CRA-024781: a novel synthetic inhibitor of histone deacetylase enzymes with antitumor activity in vitro and in vivo

Introduction

The acetylation of lysine residues on the amino terminal tails of nucleosomal histone proteins has a crucial role in the regulation of the chromatin structure. The acetylation state of histones is maintained by the opposing actions of histone acetyl transferase enzymes and histone deacetylases (HDAC). There are 11 known isoforms of the classic HDAC family, denoted HDAC1 to HDAC11 (1, 2). In addition to histones, HDAC enzymes are known to deacetylate other proteins, including α-tubulin (3), suggesting complex, multifunctional roles for HDACs in vivo.

One of the best studied roles of HDAC enzymes is their ability to function as transcriptional corepressors. HDAC enzymes are recruited to specific promoters to repress transcription by deacetylating histones proximal to the promoter. Histone deacetylation is thought to repress transcription by increasing the affinity of histones for DNA, thereby sterically preventing access of transcription factors to promoter regions. Inhibitors of HDAC enzymes have been shown to alter the expression of ∼2% to 10% of transcribed genes in the human genome. Interestingly, these changes include genes both strongly up-regulated and down-regulated, suggesting that the role of HDAC enzymes is more complex than simple corepression. Genes up-regulated by HDAC inhibitors include the tumor suppressor p21^Cip1/WAF1, p16, gelsolin, histones, and caspases, whereas down-regulated genes include Her2/neu, vascular endothelial growth factor, and cyclin proteins (4).

Alterations in the acetylation state of specific lysines have been associated with human cancer (5). It has been hypothesized that aberrant patterns of histone acetylation function to maintain the transformed state of human tumors, a state that can be reversed by inhibiting HDAC enzymes. Several studies have shown that the treatment of tumor cells with HDAC inhibitors results in growth arrest, differentiation, apoptosis, and inhibition of tumor angiogenesis (6). The precise antitumor mechanism of HDAC inhibition is unclear, and may be the result of altered transcription, or perhaps may result directly by affecting chromosome stability, assembly, or function (7, 8).

A variety of structurally distinct compounds have been described to inhibit HDAC enzymes, including hydroxamic acids, benzamides, electrophilic ketones, and cyclic peptides (9). Several hydroxamic acid–based inhibitors have been described (suberoylanilide hydroxamic acid, trichostatin A, oxamflatin, LAQ824, LBH589, and PXD101), some of which are currently in clinical trials for cancer. We have focused on the development of a hydroxamic acid–based series of compounds that has been optimized for in vivo efficacy and therapeutic index (10). Herein, we describe the in vitro and in vivo characteristics of a novel HDAC inhibitor, CRA-024781. CRA-024781 is currently undergoing evaluation in clinical trials for cancer.

Materials and Methods

Cell Lines

The following human tumor cell lines were obtained from the American Type Tissue Culture Collection: HCT116, HCT-15, BT-549, NCI-H226, CWR-22RV1, MCF-7, NCI-PC3, DLD-1, SKOV-3, and OVCAR-3. Human umbilical vein endothelial cells (HUVEC) were obtained from Cambrex (Baltimore, MD). For in vivo studies, HCT116 human colorectal tumor cells were cultured and expanded in McCoy's 5A medium (supplemented with 1.5 mmol/L l-glutamine and 10% fetal bovine serum), and DLD-1 human colorectal tumor cells were cultured and expanded in RPMI 1640 (supplemented with 2 mmol/L l-glutamine, 1 mmol/L sodium pyruvate, and 10% fetal bovine serum).

