Monitoring the effect of belinostat in solid tumors by H4 acetylation

Histone deacetylase (HDAC) inhibition is a novel entity in medical oncology, and several HDAC inhibitors are in clinical trials. One of them is the hydroxamic acid belinostat (PXD101) that has demonstrated therapeutic efficacy for several clinical indications. Acetylation of histones is a key event after treatment with HDAC inhibitors, and could thus be used as a marker for monitoring cellular response to HDAC inhibitor treatment. Here we describe the utility of a newly described monoclonal antibody against acetylated H4 for immunohistochemistry on paraffin-embedded fine needle biopsies from nude mice carrying A2780 human ovarian cancer xenografts. Acetylated H4 was monitored in vivo by immunohistochemistry during treatment with belinostat, and compared with pharmacokinetics in plasma and tumor tissue. We found an increased level of acetylated H4 15 min after a single treatment (200 mg/kg i.v.) with maximum level reached after 1 h. H4 acetylation intensity reflected the belinostat concentration in plasma and tumor tissue. The threshold level for belinostat activity, indicated by acetylated H4, correlated with belinostat plasma concentrations above 1,000 ng/ml. In conclusion, examination of H4 acetylation in fine needle biopsies using the T25 antibody may prove useful in monitoring HDAC inhibitor efficacy in clinical trials involving humans with solid tumors.

Histone deacetylase (HDAC) inhibitors are a promising group of anticancer agents that target the group of HDAC enzymes reported to be involved in cancer development (1)(2)(3). HDACs lead to transcriptional repression through alteration of chromatin structure. Remodeling of chromatin through modifications of histone molecules is a key event in gene transcription (4). Histones can be modified by acetylation, methylation, phosphorylation, and ubiquitination (4,5).
Acetylation of histones is regulated by the opposite mechanisms of HDACs and histone acetyltransferases (HATs). In general, histone acetylation through HAT activity leads to increased transcriptional activity, whereas deacetylation through HDAC activity leads to transcriptional repression (6,7).
Histone acetylation is a defining event after treatment with an HDAC inhibitor, and can thus be used as an indicator of HDAC inhibitor activity in both normal and tumor cells. This has led to the widespread use of histone acetylation in peripheral blood mononuclear cells (PBMCs) as a surrogate marker in phase I clinical trials (14,(18)(19)(20)(21). However, it would obviously be an advantage to be able to monitor acetylation in a solid tumor and correlate the efficacy of the HDAC inhibitor with concentrations found in plasma. This would provide a valuable tool for investigating alternative schedules and for optimizing the efficacy of HDAC inhibitors in the treatment of solid tumors.
In order to obtain a validated model for this approach, we describe the utility of using a new monoclonal antibody to H4 acetylation for immunohistochemistry, together with its applicability in repeated fine needle biopsies from mouse xenograft solid tumors treated intravenously with the HDAC inhibitor belinostat (PXD101). We further investigated to what extent the time course of H4 acetylation in tumor tissue correlated with the belinostat concentration in plasma.

Cell lines
The following cell lines were purchased from ATCC: HCT-116 human colon carcinoma (ATCC, CCI-247), MCF-7 human breast carcinoma (ATCC, HTB-22) and A549 human lung carcinoma (ATCC, CCL-185). PC-3 human prostate cancer cell line (ECACC, 90112714) was purchased from ECACC. A2780 human ovarian carcinoma cell line was a gift from R. Ozols, Fox Chase Cancer Center, Philadelphia, PA, and P388 murine leukemia cell line was a gift from F.M. Shabel, Southern Research Institute, Birmingham, AL.

Test of antibody specificity and reactivity
We used the novel mouse monoclonal anti-human T25 antibody against acetylated histone H4 that recognizes acetylated Lys8 and Lys12 on H4 (22). Western blotting using either whole cell lysates of HCT-383 116 cells or histones purified from P388 cells after treatment with 1 mM belinostat for 1 h was used for testing antibody specificity. Blots were incubated with T25 antibody before development. The appearance of only one band verified antibody specificity. The reactivity of T25 was evaluated according to DAKO standard IHC-procedures on different malignant tumors (head and neck, cervix, brain, colon, stomach, prostate, lymphomas, melanomas, lung, ovary, liver, and breast) as well as on normal tissue (prostate, kidney, pancreas, cerebellum, and tonsil).

