DESIGN, SYNTHESIS AND EVALUATION OF ANTIPROLIFERATIVE ACTIVITY OF MELANOMA-TARGETED HISTONE DEACETYLASE INHIBITORS

Abstract

The clinical validation of histone deacetylase inhibition as a cancer therapeutic modality has stimulated interest in the development of new generation of potent and tumor selective histone deacetylase inhibitors (HDACi). With the goal of selective delivery of the HDACi to melanoma cells, we incorporated the benzamide, a high affinity melanin-binding template, into the design of HDACi to generate a new series of compounds 10a–b and 11a–b which display high potency towards HDAC1 and HDAC6. However, these compounds have attenuated antiproliferative activities relative to the untargeted HDACi. An alternative strategy furnished compound 14, a prodrug bearing the benzamide template linked via a labile bond to a hydroxamate-based HDACi. This pro-drug compound showed promising antiproliferative activity and warrant further study.

Graphical Abstract

Melanoma is one of the most prevalent cancers among skin malignancies.¹ As with most cancer types caused by disruption in the homeostatic regulation of cellular processes, melanomas are typified by dysregulation of melanin-producing melanocytes by keratinocytes in the skin epidermis.² This dysregulation results in uncontrolled proliferation of melanocytes, hence the high melanin content associated with most melanoma cases. When detected early, melanoma can be corrected by surgery. Therefore, sustained efforts have been on developing effective new tools for early diagnosis of melanoma.

Over the past two decades, benzamides (Fig. 1A) have emerged as promising class of compounds that can be developed into theranostics for early detection and treatment of melanomas.³ What makes this class of compounds suitable for this purpose is their high affinity for melanin, which affords them a high residence time in melanoma cells. In phase II clinical trial, N-(2-(diethylamino) ethyl)-4-iodobenzamide (Fig. 1A) showed 100% specificity on a lesion-site basis with little toxicity to the patients.^3a When incorporated into the structure of a conventional alkylating agent, the resulting alkylating benzamides showed superior selectivity for melanoma cells both in vitro and in vivo.⁴ Moreover, other studies have demonstrated the suitability of the benzamide template for developing targeted proteasome degradation inhibitors⁵. Hence, benzamides have become the Holy Grail for developing melanoma-targeted therapeutics.

Fig. 1.

A) structures of selected melanin-targeting benzamides for imaging melanomas^{3b, 3c, 11}; B) structures of clinically approved HDACi; C) standard HDACi pharmacophoric model.

Epigenetic dysregulation, exemplified by aberrant regulation of expression of histone deacetylases (HDAC) and histone acetyl transferase (HAT), is one of the root causes of cancer. Several isoforms of HDAC have been implicated in the sustained proliferation of most tumors, with class I and class IIb HDACs being the most prominent.⁶ With respect to melanomas, HDAC isoforms 5 and 6 were recently reported to play significant roles in disease progression.⁷ To further exploit the relevance of HDAC 6 in melanoma progression, Kozikowski et. al.⁸ have studied the effects of HDAC 6 selective HDAC inhibitors (HDACi) on melanoma cell growth. Though a promising approach, these isoform 6 selective HDACi showed weak cytotoxic effects on B16F10 melanoma cell line.

The structures of the five clinically approved HDACi (Fig. 1B) conform to a structural model comprising a surface recognition group, linker and a zinc binding group (ZBG) as shown for SAHA in Fig. 1C. Previous SAR studies on the surface recognition group have identified this part of most HDACi to be the most amenable to modification.⁹

To further demonstrate the broad scope of application of the benzamide template for developing targeted therapies, we designed and synthesized HDACi which have melanin binding benzamide group incorporated unto the surface recognition group (Figs. 2A and B). We envisaged that this approach will result in selective localization of benzamide-bearing HDACi within the melanoma cells, and mitigate the off-target toxicities associated with most HDACi. Following our previous reports,^9a we restricted linker lengths in our design to be between 5–7 methylene groups separating the hydroxamate (ZBG) from the cap group.

