Solid-Phase Parallel Synthesis of Dual Histone Deacetylase-Cyclooxygenase Inhibitors

Multi-target drugs (MTDs) are emerging alternatives to combination therapies. Since both histone deacetylases (HDACs) and cyclooxygenase-2 (COX-2) are known to be overexpressed in several cancer types, we herein report the design, synthesis, and biological evaluation of a library of dual HDAC-COX inhibitors. The designed compounds were synthesized via an efficient parallel synthesis approach using preloaded solid-phase resins. Biological in vitro assays demonstrated that several of the synthesized compounds possess pronounced inhibitory activities against HDAC and COX isoforms. The membrane permeability and inhibition of cellular HDAC activity of selected compounds were confirmed by whole-cell HDAC inhibition assays and immunoblot experiments. The most promising dual inhibitors, C3 and C4, evoked antiproliferative effects in the low micromolar concentration range and caused a significant increase in apoptotic cells. In contrast to previous reports, the simultaneous inhibition of HDAC and COX activity by dual HDAC-COX inhibitors or combination treatments with vorinostat and celecoxib did not result in additive or synergistic anticancer activities.


Introduction
Every year, several millions of people die of cancer [1]. The development and investigation of new anticancer drugs therefore remains an important challenge. As a multifactorial disease with many signaling pathways and various dysregulated physiological processes involved, cancer requires complex therapy regimes that modulate different biological targets [2][3][4]. In addition to the well-established strategy of combination therapies, multi-target drugs (MTDs), have gained growing attention over the last years [5,6]. In combination therapy, a minimum of two drugs is applied to achieve synergistic or additive effects aiming at a higher therapeutic efficacy. Engaging different modes of action, they may also help to avoid cellular mechanisms of compensation or resistance [3,5]. MTDs, in contrast, are single molecules with multiple modes of action that specifically address several disease-relevant targets [2,5]. The advantages of MTDs include the simplification of therapy regimes and the reduction of side effects and drug-drug-interactions [3][4][5]. Another important advantage of MTDs is that only one regulatory approval is required for one molecular entity, while two or more approvals are needed for a combination therapy. The challenge in designing MTDs lies in the balancing of the affinity to each target while retaining suitable physicochemical properties and the desired pharmacological activities. Optimal efficacy may, moreover, require different doses reaching the respective targets [2][3][4][5].
coupling of the COX inhibitory cap groups followed by cleavage from the solid phase yielded a range of bifunctional inhibitors. All synthesized compounds were tested for their inhibitory activity against COX-1, COX-2, HDAC1, and HDAC6 as well as their antiproliferative activity against PDAC cell line AsPC1. The most promising dual inhibitors were further characterized in a panel of seven cancer cell lines. Furthermore, we investigated the intracellular target engagement of the best compounds by whole-cell HDAC inhibition assays and immunoblot experiments. Their ability to induce apoptosis was confirmed by annexin V/propidium iodide apoptosis assays.
The designed compounds were synthesized as outlined in Scheme 1. In the first step, a phthaloyl protected hydroxylamine resin, readily available from N-hydroxyphthalimide and a 2-chlorotrityl chloride (2-CTC) resin [39], was deprotected by treatment with hydrazine hydrate to furnish the free aminoxy group. The subsequent amide coupling reaction with the respective Fmoc-protected linker generated the preloaded resins PR1-7 in multi-gram scales. The seven preloaded resins were obtained in high purities (as indicated by test cleavages followed by purity determination by high performance liquid chromatography (HPLC)) and in loading capacities ranging from 0.48 to 0.77 mmol*g −1 . For the subsequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35].