Xenografts
Female NMRI nude mice (Taconic), 6-8 weeks of age, were used for the xenograft models. Tumor cells grown in cell cultures were trypsinized and ∂10 7 cells (100 ml) were mixed with 100 ml matrigel. Mice were anesthesized using Hypnorm/Dormicum, and cells were injected subcutaneously into the flank.

Extraction and analysis of belinostat in plasma and tissue
Extraction procedures for plasma samples were conducted to separate any tissue/protein from the sample before analysis. The extractions were carried out using Waters SiroccoTM Protein Precipitation Plates A with 96-well collection plates. First, 150 ml internal standard (13-C-Belinostat) and 50 ml plasma sample was added to each well and shaken on a vortexer. The loaded plate was centrifuged for 5 min, 2000 rpm. All liquid was evaporated under heated nitrogen and the samples were re-dissolved in 100 ml 10% HCOOH in acetonitrile (ACN) (starting eluent) before analysis.
Tumor and spleen samples were homogenized in 400 ml homogenization buffer (25:25:25 Water/ACN/ Methanol with internal standard, 13-C-Belinostat 1 mg/ml) using an Ultra Turax A mechanical homogenizer. Samples were then centrifuged at 5000 rpm for 5 min and 150 ml supernatant was removed and kept for analysis. Quantitative analysis of plasma, tumor and spleen samples was done by liquid chromatography tandem mass spectrometry using Waters UPLC A with reversed-phase analytical column (Acquity C18, 2.1¿50 mm, 1.7 mm particle size). A 5 min gradient using 0.05% HCOOH and ACN as mobile phase A and B, respectively, was used with a flow of 0.5 ml/min. All data were processed using MassLynx 4.1.

Experiment 1: Monitoring the effect of HDAC inhibitor treatment in vivo in different xenograft models
The effect of belinostat in vivo was tested in a number of different xenograft models (PC-3, HCT-116, MCF-7, A549 and A2780) in order to select a model for biopsy sampling. Mice were treated with 100 mg/ kg belinostat or vehicle control (L-arginine 200 mg/ kg in isotonic sterile saline). After 1 h, mice were sacrificed and tumors were excised and prepared for immunohistochemistry.

Experiment 2: Collection of tumor biopsies and influence on H4 acetylation
Tumor biopsies were collected by insertion of an 18G needle into the tumor tissue and carefully aspirating while rotating the needle. It was investigated whether the biopsies taken were representative for the whole tumor and whether the method for biopsy collection had any influence on H4 acetylation. Mice with HCT-116, A2780, PC-3, or MCF-7 tumors were treated with belinostat, 100 mg/kg, and after 1 h they were sacrificed, tumor biopsies were collected, and 384 biopsies and remaining tumor tissue were prepared for immunohistochemistry. For the A2780 model, the effect of repeated biopsy sampling was also investigated. The mice were anesthetized by isoflurane inhalation while tumor biopsies were collected.

Experiment 3: Time dependence of belinostat treatment on H4 acetylation in solid tumor
The A2780 tumor model exhibited a limited amount of necrosis, and this model was thus selected for further investigation of the relationship between exposure time and H4 acetylation in tumor tissue. Furthermore, the expression of p21 was examined to investigate a possible correlation between H4 acetylation and activation of gene transcription. 16 mice with either small (300-400 mg) or large (1500-1800 mg) A2780 tumors were treated with 100 mg/kg belinostat i.v. at time zero. One pretreatment (only large tumor) and one post-treatment biopsy were collected from the mice. Biopsies were collected at different time points from 1 to 6 h after treatment, as described in Table 1. At sacrifice (3, 6, or 24 h) the entire tumor was removed. Sets of biopsies and corresponding tumors were prepared for immunohistochemistry.

