Abstract
JQ1, a cell-permeable small-molecule inhibitor of bromodomain and extra-terminal protein (BET) function with reportedly good CNS penetration, however, unbound and pharmacologically active CNS JQ1 exposures have not been characterized. Additionally, no quantitative bioanalytical methods for JQ1 have been described in the literature to support the CNS penetration studies. In the present article, we discuss the development and validation of sensitive and reliable liquid chromatography-tandem mass spectrometry (LC-MS/MS) quantitative methods to determine JQ1 in mouse plasma and brain microdialysate. JQ1 and the internal standard, dabrafenib (ISTD), were extracted from plasma and microdialysate samples using a simple solid phase extraction protocol performed on an Oasis HLB μElution plate. Chromatographic separation of JQ1 and ISTD was achieved on a reversed phase C12 analytical column with gradient elution profile of mobile phases (MP A: water containing 0.1% formic acid and MP B: acetonitrile containing 0.1% formic acid) at a flow rate of 0.6 mL/min. The mass spectrometric detection was performed in the positive MRM ion mode by monitoring the transitions 457.40 > 341.30 (JQ1) and 520.40 > 307.20 (ISTD). The calibration curves demonstrated good linearities over the concentration range of 5–1000 ng/mL for the mouse plasma method (r2 ≥ 0.99) and 0.5–500 ng/mL for the microdialysate method (r2 ≥ 0.99). The experimental limit of quantification obtained was 5 and 0.5 ng/mL for the mouse plasma and microdialysate method, respectively, with the coefficient of variation less than 10% for the analyte peak area. All the other validation parameters, including intra-and inter-day accuracy and precision, matrix effect, selectivity, carryover effect, and stability, were within the USFDA bioanalytical guidelines acceptance limits. The LC-MS/MS method was successfully applied to a mouse pharmacokinetic and cerebral microdialysis study to characterize the unbound JQ1 exposure in brain extracellular fluid and plasma.
Keywords: JQ1, BET inhibitor, LC-MS/MS, Solid Phase Extraction, Pharmacokinetics, Cerebral Microdialysis
1. Introduction
Medulloblastoma (MB), the most common embryonal brain tumor of childhood, has been categorized into four molecular subgroups, winged (WNT), sonic hedgehog (SHH), Group 3, and Group 4 [1, 2]. Of the four subgroups, Group 3 MB accounts for approximately 25% of all MB, occurs more frequently in males and infants, and has a high likelihood to have metastatic disease at presentation. Patients with Group 3 MB are considered to have the most aggressive form of MB and have a poor 5-year survival of < 60% [1, 2]. Thus, it is essential that new therapeutic approaches be developed for this subgroup of MB.
With an improved understanding of the molecular composition of pediatric CNS tumors has come an increased opportunity for the novel use of molecularly targeted drugs to treat these tumors. Group 3 MB is an excellent example as MYC amplification occurs in ~17% of patients but is rare in other MB subgroups [2]. Patients with MYC amplified MB have a particularly poor clinical outcome. Thus, a drug that could target MYC amplified MB could possibly represent a new therapeutic approach to treat Group 3 MB.
Although JQ1 represents a class of compounds with potential to treat Group 3 MB, little is known regarding its CNS penetration. Matzuk and colleagues, reported a total brain-to-plasma exposure ratio (Kp) of 0.98 in mice after a single JQ1 dose (50 mg/kg, IP) [3]. The unbound brain exposure of JQ1, a much more relevant metric assessing the CNS drug penetration, has not been reported yet. Thus, to gain a better understanding of JQ1 CNS penetration, we desired to perform CNS microdialysis studies, which required that we develop and validate sensitive and specific liquid chromatography-tandem mass spectrometric (HPLC-MS/MS) methods for JQ1 in murine plasma and dialysate.
