Sulfonamide Anilides, a Novel Class of Histone Deacetylase Inhibitors, Are Antiproliferative against Human Tumors

In eukaryotic cells, histone acetylation/deacetylation is important in transcriptional regulation (1). Transcriptionally active genes associate with hyperacetylated chromatin, whereas transcriptionally silent genes associate with hypoacetylated chromatin (2). Chromatin acetylation is controlled by the opposite effects of two families of enzymes, histone acetyltransferases and HDACs² (3, 4). Histone acetyltransferases, as transcription coactivators, catalyze the addition of acetyl groups on the ε-amino group of lysine residues in the NH₂-terminal tails of core histones. Conversely, HDACs, as transcription corepressors, remove the acetyl groups from the acetylated lysines in histones (3, 4).

There are three subclasses of HDAC enzymes in humans (5). Class I enzymes (HDAC1, HDAC2, HDAC3, and HDAC8; Refs. 6, 7, 8, 9) are homologues of yeast Rpd3 deacetylase, whereas class II enzymes are homologues of yeast Hda1 deacetylase (HDAC4, HDAC5, HDAC6, HDAC7, HDAC9, HDAC10, and HDAC11; Refs. 10, 11, 12, 13, 14, 15). Class III enzymes are homologues of yeast Sir2 (16). Unlike class I and class II enzymes, class III HDAC activities require NAD⁺, and they are not sensitive to TSA (16). Deregulation of HDAC activity can cause malignant diseases in humans (17, 18).

Small molecule inhibitors of HDACs have emerged as antiproliferative agents because of their ability to inhibit proliferation, induce apoptosis, and/or cell differentiation in cancer cells (5, 19, 20). These HDAC inhibitors include sodium butyrate, TSA, suberoylanilide hydroxamic acid, oxamflatin, MS-275 (21), trapoxin, depudecin, apicidin, and FR901228 (see Refs. 5, 19, 20 for reviews). It is believed that induction of p21^WAF1/Cip1 and blockade of tumor cell cycle progression are critical for antitumor activities of all of the HDAC inhibitors (22, 23, 24, 25).

In the course of developing agents with antitumor activities, we aimed to develop novel HDAC inhibitors with potency and safety. In previous publications, we described our efforts to develop novel hydroxamates as HDAC inhibitors (26, 27). As nonhydroxamates generally have more pharmaceutically desirable properties than hydroxamates (19), we developed a novel class of HDAC inhibitors, sulfonamide anilides, based on their corresponding hydroxamates (28). The results presented here show that sulfonamide anilides exhibit antitumor activity both in vitro and in vivo, and compared with MS-275, another anilide-based HDAC inhibitor in clinical trial, our compound (Compound 2) does not exhibit gross toxicity in mice. Thus, sulfonamide anilides could be potential therapeutic agents in treating human cancers.

All sulfonamide anilides and MS-275 were prepared in-house (26, 28, 29). Compound 1 (N-2-aminophenyl-3-[4-(3,4-dimethoxy benzenesulfonylamino)-phenyl]-2-propenamide), Compound 2 (N-2-aminophenyl-3-[4-(4-methyl benzenesulfonylamino)-phenyl]-2-propenamide), Compound 3 (N-2-aminophenyl-3-[4-(4-phenyl benzenesulfonylamino)-phenyl]-2-propenamide), and Compound 4 (N-phenyl-3-[4-(4-phenylbenzenesulfonylamino)-phenyl]-2-propenamide) were all prepared with >95% purity. Other chemicals were purchased from Sigma. Each compound’s cLogP was calculated using Pallas program.

HMEC cells were obtained from BioWhittaker (Walkersville, MD). All other cell lines used were from American Type Culture Collection (Manassas, VA). Human normal and cancer cells were all cultured after the vendor’s instructions. In growth curve analysis, cells were plated 24 h before treatment with HDAC inhibitors. Cells were counted by trypan blue exclusion in triplicates at various time points.