Fig. 2.

General structures of designed compounds.

Synthesis of target compounds started with the esterification of 4-iodobenzoic acid 1 and 5-bromopyridine-2-carboxylic acid 2, both of which are commercially, using TMSCl in MeOH to give methyl esters 3 and 4 (Scheme 1). Sonogashira coupling¹² of the respective halo-methyl ester (compound 3 or 4) with TMS-acetylene gave the appropriate TMS-protected ethynyl intermediates 5 and 6. At this stage, different synthetic pathways were employed to access alkynes 8 and 9. Subjecting TMS-protected methyl-4-(ethynyl)-benzoate 5 to alkaline hydrolysis, using KOH in isopropanol, resulted in TMS removal and gave the potassium salt of 4-(ethynyl) benzoate 7. Coupling of 7 to N,N-diethyl ethane-1,2-diamine using TSTU^3c afforded compound 8. In the case of compound 9, TMS-protected methyl-5-ethynyl picolinate 6 was refluxed with N,N-diethyl ethane-1,2-diamine in MeOH to give compound 9. Cu (I) catalyzed cycloaddition reaction¹³ between alkyne 8 or 9 and O-trityl protected azido hydroxamates with appropriate linker lengths, followed by trityl deprotection using TFA, gave the target compounds 10a–b and 11a–b.

Scheme 1.

Synthesis of Benzamide-bearing HDACi. a) TMSCl, MeOH, rt; b) TMS acetylene, Pd(PPh₃)₂Cl₂, CuI, TEA, CH₃CN, 90°C, overnight; c) KOH, isopropanol, rt, overnight; d) TSTU, DIPEA,DCM, rt; e) MeOH, 120°C; f) CuI, DIPEA,THF, rt; g) TFA/TIPS.

Taking into consideration the previously reported roles of HDAC6 in sustaining melanoma growth^7–8 and the absolute requirement of HDAC1 inhibition for HDACi to elicit robust antiproliferative effects,⁸ we profiled the HDAC1 and HDAC6 inhibition activities of compounds 10a–b and 11a–b in a cell-free assay. All the compounds showed strong inhibitory activities towards the two HDAC isoforms tested, with compound 10a being ten-fold more potent towards HDAC1 compared to SAHA (Table 1). We observed these compounds to be more potent towards HDAC1 compared to HDAC6, suggesting that they may possess potent in vitro anticancer activities. The most potent compound in this series, compound 10a, showed about four-fold selectivity for HDAC1 compared to HDAC6 while the weakest in the series, compound 11a, is almost indistinguishable in terms of inhibitory activity towards HDAC1 and HDAC6. Compounds 10b and 11b, despite having seven- and six methylenes, respectively, as linkers are both equipotent towards HDAC1 and HDAC6. Interestingly, the phenyl-based compound with six methylene linker, 10a, is more potent than the analogous pyridyl-based compound, 11b, in both HDAC1 and HDAC6. This observation is similar to a previous study in our lab in which HDAC inhibitory activities of phenyl-based HDACi and their analogous pyridyl-based HDACi were compared.¹⁴ Likewise, linker-length effect, similar to previous observations,^14–15 was noticeable in the two series of compound.

Table 1.

HDAC isoforms inhibition study

Compound	n	IC50 (µM)*
		HDAC1	HDAC6
10a	2	0.0038	0.016
10b	3	0.012	0.027
11a	1	0.100	0.170
11b	2	0.017	0.034
SAHA	-	0.031	0.011

^*values indicate average of three independent experiments

Encouraged by the potent cell-free HDAC inhibitory activities of these benzamide-bearing HDACi, we assayed their growth inhibitory effects on two melanin-producing melanoma cell lines, B16F10 and A375. We observed that compounds 10a and 11b were weakly cytotoxic towards the two cell lines while 10b and 11a were almost devoid of growth inhibitory effects at the maximum concentration tested (Table 2). The most active compound, 10a, was about 3- and 6-fold less active relative to SAHA (control HDACi) against B16F10 and A375 cell lines respectively. This weak cytotoxic activity of these compounds against melanin-producing melanoma cell lines was surprising to us.

