Hydroxamic Acids Immobilized on Resins (HAIRs): Synthesis of Dual‐Targeting HDAC Inhibitors and HDAC Degraders (PROTACs)

Abstract Inhibition of more than one cancer‐related pathway by multi‐target agents is an emerging approach in modern anticancer drug discovery. Here, based on the well‐established synergy between histone deacetylase inhibitors (HDACi) and alkylating agents, we present the discovery of a series of alkylating HDACi using a pharmacophore‐linking strategy. For the parallel synthesis of the target compounds, we developed an efficient solid‐phase‐supported protocol using hydroxamic acids immobilized on resins (HAIRs) as stable and versatile building blocks for the preparation of functionalized HDACi. The most promising compound, 3 n, was significantly more active in apoptosis induction, activation of caspase 3/7, and formation of DNA damage (γ‐H2AX) than the sum of the activities of either active principle alone. Furthermore, to demonstrate the utility of our preloaded resins, the HAIR approach was successfully extended to the synthesis of a proof‐of‐concept proteolysis‐targeting chimera (PROTAC), which efficiently degrades histone deacetylases.


In memory of Victoria Michler-Hansen
Abstract: Inhibition of more than one cancer-related pathway by multi-target agents is an emerging approach in modern anticancer drug discovery. Here, based on the well-established synergy between histone deacetylase inhibitors (HDACi) and alkylating agents, we present the discovery of a series of alkylating HDACi using a pharmacophore-linking strategy. For the parallel synthesis of the target compounds, we developed an efficient solid-phase-supported protocol using hydroxamic acids immobilized on resins (HAIRs) as stable and versatile building blocks for the preparation of functionalized HDACi. The most promising compound, 3 n, was significantly more active in apoptosis induction, activation of caspase 3/7, and formation of DNA damage (g-H2AX) than the sum of the activities of either active principle alone. Furthermore, to demonstrate the utility of our preloaded resins, the HAIR approach was successfully extended to the synthesis of a proof-of-concept proteolysis-targeting chimera (PROTAC), which efficiently degrades histone deacetylases.
Difficulties in developing new drugs for multifactorial diseases like neurological disorders or cancer have led to rethinking of the "one disease-one target-one drug" paradigm to a "multi-target drug" concept. [1] Compared to combination therapies using two or more drugs, a single multi-target molecule has the advantages of no drug-drug interactions, a more predictable pharmacokinetic profile, and improved patient compliance. [2,3] Histone acetyltransferases (HATs) and histone deacetylases (HDACs) are key enzymes controlling the acetylation level of histones and non-histone proteins. [4] Due to their repressive effect on gene transcription and their essential influence on drug resistance mechanisms of tumor cells, [5][6][7] HDACs are validated drug targets in epigenetic cancer therapy with four inhibitors already approved by the FDA. [8] Typically common HDAC inhibitors (HDACi) comprise a cap group, a linker region and a zinc binding group (ZBG), which is crucial for the chelation of the zinc ion inside the active site tunnel ( Figure 1). [9] Fortunately, the cap group can be subjected to various structural modifications providing sufficient scope for hybridization approaches towards HDACi-based multi-target drugs. [10,11] Recent preclinical results provide evidence that combinations of alkylating agents and HDACi exhibit efficacy against drug-resistant glioblastoma multiforme (GBM), the most common and aggressive primary brain tumor, [12,13] by increasing the DNA damage of alkylating agents via HDACimediated chromatin relaxation. [14,15] Consequently, the firstin-class nitrogen mustard-HDACi hybrid molecule tinostamustine ( Figure 1) was developed by fusing the pharmacophores of the DNA-alkylating drug bendamustine and the HDACi vorinostat (SAHA). [16] Tinostamustine demonstrated superior in vivo activity compared to bendamustine, temozolomide and radiotherapy indicating that DNA/HDAC dualtargeting inhibitors could be promising drug candidates for cancer therapy.
Since HDACi-based multi-target drugs show great promise in preclinical and early clinical studies, [10,11] there is an urgent need for efficient synthetic protocols allowing the synthesis of focused compound libraries with linked or merged pharmacophores. Herein we present the development of a series of preloaded resins for solid-phase synthesis termed hydroxamic acids immobilized on resins (HAIRs). To demonstrate the utility of these HAIRs and to take advantage of the synergism between HDACi and alkylating agents, we successfully prepared a set of HDACi with DNA-alkylating properties via a fast and straightforward parallel synthesis approach. The scope of the HAIR technology was further extended to the synthesis of a proof-of-concept proteolysistargeting chimera (PROTAC), that is, a protein degrader.