HDAC Activity

HDAC activity was measured using a continuous trypsin-coupled assay that has been previously described in detail (11). For inhibitor characterization, measurements were done in a reaction volume of 100 μL using 96-well assay plates. For each isozyme, the HDAC protein in reaction buffer [50 mmol/L HEPES, 100 mmol/L KCl, 0.001% Tween 20, 5% DMSO (pH 7.4), supplemented with bovine serum albumin at concentrations of 0% (HDAC1), 0.01% (HDAC2, 3, 8, and 10), or 0.05% (HDAC6)] was mixed with inhibitor at various concentrations and allowed to incubate for 15 minutes. Trypsin was added to a final concentration of 50 nmol/L, and acetyl-Gly-Ala-(N-acetyl-Lys)-AMC was added to a final concentration of 25 μmol/L (HDAC1, 3, and 6), 50 μmol/L (HDAC2 and 10), or 100 μmol/L (HDAC8) to initiate the reaction. Negative control reactions were done in the absence of inhibitor in replicates of eight. Reactions were monitored in a fluorescence plate reader. After a 30-minute lag time, the fluorescence was measured over a 30-minute time frame using an excitation wavelength of 355 nm and a detection wavelength of 460 nm. The increase in fluorescence with time was used as the measure of the reaction rate. Inhibition constants K_i(app) were obtained using the program BatchKi (12).

Cell Proliferation Assay

Ten tumor cell lines and HUVEC were cultured for at least two doubling times, and growth was monitored at the end of compound exposure using an Alamar blue (Biosource, Camarillo, CA) fluorometric cell proliferation assay as previously described (13). The compound was assayed in triplicate wells in 96-well plates at nine concentrations using half-log intervals ranging from 0.0015 to 10 μmol/L. The final DMSO concentration in each well was 0.15%. The concentration required to inhibit cell growth by 50% (GI_50%) and 95% confidence intervals were estimated from nonlinear regression using a four-parameter logistic equation.

Histone and Tubulin Acetylation, p21^Cip1/WAF1 Accumulation, Poly(ADP-Ribose) Polymerase Cleavage, and Phosphorylated Histone Variant H2AX

Acetylated histone, acetylated tubulin, p21^Cip1/WAF1, poly(ADP-ribose) polymerase (PARP) cleavage, and phosphorylated histone variant H2AX (γH2AX) proteins were detected by Western blotting in cells treated with CRA-024781. Tumor cells and subconfluent HUVEC were cultured for 18 hours in the presence of CRA-024781 concentrations ranging from 0.01 to 10 μmol/L. Cells were then collected and lysed in lysis buffer (M-Per, Pierce, Rockford, IL) containing protease inhibitors (Complete, Mini, EDTA-free; Roche, Indianapolis, IN) and phosphatase inhibitors (phosphatase inhibitor cocktail set II; Calbiochem, La Jolla, CA). Lysates were solubilized in SDS-PAGE reducing sample buffer, boiled, and electrophoresed in Novex Tris-glycine gels (Invitrogen, San Diego, CA). The gels were blotted onto nitrocellulose and probed with either an antiacetyl lysine antibody (Upstate, Charlottesville, VA) to detect acetylated histone, an anti-acetylated tubulin antibody (Sigma, St. Louis, MO), an anti-p21^Cip1/WAF1 antibody (BD Biosciences, San Jose, CA), an anti-PARP antibody (Cell Signaling, Danvers, MA), or an anti-γH2AX antibody (Cell Signaling). The blots were washed, incubated with an appropriate horseradish peroxidase–conjugated secondary antibody (The Jackson Laboratory, Bar Harbor, ME) and the blots were developed for enhanced chemiluminescence (Pico Substrate Kit, Pierce).