Table 2.

Antiproliferation study in cancer cell lines.

Compound	n	IC50(µM)*
		B16F10	A375	LNCaP
10a	2	34.16	23.60	1.2
10b	3	75.91	>100	3.0
11a	1	>100	>100	10.5
11b	2	55.67	47.18	1.5
SAHA	-	13.1	4.1	1.8

^*values indicate average of three independent experiments

Previously, intricate drug efflux system in melanoma, involving melanosomal sequestration and subsequent efflux of cytotoxic agents, was identified to be responsible for resistance of melanoma to chemotherapeutic agents.^{10, 16} It is quite possible that the weak cytotoxic effects observed with the benzamide-bearing HDACi in melanoma cell lines is a reflection of this phenomenon. Alternatively, melanin, being a non-biological polymer,¹⁷ may be unable to release sufficiently high concentration of the benzamide-bearing HDACi upon binding, resulting in their inability to engage HDACs, the therapeutic targets. Hence, melanin binding may not provide the anticipated advantage. It is also possible that the observed weak antiproliferative activity of the benzamide-bearing HDACi may just be a result of the inability of the tested compounds to penetrate the cell membrane.

Intrigued by this observation and determined to unravel the basis for the weak in cellulo activities of these compounds, we extended cell growth inhibition study to LNCaP, an androgen dependent prostate cancer cell line not known to produce melanin. Results obtained from this study should provide insights into the role (s) of melanin in the observed cellular inactivity, and also confirm the ability of these compounds to cross the cell membrane. As seen in Table 2, all of the tested compounds inhibited the growth of LNCaP in a pattern similar to the HDAC inhibitory activity, confirming that these compounds can cross the cell membrane easily. This makes it highly probable that either of melanin binding or effective drug efflux is impeding access to HDAC enzymes. To confirm the role of drug efflux on diminishing the cell growth inhibitory activity of these compounds, we treated B16F10 with compounds 10a and 11b followed by verapamil, an established efflux inhibitor,¹⁸ at 100 µM. This verapamil co-treatment did not improve the cytotoxic effect of 10a and 11b in B16F10 (data not shown), strongly suggesting that the high affinity for melanin is the likely reason for the weak cytotoxicity of these compounds.

We then considered exploiting benzamides’ affinity for melanin in a slightly different way. For this, a prodrug approach, in which the melanin-binding benzamide is attached to the hydroxamate-moiety of a HDACi via a labile ester bond that could provide a means of selective delivery of such HDACi to melanoma, was developed. When such compound binds to melanin in melanoma cells, the HDACi component could, thereafter, be released by spontaneous hydrolysis and/or cellular esterase-facilitated hydrolysis. This approach has been successfully used in the past to improve the cellular accessibility of hydroxamate-based HDACi.¹⁹ In addition to this; such a design could mask potential off target interaction of the hydroxamate group with other biological targets in the cells.

Previous work from our lab has identified triazolyl hydroxamate 7u (scheme 2) as a promising HDACi with robust growth inhibitory activity in prostate cancer cell line.¹⁴ Encouraged by this, we incorporated compound 7u into the design of a releasable benzamide-HDACi prodrug (Fig. 3).

Scheme 2.

Synthesis of releasable BZA-based HDACi. a) KOH, MeOH, rt, overnight; b) Compound 7u, TSTU, DIPEA, THF/DMF, rt, overnight.

Fig. 3.

Proposed prodrug release mechanism

Potassium 4-((2-(diethylamino) ethyl) carbamoyl) benzoate (13) used in the synthesis of the prodrug was made as described previously^5c. Coupling of compound 13 to 7u using TSTU gave the desired prodrug 14 (Scheme 2).