Determination of COX Inhibition
In general, cyclooxygenase inhibitors exert anti-inflammatory, antipyretic, and analgesic effects by binding in the cyclooxygenase-active site and hence inhibiting the synthesis of prostaglandins [21]. COX-2 inhibitors, also called coxibs, selectively bind to the COX-2 isoform whose binding pocket differs from COX-1 because of an amino acid exchange in the sequence from isoleucine to valine, which facilitates the binding of taller substrates [21]. As lead compounds of this study, indometacin happens to be unselective, whereas celecoxib represents a selective inhibitor for the isoenzyme. The ability to inhibit both isoforms, ovine COX-1 and human COX-2, was determined using the "COX Fluorescent Inhibitor Screening Assay Kit" (Cayman Chemical, Ann Arbor, MI, USA) in a concentration range up to 100 µM [40,41]; the results are shown in Table 2. In general, modification of indometacin (A) and celecoxib derivatives (B and C) with linker and zinc binding motifs was tolerated by COX-2 leading to inhibitory potencies in the low micromolar range of IC 50 between 1-10 µM with exception of compound B1 (IC 50 = 43.0 µM). More pronounced differences were obtained in the selectivity between both isoforms due to differences in the inhibition profile for COX-1 ranging from IC 50 of 1.8 µM up to >100 µM. A definite structure-activity relationship could not be derived from this set of substances. The highest selectivity and most potent COX-2 inhibition was obtained for the methylsulfonyl-substituted celecoxib derivative C3 comprising a hexyl linker which showed a 96-fold selectivity for COX-2 and IC 50 (COX-2) of 0.98 µM. The respective aminosulfonyl-substituted celecoxib derivative B3 showed unselective inhibition, while the indometacin derivative A3 demonstrated a 14-fold preference for COX-2. The COX-2-selective inhibition for the alkyl substituted derivatives A2, A3, B2, B3, and C3 reported by Raji et al. [35] could be verified in our study, although the observed trend was not completely in line with our results. Compound A3, which Raji et al. found to be a very potent and selective inhibitor of COX-2 (IC 50 COX-2 = 0.33 µM and IC 50 COX-1 = 21.00 µM [35]), appeared to be no more than a moderate inhibitor in this set. Compound C3 (IC 50 COX-2 = 4.44 µM and IC 50 COX-1 = 23.09 µM [35]), on the other hand, showed opposite results. For the aromatic linker derivatives 4-7 which were designed for this study, we found rather non-selective COX inhibition in the low micromolar concentration range. The two benzimidazole-fluorobenzyl-based compounds A6 and B6 exerted nearly equipotent inhibition of both isoforms, while the derivatives with benzyl linkers 4 and 5, as well as the phenylvinyl linker 7 favored COX-2 inhibition with selectivity indices ranging between 2-15. respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. sequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. quent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the spective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling action with the cap group A, B, or C. Cleavage from the solid support followed by pufication with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-4 ( sequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. sequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. sequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. sequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. sequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. raphy (HPLC)) and in loading capacities ranging from 0.48 to 0.77 mmol*g . For the sub-sequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. raphy (HPLC)) and in loading capacities ranging from 0.48 to 0.77 mmol*g . For the sub-sequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. test cleavages followed by purity determination by high performance liquid chromatog-raphy (HPLC)) and in loading capacities ranging from 0.48 to 0.77 mmol*g −1 . For the subsequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. raphy (HPLC)) and in loading capacities ranging from 0.48 to 0.77 mmol*g −1 . For the subsequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35]. raphy (HPLC)) and in loading capacities ranging from 0.48 to 0.77 mmol*g −1 . For the subsequent parallel synthesis of the desired dual HDAC-COX inhibitors, an aliquot of the respective resin (PR1-7) was subjected to Fmoc deprotection followed by amide coupling reaction with the cap group A, B, or C. Cleavage from the solid support followed by purification with preparative HPLC yielded the target compounds A1-A7, B1-B7, and C2-C4 (Table 1). Compounds A6 and B6 were isolated as TFA salts due to the acidic modifier during HPLC purification. Compounds A1, A4-A7, B1, B4-B7, C2, and C4 are novel compounds, while compounds A2, A3, B2, B3, and C3 were previously prepared by Raji et al. using solution-phase methods [35].

Determination of COX Inhibition
In general, cyclooxygenase inhibitors exert anti-inflammatory, antipyretic, and analgesic effects by binding in the cyclooxygenase-active site and hence inhibiting the synthesis of prostaglandins [21]. COX-2 inhibitors, also called coxibs, selectively bind to the COX-2 isoform whose binding pocket differs from COX-1 because of an amino acid exchange in the sequence from isoleucine to valine, which facilitates the binding of taller substrates [21]. As lead compounds of this study, indometacin happens to be unselective,

