistical analysis was performed by Student's t test using the Bonferoni correction for multiple test when appropriate. In all cases, only probability (p) values below 0.05 were considered significant.