Pharmacokinetic Analysis

CRA-024781 was formulated in 30% HP-cyclodextrin in water and a 10 mg/kg bolus dose was administered i.v. to female BALB/c nu/nu mice (16–22 g). Blood samples were collected over 24 hours. Plasma was prepared from each blood sample with lithium heparin and was assayed for CRA-024781 by liquid chromatography with tandem mass spectrometry. Mouse plasma samples were assayed for CRA-024781 by high-performance liquid chromatography with tandem mass spectrometric detection. Plasma samples were extracted by protein precipitation using acetonitrile (MeCN). A Hypersil C-18 column, 50 × 2.1 mm was used for sample separation. Samples were separated on reversed-phase high-performance liquid chromatography under linear gradient conditions using water/MeCN as mobile phases, at a flow rate at 0.5 mL/min. Samples were ionized under electrospray and quantified by multiple reactive monitoring, recording the transition from the molecular ion to the product ion: 398 → 200 m/z. Linearity (>0.25–1,000 ng/mL) was achieved with interday and intraday coefficient of variation (%) and deviation of the mean from theoretical (%) ±15%. The lower limit of quantitation was 0.25 ng/mL. Pharmacokinetic variables were estimated with compartmental methods using WinNonlin-Pro version 4.1 (Pharsight Corp., Mountain View, CA). Pharmacokinetic calculations were done using nominal doses and nominal collection times.

In vivo Efficacy Studies

Female BALB/c nu/nu mice were purchased from Charles River Laboratories and acclimated for 3 to 5 days prior to tumor implantation. All mice were maintained in sterilized translucent polycarbonate microisolator cages under specific pathogen–free conditions and were used in compliance with protocols approved by the Institutional Animal Care and Use Committee. HCT116 and DLD-1 tumor cells were implanted s.c. in nude mice at 3 × 10⁶ per mouse. Tumor-bearing mice were randomized based on tumor volume prior to the initiation of treatment. Treatment with CRA-024781 started when the average tumor volume was ∼100 mm³. Tumor volume was calculated as follows: volume = 0.5 × X² × Y, where X, tumor width; and Y, tumor length. The vehicle used in all studies was 20% HP-β-cyclodextrin in water. Dosages were given i.v. either every other day (q.o.d.) or for 4 consecutive days followed by 3 days without treatment for each week of the study (q.d. × 4 per week) as indicated. Inhibition of tumor growth was calculated as follows: 100 × (1 − dT/dC), where dT was the change in average tumor volume since the first dose in the treatment group and dC was the change in average tumor volume since the first dose in the control group. Statistical analysis was done with one-way ANOVA on log-transformed data to meet the assumption of underlying normal distribution for the test to be valid, and P values were corrected for multiple comparisons to control by Dunnett's method. For single-group comparison, statistical analyses were done with t test on log-transformed data to meet the assumption of an underlying normal distribution for the test to be valid.

Detection of Acetylated Tubulin in Whole Blood

Blood samples collected at various time points after dosing were processed in Red Blood Cell Lysis Buffer (Roche) to isolate nucleated blood cells as per the manufacturer's instruction. Cells were pelleted and stored at −80°C until analysis. Frozen nucleated blood cell pellets were lysed in protein extraction buffer (M-Per, Pierce) containing protease inhibitors (Roche) and phosphatase inhibitors (Calbiochem). Acetylated tubulin was detected in blood lysates by Western blotting. Lysates (20 μg total protein) were solubilized in SDS-PAGE reducing sample buffer, boiled, and electrophoresed. The gels were blotted onto nitrocellulose and probed with an anti-acetylated tubulin antibody (Sigma) to detect acetylated tubulin. The blots were washed, incubated with an appropriate horseradish peroxidase–conjugated secondary antibody (The Jackson Laboratory), and the blots were developed for enhanced chemiluminescence (Pico Substrate Kit, Pierce).

Microarray Analysis

Transcriptional profiles of tumor samples were obtained on Codelink Human Uniset 1 oligonucleotide microarrays (GE-Amersham, Piscataway, NJ). Tumors from compound- and corresponding vehicle-treated tumors were lyophilized and one part of the tumor powders were used to prepare the RNA using Qiagen RNeasy kits (Qiagen, Inc., Valencia, CA). Equal amounts of RNA from at least three animals per dose/drug/time point were pooled together for analysis. Biotin-labeled cRNA probes were prepared from the total RNA by standard GE-Codelink IVT protocols. Ten micrograms of biotinylated cRNA were fragmented and hybridized to the arrays. Arrays were hybridized for 18 hours at 37°C, washed and detected with Strepatavidin-Alexa 647 (Codelink protocol v2.1). They were scanned with an Axon GenePix 4000B scanner (Axon Instruments, Union City, CA) and the images were processed with Codelink 4.0 batch processing software. The data were then transferred to a Signet database and analyzed in Genespring (Agilent, Inc., Palo Alto, CA). Only genes flagged as “good” by the Codelink quality control software were used in the analyses. Each treated sample was normalized to the corresponding vehicle control.