Determination of COX Inhibition
In general, cyclooxygenase inhibitors exert anti-inflammatory, antipyretic, and analgesic effects by binding in the cyclooxygenase-active site and hence inhibiting the synthesis of prostaglandins [21]. COX-2 inhibitors, also called coxibs, selectively bind to the COX-2 isoform whose binding pocket differs from COX-1 because of an amino acid exchange in the sequence from isoleucine to valine, which facilitates the binding of taller substrates [21]. As lead compounds of this study, indometacin happens to be unselective, )) for COX-2 was found to be 7.48 ± 0.27 (mean ± SD, n = 6; IC 50 = 38.1 ± 20.7 nM) with an inter-assay coefficient of variation of 3.6%. 3 All compounds were measured in duplicate and standard deviation was <0.02 for all compounds. The retention time, mean, and standard deviation of retention time is given in the electronic supporting information (Table S1) for every compound. 4 n.e. = no effect up to 100 µM.

Inhibition of HDAC1 and HDAC6
All synthesized compounds and vorinostat as positive control were screened in a fluorogenic assay for their in vitro inhibitory activity against HDAC1 (as a representative class I isoform) and HDAC6 (as a representative class IIb isoform); the results are presented in Table 2. As expected, the compounds with a short propyl linker (A1 and B1) showed low inhibitory activities against both isoforms. In agreement with the literature data on vorinostat and panobinostat derivatives [42], the compounds with a pentyl (A2, B2, and C2), hexyl (A3, B3, and C3), and phenylvinyl linker (A7 and B7) turned out to be potent but rather unselective HDACi. The introduction of a benzyl linker provided compounds A4, B4, and C4 with a slightly improved preference for HDAC6. This effect was more pronounced when a fluorine atom was introduced in meta-position to the hydroxamic acid (A5 and B5). The beneficial effect of a fluorinated benzyl linker with regard to HDAC6 selectivity is in line with recent results disclosed by us and others [38,43,44]. Notably, the benzimidazole-fluorobenzyl-based compounds A6 (IC 50 HDAC1 = >2.80 µM; IC 50 HDAC6 = 0.011 µM; SI = >254) and B6 (IC 50 HDAC1 = 0.829 µM; IC 50 HDAC6 = 0.005 µM; SI = 166) demonstrated potent and highly selective HDAC6 inhibition, which is in excellent agreement with a recent report on HDAC6-selective proteolysis-targeting chimeras (PRO-TACs) [45]). With respect to the impact of the different cap groups, it can be concluded that the indometacin-derived derivatives (A series) are suitable for less potent but more selective HDAC6 inhibition. In contrast, the celecoxib-based compounds (B and C series) displayed more potent but less selective HDAC6 inhibition.

Determination of Lipophilicity
As a measure of lipophilicity, logD 7.4HPLC as the distribution coefficient at pH 7.4 was determined for all compounds (Table 2) by an HPLC method originally described by Donovan and Pescatore [46]. The logD 7.4HPLC was found to consistently increase with the respective linker lipophilicity in each sub-set (A, B, or C) of compounds. Covering a range from 1.23 for B1 to 3.31 for A7, the compound lipophilicities lie within the range predicted for potential good absorption and bioavailability in vivo and moderate permeability and solubility in general [47].

Antiproliferative Properties of Dual HDAC-COX Inhibitors
The antiproliferative potential of the dual HDAC-COX inhibitors A1-A7, B1-B7, and C2-C4 was determined using MTT assays in the PDAC cell line AsPC1. The PDAC cell line was chosen due to a report by Peulen et al., in which it was demonstrated that the anti-cancer effects of HDACi can be enhanced by the simultaneous inhibition of COX-2 [34]. Vorinostat and celecoxib were included as control compounds; the results are summarized in Table 2. Five out of the 17 tested dual inhibitors showed no effect against the AsPC1 cell line. The remaining compounds revealed IC 50 values ranging from 2.39 to 59.32 µM. Vorinostat (IC 50 AsPC1 = 1.04 µM) turned out to be the most active compound in this screening, while celecoxib displayed no antiproliferative activity.
Compounds A7, B7, C3, and C4 were selected as representative compounds to assess their antiproliferative potential in a larger panel of cell lines including melanoma (MV3), pancreatic cancer (PC-3), breast cancer (MDA-MB-231), squamous cell carcinoma (FaDu), colorectal cancer (HT-29), and glioblastoma (U-87). This initial selection of candidates for a more detailed analysis in different cell lines was based both on lipophilicity and cytotoxic effects in the AsPC1 cell line. Based on the IC 50 values from the COX and HDAC inhibition assays in comparison to the effects on AsPC1 cells, we concluded that some of the compounds were potent inhibitors against the isolated enzymes but not able to enter cells and exert their effect there. B4, for example, was one very potent HDAC inhibitor in the low nanomolar range which showed, however, only a IC 50 Table 3. Interestingly, the dual HDAC-COX inhibitor C3 exerted the most pronounced antiproliferative activities across the seven cell lines with IC 50 values ranging from 2.21 µM (FaDu) to 6.91 µM (U-87). Vorinostat displayed the highest activity in all cell lines, while celecoxib was inactive. To investigate the potential of the combined inhibition of HDAC and COX activity by single target drugs, we also included 1:1 combinations of vorinostat and celecoxib into the assays. The combinations showed similar IC 50 values compared to the treatment with vorinostat alone (Table 3), thus indicating that the activity of vorinostat is not boosted by celecoxib. On the whole, our data indicate that, at least in the seven cell lines used in this study, the simultaneous inhibition of HDAC and COX activity by dual HDAC-COX inhibitors or a combination treatment does not lead to additive or synergistic anticancer activity.