Experiment 4: Relationship between H4 acetylation in solid tumor and belinostat in plasma, tumor and spleen tissue
Groups of 3-4 mice with subcutaneous A2780 xenografts were treated with 200 mg/kg¿1 belinostat i.v. or vehicle control (L-arginine 400 mg/kg in isotonic sterile saline) at time zero. After 15 min to 3 h, mice were sacrificed and plasma (lithium heparin stabilized), tumor, and spleen tissue were collected. Spleen and tumor were divided into two and prepared for analysis for belinostat content and H4 acetylation by IHC analysis. The degree of H4 acetylation was  1  3  237  2  3  231  2  3  225  3  3  236  3  3  228  1  6  230  6  6  552 24 Seven mice with large A2780 tumors and nine mice with small A2780 tumors were included in the study. The table shows time points for collection of biopsies and removal of whole tumors. Pretreatment biopsies were collected only from mice with large tumors. correlated with belinostat in plasma and tumor tissue. The spleen representing peripheral blood mononuclear cells (PBMCs) was included as reference tissue.

Reactivity of T25 anti-acetylated H4 antibody
The antibody T25 performed well in immunohistochemistry on formalin-fixed paraffinembedded tissue samples with no background staining; it showed nuclear labeling of epithelial and cancer cells in all tested specimens. Stromal cells showed no H4 acetylation. Acetylation of H4 in cancer tissues is shown in Fig. 2.

Belinostat effect in different xenograft models (Experiment 1)
The effect of belinostat on H4 acetylation was tested in the PC-3, HCT-116, MCF-7, A549 and A2780 model. A single i.v. injection of 100 mg/kg caused H4 acetylation after 1 h in all models. The strongest signal for H4 acetylation was found in the A2780 ovarian cancer model, followed by the HCT-116 colon cancer model 385 and the A548 lung cancer model. Weak acetylation was seen in the PC-3 prostate cancer and the MCF-7 lung cancer model (Fig. 3).

Collection of tumor biopsies and influence on H4 acetylation (Experiment 2)
Biopsies and corresponding solid tumor (500-1000 mg) from A2780 (15 mice), HCT-116 (12 mice), PC-3 (6 mice), and MCF-7 (4 mice) were compared to ensure that the biopsies taken were representative for the whole tumor, and that the models were suitable for biopsy collection.
All 15 biopsies collected from A2780 xenograft mice contained tumor tissue with preserved or partly preserved morphology, and only minimal necrosis. All 12 HCT-116 tumors contained liquid necrosis in their center, and viable tumor tissue was only observed in the periphery of the tumor. Consequently, it was only possible to collect a tumor tissue sample from 1 of 12 biopsies taken. All six PC-3 tumors contained large necrotic areas and only one biopsy contained tumor cells. Two of four biopsies from MCF-7 tumors contained tumor tissue.  However, one of these biopsies contained necrosis as well.
H4 acetylation was compared between the biopsies and the representative tumors to ensure that biopsy collection itself did not have an impact on H4 acetylation. Of the 15 A2780 biopsies, 7 were treated with belinostat and the staining profile of acetylated H4 was similar in tumors and representative biopsies. In contrast, no or only weak H4 acetylation was observed in A2780 tumors and biopsies from eight vehicletreated control mice. Examples of the staining are found in Fig. 4A. It was also investigated whether the repeated biopsy sampling had any effect on acetylation in A2780 tumors. Even though the repeat biopsies contained blood, this did not interfere with the IHC staining or interpretation of the results.