Human HDAC1 cDNA was generated by reverse transcriptase-PCR reactions based on human HDAC1 coding sequence (GenBank accession no. U50079). Insect High Five cells (Invitrogen) were used to produce recombinant Flag-tagged HDAC1 using pBlueBAC vector (Invitrogen). HDAC1 recombinant enzymes were partially purified by a Q-Sepharose column (Pharmacia) followed by purification with anti-Flag M2 affinity gels (Sigma). Deacetylase enzyme assays were carried out in the assay buffer [40 mm Tris-Cl (pH 7.6), 20 mm EDTA, and 50% glycerol] using [³H]-labeled acetylated histones as described in literatures (30). The IC₅₀ for inhibitors were determined by analyzing dose-response inhibition curves.

Whole cell extracts or acid extracted histones prepared from inhibitor-treated cells were analyzed by SDS-PAGE. Proteins were transferred on the polyvinylidene difluoride membrane and probed with various primary antibodies. Primary antibodies were ordered from Santa Cruz Biotechnology, except for acetylated H4 or acetylated H3 antibodies (Upstate Biotechnology), or antibodies against p21^WAF1/Cip1 (Transduction Laboratories). Horseradish peroxidase-conjugated secondary antibodies (Sigma) were used, and the enhanced chemiluminescence (Amersham) was followed for detection.

Cells seeded in 96-well plates were incubated for 72 h at 37°C in a 5% CO₂ incubator. MTT (Sigma) was added at a final concentration of 0.5 mg/ml and incubated with the cells for 4 h before an equal volume of solubilization buffer [50% N,N-dimethylformamide, 20% SDS (pH 4.7)] was added onto cultured cells. After overnight incubation, solubilized dye was quantified by colorimetric reading at 570 nm using a reference at 630 nm. A values were converted to cell numbers according to a standard growth curve of the relevant cell line. The concentration, which reduces cell numbers to 50% of those of DMSO-treated cells, is determined as MTT IC₅₀.

Cells were treated with inhibitors for 16 h, harvested, and fixed by 70% ethanol at −20°C. Nucleic acids from fixed cells were stained with propidium iodide (50 μg/ml). DNA content was measured by using a fluorescence-activated cell sorter.

Female BALB/c nude mice (obtained from Charles River Laboratories) were used at ages 8–10 weeks. Human colon carcinoma HCT116 cells (2 million) were injected s.c. in the animal flank and allowed to form solid tumors. Tumor fragments were serially passaged a minimum of three times before use in the experiments described. Tumor fragments (∼30 mg) were implanted s.c. through a small surgical incision under general anesthesia. HDAC inhibitors were dissolved as clear solutions in vehicles (either 100% DMSO or 20% DMSO, 15% ethanol, and 65% PBS acidified with 0.2 N HCl). The pH value for the latter vehicle was 2.0. Vehicles alone and HDAC inhibitors in vehicles were administered daily i.p. by injection. The maximum volume for daily injection was 100 μl/animal. Tumor volumes and gross body weight of animals were monitored twice weekly for up to 3 weeks. Each experimental group contained at least 8 mice. At the end of each experiment, blood and spleen were collected from mice. Spleens were weighed and routine hematological parameters were determined. Student’s tests were used to analyze the statistical significance between numbers in data sets.

Total RNAs from HCT116 human cancer cells were extracted using RNeasy mini kit (Qiagen). Generally, 50 μg of total RNA were used to synthesize cDNA probes using Atlas Pure Total RNA Labeling System (Clontech) before their hybridization on the Atlas Human Cancer cDNA Expression Array membranes (Clontech). After hybridization and washing, array membranes were exposed to Cyclone Phosphor-Screen (Packard) for data analysis.

We designed and synthesized a family of sulfonamide anilides (28). In Table 1, four of these compounds are presented. Because HDAC1 is ubiquitously expressed in many cancer cell lines and is important for transcriptional regulation, Compounds 1–4 were tested against partially purified recombinant human HDAC1 enzymes. As shown in Table 1, Compounds 1–3 can all significantly inhibit human HDAC1 activity in vitro with IC₅₀s between 1 and 6 μm, whereas MS-275 can inhibit HDAC1 with an IC₅₀ of 1.2 μm. Compound 4, the des-amino analogue of Compound 3, has an IC₅₀ > 100 μm against the human HDAC1 enzyme, suggesting that the integrity of the aniline group is essential for the inhibitory activity of this class of compounds.