While we do not anticipate any appreciable inhibition of HDAC (because the ZBG is masked), compound 14, when profiled against HDAC isoforms 1, 6 and 8 at 5 µM, showed 100 percent inhibition against HDAC1, 99 percent inhibition against HDAC6, and 52 percent inhibition against HDAC8. A similar observation was previously reported for some of the compounds made by Miller et al. in their patent describing prodrugs of SAHA-like molecules.^19b

Compounds 7u and 14 were tested in melanoma cell lines, A375 and B16F10, as shown in Table 3. In B16F10, the prodrug 14 is slightly more potent than compound 10a and about half as potent as compound 7u. Similarly, compound 10a is about half as potent as compound 7u and equipotent to prodrug 14 in A375 cell line. When dosed at concentration just below IC₅₀ (25 µM), prodrug 14 showed the same level of potency as compound 10a. The advantage to having the prodrug becomes more pronounced at 50 µM, where prodrug 14 becomes much more potent than compound 10a (Fig. 4). This data implies that there may be an advantage to using the benzamide template to design prodrugs rather than having it “permanently” incorporated into the design of compounds targeted towards melanin-producing melanoma cells.

Table 3.

Anti-proliferation study of prodrug in cancer cell lines.

Compound	IC50 (µM)*
	B16F10	A375	LNCaP
14	27.75	25.95	6.46
7u	10.50	12.61	5.10
10a	34.16	23.60	1.20
SAHA	13.44	4.31	1.80

^*values indicate average of three independent experiments

Fig. 4.

Comparison of growth inhibitory effects of prodrug 14, and compounds 7u and 10a in B16F10 cell line.

To gain further insight into the cellular accessibility of the synthesized compounds, and confirm that prodrug 14 releases compound 7u upon getting into the cells, we used western blot to probe for accumulation of acetylated tubulin upon treating B16F10 cells with compounds 10a, 7u, 14, and SAHA. Accumulation of acetylated tubulin in the cytoplasm is a marker of HDAC 6 inhibition in cells.²⁰ SAHA and compound 7u as expected showed significant accumulation of acetylated tubulin at 20 µM (Fig. 6). Likewise, compound 10a gave a similar effect at increasing drug concentrations (5–50 µM). Gratifyingly, prodrug 14 also caused increase in the level of acetylated tubulin in a concentration dependent manner (Fig. 6). This implies that the prodrug is capable of releasing compound 7u within the cells, validating our prodrug design hypothesis.

Fig. 6.

Western blot analysis confirms that prodrug 14 caused intracellular accumulation of acetylated tubulin.

In conclusion, we reported a series of HDACi 10a–b and 11a–b designed to selectively target melanoma cells. These compounds displayed potent HDAC inhibitory activities that are largely dependent on the nature of the cap group and the linker length. Preliminary evaluation of the growth inhibitory effects of these compounds on two melanin-rich melanoma cell lines, B16F10 and A375, showed that they have attenuated antiproliferative activities relative to the untargeted HDACi 7u and SAHA. Against melanin null prostate cancer cell line LNCaP cell, however, all the compounds showed anticancer activities consistent with their strong HDAC enzyme inhibitory effects. We believe that the attenuated cytotoxicity of 10a–b and 11a–b towards melanoma cell lines may be due to their being bound to melanin in melanoma cells, a scenario which hinders their access to HDACs for therapeutic effect. An alternative prodrug strategy showed some promise as prodrug 14 is slightly more potent than 10a, a compound having benzamide moiety covalently incorporated into the surface recognition group, in high melanin-producing B16F10 cell line. Further study will be required to optimize benzamide as promoiety to achieve selective delivery of HDACi prodrugs into melanin-producing melanoma cells.

Abbreviations

HDAC

Histone deacetylase

HDACi

Histone deacetylase inhibitors

ZBG

zinc binding group