In the present study, we aimed to characterize unbound CNS exposure of JQ1 by performing cerebral microdialysis studies in non-tumor bearing CD1 nude mice. The first objective was to develop and validate sensitive and robust LC-MS/MS methods to quantify JQ1 concentrations in mouse plasma and Ringer’s solution. Then, these methods were applied to measure JQ1 in plasma and extracellular fluid (ECF) samples collected in non-tumor bearing mice using a cerebral microdialysis technique.
2. Materials and Methods
2.1. Chemicals
JQ1 (M.W.: 456.99 g/mol; M.F.: C23H25ClN4O2S; Purity by HPLC: 98.80%; Fig. 1A) was synthesized in-house by the Chemical Biology & Therapeutics Department, St. Jude Children’s Research Hospital (Memphis, TN, USA). The internal standard (ISTD), dabrafenib (M.W.: 519.56 g/mol; M.F.: C23H20F3N5O2S2; Purity by HPLC & TLC: 100%; Fig. 1B), was purchased from Cayman Chemical (Ann Arbor, Michigan, USA). The chemicals, Optima™ LC/MS grade Acetonitrile, Optima™ LC/MS grade Methanol, ACS grade Dimethyl Sulfoxide (DMSO) and Honeywell Fluka™ Formic acid for mass spectrometry, were procured from Fisher Scientific (Waltham, MA, USA). Pharmaceutical grade Trappsol® 1-Hydroxypropyl-β-Cyclodextrin (BCD) was obtained from Cyclodextrin Technologies Development (Gainesville, FL, USA). Double distilled water was prepared in-house using a MilliporeSigma water purification system (Burlington, MA, USA). Blank K2EDTA CD-1 murine plasma was bought from BioIVT (Westbury, NY, USA). Laboratory grade Ringer’s solution was purchased from Frey Scientific (Nashua, NH, USA).
Figure 1.
Chemical Structures of JQ1 (A) and dabrafenib, ISTD (B).
2.2. Preparation of Primary Stock and Working Solution
Based on the solubility data available for JQ1 and dabrafenib, both analyte and ISTD primary stock solutions were prepared in DMSO at 1 mg/mL concentration by independent measurements. The calibrator working solutions at concentrations 10, 20, 50, 100, 200, 500, 1000, and 2000 ng/mL; and the quality control working solutions at concentrations 30, 300, and 1600 ng/mL were prepared in methanol: water (1:1, v/v) for the mouse plasma method. For the microdialysate method, the calibrator working solutions at concentrations 1, 2, 10, 50, 100, 250, 500, and 1000 ng/mL; and the quality control working solutions at concentrations 3, 200, and 800 ng/mL were prepared in methanol: water (1:1, v/v). The calibrant and quality control working solutions were stable for three months when stored at 4°C ± 2°C. The ISTD working solution (1 μg/mL) was made freshly on the day of the assay by serially diluting the ISTD primary stock with methanol: water (1:1, v/v).
2.3. Instrumentation
The chromatographic system comprised of a Shimadzu Nexera X2 high performance liquid chromatograph (HPLC, Kyoto, Japan) which was equipped with a binary pump (Model: LC-30AD), a degasser (Model: prominence DGU-A5), an autosampler (Model: SIL-30AC), a communication bus module or controller (Model: prominence CBM-20A) and a column oven (Model: Phenomenex Thermasphere TS-130). This HPLC system was interfaced with an Applied Biosystems Hybrid Q-Trap 4000 mass spectrometer (Framingham, MA, USA) equipped with electrospray ionization (ESI) source. The Sciex Analyst® Software (version 1.7.1) was used for the LC-MS/MS spectral data acquisition, peak integration, and quantification. Solid phase extraction (SPE) was performed on a Waters™ 96-well plate vacuum manifold (Milford, MA, USA). A Mettler Toledo®-XP26 analytical balance (Columbus, OH, USA) was used for weighing standards and buffers. Additionally, a multi-tube vortex shaker from Fisher Scientific (Waltham, MA, USA), Eppendorf centrifuge 5804R (Enfield, CT, USA) and Eppendorf microcentrifuge 5415R (Enfield, CT, USA) were used for biological sample preparation.