TaqMan Analysis

TaqMan gene expression assays for selected genes were obtained from Applied Biosystems, Inc. One-step reverse transcription-PCR was carried out in triplicate on 25 ng of total RNA from each sample on an ABI PRISM 7700 instrument according to standard protocols. The mRNA levels for each gene were normalized to the amount of RNA in the well as measured in parallel using Ribogreen (Invitrogen). The treated samples were then normalized to the vehicle control at that time point.

Results

Chemical Design and Synthesis

In our efforts to design and develop small-molecule inhibitors of HDACs, we identified a promising N-hydroxy-benzamide lead series (Fig. 1 ). Design and synthesis led to the benzofuran amide analogues exemplified by CG-003 (K_i HDAC1 = 0.004 μmol/L) and further optimization of this series for in vivo efficacy and pharmacokinetics provided our lead inhibitor CRA-024781 (K_i HDAC1 = 0.007 μmol/L). The hydroxamic acid functionality was essential for potency because replacement with a carboxylic acid led to an inactive compound, as seen with CG-004 (K_i HDAC1 > 50 μmol/L). CRA-024781, a compound chosen for further preclinical development, inhibited purified recombinant HDAC isoforms in the nanomolar range as shown in Table 1 . Inhibition constants ranged from 0.007 (HDAC1) to 0.28 μmol/L (HDAC8).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Structure-activity relationship of HDAC inhibitors. Compound numbers, chemical structure, and inhibition constant (K_i) for HDAC1.

Effect on Tumor Cell Proliferation

In order to determine if HDAC inhibition by CRA-024781 affects the proliferation of tumor cells, a panel of human tumor cell lines was treated in vitro with various concentrations of inhibitor. Cells were treated with CRA-024781 for durations based on the intrinsic doubling times of each tumor cell line, and these data were used to calculate a GI_50% value as shown in Table 2 . Antitumor activity was observed in all 10 tumor cell lines tested, with GI_50% values ranging from 0.15 to 3.09 μmol/L. In addition, CRA-024781 had an antiproliferative effect on HUVEC endothelial cells with a GI_50% value of 0.43 μmol/L. Based on these data, we decided to focus on the further characterization of the effects of CRA-024781 in two colon tumor cell lines, HCT116 (GI_50% = 0.2 μmol/L) and DLD-1 (GI_50% = 0.53 μmol/L).

Effect on Protein Acetylation, p21^Cip1/WAF1 Induction, and the Apoptosis Markers PARP and γH2AX

We further characterized the antitumor mechanism of CRA-024781 by treating both HCT116 and DLD-1 colon tumor cells with increasing concentrations of compound and measuring the accumulation of various mechanistic biomarkers proposed to be involved in the antitumor activity of HDAC inhibitors. Treating either HCT116 or DLD-1 cells with CRA-024781 resulted in the dose-dependent accumulation of both acetylated histones and acetylated tubulin (Fig. 2A and B ) suggesting that HDAC enzymes are inhibited in these cells. In addition, CRA-024781 induced expression of the cyclin-dependent kinase inhibitor, p21^Cip1/WAF1, a protein postulated to play a role in the antitumor effect of HDAC inhibition (14). In order to determine if CRA-024781 treatment of these tumor lines led to apoptosis, PARP cleavage and the accumulation of the γH2AX were assayed after compound incubation for 18 hours. γH2AX formation is an early chromatin modification that occurs following initiation of DNA fragmentation during apoptosis (15). PARP cleavage and γH2AX were detected in both cell lines in the low micromolar range. Interestingly, treatment of HUVEC cells showed no accumulation of PARP or γH2AX, suggesting that these endothelial cells do not undergo apoptosis following treatment with CRA-024781 for 18 hours (Fig. 2C). The appearance of the acetylated histone and tubulin biomarkers roughly tracked with the differential sensitivity of HCT116 and DLD-1 to growth inhibition by CRA-024781.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