Intracellular Target Engagement and Apoptosis Induction
On average across seven cell lines, compounds C3 and C4 demonstrated the most pronounced antiproliferative activities. Consequently, both compounds were characterized in more detail using vorinostat and celecoxib as controls. First, the cellular HDAC inhibitory activities of the selected compounds were assessed in whole-cell HDAC inhibition assays in PC-3 cells using Boc-Lys(Ac)-AMC as broad-spectrum HDAC substrate; the results are summarized in Figure 1A. In good agreement with the results from the MTT assays (Table 3) had no effect on HDAC activity up to a concentration of 100 µM. To confirm the HDAC inhibition in a cellular environment, PC-3 cells were treated with 5 µM of C3, C4, vorinostat, and celecoxib and subsequently subjected to immunoblot analysis. C3, C4, and vorinostat induced prominent acetylation of histone H3 (a marker of HDAC1-3 inhibition) and αtubulin (a marker of HDAC6 inhibition), while celecoxib did not increase the acetylation levels of both proteins ( Figure 1B). Taken together, the results from the whole-cell HDAC inhibition assays and immunoblot experiments confirmed the intracellular unselective HDAC inhibition by C3 and C4. n.e. = no effect up to 100 µM. 2 1:1 combination of vorinostat and celecoxib.

Intracellular Target Engagement and Apoptosis Induction
On average across seven cell lines, compounds C3 and C4 demonstrated the most pronounced antiproliferative activities. Consequently, both compounds were characterized in more detail using vorinostat and celecoxib as controls. First, the cellular HDAC inhibitory activities of the selected compounds were assessed in whole-cell HDAC inhibition assays in PC-3 cells using Boc-Lys(Ac)-AMC as broad-spectrum HDAC substrate; the results are summarized in Figure 1A. In good agreement with the results from the MTT assays (Table 3), compound C3 (IC50 = 1.64 µM) displayed a more potent inhibition of cellular HDAC activity compared to C4 (IC50 = 2.72 µM). The positive control vorinostat showed the strongest HDAC inhibitory capacity (IC50 = 0.50 µM), while, as expected, celecoxib had no effect on HDAC activity up to a concentration of 100 µM. To confirm the HDAC inhibition in a cellular environment, PC-3 cells were treated with 5 µM of C3, C4, vorinostat, and celecoxib and subsequently subjected to immunoblot analysis. C3, C4, and vorinostat induced prominent acetylation of histone H3 (a marker of HDAC1-3 inhibition) and α-tubulin (a marker of HDAC6 inhibition), while celecoxib did not increase the acetylation levels of both proteins ( Figure 1B). Taken together, the results from the whole-cell HDAC inhibition assays and immunoblot experiments confirmed the intracellular unselective HDAC inhibition by C3 and C4. Whole-cell HDAC inhibition assays of C3, C4, vorinostat, and celecoxib in PC-3 cells; n.e. = no effect up to 25 µM. (B). Representative immunoblot analysis of acetylated α-tubulin and acetylated histone H3. PC-3 cells were incubated for 24 h with C3, C4, vorinostat, and celecoxib at the indicated concentration. Afterwards, cell lysates were immunoblotted with anti-acetyl-α-tubulin and acetylhistone H3 antibodies. GAPDH was used as a loading control. (C). Densitometric analysis of acetylated α-tubulin and acetylated histone H3 of C3, C4 and vorinostat. Intensity was normalized to GAPDH and depictured in relation to vorinostat. . Representative immunoblot analysis of acetylated α-tubulin and acetylated histone H3. PC-3 cells were incubated for 24 h with C3, C4, vorinostat, and celecoxib at the indicated concentration. Afterwards, cell lysates were immunoblotted with anti-acetyl-α-tubulin and acetyl-histone H3 antibodies. GAPDH was used as a loading control. (C). Densitometric analysis of acetylated α-tubulin and acetylated histone H3 of C3, C4 and vorinostat. Intensity was normalized to GAPDH and depictured in relation to vorinostat.
