Heterocycle-containing tranylcypromine derivatives endowed with high anti-LSD1 activity

Abstract As regioisomers/bioisosteres of 1a, a 4-phenylbenzamide tranylcypromine (TCP) derivative previously disclosed by us, we report here the synthesis and biological evaluation of some (hetero)arylbenzoylamino TCP derivatives 1b-6, in which the 4-phenyl moiety of 1a was shifted at the benzamide C3 position or replaced by 2- or 3-furyl, 2- or 3-thienyl, or 4-pyridyl group, all at the benzamide C4 or C3 position. In anti-LSD1-CoREST assay, all the meta derivatives were more effective than the para analogues, with the meta thienyl analogs 4b and 5b being the most potent (IC50 values = 0.015 and 0.005 μM) and the most selective over MAO-B (selectivity indexes: 24.4 and 164). When tested in U937 AML and prostate cancer LNCaP cells, selected compounds 1a,b, 2b, 3b, 4b, and 5a,b displayed cell growth arrest mainly in LNCaP cells. Western blot analyses showed increased levels of H3K4me2 and/or H3K9me2 confirming the involvement of LSD1 inhibition in these assays.


Introduction
Protein post-translational modifications (PTMs), such as ac(et)ylation, methylation, and phosphorylation, play an important role in regulating many protein functions. Among such PTMs, reversible methylation/demethylation of lysine residues is typically involved in regulating the chromatin environment and eukaryotic gene expression 1,2 . mono-and dimethylated Lys4, although its activity is weaker than that of LSD1 in vitro 8 .
LSD1 plays a vital role in a broad spectrum of biological processes, including embryonic development, stem cell maintenance and differentiation, and is also a key player in oncogenic processes, including cancer cell growth and metastasis. LSD1 has been reported to be overexpressed in a variety of cancers, and its inactivation or downregulation of its expression inhibits the development of cancer cells 2,9 . LSD1 regulates cellular signalling pathways in various cancerous entities such as breast cancer, where it interacts in a multiprotein complex containing deacetylase (NuRD) 10 . In prostate cancer, high levels of LSD1 are associated with cancer progression and metastasis and therefore LSD1 levels could be a useful biomarker 11,12 . LSD1 and LSD2 share a similar catalytic domain (45% sequence identity) that is structurally homologous with the amine oxidases, a class of flavin-dependent enzymes that act on biogenic amines 5,13 . Among these proteins, human monoamine oxidases (MAOs) A and B have been the subject of more than 50 years of research that has led to the development of a multitude of inhibitors including antidepressant and antiparkinson drugs 14 . Their similarity in the catalytic and structural properties prompted the investigation of anti-MAO drugs as potential LSD1 inhibitors 15 . Among them, tranylcypromine (TCP) is one of the most reliable molecules able to inhibit LSD1 irreversibly via a covalent interaction with the FAD cofactor and has been widely used as a fragment to create valuable LSD1 inhibitors 16 .
Based on these data, we designed and synthesised new analogs of 1a as potential inhibitors of LSD1, by shifting the 4-phenyl ring from the 4-to the 3-position of the benzoylamino moiety (1b) ( Figure 2) and replacing this 4-or 3-phenyl ring with different heterocyclic moieties such as furan, thiophene, and pyridine (2-6) ( Figure 2), aiming to improve their potency and LSD1 selectivity. Selected compounds were tested against U937 AML and prostate LNCaP cancer cells to assess their antiproliferative potential.