CRA-024781 activates biomarkers of HDAC inhibition. Accumulation of acetylated histone, acetylated tubulin, p21^Cip1/WAF1, cleaved PARP, and γH2AX are detected by Western blotting of lysates from treated HCT116 (A) and DLD-1 (B) cells. C, HUVEC cells.

In vivo Pharmacokinetics

Having established that CRA-024781 displayed HDAC inhibitory and antitumor properties in vitro, we assessed the pharmacokinetics of the compound to evaluate exposure in order to enable further testing in efficacy models. CRA-024781 was delivered i.v. to female mice and plasma concentrations were monitored over time (Fig. 3 ). Based on these data, the clearance was calculated to be 297 mL/min/kg, the volume of distribution in the central compartment was 3,750 mL/kg, the steady-state volume of distribution was 9,230 mL/kg, the predominant plasma half-life was 6.7 minutes (73% area under the curve), and the mean residence time was 31 minutes. These data suggest that CRA-024781 had sufficient in vivo exposure when given by the i.v. route of administration for use in the study of the biological effects of HDAC inhibition in vivo.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Pharmacokinetics of CRA-024781. Plasma concentration versus time following i.v. administration at 10 mg/kg in female BALB/c nude mice (n = 3/time point).

Antitumor Activity and Induction of Acetylation by CRA 024781 In vivo

In order to assess the antitumor activity of CRA-024781 in vivo, mice bearing human colon tumor xenografts were dosed i.v. with the compound using various dosages and schedules. I.v. administration of CRA-024781 in a previous dose scheduling study in HCT116 xenografts identified two regimens with good therapeutic indices: (a) once daily every other day (q.o.d.) or (b) once daily for 4 consecutive days followed by 3 days without treatment each week (q.d. × 4 per week; dose scheduling data not shown). When CRA-024781 was evaluated in HCT116 or DLD-1 xenografts according to the first regimen at a dosage of 200 mg/kg i.v. q.o.d., statistically significant inhibition of tumor growth was observed (Fig. 4A and B ). The inhibition of tumor growth was 69% (P < 0.000001) and 59% (P < 0.01) for HCT116 and DLD-1 models, respectively. Although some body weight loss approaching 13% relative to vehicle controls was observed in the DLD-1 study, no body weight loss was observed in the HCT116 study, suggesting that there was some tolerability variation across experiments, and that 200 mg/kg was near the maximum tolerated dose for q.o.d. administration (data not shown). CRA-024781 was then evaluated at multiple doses using the second regimen (q.d. × 4 per week). In the HCT116 model, inhibition of tumor growth was 48% (P < 0.05), 57% (P < 0.01), 82.2% (P < 0.0001), and 80.0% (P < 0.0001) for 20, 40, 80, and 160 mg/kg, respectively. None of these doses led to reductions in animal weight relative to vehicle by the end of the study (data not shown). In the DLD-1 model, CRA-024781 administered q.d. × 4 per week did not significantly inhibit tumor growth, although a marginally significant trend towards inhibition was observed with the highest dose of 160 mg/kg, showing 43% inhibition (P = 0.09) with no associated body weight loss (data not shown). In order to determine if the plasma concentrations of CRA-024781 attained in the studies were sufficient to inhibit HDAC enzymes in vivo, peripheral blood cells were examined ex vivo. As shown in Fig. 5 , each treatment group had a measurable increase in tubulin acetylation at 2 and 6 hours after dosing. In summary, CRA-024781 exhibited statistically significant antitumor activity against both HCT116 and DLD-1 human colorectal tumor xenografts, although antitumor activity overall seemed to be more pronounced in the HCT116 xenograft model.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