In the next step, the induction of apoptosis following the treatment of PC-3 cells with 5 or 10 µM of C3, C4, vorinostat, and celecoxib for 72 h was determined (Figure 2). The treated cells were stained with annexin V and propidium iodide (PI) and afterwards analyzed by flow cytometry. The percentage of cells that were annexin V positive but PI negative were considered early apoptotic, and the percentage of cells that were both annexin V and PI positive were considered late apoptotic. The treatment with C3 and C4 (5 and 10 µM) led to a significant (p < 0.001) increase in apoptotic cells. These results indicate that the induction of apoptosis contributes to the antiproliferative properties of C3 and C4.
Molecules 2023, 28, x FOR PEER REVIEW 9 of 23 In the next step, the induction of apoptosis following the treatment of PC-3 cells with 5 or 10 µM of C3, C4, vorinostat, and celecoxib for 72 h was determined (Figure 2). The treated cells were stained with annexin V and propidium iodide (PI) and afterwards analyzed by flow cytometry. The percentage of cells that were annexin V positive but PI negative were considered early apoptotic, and the percentage of cells that were both annexin V and PI positive were considered late apoptotic. The treatment with C3 and C4 (5 and 10 µM) led to a significant (p < 0.001) increase in apoptotic cells. These results indicate that the induction of apoptosis contributes to the antiproliferative properties of C3 and C4. Figure 2. Apoptosis of PC-3 cells induced by C3, C4, vorinostat, and celecoxib. C3, C4, vorinostat, and celecoxib were incubated with PC-3 cells for 72 h at the indicated concentrations and apoptosis was detected by annexin V and PI staining; representative data of n = 2 independent experiments performed in triplicates. Statistical analysis to compare annexin V positive cells with control was performed using one-way ANOVA (* p < 0.05, *** p < 0.001).

High Performance Liquid Chromatography (HPLC)
For analytical purposes, a Thermo Fisher Scientific GmbH (Waltham, MA, USA) UltiMate TM 3000 UHPLC system with a Macherey Nagel (Düren, Germany) Nucleodur 5 µm C18 100 Å column (250 × 4.6 mm) was used with bidistilled H 2 O with 0.1% TFA (A) and ACN (B) as mobile phase. After the column was equilibrated for 5 min with the initial conditions A/B 95:5, a linear gradient from 5 to 95% B over 15 min was performed followed by a 5 min isocratic phase of 95% B. The method was carried out with a flow rate of 1 mL/min at 25 • C and the chromatograms were acquired via UV absorption detection at 254 nm.
For the preparative purification a Varian ProStar HPLC system, either a preparative Macherey Nagel Nucleodur 5 µm C18 HTec column (150 × 32 mm) or a semi-preparative Phenomenex (Aschaffenburg, Germany) Jupiter 5 µm C18 100 Å column (250 × 10 mm) was used with bidistilled H 2 O with 0.1% TFA (A) and ACN (B) as mobile phase. After the column was equilibrated for 5 min with the initial conditions A/B 95:5, a linear gradient from 5 to 95% B over 15 min was performed followed by a 10 min isocratic phase of 95% B (method A). For method B, the linear gradient from 5 to 95% B was extended to 20 min. The methods were carried out with flow rates of 15 mL/min for the preparative and 4 mL/min for the semi-preparative column at room temperature and the chromatograms were acquired via UV absorption detection at 220 and 254 nm.
Furthermore, a Knauer (Berlin, Germany) AZURA preparative HPLC system with a preparative Macherey Nagel Nucleodur 5 µm C18 HTec column (250 × 32 mm) and the fraction collector Foxy R1 was used. The eluents were the same as mentioned above. For method C, linear gradients from 5-95 % B over 20 min and for method D over 25 min were used. The methods were carried out with a flow rate of 20 mL/min at room temperature and the chromatograms were acquired via UV absorption detection at 220 and 254 nm.