2.4. Chromatographic and Mass Spectrometric Conditions
The chromatographic separation of JQ1 and ISTD was achieved on a Synergi Max-RP column (80A, 75 X 2.00 mm, 4 μm particle size) attached to a KrudKatcher™ Ultra HPLC In-Line Filter (2 μm Depth Filter X 0.004in ID) from Phenomenex (Torrance, CA, USA). A column oven was used to enclose and maintain the temperature of the analytical column at 45° throughout the sample analysis. Mobile phase A and B were comprised of distilled water and acetonitrile, respectively, each containing 0.1 % formic acid. A gradient elution method was employed to elute the compounds starting with 25% B, rapidly increased to 75% B in 0.2 min, maintained at 75% B for next 2 min followed by gradual increase to 98% for 0.8 min, and finally decreased to 25% B over the next 2 min for equilibrating the column for next sample analysis. The flow rate was maintained constant at 0.6 mL/min throughout the sample run and the total run time/sample was 5 min. The injection volume was 5 μL/ sample and the autosampler temperature was set to 4°C ± 1°C. A flushing solution (acetonitrile: distilled water (1:1, v/v)) was used for rinsing the autosampler needle and port before and after sample injection.
ESI and MS detection parameters were optimized by direct infusion of JQ1 and dabrafenib (0.001 mg/mL solution in methanol) at a flow rate of 10 μL/ min using an external Harvard syringe pump Model 11 plus (Holliston, MA, USA). Full-mass scan spectra were acquired in the positive ionization mode over the mass range 200– 700 m/z and the mass transition optimized were 457.40 > 341.30 (JQ1; Fig. 2A) and 520.40 > 307.20 (dabrafenib; Fig. 2B). The compound dependent parameters such as declustering potential (DP), entrance potential (EP), collision energy (CE) and collision exit potential (CXP) were set at 60.0 V, 10.0 V, 45.0 eV and 10.0 V, respectively for JQ1; while for dabrafenib, it was 125.0 V, 9.0 V, 30.0 eV and 8.0 V, respectively. The compound independent parameters including curtain gas (CUR), ion spray voltage (ISP), nebulizer gas (GS1) and heater gas (GS2) were set at 20 psi, 5500 V, 40 psi and 60 psi, respectively for both JQ1 and dabrafenib. The ion source temperature was maintained at 600°C, while the dwell time for monitoring both the transition was set to 200 ms.
Figure 2.
Typical full scan precursor and product ion mass spectrum of JQ1 (A and B) and dabrafenib (C and D).
2.5. Preparation of calibrator standards and quality control samples
For the mouse plasma curve, the calibrator standards were prepared by spiking 10 μL of calibrator working solutions into 20 μL aliquots of blank CD1 mouse plasma to yield JQ1 concentrations of 5,10, 20, 50, 100, 200, 500, and 1000 ng/mL. In a similar manner, the quality control (QC) samples were prepared from respective QC working solutions in blank CD1 mouse plasma at JQ1 concentrations of 15 ng/mL (LQC), 150 ng/mL (MQC), and 800 ng/mL (HQC). The perfusate used for cerebral microdialysis study was Ringer’s solution modified with 10% BCD and therefore, this solution was used as blank microdialysate for the method validation studies. For the microdialysis curve, the calibrator standards were prepared by spiking 10 μL of calibrator working solutions into 20 μL aliquots of blank microdialysate to yield JQ1 concentrations of 0.5,1, 5, 25, 50, 125, 250, and 500 ng/mL. Likewise, the quality control samples were prepared from respective QC working solutions in blank microdialysates at JQ1 concentrations of 1.5 ng/mL (LQC), 100 ng/mL (MQC), and 400 ng/mL (HQC).