To determine whether Compounds 1–4 can induce core histone acetylation in cultured human cancer cells, human bladder carcinoma T24 cells, which do not express p53, were treated with escalating doses of these compounds. T24 cells were chosen because their basal levels of histone acetylation are low. As shown in Fig. 1, Compounds 1–3 can all induce core histone H4 or H3 acetylation in a dose-dependent manner. Compound 4, which did not inhibit HDAC1, did not induce histone acetylation even at 25 μm (Fig. 1). Among Compounds 1–3, Compound 2 can induce histone acetylation most strongly, whereas the ability of MS-275 in induction of histone acetylation in cells is about 2–3 times stronger than that of Compound 2 (Fig. 1). Thus, the ability of these compounds to induce histone acetylation in cancer cells largely correlated with their ability to inhibit HDAC1 in vitro. Although Compound 3 was more potent than Compounds 1–2 on inhibition of HDAC1 enzyme activity in vitro, Compound 3 was less effective on induction of histone acetylation in whole cells, possibly because of its relatively high cLogP value (cLogP = 4.8) compared with that of Compound 1 or Compound 2 (cLogP = 3.2 or 3.6, respectively.) Induction of histone acetylation by this class of compounds was independent of p53 and was not cell line specific. Similar induction of histone acetylation by these compounds was also observed in other human cancer cell lines that express p53 (A549 and HCT116, data not shown).

Antiproliferative activities of Compounds 1–3 against human cancer cells correlated well with their abilities to induce histone acetylation in intact cells (Fig. 1). Among these compounds, Compounds 1 and 2 exhibited significant antiproliferative activity against a broad panel of human cancer cells as shown in Table 2. Antiproliferative activities of Compounds 1 and 2 were not limited to a specific tumor type and were p53 independent. In contrast, Compound 3 was less potent in inhibiting growth of human cancer cell lines, whereas Compound 4 could not inhibit the growth of any of the human cancer cell lines.

Compounds 1 and 2 selectively inhibited growth of human cancer cells but not normal human diploid fibroblasts (MRHF) or normal human epithelial cells (HMEC) as shown in Table 2. This effect is consistent with a report that HDAC inhibitors do not cause growth inhibition in human normal dermal fibroblasts (31).

To determine whether sulfonamide anilides blocked growth of human cancer cells in vitro in a reversible or irreversible manner, we performed a growth curve analysis of HCT116 cancer cells treated with Compound 2 (5 μm). As shown in Fig. 2,A, although the continuous presence of Compound 2 in cell culture media significantly blocked growth of HCT116 cells and induced histone acetylation, the wash-out of Compound 2 from cells after 24-h initial treatment allowed the growth of the cells to recover and histone acetylation decreased to basal level (Fig. 2,B). Induction of histone acetylation by Compound 2 in HCT116 cells directly correlated with its ability to block proliferation of these cancer cells (Fig. 2 B).

Compounds 1 and 2 induced a dose-dependent G₂-M arrest in the human cancer cell lines HCT116 (Fig. 3, A and B), T24, A549, Du145, and MCF-7 (data not shown) but not in normal HMEC cells (Fig. 3, C and D). However, unlike in a previous report where MS-275 (1 μm) can induce G₀-G₁ arrest in human cancer A2780 cells (21), we found that MS-275 (at 1–10 μm) similarly induced G₂-M arrest as Compound 2 in those human cancer cell lines we have tested (data not shown).

We used a human colon cancer xenograft model (HCT116) to assay in vivo antitumor activity of Compounds 1 and 2. Compound 1 exhibited weak antitumor activity in vivo. From two independent experiments, the growth of tumors in mice treated with Compound 1 at 40 mg/kg body weight (MPK) for 18 days by daily i.p. injection was inhibited by 33 and 26% compared with that of control mice treated with only the vehicle (DMSO).