Chemistry
Melting points were determined on a Buchi 530 melting point apparatus and are uncorrected. 1 H and 13 C NMR spectra were recorded at 400 MHz and 100 MHz, respectively, on a Bruker AC 400 spectrometer; chemical shifts are reported in d (ppm) units relative to the internal reference tetramethylsilane (Me 4 Si). EIMS spectra were recorded with a Thermo MSQ Plus spectrometer;  only molecular ions (M þ ) and base peaks are given. All the compounds were routinely checked by TLC, 1 H NMR and 13 C NMR spectra. TLC was performed on aluminum-backed silica gel plates (Merck DC, Alufolien Kieselgel 60 F 254 ) with spots visualised by UV light. All solvents were reagent grade and, when necessary, were purified and dried by standard methods. The concentration of solutions after reactions and extractions involved the use of a rotary evaporator operating at reduced pressure of ca. 20 Torr. Organic solutions were dried over anhydrous sodium sulphate. Elemental analysis has been used to determine the purity of the described compounds, that is >95%. Analytical results are within ± 0.40% of the theoretical values. All chemicals were purchased from Sigma Aldrich, Milan (Italy), or from TCI Europe, Zwjindrecht (Belgium), and were of the highest purity.

Lsd1-CoREST binding and inhibition assays
Thermal stability, activity on histone H3K4me peptide, and inhibition of human LSD1-CoREST were measured using established protocols 17,28 .
The complex of human recombinant LSD1/CoREST protein was produced in E. coli as separate proteins and co-purified following previously reported procedures 13,29 . The experiments were performed in 96 well half area white plates (cat. 3693, Corning, Corning, NY) using a di-methylated H3K4 peptide containing 21 amino acids (custom synthesis done by Thermo Scientific) as substrate in 100 lL volume of 50 mM HEPES, pH 7.5. The peptide purity was >95% as checked by analytical high-pressure liquid chromatography and mass spectrometry. The demethylase activity was estimated under aerobic conditions at room temperature by measuring the release of H 2 O 2 produced during the catalytic process by the Amplex UltraRed detection system coupled with horseradish peroxidase (HRP). Briefly, 300 nM of LSD1/CoREST complex was incubated at room temperature for 10 min in the absence and/or the presence of various concentrations of the inhibitors, 50 mM Amplex UltraRed (Life Technologies) and 0.023 mM HRP (Sigma-Aldrich, Schnellendorf, Germany) in 50 mM HEPES pH 7.5 and 0.05 mg/mL BSA. The inhibitors were tested twice in duplicates at each concentration. Tranylcypromine (Sigma) was used as a control. After preincubation of the enzyme with the inhibitor, the reaction was initiated by adding 10 l of the dimethylated H3K4 peptide. The conversion of the Amplex UltraRed reagent to resurfin was monitored by fluorescence (excitation at 510 nm, emission at 595 nm). Arbitrary units were used to measure the level of H 2 O 2 produced in the absence and/or in the presence of inhibition. The maximum demethylase activity of LSD1/CoREST was obtained in the absence of inhibitors and corrected for background fluorescence in the absence of the substrate. The IC 50 values were calculated using GraphPad Prism version 4.0 (GraphPad Software, San Diego, CA).