2.6. Plasma and Ringer’s/BCD Sample Preparation
A simple solid phase extraction protocol was employed to extract JQ1 and ISTD from mouse plasma and microdialysate samples. Initially, the study samples were thawed at room temperature and vortex mixed. A 20 μL aliquot of the mouse plasma or microdialysate sample was transferred to a 1.5 mL siliconized low-retention microcentrifuge tube (Fisher Scientific, Waltham, MA, USA) and spiked with 10 μL of ISTD working solution (1 μg/mL). For volume correction, 10 μL of methanol: water (1:1, v/v) was added to mouse plasma and microdialysate study samples. After vortex mixing, the samples were protein precipitated with 0.2 mL of methanol followed by centrifugation at 12000 g at 4°C. The supernatant was transferred to another tube containing 0.2 mL of water (modified with 2% formic acid) and once again vortex mixed. This sample mixture was loaded on to an Oasis HLB μElution 96-well plate 30 μm (Waters, Milford, MA, USA), which was pre-conditioned with 0.2 mL methanol and 0.2 mL distilled water. Next, the well plate was flushed with 0.2 mL of distilled water after elution of the loaded mixture. In the final step, JQ1 and ISTD were eluted with 0.1 mL methanol into a clean 96 well collection plate. The collection plate was vortex mixed and centrifuged at 1000 rpm at 4°C after the addition of 0.1 mL of distilled water. This plate was transferred to the autosampler and, 5 μL sample extracts were injected onto the LC-MS/MS system.
2.7. Bioanalytical Method Validation
The bioanalytical method was validated to meet the acceptance criteria as defined in the U.S. Food and Drug Administration (USFDA) guidelines [4]. The major validation parameters assessed were linearity, precision, accuracy, selectivity, limit of detection (LOD), limit of quantitation (LOQ), matrix effect, recovery, carryover effect, dilution integrity, and stability.
2.7.1. Selectivity and Sensitivity
Selectivity studies were performed by analyzing six replicates at blank (no analyte or ISTD spiked) and LOQ level in plasma and microdialysate. Selectivity investigations were carried out in mouse plasma from two different mice species viz. CD1 mice and CD1 nude mice. For the microdialysate method, the Ringer’s/BCD solution was used to evaluate selectivity. The method selectivity was determined by comparing the blank and LOQ level chromatograms for potential interferences from the plasma and microdialysate matrix components. Signal-to-noise (S/N) ratio was computed to determine the sensitivity of the developed method. The LOD and LOQ were defined based on the S/N ratio of ≥ 3 and ≥ 10, respectively.
2.7.2. Calibration Curve
Eight-point concentration levels were prepared and analyzed on three successive days to assess the linearity of the developed LC-MS/MS method. The concentration ranges used for the mouse plasma and microdialysate method were 5–1000 ng/mL and 0.5–500 ng/mL, respectively. A weighted linear regression (1/x2) analysis, which is a plot of peak area ratios of JQ1 and ISTD against the JQ1 concentrations, was used to construct the calibration curves. A deviation of ± 15% was considered acceptable for the back-calculated calibrator standard concentration with an ± 20% exception for the LOQ level.
2.7.3. Precision and accuracy
Intra-day and inter-day precision and accuracy were assessed within the same day and on three consecutive days at the LOQ, LQC, MQC, and HQC levels with six replicates for each concentration in mouse plasma as well as microdialysate. The precision and accuracy data were expressed in terms of relative standard deviation (RSD) and relative error (RE). The acceptance criteria for low-, mid-, and high-quality control samples were within ± 15% RE and ≤ 15% RSD. For the LOQ level, the acceptance criteria were within ± 20 % RE and ≤ 20% RSD.
2.7.4. Matrix Effect and Recovery
Matrix Effect was quantified by dividing the peak areas of the post-extracted spiked samples to that of neat standard samples at equivalent concentrations, and this peak ratio was defined as Matrix Factor (MF). To prepare the post-extracted spiked samples, blank microdialysate and mouse plasma from two different strains, i.e. CD1 and CD1 nude mice, were initially extracted as per the sample preparation protocol and later spiked at LQC and HQC concentration with ISTD. The value of MF=1, MF ≤ 0.8, and MF ≥ 1.2 signified no matrix effect, ion suppression, and ion enhancement, respectively [5]. The mean extraction recoveries of JQ1 and ISTD were calculated by comparing peak area responses of JQ1 and ISTD in extracted samples with that of peak responses in post extracted spiked samples. The matrix effect and recovery studies were both performed in triplicates at low- and high-quality control levels for the mouse plasma and microdialysate method.