CRA-024781 inhibits the growth of human tumor xenografts in vivo. The growth of tumors over time is plotted in (A) HCT116 colon tumor xenograft model [vehicle alone (▵), 200 mg/kg (▴); schedule, q.o.d.]. B, DLD-1 colon tumor xenograft model [vehicle alone (▵), 200 mg/kg (▴); schedule, q.o.d.]. C, HCT116 colon tumor xenograft model [vehicle alone (▵), 20 mg/kg (▴) 40 mg/kg (□) 80 mg/kg (▪) 160 mg/kg (×); schedule, q.d. × 4]. D, DLD-1 colon tumor xenograft model [vehicle alone (▵), 60 mg/kg (▪), 100 mg/kg (□), 160 mg/kg (▴); schedule, q.d. × 4]. *, P < 0.05; **, P < 0.01; bars, calculated as described in Materials and Methods.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

In vivo pharmacodynamics of CRA-024781. Peripheral blood mononuclear cells were prepared from blood drawn at the time points indicated from mice dosed in the HCT116 efficacy study shown in Fig. 4C. Acetylation of α-tubulin was detected by Western blot.

Transcriptional Profile of CRA-024781

To gain insights into the mechanism by which CRA-024781 inhibits tumor growth, mRNA profiles of HCT-116 and DLD-1 tumors were compared from animals treated with CRA-024781 at efficacious doses (80 and 160 mg/kg for HCT-116, and 160 mg/kg for DLD-1), and at different times (from 2 to 48 hours) using microarray analysis. Hierarchical clustering was done on the microarray data using the standard correlation algorithm in GeneSpring as shown in Fig. 6 . Expression changes were seen in ∼15% of all expressed genes (990 genes out of the 6,640 total genes that passed quality control criteria). In general, maximal differential expression was seen at 2 and 4 hours, and more minimal changes occurred at the 24-hour and 48-hour time points, consistent with both the pharmacokinetics and with the tubulin acetylation data. A pattern of up-regulated genes was observed at some early time points (Fig. 6, red box) and another pattern of down-regulated genes were found at the mid (4- to 6-hour) time points (yellow box). Interestingly, this pattern of down-regulated genes was also seen at the 2-hour time point of the highest (160 mg/kg) dose in HCT-116 tumors. A subset of these differentially expressed genes is shown in Table 3 . Genes were selected that met a statistical filter of P < 0.05, represented a >1.8-fold change in gene expression averaged from both the early (2-hour) and mid (4- to 6-hour) time points.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Effect of CRA-024781 on gene expression. A hierarchical clustering analysis of tumors from mice treated with CRA-024781. Red, genes up-regulated >1.8-fold; green, genes down-regulated >1.8-fold; black, others. Each column corresponds to a CRA-024781-treated sample which was normalized to its corresponding vehicle treatment at the same time point. Clusters of similarly up-regulated (red box) or down-regulated (yellow box) genes.

Discussion

Herein, we have described CRA-024781, a novel hydroxamic acid–based inhibitor of HDAC enzymes with potent antitumor activity in vitro and in vivo. This compound was derived from a series of compounds designed around an N-hydroxy-benzamide scaffold, and was optimized for potency against HDAC enzymes and for in vivo antitumor activity. CRA-024781 inhibited all of the HDACs tested with K_i values ranging from 0.007 to 0.28 μmol/L.