Anti-MAO assays
The tested compounds were dissolved in DMSO (Sigma-Aldrich, Schnellendorf, Germany) to prepare 10 mM stock solutions which were kept for storage at À20 C. The percentage of DMSO used in the experiments was never higher than 1%. Clorgyline and selegiline, used as reference inhibitors, have been acquired from Sigma-Aldrich, Schnellendorf, Germany. Human recombinant MAO isoforms, used in the experiments, was purchased from Sigma-Aldrich. Resorufin sodium salt, p-tyramine hydrochloride, sodium phosphate buffer, horseradish peroxidase, and Amplex Red reagent have been supplied in the assay kit of Amplex Red MAO Molecular Probes (Molecular Probes Inc., Eugene, OR). Determination of MAO isoforms enzymatic activity: Briefly, 0.1 ml of sodium phosphate buffer (0.05 M, pH 7.4) containing different concentrations of the test drugs (new compounds or reference inhibitors) in various concentrations and adequate amounts of recombinant hMAO-A or hMAO-B required and adjusted to obtain in our experimental conditions the same reaction velocity, that is, to oxidise (in the control group) the same concentration of substrate: 165 pmol of p-tyramine/min (hMAO-A: 1.1 lg protein; specific activity: 150 nmol of p-tyramine oxidised to p-hydroxyphenylacetaldehyde/min/mg protein; hMAO-B: 7.5 lg protein; specific activity: 22 nmol of p-tyramine transformed/min/mg protein) were incubated for 15 min at 37 C in a flat-black-bottom 96-well microtest plate, placed in the dark fluorimeter chamber. After this incubation period, the reaction was started by adding (final concentrations) 200 lM Amplex Red reagent, 1 U/mL horseradish peroxidase, and 1 mM p-tyramine. The production of H 2 O 2 and, consequently, of resorufin was quantified at 37 C in a multidetection microplate fluorescence reader (FLX800, Bio-Tek Instruments Inc., Winooski, VT) based on the fluorescence generated (excitation, 545 nm, emission, 590 nm) over a 15 min period, in which the fluorescence increased linearly. Control experiments were carried out simultaneously by replacing the tested drugs (new compounds and reference inhibitors) with appropriate dilutions of the vehicles. In addition, the possible capacity of the above test drugs to modify the fluorescence generated in the reaction mixture due to non-enzymatic inhibition (e.g. for directly reacting with Amplex Red reagent) was determined by adding these drugs to solutions containing only the Amplex Red reagent in a sodium phosphate buffer. To determine the kinetic parameters of hMAO-A and hMAO-B (K m and V max ), the corresponding enzymatic activity of both isoforms was evaluated (under the experimental conditions described above) in the presence of a wide range of p-tyramine concentrations. The specific fluorescence emission (used to obtain the final results) was calculated after subtraction of the background activity, which was determined from wells containing all components except the hMAO isoforms, which were replaced by a sodium phosphate buffer solution. In our experimental conditions, this background activity was practically negligible. MAO activity of the test compounds and reference inhibitors is expressed as IC 50 , i.e. the concentration of each drug required to result in a 50% decrease in respect to the control value activity of the relative MAO isoforms. The corresponding IC 50 values were calculated by using the Origin 5.0 software (Microcal Software Inc., Northampton, MA).

Chemicals
The tested compounds have been dissolved in DMSO at 50 lM as stock concentration. All compounds have been tested at 10 lM and 50 lM as final concentrations for WB analysis. All compounds have been used at 0.1 lM, 5 lM, 10 lM, 25 lM, 50 lM as final concentration for proliferation assays. ORY-1001 (Selleck Catalog No. S7795) was used as reference compound at the concentration of 25 lM.

Histone extraction, Western blot analysis and protein extraction
After stimulation for 48 h with the compounds at 10 lM and 50 lM, ORY-1001 (Selleck Catalog No. S7795), commercially available LSD1 inhibitor, was used as positive controls. ORY-1001 was used at the final concentration of 25 lM. Cells were collected and washed 2 times with PBS then processed for histone extraction. Pellets were resuspended in triton extraction buffer [TEB; PBS containing 0.5% Triton X 100 (v/v), 2 mmol/L PMSF, 0.02% (w/v) NaN 3 ], and the lysis was performed for 10 min at 4 C. The samples were centrifuged at 2000Â g for 10 min at 4 C and pellets were washed in TEB (half volume). Samples were then resuspended in 0.2 N HCl, and acid histone extraction was carried out overnight at 4 C. The supernatants were recovered, and protein concentration was quantified by Bradford assay (Bio-Rad). For each sample, 4 lg of proteins were loaded on 15% polyacrylamide gels. The nitrocellulose filters were stained with Ponceau red (Sigma-Aldrich, Schnellendorf, Germany) as an additional control for equal loading. H3K4me2, H3K9me2 (Diagenode, Ougr ee, Belgium; pAB-035-050, pAb-060-050), and H4 (Cell Signalling #2592) were used according to the manufacturer's instructions.

Statistical analysis and software tool analysis
Raw data were acquired by using ImageJ software and a semi-quantitative analysis was performed. Relative intensities were normalised on the control and reported above images, in Figure 5.