2.7.5. Stability
The stability of JQ1 in mouse plasma and microdialysate samples was investigated by comparing freshly processed samples against experimental samples kept under the following storage conditions: bench-top stability at room temperature and 4°C for 24 h, freeze (at −80°C) - thaw (at room temperature) stability over three cycles, and autosampler stability at 4°C for 48h. The long-term stability in mouse plasma and microdialysates was performed at −80°C for 91 and 21 days, respectively. The stability studies were carried out in triplicates at LQC and HQC levels for the mouse plasma and microdialysate method. These experimentally stored samples were considered stable if the percentage difference in concentration were within 15% of the freshly processed quality control samples.
2.7.6. Carry-over Effect and Dilution Integrity
The carry-over effect was evaluated by serial injection of the highest concentration of the plasma (1000 ng/mL) and microdialysate (500 ng/mL) calibration curve followed by blank plasma/microdialysate injections to elute any residual analyte. A carryover was considered if the peak area for JQ1 in the blank injections was greater than 20% LOQ.
Dilution integrity was determined for animal study samples whose concentrations were higher than the upper limit of quantification (ULOQ). To investigate the dilution effect, mouse plasma samples with initial concentrations of 16000 ng/mL (20 times QC3 concentration) were diluted 20-fold with blank plasma. For the microdialysate method, the microdialysate samples with initial concentrations of 4000 ng/mL (10 times QC3 concentration) were diluted 10-fold with the blank microdialysate. Appropriate dilution factors were applied to calculate the diluted sample concentrations on a freshly prepared plasma and microdialysate calibration curve.
2.8. Non-specific binding studies
Preliminary non-specific binding studies suggested that JQ1 in Ringer’s solution has an affinity to bind to components of the microdialysis system: syringes, tubing, probe and/or collection vials. Thus, several experiments were performed to evaluate and reduce JQ1 non-specific binding.
First, we evaluated non-specific binding of the compound to the reconstitution vessel, collection vial, or transfer device (e.g., glass beaker, Eppendorf tube, pipette tip). JQ1 was reconstituted in blank Ringer’s solution as well as Ringer’s solution with the addition of 10% BCD. The samples containing 10% BCD in Ringer’s solution were transferred to lo-bind tubes containing methanol before being stored in the freezer. Spiked solutions were prepared in plastic and glass reconstitution vessels, and samples from these vessels were transferred to different collection vials: plastic Eppendorf tubes, glass vials, and lo-bind tubes. Variance greater than 15% from the expected concentration in any of these samples suggested non-specific binding to one or several of these components.
Then, we determined whether JQ1 displayed non-specific binding properties through adsorption or retention to the tubing component of the microdialysis system. A spiked solution was perfused through the microdialysis tubing at a rate of 1.0 μL/min and three samples were collected, each for 30 min. If the sample collected was >15% different from the original solution that was perfused, this indicated that the drug was adsorbing to the tubing. Then, a blank solution was perfused through the tubing and two samples were collected for 30 min each. If any drug was detected in these samples, this suggested that the drug was retained on the tubing.
2.9. Plasma and cerebral microdialysis studies
Plasma and cerebral microdialysis studies were performed in female CD-1 nude mice (Charles River, Wilmington, MA). All the animal procedures were approved by the St. Jude Institutional Animal Care and Use Committee and met the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). The CNS penetration of JQ1 was assessed using a pharmacokinetic modeling and simulation method described previously and summarized below [6, 7].