Nuclear Magnetic Resonance Spectroscopy (NMR)
For the characterization of the compounds proton ( 1 H), carbon ( 13 C), and fluorine ( 19 F), NMR spectra were acquired on a Bruker (Billerica, MA, USA) Avance III HD 400, a Varian/Agilent (Santa Clara, CA, USA) Mercury Plus 400, or a Varian/Agilent Mercury Plus 300 at room temperature. The used frequencies were 300 or 400 MHz for the 1 H spectra, 75, 76, or 101 MHz for the 13 C spectra and 377 MHz for the 19 F spectra. The chemical shifts δ were given in parts per million (ppm) and normalized on the residual solvent signal of the deuterated solvents (CDCl 3 : 1 H-NMR: 7.26 ppm, 13 C-NMR: 77.16 ppm; DMSO-d 6 : 1 H-NMR: 2.50 ppm, 13 C-NMR: 39.52 ppm). The coupling constants J were given in Hertz (Hz), and the multiplicities of the signals were described as singlet (s), doublet (d), triplet (t), quartet (q), quintet (p), multiplet (m), and their corresponding combinations. They were given as they were measured and thus might disagree with the expected values.

Mass Spectrometry (MS)
For the characterization of the compounds, high resolution electrospray ionisation mass spectra (HR-ESI-MS) were recorded with a Bruker Daltonics (Bremen, Germany) micrOTOF coupled to a LC Packings Ultimate HPLC system and controlled by micrOTOF-Control3.4 and HyStar 3.2-LC/MS.

TNBS-Test
One drop of 2,4,6-trinitrobenzenesulfonic acid in DMF (1%) and DIPEA in DMF (10%) were added to some resin beads and incubated for 3 min. A red colouring of the resin beads confirmed the presence of free amino groups.

Determination of the Loading
The loading was determined on a Shimadzu (Kyōto, Japan) double beam photometer UV-160A. To a defined amount of the resin (exact weight 5 mg), a solution of 20% piperidine in DMF (500 µL) was added and incubated for 5 min at room temperature. The solution was transferred to another tube and the remaining resin was treated with the mentioned deprotection solution (500 µL) again for 5 min. The solutions were combined and the absorbance was measured at 300 nm at room temperature in a quartz cuvette (volume: 3500 µL, path length: 10 mm, 100-QS, Hellma Analytics (Müllheim, Germany)). By using the Lambert-Beer law A = ε*c*d (A = absorbance, ε = molar extinction coefficient with ε 300 nm (dibenzofulvene) = 7800 L/(mol*cm), c = concentration in mol/L, d = optical path length in cm), the concentration was calculated, and by including the mass of the weighed resin, the loading in mmol/g was determined.

General Procedures General Procedure A: Coupling of the Cap Group
The given amounts refer to a synthesis scale of 0.1 mmol of the linker-modified resins PR1-7. The linker-modified resin PR1-7 (1.00 eq., defined loading) was swollen in DMF (1.5 mL) for 30 min. Then, the Fmoc protecting group was cleaved by treatment with 20% piperidine in DMF (1.5 mL) for 5 min. This step was repeated once. Afterwards, the resin was washed with DMF (5 × 2 mL) and DCM (5 × 2 mL). To confirm that the Fmoc deprotection was successful, a TNBS-test was carried out with some resin beads, and afterwards the resin was washed with DMF (5 × 2 mL) again. Next, a solution of the respective acid Cap-COOH A-C (2.00 eq.), HATU (2.00 eq.), and DIPEA (3.00 eq.) in DMF (1 mL/mmol acid) was agitated for 5 min, added to the resin, and incubated for 18 h at room temperature. Subsequently, the resin was washed with DMF (10 × 2 mL) and DCM (10 × 2 mL). The completion of the reaction was monitored by the TNBS-test and the resin was washed with DCM (10 × 2 mL) again and dried in vacuo. For the cleavage of the coupling product, the resin was treated with 5% TFA in DCM (1 mL/40 mg resin) for 1 h at room temperature. The resulting solution was concentrated under reduced pressure followed by the purification of the crude products A1-A7, B1-B7, and C2-C4 via (semi)-preparative HPLC. The product fractions were lyophilized, and the products A1-A7, B2-B7, and C2-C4 were isolated in purities of 95% or higher, while B1 was obtained in a purity of 89%.