Consistent with the reported biological effects of other HDAC inhibitors, CRA-024781 induced histone and tubulin acetylation, and p21^Cip1/WAF1 protein, as well as inhibited the growth of 10 human tumor cell lines and nonneoplastic HUVEC in vitro. Two cell lines derived from colorectal adenocarcinomas, HCT116 and DLD-1, were further characterized. These two cell lines differ in several respects, including doubling time (see Table 2) and p53 status: HCT116 cells contain wild-type p53, whereas DLD-1 cells express mutant p53 (Ser²⁴¹Phe). Despite these differences, inhibition of cell growth by CRA-024781 in vitro was similar in these lines. In both cell lines, growth inhibition was accompanied by changes in several known biomarkers of HDAC response. These biomarkers included the accumulation of acetylated histones and acetylated tubulin. In both tumor cell lines, acetylated tubulin was detectable at lower concentrations than acetylated histones, which might reflect differing sensitivities for specific HDACs or HDAC substrates in the cell, or may reflect differing sensitivities of the specific detection reagents used. Other biomarkers included the accumulation of the tumor suppressor protein p21^Cip1/WAF1, and the generation of cleaved PARP fragment and phosphorylated H2AX (γH2AX), two known markers of apoptosis. Interestingly, treatment of nonneoplastic HUVEC with CRA-024781 under the same conditions did not lead to PARP accumulation or γH2AX, despite a cellular GI_50% and doubling time in the same range as DLD-1 and HCT116 tumor cells. These data suggest that CRA-024781 may induce apoptosis more effectively or rapidly in tumor cells relative to HUVEC. If generally true in nonneoplastic cells, this observation could partly explain the ability to inhibit tumor growth in vivo at dosages that are well tolerated.

CRA-024781 had an in vivo pharmacokinetic profile that suggested adequate drug exposure to allow HDAC inhibition in vivo. The 10 mg/kg i.v. dose achieved a plasma concentration above the measured cellular GI_50% for HCT116 tumor cells for 15 minutes. However, it is difficult to predict a priori the in vivo drug exposure (area under the curve) that will be adequate for efficacy based on GI_50% or on the appearance of mechanistic biomarkers (protein acetylation, etc.), because these values are based on continuous exposure of tumor cells to the drug in vitro, on the other hand, tumors grown in vivo are exposed to the drug on a certain schedule (e.g., q.d. × 4 for 16 days) at varying concentrations of compound based on the pharmacokinetics and degree of protein binding of the drug in plasma. Also, the drug concentration in plasma does not necessarily reflect the concentration distributed to the tumor. Indeed, the relatively high steady-state volume of distribution (9,230 mL/kg) suggests that CRA-024781 may readily distribute into tissues. Nevertheless, the doses required for maximal efficacy (≥80 mg/kg) are >10 mg/kg, perhaps because HDAC enzymes must be inhibited for a more prolonged period of time or to a greater extent following each dose in order to achieve antitumor efficacy.

Antitumor activity was shown in two independent colorectal xenograft models, HCT116 and DLD-1. The degree of tumor growth inhibition observed with CRA-024781 is similar to that seen with other published HDAC inhibitors, including trichostatin A (16), LAQ-824 (6, 17), MS-275 (18, 19), PXD101 (20), Cpd 2 (21), cyclic hydroxamic acid containing peptide 31 (22), and Valproate (23). In order to follow the inhibition of HDAC activity over time, we assayed the acetylation of tubulin in blood as a pharmacodynamic marker. The efficacious doses of CRA-024781 delivered to the mice were accompanied by an accumulation of tubulin acetylation in peripheral blood as early as 2 hours after dosing, and levels persisted for at least 6 hours following dosing. Therefore, the kinetics of appearance and disappearance of this biomarker can be used to confirm HDAC inhibition in vivo, and may provide a reasonable surrogate marker for clinical monitoring. The acetylation profile in the HCT116 dose-response experiment suggests that in order to achieve robust in vivo efficacy, HDACs must be inhibited for a period of 6 hours or longer, because at that time point, the more efficacious doses of 160 and 80 mg/kg resulted in greater accumulation of tubulin than the lower doses, which were not as effective.