First, a plasma pharmacokinetic study was performed in N=9 CD-1 nude mice dosed with 30 mg/kg JQ1 intraperitoneally. The JQ1 solution was prepared by adding 0.107 mL of NMP (5%), 0.107 mL of Solutol HS-15 (5%), and 1.934 mL of normal saline to 6.79 mg of JQ1. The final formulation was vortexed for 2 minutes to obtain a clear solution. Three blood samples (~75 μL each) were collected per mouse per retro-orbital bleed or cardiac stick, using a population-based study design. Mice were divided in three groups (3 mice per group), and blood samples were collected at 30 min, 2 h, and 18 h post-dose (group 1); 15 min, 1.5 h, and 6 h post-dose (group 2); or 1 h, 4 h, and 24 h post-dose (group 3). Blood samples were centrifuged to plasma which was separated stored at −80°C. Plasma concentration-time data were analyzed using population-based modeling (Monolix v2019R2. Antony, France: Lixoft SAS, 2019). The pharmacokinetic parameter estimates were used to develop a limited plasma sampling model for JQ1 using the D-optimality method implemented in ADAPT 5 (BSMR, Los Angeles, CA, USA).
In vivo cerebral microdialysis experiments were conducted in N=4 CD-1 nude mice dosed with 30 mg/kg JQ1 intraperitoneally to sample unbound JQ1 in the brain ECF. Microdialysis guide cannula (MD-2255, BASi) were implanted into the cerebral cortex of mice using a stereotactic instrument as described previously. The day of the experiment, microdialysis probes were primed and flushed with Ringer’s solution containing 10% BCD, and then slowly inserted through the guide cannula into the cortex and allowed to equilibrate for 1 h at 0.5 μL/min. After JQ1 dosing, dialysate fractions consisting of Ringer’s with BCD were collected over 1 h intervals for 6 h at a flow rate of 0.5 μL/min for a total of 30 μL per sample. These samples were collected continuously, so the equilibrium was maintained. The use of Ringer’s/BCD dialysate over simple Ringer’s was justified by issues associated with nonspecific binding of JQ1 to components of the microdialysis system that were identified by our group during preliminary studies (data not shown). Blood samples were collected at the time-points determined by the limited sampling model via retro-orbital bleeding and plasma was extracted as described above. Recovery of each microdialysis probe was determined in-vitro using the zero-flow rate method with dialysates collected at different flow rates 0.5, 1,1.5, and 2 μL/min, as previously described [8].
JQ1 plasma and ECF concentration-time data were sequentially analyzed with a population-based model in which the plasma parameters were fixed to the values estimated in the first step. For each mouse, JQ1 total plasma and unbound ECF exposures were determined as the area under plasma and ECF concentration–time curves up to 6 h (AUC0–6h) and were estimated by integration of the model-predicted pharmacokinetic profiles.
3. Results and Discussion
3.1. Mass Spectrometric Detection and Optimization
Syringe pump infusion experiments for JQ1 (0.001 mg/mL) and ISTD (0.001 mg/mL) prepared in methanol were performed in the positive and negative ionization modes for the development of a sensitive ESI-MS/MS method. The mass spectra for JQ1 and ISTD revealed that the intensity of the protonated molecular ion peak, [M+H]+, was significantly greater than the deprotonated molecular ion peak, [M-H]−. The [M+H]+ ion peaks at m/z 457.3 for JQ1 and at m/z 520.4 for ISTD were selected as precursor ions. The fragmentation of precursor ion for JQ1 using a collision energy of 45 eV resulted in the formation of high intensity product ion at m/z 341.3, most likely due to the loss of tert-butyl acetate ion. The other prominent product ion peaks detected were at m/z 383.4 and 274.2. Figure 3 illustrates the proposed fragmentation pattern for JQ1 when a CE of 45 eV is applied. Likewise, a high intensity ISTD product ion peak was detected at m/z 307.2 when using a collision energy of 25 eV. The fragmentation product structure for ISTD has been described previously [9]. The SRM transitions used for method quantitation of JQ1 and ISTD were m/z 457.3 > 341.3 and m/z 520.4 > 307.2, respectively. Subsequently, the compound dependent as well as gas parameters were optimized to achieve consistent and stable mass response for JQ1 and ISTD SRM transitions. Figure 2A-D illustrates the representative precursor and product ion mass spectrum for JQ1 and ISTD.