Lsd1 and MAO enzymes evaluation assays
First, the new TCP-based compounds 1b-6 were tested to detect their binding and inhibition of LSD1. ThermoFAD assay 28 was used to detect the effects of 1b-6 on the thermal stability of the LSD1-CoREST complex, and the largest shifts in the Tm of the enzyme usually indicate the strongest LSD1 binding abilities (Table 1). Representative curves relative to ThermoFAD assay and FAD spectral bleaching using LSD1-CoREST (LC305) with selected compounds have been depicted in Figure 3(A,B), respectively.
The inhibition of LSD1 was assessed by protein expression in Escherichia coli and co-purification of LSD1D124-CoREST1D305 (LSD1-CoREST) and using H3K4me1 (residues 1-21) peptide as the substrate, as previously described 17 . The IC 50 values of 1b-6 against the LSD1-CoREST complex are reported in Table 1, and the relative curves are depicted in Figure 4. TCP and 1a were used as reference compounds.
The shift of the 4-phenyl ring of the benzoylamino moiety of 1a from the para to meta position led to slightly increased potency (1.3-fold) against LSD1 (compare 1b with 1a). This trend of inhibition potency is confirmed for all the following derivatives, in which the phenyl ring has been replaced by a 2-furyl (2a,b), 3-furyl (3a,b), 2-thienyl (4a,b), 3-thienyl (5a,b), and 4-pyridyl (6a,b) group. In general, the meta regioisomers 2b, 3b, 4b, 5b and 6b displayed 10.6to 71.8-fold higher potency against LSD1 than the para counterparts. When compared to the phenyl-containing 1a and 1b, the 2furyl (2a,b) compounds decreased the inhibitory potency when placed in para (2a), and gave little improvement when inserted in meta (2b). Differently, the replacement with a 3-furyl (3a,b) ring was detrimental for the anti-LSD1 activity regardless of the para or meta position. The introduction of the 2-or 3-thienyl (4a,b and 5a,b) ring led to a severe (1.8-to 5.7-fold) drop of potency when the heterocycle was placed in para (compare 4a and 5a with 1a), whereas the meta corresponding analogs 4b and 5b showed 10-to  30-fold improved potency respect to 1b in the anti-LSD1 assay. At last, the 4-pyridyl (6a,b) ring abated the inhibitory potency from 73 (6a) to 4 (6b) times when compared with 1a and 1b, respectively.
Since 1b-6 contain in their structure the TCP fragment, they were tested against MAOs to assess their eventual LSD1 selectivity. The IC 50 values of 1b-6 against MAO-A and MAO-B have been    Table 2. Clorgyline and R-(-)-deprenyl were used as reference MAO-A and -B selective inhibitors, respectively.
The tested compounds displayed IC 50 values against MAO-A in the range 0.024-1.120 lM, and against MAO-B they were often less potent, with a range of inhibition between 0.032 and 10.42 lM. Among the para and meta (a and b) substituted analogs, the first generally were more potent than the latter against the two MAO isoforms, reversing the behaviour observed with LSD1. Exceptions to this rule are the thienyl derivatives (4a,b and 5a,b) versus MAO-A: indeed, against this MAO isoform, the 2thienyl derivatives 4a and 4b showed the same inhibitory activity, and the 3-thienyl meta derivative 5b was 1.5-fold more potent than the corresponding para analog 5a.
3.3. Effects of 1a,b, 2b, 3b, 4b, and 5a,b in U937 AML and prostate cancer LNCaP cells Selected compounds among the most potent against LSD1 at enzyme level (1a,b, 2b, 3b, 4b, and 5a,b, with IC 50 values in the range 0.005-0.341 lM) were tested at 0.1, 5, 10, 25, and 50 lM for 48 and 72 h in U937 AML and prostate cancer LNCaP cells to determine their effects on cell viability. Tables 4 and 5 show the dose-dependent decrease of cell viability observed after treatment in U937 and LNCaP cells, respectively.
In U937 cells the new LSD1 inhibitors showed moderate activities, with 1b, 3b, 5a, and 5b being the most potent and 1a, 2b, and 4b almost ineffective at both time points. In this cell line, the meta phenyl and 3-thienyl derivatives 1b and 5b exerted the highest potency at 48 and 72 h. LNCaP cells were more sensitive than U937 cells to these inhibitors, likely due to the ability of the tested LSD1 inhibitors to also inhibit MAO-A, that is known to play a role in prostate cancer cell proliferation [30][31][32][33] . In LNCaP cells, all the tested compounds exhibited dose-dependent decreased cell viability. After 48 h of treatment 3b was the most potent, with IC 50 ¼ 12.4 lM, and after 72 h all the tested inhibitors displayed IC 50 values comprised between 9.9 and 25.4 lM, with 1b and 3b being the most effective (Table 6).
To link the reduction of cell viability to the inhibition of LSD1, western blot analyses have been performed on the new TCPbased inhibitors at 10 and 50 lM after 48 h of treatment to determine the changes in H3K4me2 (U937) or H3K4me2 and H3K9me2 (LNCaP) levels, typical marks for LSD1 demethylase activity ( Figure  5). In LNCaP cells we also detected the levels of H3K9me2 because LSD1 is known to interact with the androgen receptor and to   promote androgen-dependent transcription of target genes by ligand-induced demethylation of H3K9me1/2 4,34 . ORY-1001 (iadademstat) 35 , a known covalent LSD1 inhibitor, was used as a reference drug (at 25 lM). Western blot analyses in both cell lines confirmed the capability of the tested compounds to inhibit LSD1 in cellular contexts, showing a moderate to huge increase of methylation levels of all the investigated marks.