CRA-024781 displayed similar inhibition of cell growth and induction of apoptosis in HCT116 and DLD-1 cell lines in vitro and inhibited HDACs in vivo, as measured by acetylated tubulin in blood, yet the compound seemed to be more inhibitory in the HCT116 xenograft model relative to the DLD-1 model, particularly when the compound was dosed according to the q.d. × 4 per week regimen. The differing sensitivities of HCT116 and DLD-1 xenografts to CRA-024781 in vivo may be due to adaptive responses in vivo which render DLD-1 less dependent on HDACs for growth and survival.

To investigate the mechanism of action of CRA-024781 in vivo, we focused efforts on the analysis of mRNA expression patterns over time in the colorectal xenograft models at doses where efficacy was observed. In the HCT116 xenograft study, a total of 990 genes (or ∼15% of the genes analyzed on the array) were found to be differentially expressed by at least 1.8-fold at any time point, and over half of these were down-regulated. This suggests that (a) in vivo, down-regulation of genes that may be essential for cell proliferation or survival may contribute to the antiproliferative effect of HDAC inhibitors, and (b) based on the large number of differentially regulated genes, the antitumor mechanism is likely to be complex. A pattern of similarly up-regulated genes was observed 2 hours following dosing at 80 mg/kg (HCT-116) and 160 mg/kg (DLD-1; Fig. 6, red box). Interestingly, the up-regulated (red box) genes were not observed at 2 hours in the sensitive HCT-116 tumors treated at the highest dose of 160 mg/kg, consistent with a more rapid onset of events under these conditions. At later (4- to 6-hour) time points, the dominant pattern is a set of genes that become down-regulated (yellow box). Therefore, it seems that an early pattern of genes is induced within 2 hours followed by a later pattern of down-regulated genes at 4 to 6 hours.

A selection of genes with at least 1.8-fold differential expression (P < 0.05) averaged over the early and mid time points is shown in Table 3. The list includes genes with the highest fold change that also have a known biological function, and specifically includes many genes that have previously been shown to be regulated by HDAC inhibitors such as suberoylanilide hydroxamic acid and trichostatin A (4), for example, p21^Cip1/WAF1 (CDKN1A) and caspases (CASP1, CASP3), as well as TNFR superfamily members including TNFRSF 7 (CD27) and 14 (LIGHT), as well as serpins, and a histone subunit. The up-regulated pattern is consistent with the induction of apoptosis being one of the mechanisms of action of this compound. Down-regulated genes include cyclins and other cell division control proteins, several HDACs, c-MYC, and Her2/NEU (ERBB2). The down-regulation of several HDAC isoforms suggests the presence of transcriptional feedback loops resulting from HDAC inhibition, which may result in prolonging the re-expression of genes. The observation that the prominent down-regulated genes include MYC, cyclins, and other proteins required for cell division, and that the cyclin inhibitor p21^Cip1/WAF1 is up-regulated, suggests that cell cycle arrest leading to reduced proliferation is an important part of the mechanism of tumor growth inhibition in vivo, in agreement with previous in vitro observations (21, 24). Thus, growth arrest and apoptosis in both HCT116 and DLD-1 tumors could occur in part by up-regulating p21^Cip1/WAF1 to induce arrest in G₁ and apoptosis. In addition, the tumor suppressor p19 (INK4d), a gene previously shown to be an important mechanism of growth arrest in HCT116 cells in vitro (25), is also up-regulated in these tumors.

In summary, we have identified a novel inhibitor of HDAC, CRA-024781, with antitumor activity in vitro and in vivo. The results presented here show that CRA-024781 inhibits HDAC enzymes and has antitumor properties in vivo at doses that are well tolerated and may therefore be active in a clinical setting in patients with cancer. Antitumor activity without associated toxicity has been reported for other HDAC inhibitors as well (20, 26, 27). Based on these results, CRA-024781 has entered phase I clinical trials in cancer to determine its safety in humans and suitability for further development as a novel oncolytic agent.