The preparation of HAIRs A-E as HDACi precursors is summarized in Scheme 1. Initially, we modified the commercially available 2-chlorotrityl chloride (2-CTC) resin with the immobilization of hydroxylamine by treatment of the resin with N-hydroxyphthalimide and triethylamine for 48 h. Next, after deprotection of the Phth-group using hydrazine hydrate, different Fmoc-protected HDACi linkers were loaded to the functionalized resin to provide the preloaded resins HAIRs A-E. For the linker a series of well-established HDACi linkers such as benzyl, alkyl and cinnamyl was selected. For all synthesized resins, high loadings between 0.81-0.90 mmol g À1 were achieved. All prepared HAIRs showed excellent crude purities and reproducible loadings (AE 5 %). Furthermore, after storage for > 6 months the weight and purity of all modified resins were retested after swelling and drying. All HAIRs showed stable weight (AE 5 %) and high crude purities (> 95 % after > 6 months of storage) ( Table 1).
After establishing the preloaded HAIRs A-E, we utilized them to incorporate the DNA-alkylating part in the cap group region. Due to the synergistic activity of HDACi with alkylating agents, we chose the DNA-alkylating drugs temozolomide (TMZ), mitozolomide (MTZ), and chlorambucil (CAB) as suitable scaffolds for a hybridization approach. All three examples can be considered as challenging, because TMZ and MTZ are sensitive to base treatment, which leads to ring opening of the imidazotetrazinone ring system, whereas CAB contains a very reactive nitrogen mustard group. [17,18] For library synthesis, the Fmoc protecting group was removed and each linker was coupled with the three chosen alkylators in a concentration range of 0.5-1.0 m. Using a parallel synthesis strategy, we prepared a series of compounds combining each linker with each alkylator yielding a library comprising 15 hybrid molecules. This approach provided the target compounds in total yields of up to 83 %. Purification by preparative HPLC afforded all compounds in > 95 % purity. Thus, only one purification step at the very end of the synthesis is required for each inhibitor allowing for rapid and time-efficient library expansion.
Compounds 3 a-n were tested in a fluorogenic assay for their in vitro inhibitory activity against HDAC1 (class I) and HDAC6 (class IIb). Results of the inhibition assay are shown in Table 2. Compounds containing a cinnamyl (3 l-n) and hexyl (3 i-k) linker showed inhibitory activity in the nanomolar range against both HDAC1 and HDAC6, which was comparable to the control compound vorinostat. Notably, inhibitors utilizing a benzyl linker revealed potent and preferential inhibition of HDAC6 with up to 17-fold selectivity over HDAC1 (see compound 3 a, Table 2). All compounds with a short propyl linker (3 d-f) showed very low HDAC inhibition, which suggests that this linker length is too short to chelate the Zn 2+ ion in the active site. To investigate the anticancer properties of the hybrid HDACi, the antiproliferative effects of 3 a-n were determined in three human cancer lines: the human tongue squamous cell carcinoma cell line Cal27 and the human primary glioblastoma cell lines U87 and U251. Results are summarized in Table 2. Based on data from MTT assays, compound 3 n, which contains a cinnamyl linker and CAB as cap group, emerged as the most promising compound. 3 n showed HDAC1 and 6 inhibition in a similar range as vorinostat and 3 to 5.6-fold weaker inhibition improved antiproliferative effect in comparison to its parent compound CAB. Compounds bearing the cinnamyl linker (3 l-n) showed the most effective combination of antiproliferative effect as well as HDAC inhibitory activity and were therefore chosen for evaluation of their DNA-damaging effects. 3 l-n were tested for induction of DNA double strand breaks by a g-H2AX assay [19] using cisplatin, vorinostat, TMZ, MTZ, and CAB as controls (Figure 2 A). Cal27 cells were treated with an IC 50 or fivefold IC 50 (from MTT assay) for 24 h. Among the cinnamyl linker-containing compounds 3 l-n, 3 n caused the largest increase in g-H2AX levels in Cal27 cells. Using an IC 50 from MTT assay, only 3 n achieved a significant increase in g-H2AX formation, whereas 3 l and 3 m were not different from control. At fivefold IC 50 , 3 l-n showed DNA damage with 3 n being significantly more active than 3 l and 3 m, Because 3 l and 3 m showed only weak DNA damage, they were excluded from further studies.
Based on these results, we investigated whether the nitrogen mustard moiety is contributing to the antiproliferative activity and DNA-damage of 3 n. We thus synthesized the control compound 3 o omitting the nitrogen mustard group (see Scheme S1, Supporting Information). The anti-  [a] n ! 2, each in duplicate wells.