Conclusions
The known anti-MAO drug TCP is the most reliable fragment to design potent, covalent LSD1 inhibitors, active in cancer cells as antiproliferative agents 16 . In our previous works, we identified MC2580 17 and MC2584 18 as TCP analogs with higher potency against LSD1. The insertion of a phenyl ring at the para position of the benzoylamino moiety of MC2584 led to 1a, which retained potency against LSD1 and displayed high effects in MV4-11 and NB4 leukaemia cells 25 . Starting from these findings, we designed and prepared new 1a analogs by shifting the phenyl ring from para to meta position (1b), and by inserting 5-or 6-membered heterocyclic rings either at para or meta positions (2a,b-6a,b).
Compounds 1b-6 were tested against LSD1: the inhibition assay showed the meta derivatives (b) to be more potent than the corresponding para counterparts (a), with the meta 2-and 3-thienyl analogs 4b and 5b being the most potent (IC 50 values ¼ 0.015 Table 4. Effects of 1a,b, 2 b, 3 b, 4 b, and 5a,b on cell viability in U937 cells after 48 and 72 h of treatment. (4b) and 0.005 (5b) lM). When tested against MAOs, 1b-6 were found active at submicromolar/nanomolar levels mainly against MAO-A. The comparison between the inhibitory activities of 1b-6 against the two families of flavoenzymes (LSD1 and MAOs) showed an LSD1-selective behaviour for all the meta regioisomers over MAO-B and in some cases respect to MAO-A, with the 2-and 3-thienyl derivatives 4b and 5b being the most LSD1-selective respect to MAO-B (selectivity indexes: 24.4 and 164, respectively).
Selected compounds 1a,b, 2b, 3b, 4b, and 5a,b were tested in U937 AML and prostate cancer LNCaP cells to detect their antiproliferative activities after 48 and 72 h of treatment. LNCaP cells turned out to be more sensitive than U937 to these LSD1 inhibitors because all the tested compounds displayed a dosedependent decrease of cell viability with 1b and 3b being the most effective. In U937 cells, only 1b, 3b, 5a, and 5b gave an effect, while the remaining compounds were less active or even inactive. Western blot analyses, performed in both the cell lines after 48 h of treatment with the same compounds to detect the levels of H3K4me2 (U937 and LNCaP) and H3K9me2 (LNCaP), confirmed the involvement of LSD1 inhibition in these cellular assays showing a general increase of the methylation levels.  Effects of 1a,b, 2 b, 3 b, 4 b, and 5a,b on cell viability in LNCaP cells after 48 and 72 h of treatment.

h 72 h
Further studies will be performed with these compounds in different cancer contexts to assess their anticancer potential.

Disclosure statement
No potential conflict of interest was reported by the authors.