For Compound 2, three independent experiments were performed. In the first two experiments, Compound 2 was administered by i.p. injection into animals at 40 MPK daily for 18 days. Growth of tumors in treated mice was inhibited by 46 and 51% compared with that of control mice treated with only the vehicle (DMSO). In the third experiment, Compound 2 or MS-275 were given to mice at 20 MPK/day for 20 days by i.p. injection. We found that the growth of tumors in mice treated with Compound 2 was inhibited by 48%, whereas the growth of tumors in mice treated with MS-275 was inhibited by 54% (Fig. 4 A) compared with that of tumors of vehicle-treated animals. Thus, Compound 2 has significant antitumor activity in this human colon cancer xenograft model in vivo. This antitumor activity is comparable with that of MS-275 under the same experimental condition.

All mice treated with Compound 2 or MS-275 (both at 20 MPK) in Fig. 4,A did not show gross body weight loss. However, MS-275 reduced the number of peripheral RBCs by 12% compared with vehicle (P < 0.0005; Fig. 4,B). The RBC counts of Compound 2-treated animals were comparable with those of animals treated with vehicle only (P < 0.5; Fig. 4,B). MS-275 also reduced total peripheral WBC counts in mice by almost 50% compared with that of vehicle-treated mice (P < 0.0005; Fig. 4,C), whereas Compound 2 did not significantly alter the WBC counts of mice (P < 0.1; Fig. 4 C).

Spleen weights from animals treated with MS-275 or Compound 2 (both at 20 MPK for 20 days by daily i.p. injection) in two independent experiments were pooled and analyzed. As shown in Fig. 5, spleen weights of animals treated with MS-275 were reduced to 77% of those of vehicle-treated animals (P < 0.05). However, Compound 2 did not affect spleen weights of animals compared with those of mice treated with vehicle alone (P > 0.5). Our results suggest that Compound 2 has significantly less gross toxicity than MS-275 in vivo.

To understand the mechanism of antitumor activity of Compound 2 and MS-275, we performed cDNA array expression analysis in HCT116 cells treated with Compound 2 or MS-275. In the analysis, only alterations of gene expression of 2-fold or greater were considered significant. After 24 h of treatment, Compound 2 (5 μm) altered expression of 30 of 588 genes, which we analyzed, whereas MS-275 (1 μm) affected expression of 75 genes under the same experimental condition.

From the cDNA expression analysis, the antitumor mechanism of MS-275 overlaps, to some degree, with that of Compound 2. Among the 30 genes that were affected by Compound 2, transcriptions of 14 genes were also altered by MS-275. In Table 3, 5 of these genes, the expression of which was commonly altered by both MS-275 and Compound 2, were listed.

In Fig. 2,A, growth inhibitory effect of Compound 2 on cancer cells was reversible. Does the gene expression profile of HCT116 cells treated with Compound 2 correlate with the phenotype response? We found that after 48 h of treatment, Compound 2 altered expression of more genes (67 genes) in HCT116 cells compared with that after 24 h of treatment (30 genes). Expression of 46 of 67 genes was dependent on the continuous presence of Compound 2. For example, Compound 2 significantly induced gene expression of p21^WAF1/Cip1 by 2- or 3-fold after 24 or 48 h of treatment, respectively (Table 3). Transcription of p21^WAF1/Cip1 returned to basal level if Compound 2 was depleted from the media for the last 24 h after the initial 24-h treatment. Similarly, expression of cyclins A or B1 was down-regulated significantly by Compound 2 after 24- to 48-h treatment, but the depletion of Compound 2 reversed the down-regulation of expression of these two cyclins (Table 3). Thus, gene expression changes correlated with the phenotypic responses of cells treated with Compound 2 (Fig. 2).

At the protein level, Compound 2 significantly induced the expression of p21^WAF1/Cip1 while down-regulating the expression of cyclin A and cyclin B1 in HCT116 cells (Fig. 6). Induction of p21^WAF1/Cip1 expression by Compound 2 was already dramatic at 24 h and peaked at 48 h posttreatment in HCT116 cells (Fig. 6). Down-regulation of the expression of cyclins A and B1 by Compound 2 in HCT116 cells peaked at ∼48 h posttreatment (Fig. 6). Alteration of expression of these three genes was largely maintained even at 72 h posttreatment. The ability of Compound 2 to alter the expression of these cell cycle regulators was consistent with its ability to induce cell cycle arrest and block cell proliferation of HCT116 cells (Fig. 2, Fig. 3, and Table 2).