Time dependence of belinostat treatment on H4 acetylation in solid tumor (Experiment 3)
The A2780 xenograft model was selected for monitoring time dependence of belinostat effect in vivo. 16 mice with either small (300-400 mg) or large (1500-1800 mg) A2780 tumors were treated with 100 mg/kg belinostat i.v. at time zero. Pretreatment (only large tumors) and post-treatment biopsies were collected from mice after 1, 2, 3, or 6 h. At sacrifice the whole tumor was excised.
22 biopsies from 16 mice were collected during the study, and 21 of these biopsies contained tumor tissue. Results of H4 acetylation are shown in Table 2. Examples of staining are shown in Fig. 4B. Pretreatment biopsies (nΩ7) showed no (nΩ1), weak (nΩ3) or moderate (nΩ 2) H4 acetylation. One pretreatment biopsy did . The pretreatment biopsy shows weak H4 acetylation, whereas strong H4 acetylation is seen 1 h after treatment. After 3 h, the H4 acetylation is weak and has decreased to a level similar to that in the pretreatment biopsy. not contain tumor tissue. All biopsies collected 1 h (nΩ5) and 2 h (nΩ4) after belinostat treatment showed strong H4 acetylation; 3 h after treatment strong acetylation was observed in 1/10 removed tumors as well as in 2 biopsies from 3 of these tumors. The general findings at 3 h were weak or moderate acetylation similar to the levels observed in the pretreatment biopsies. After 6 and 24 h, weak or no acetylation was observed. Results of the p21 expression are shown in Fig. 5. p21 showed strong nucleic expression in occasional cells from biopsies collected pretreatment, and 1 and 2 h after treatment. In addition, weak cytoplasmic expression was observed in two of six pretreatment biopsies 387 (33%), while one of four biopsies collected after 2 h showed moderate cytoplasmic expression (25%). The cytoplasmic expression of p21 was increased after 3 and 6 h, respectively. Thus, of 10 tumors removed after 3 h, 5 (50%) showed weak expression, 2 (20%) showed moderate expression, and one (10%) showed strong expression. After 6 h, three of four tumors (75%) showed moderate expression and one of four (25%) showed strong expression. All biopsies removed after 6 h showed moderate-to-strong cytoplasmic expression as well. After 24 h, the cytoplasmic expression had disappeared. In contrast, nucleic staining was more pronounced than in the earlier samples. No Strong H4 acetylation is observed in biopsies collected 1 and 2 h after treatment. In most samples, acetylation is decreased after 3 h to a level similar to that observed in the pretreatment biopsies. Only weak acetylation is observed 6 h after treatment. (-, missing).

Relationship between H4 acetylation in solid tumor and belinostat levels in plasma, tumor and spleen tissue (Experiment 4)
Mice with subcutaneous A2780 xenografts (500-1000 mg) were treated with belinostat (200 mg/kg i.v.) or vehicle control and sacrificed after 15 min, 30 min, 1 h, 2 h, or 3 h for collection of plasma, tumor and spleen for analysis of belinostat content and staining for acetylated H4 in tissue. The belinostat dose was increased from 100 to 200 mg/kg to increase the possibility of detection of belinostat in tumor and spleen tissue and to investigate whether the acetylation Fig. 5. Expression of p21 in A2780 xenograft mice before and after belinostat treatment. p21 is observed in the nucleus of occasional cells in samples collected pretreatment, and up to 3 h after treatment. Weak cytoplasmic expression of p21 is observed 3 h after treatment and is strongly increased 6 h after treatment. The cytoplasmic expression of p21 disappears after 24 h. However, the nucleic expression is now more pronounced. 388 period could be extended beyond the 1-2 h period observed in Experiment 3.
Analysis of belinostat in plasma, tumor and spleen showed that its distribution to tumor and spleen tissue was similar, and that tissue and plasma distribution curves were parallel (Fig.  6). Thus, belinostat plasma concentration seems to reflect the actual belinostat concentrations in subcutaneous solid tumor tissue.
Results of H4 acetylation are shown in Table  3. After 15 min, weak-to-moderate acetylation was observed in three of four tumors. Maximal H4 acetylation levels were reached at 1 h, with Fig. 6. Belinostat in plasma (ng/ml), A2780 subcutaneous tumor (ng/g) and spleen (ng/g) after intravenous treatment with 200 mg/kg single dose at 0 min (Experiment 4). The distribution of belinostat to tumor tissue follows the plasma pharmacokinetic curve the first 2 h after treatment. After 3 h, the concentration of belinostat in plasma seems to decline more than that in the tumor and spleen tissue. The distribution of belinostat to subcutaneous solid tumor is comparable to the spleen distribution.
four of four tumors showing strong H4 acetylation. After 2 h, one of three tumors showed strong acetylation, but after 3 h, the acetylation had decreased to weak and moderate levels. Similar levels of H4 acetylation were observed for spleen tissue (Fig. 7). No H4 acetylation was detected in spleen and tumor tissue from vehicle control mice. When correlating the plasma pharmacokinetic profile with H4 acetylation in solid tumor tissue, the present study found that belinostat activity, as indicated by the acetylated H4 surrogate marker, was present as long as the plasma concentration was above the 1000 ng/ml level (Fig. 8). As belinostat plasma concentrations decrease, the H4 acetylation in tumor tissue also declines.