Angewandte Chemie
Communications 22496 www.angewandte.org proliferative potency of 3 o was slightly reduced compared to 3 n (Figure 2 B). The most pronounced reduction in antiproliferative potency was observed in Cal27 cells (3 n IC 50 : 2.68 mM vs. 3 o IC 50 : 6.59 mM). More importantly, 3 n caused significantly greater DNA damage (g-H2AX assay) than 3 o at a concentration of five times its IC 50 (Figure 2 A) indicating an additive effect of the nitrogen mustard group. g-H2AX formation is a marker for DNA damage, which usually results in induction of apoptosis. [19] Figure 2 C displays apoptosis, shown as subG1 nuclei, induced by 3 n and the nitrogen mustard-free control 3 o. Notably, 3 n is more potent than 3 o, cisplatin, vorinostat, and CAB in inducing apoptosis at the respective IC 50 values. Comparable results were observed in a caspase 3/7 activation-apoptosis assay demonstrating the superior utility of 3 n to kill cancer cells via induction of the programmed cell death (Figure 2 D). Notably, although the combination of 3 o and CAB showed synergism in caspase activation assay as analysed by the method of Chou-Talalay (combination index (CI) values < 1), equal concentrations of compound 3 n demonstrated significantly stronger effects and lower CI values (< 0.9) (see Figure S1, Table S1, Supporting Information). In addition, the effect of 15 mM 3 n was significantly stronger than the sum of the effects of even 100 mM CAB and 30 mM 3 o, again indicating a superadditive effect of the dual-targeting compound 3 n (Figure 2 E, left). The same holds true for results from the g-H2AX assay (Figure 2 E, right). The superior antiproliferative activity, apoptosis induction, activation of caspase 3/7, and formation of DNA damage of 3 n vs. 3 o could result from an altered HDAC inhibition profile. 3 n and 3 o were therefore screened for their inhibitory activity against all class I HDACs and HDAC6. However, 3 n showed similar or less potent HDAC inhibitory activity than 3 o (Figure 3 A) highlighting the importance of the DNA-alkylating feature of the hit compound 3 n for its anticancer activity. Based on the biochemical HDAC inhibition data, it can be assumed that the nitrogen mustard in p-position of the cap group has little impact on the HDAC inhibitory activity of 3 n. To test the hypothesis that the DNA alkylating moiety fails to form specific interactions with the HDAC proteins, compounds 3 n and 3 o were docked into HDAC6. During the docking study the hydroxamic acid group was restrained in proximity to the zinc ion in the . T-test was used to analyze for significant differences. ** (p 0.01); *** (p 0.001).The analysis was performed as previously described. [20] catalytic center, resulting in very similar binding poses for both compounds (Figure 3 B). The addition of the N-lost group did not alter the binding pose of the compounds as it points outwards and only interacts loosely with the protein surface. Taken together, our results demonstrate that the incorporation of a DNA-alkylating feature in the p-position of the cap group has little impact on the HDAC inhibition and can thus be utilized to enhance the anticancer effects of HDACi. Encouraged by the successful and straightforward parallel synthesis of a focused library of DNA-alkylating HDACi, our aim was to extend the HAIR approach to a second class of chimeric small molecules: proteolysis-targeting chimeras (PROTACs). PROTACs are bifunctional small molecules that are able to hijack the cellular protein degradation system by recruiting the protein of interest (POI) to E3 ubiquitin ligases, which leads to polyubiquitinylation of the POI and induction of its proteasomal degradation. [21,22] Developments in PROTAC technology give new opportunities to address and study epigenetic targets such as HDACs, with already a few HDAC PROTACs reported recently. [23][24][25][26] However, due to their high molecular weight and bifunctional nature, the synthesis of HDAC PRO-TACs is usually cumbersome and involves multi-step protocols. [26] Consequently, we undertook the first reported solidphase synthesis of an HDAC degrader. HAIR D was chosen as a suitable starting point for the preparation of the proof-ofconcept HDAC PROTAC 4. Iterative cycles of Fmoc deprotection and amide coupling allowed to introduce the HDAC cap, PROTAC linker, and thalidomide-based ubiquitin E3 ligase ligand in a modular fashion (see Scheme S2, Supporting Information for synthetic details) to generate PROTAC 4 (Figure 4 A). This straightforward and rapid HAIR-supported synthesis provided 4 in an excellent crude purity of 91 % (Figure 4 B) and > 95 % after purification.