MTT Cell Viability Assay
The intrinsic cytotoxicity of test compounds was determined by a MTT assay as previously described [42,49]. AsPC1, MDA-MB-231, FaDu, HT-29, and U87-MG cells were seeded at a density of 5000 cells/well, and MV3 and PC-3 cells were seeded at a density of 3000 cells/well in 96-well plates (Starlab GmbH, Hamburg, Germany or Greiner Bio-One, Frickenhausen, Germany); all were allowed to attach overnight. Subsequently, cells were exposed to increasing concentrations of the test compounds. After 72 h, MTT solution (5 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, in phosphatebuffered saline, Applichem, Darmstadt, Germany or Sigma Aldrich, Steinheim, Germany) was added to determine cell survival. The formazan dye was dissolved in DMSO after 1 h and absorbance was measured at 570 nm and 690, respectively, nm in a Multiskan microplate photometer (Thermo Fisher Scientific, Waltham, MA, USA) and a Cytation 5 Imaging Reader (BioTEK, Santa Clara, CA, USA). Concentration effect curves were generated using nonlinear regression with GraphPad Prism and mean IC 50 values were calculated based on at least three independent experiments in duplicates.

Apoptosis Assay
PC-3 cells (3 × 10 4 cells/mL) were seeded into 6-well plates and incubated with different concentrations of compounds or vehicle control (DMSO) for 72 h. Cells were stained with annexin V and propidium iodide and analyzed by flow cytometry (Guava ® easyCyte TM , Luminex, Austin, TX, USA), according to the manufacturer's protocol (Cat-alog#640914, BioLegend, SanDiego, CA, USA). Two independent experiments were performed in triplicates.
Statistical analysis was performed with GraphPad Prism (Graph-Pad Software, San Diego, CA, USA). The results were expressed as mean with standard deviation (SD). The normality of the data was determined using Shapiro-Wilk's test. A one-way analysis of variance (ANOVA) was performed to determine statistical differences between means. If the difference between means was significant (p < 0.05), the mean of each column was compared with the mean of the control and corrected for multiple comparisons using Dunnett's post-hoc test. Differences were considered statistically significant at * p < 0.05, *** p < 0.001.

Determination of COX Inhibition
The COX inhibition potency against ovine COX-1 and human COX-2 was determined using the fluorescence-based COX assay "COX Fluorescent Inhibitor Screening Assay Kit" (catalog number 700100; Cayman Chemical, Ann Arbor, MI, USA) according to the manufacturer's instructions as previously reported by us [40,41]. All compounds were assayed in a concentration range of 10 nM to 100 µM in a 10-fold dilutions series, and every inhibitor concentration was assayed in duplicate. If necessary, a narrower concentration range with further concentrations in between was applied to determine IC 50 . Celecoxib was used as internal control. IC 50 values were estimated using a nonlinear logistic regression fitting procedure (sigmoidal dose−response model) with GraphPad Prism.

Conclusions
In this work, we have prepared seven different preloaded resins, which were utilized for the solid-phase parallel synthesis of a set of 17 dual HDAC-COX inhibitors. All synthesized compounds were evaluated for the inhibition of COX-1, COX-2, HDAC1, HDAC6, and antiproliferative activity against the PDAC cell line AsPC1. Several of the compounds under study demonstrated a pronounced COX and HDAC inhibitory activity. The selectiv-ity of respective dual inhibitor for the different isoforms turned out to be highly dependent on the nature of the linker and COX inhibitor scaffold used. Whole-cell HDAC inhibition assays and immunoblot analysis confirmed that C3 and C4 are capable of inhibiting HDAC activity in a cellular environment. In addition, both C3 and C4 caused a significant increase in apoptotic cells, which indicates that the induction of apoptosis contributes to the anticancer properties of C3 and C4. However, at least in the seven cancer cell lines used in this study, the simultaneous inhibition of HDAC and COX activity by dual HDAC-COX inhibitors or combination treatments did not result in additive or synergistic anticancer activities, which disagrees with previous reports.