DISCUSSION
Histone acetylation in PBMCs is often used as a surrogate marker for HDAC inhibitor activity and has shown a correlation with plasma concentration (19). Pharmacokinetic studies have shown linear pharmacokinetics with dose proportional increases in C max and AUC for vorinostat (12) and linear, dose-independent pharmacokinetics for MS-275 (18). Although time to C max has shown similarity between different HDAC inhibitors (C max reached after 1-2 h), there have been large variations in the terminal half-life between HDAC inhibitors (ranging from 21 min to 74 h) (12,14,18,23).
Evaluation of five cell lines grown as xenografts on nude mice revealed differences in their utility as biopsy models (Experiments 1 and 2), which was probably due to differences in the necrotic tendency of the tumors. PC-3 and MCF-7 tumors showed less acetylation after 1 h compared to A2780 tumors. It is unknown whether this difference in acetylation is due to decreased efficacy of the drug in certain tissues or due to differences in cell viability. The PC-3, MCF-7, and HCT-116 model contained necrotic and fluid areas, which hindered biopsy collection. Biopsies from A2780 tumors contained vital tumor tissue, making this a more suitable model. In addition, A2780 biopsies were representative of the tumors regarding level of H4 acetylation. Our initial studies showed that the method of biopsy collection had no influence on histone acetylation.
H4 acetylation was detected in tumor tissue 15 min after belinostat treatment and the maximum level was reached after 1 h. In a previous study, increased H4 acetylation was found in peripheral canine blood 10 min after i.v. infusion of belinostat given over 45 min (total dose 50 mg/kg) (24). In addition, the investigators found increased H4 acetylation 30 min after treatment with 1 mM belinostat in histones extracted from A2780 cells (24). We observed that two biopsies collected after 3 h showed stronger acetylation than the corresponding tumors. Staining of the tumors was uneven with weaker acetylation in the middle than in the periphery. This problem might be a result of insufficient fixation or too short epitope retrieval of the tumors. Nevertheless, our results showed that H4 acetylation reached base line level 3 h after treatment. Similarly, Plumb et al. found that H4 acetylation had returned to base line levels in histones extracted from A2780 cells and peripheral blood from A2780 xenografts 3 h after treatment with belinostat (1 mM or 40 mg/kg i.p.) (24). Our results from the monitoring study (Experiment 3) showed different levels of H4 acetylation in pretreatment biopsies from different mice (Table 2), which indicates that it is important to collect pretreatment biopsies in order to have an internal control for baseline H4 acetylation.
We found comparable belinostat distribution to tumor and spleen tissue, and in addition that the distribution in tumor tissue followed the belinostat plasma pharmacokinetics. This indicates that belinostat distribution is not tumor specific and that plasma concentrations of belinostat may be used to predict belinostat distribution to solid tumor tissue. In the present study, the belinostat concentration in plasma and tumor correlated with levels of acetylated H4 in the solid tumor tissue and PBMCs. Acetylated H4 is a defining event after treatment with an HDAC inhibitor, and can thus be used as an indicator of belinostat activity. We found that H4 acetylation was present as long as the plasma concentration was above 1000 ng/ ml. This supports that maintenance of belinostat plasma levels above a certain level is required in order to achieve a prolonged effect of belinostat treatment in solid tumors.
In conclusion, the monoclonal antibody T25 against acetylated H4 worked well on formalinfixed paraffin-embedded tissue samples. The effect of belinostat on H4 acetylation using the T25 antibody was evident 1 and 2 h after dosing, with a return to baseline levels after 3 h. H4 acetylation followed the belinostat concentration in tumor tissue. Examination of H4 acetylation in fine needle biopsies using the T25 antibody may thus prove useful in monitoring HDAC inhibitor efficacy in clinical trials involving solid tumors.