Induction of p21^WAF1/Cip1 and down-regulation of cyclins A and cyclin B1 expression at the protein level by Compound 2 in cancer cells was independent of p53 status and was not cell line specific; similar observations were obtained in all human cancer cell lines we tested (T24, A549, and Du145 cells; data not shown). These observations are consistent with other reports that HDAC inhibitors can induce p21^WAF1/Cip1 expression and down-regulate cyclin protein expression (21, 22, 23, 24, 25).

Both MS-275 (21, 29) and the compounds presented in this paper are anilides. The integrity of the aniline group is essential for these anilides to inhibit HDAC enzymes. It is not clear how anilides inhibit HDAC enzyme activity. According to structural data on bacterial HDAC-like enzymes, members of the HDAC enzyme family may possess a tube-like active site (32). The hydroxamate groups of TSA and oxamflatin can coordinate the zinc ion located in the active site and contact amino acid side chains deep in the pocket (32). Compared with its corresponding hydroxamate (26), Compound 2 is at least a 10 fold weaker inhibitor of HDAC1 in vitro. It is possible that the aniline group can function as a weak zinc-chelating group (33), or that the amide group and/or 2-amino group may contact key amino acids in the active site but may not coordinate the Zn²⁺ ion. Additional studies are required to understand the exact mechanism of action of anilide-based HDAC inhibitors.

Compound 2 induces morphological change in human cancer cells (i.e., HCT116) at concentrations > 5 μm (data not shown). The expression of several keratins (i.e., keratins 8 and 19) was significantly induced in HCT116 cells at the RNA level (Table 3) by Compound 2 in a reversible manner. At the protein level, we found that Compound 2 induced the expression of keratin 19 and gelsolin (Fig. 6). We found that other HDAC inhibitors such as MS-275 and TSA could also induce expression of keratin genes not only in HCT116 cells but also in other cancer cell lines tested (data not shown). It is possible that keratins, similarly to gelsolin, are mediators of HDAC inhibitor-induced morphological changes in human cancer cells. In the literature, other HDAC inhibitors can similarly induce expression of gelsolin and keratins in various human cancer cells (21, 24, 34).

Compound 2 induced apoptosis at higher concentrations (e.g., 25 μm) in HCT116 cells (data not shown). Consistent with its ability to induce apoptosis, expression of an antiapoptotic protein, survivin (35), was significantly down-regulated at both the RNA and protein level in these cells (data not shown). Expression of BAD (36), a proapoptotic protein, was elevated by Compound 2 at the RNA level in HCT116 cells. It is not clear whether the induction of apoptosis by Compound 2 contributes to its antitumor activity in human cancer cells.

Compound 2 appears to have comparable antitumor activity in vivo to MS-275 (Fig. 4,A), a compound currently in Phase I clinical trial. However, Compound 2 appears to have an improved safety profile compared with MS-275 (Figs. 4, B and C, and 5), although Compound 2 and MS-275 are both aniline-based HDAC inhibitors. The basis for the improved therapeutic window in vivo for Compound 2 needs to be investigated further to develop potent and safe cancer therapeutic agents of this class.

In summary, we demonstrated that sulfonamide anilides, especially Compound 2, are novel HDAC inhibitors with significant antitumor activity in vitro and in vivo. The antitumor activity of Compound 2 correlates well with its ability to arrest cell cycles and to alter expression of cell cycle regulators, i.e., p21^WAF1/Cip1, cyclin A, and cyclin B1. We conclude that sulfonamide anilides may be useful as antiproliferative agents in future human cancer therapy.

We thank Marie-France Robert, Isabelle Dupont, and Judith Cotton-Montpetit for production of recombinant enzymes, and Cindy Lalancette, Eric Chan, and Przemyslaw (Mike) Sapieha for technical assistance. We also thank Dr. Arkadii Vaisburg, Dr. Gabi Rahil, and Normand Beaulieu for their valuable suggestions and support.