Figure 3.
The proposed fragmentation pattern for the most abundant product ions of JQ1.
3.2. Chromatographic Conditions Optimization
The elimination of the evaporation and reconstitution step during solid phase extraction of mouse plasma and microdialysate samples meant that the selection of analytical column and mobile phase was paramount to produce symmetrical chromatographic peaks for JQ1 and ISTD. During the initial chromatographic method development, varying gradients of methanol-water and acetonitrile-water mobile phases were evaluated on a Kinetix® C18 analytical column (100A, 50 X 2.1 mm, 2.6 μm particle size) from Phenomenex (Torrance, CA, USA). Methanol in the mobile phase resulted in a significantly high background noise as well as a slightly lower JQ1 peak area response when compared to the acetonitrile phase. However, a slight analyte peak tailing was observed with acetonitrile in the mobile phase. Subsequently, the effect of mobile phase additives including, ammonium acetate (1, 5, and 10 mM) and formic acid (0.05 and 0.1%), were investigated. It was found that the addition of 0.1% formic acid to mobile phase A and B significantly enhanced JQ1 sensitivity but failed to address the analyte peak tailing problem. Therefore, we tested a C12 column known to reduce analyte peak tailing and offer hydrophobic retention properties akin to a C18 column. Under the optimized chromatographic conditions discussed above, the Synergi Max-RP (80A, 75 X 2.00 mm, 4 μm particle size) C12 column yielded sharper and symmetrical analyte peaks with reduced analyte peak tailing. The typical retention times observed for JQ1 and ISTD on this C12 column were 1.77 ± 0.05 min and 1.41 ± 0.05 min, respectively.
3.3. Sample Pretreatment
The sample preparation protocol was optimized keeping in mind the complexity of the matrices under investigation (i.e., mouse plasma and microdialysates with high salt composition). Considering the relative ease and simplicity to perform protein precipitation, the first approach was to develop an extraction protocol by employing precipitating agents such as methanol and acetonitrile. Even though the methanol precipitation protocol was able to detect the LOQ levels in both matrices, the results suggested inconsistent recoveries as well as a high mass spectral background interference. Therefore, it was decided to employ a SPE after protein precipitation to have a more effective sample clean up, decreased ion suppression, and excellent recoveries. Oasis HLB, MCX, and WAX μElution 96-well plate 30 μm (Waters, Milford, MA, USA) were compared. A significantly higher and more consistent recovery for JQ1 ensued the use of Oasis HLB μElution well plate for pretreatment of both mouse plasma and microdialysate samples. It was also observed that dilution of the methanol precipitated sample with water containing 2% formic acid resulted in better method sensitivities which might be due to better retention of analyte to the SPE extraction plate. Additionally, flushing the extraction plate with 0.2 mL water and analyte elution with 0.1 mL methanol resulted in optimum JQ1 recovery and minimum matrix interferences.
3.4. Selection of Internal Standard
One of the prerequisites for quantitative LC-MS/MS analysis is a stable labelled internal standard (SIL) that would mimic the analyte during the entire sample analysis from sample pretreatment to mass spectrometric detection. However, SIL for JQ1 was not commercially available at the time of this method development. Additionally, factoring in the cost and time associated with synthesizing the JQ1 SIL, we decided to choose a structural analog as an ISTD for the LC-MS/MS assay. Among the several tested compounds, dabrafenib did not interfere with the measurement of JQ1 and possessed characteristics such as extractability from plasma and microdialysate samples as well as chromatographic retention akin to JQ1. Furthermore, the method validation data confirmed the use of dabrafenib as an ISTD for the current LC-MS/MS assay.
3.5. LC-MS/MS Method Validation
Figure 4 and 5 illustrates the typical SRM chromatograms generated for the analysis of blank, LOQ control and in vivo study mouse plasma and microdialysate samples, respectively. During the mouse plasma and microdialysate sample analysis, no significant interferences from the endogenous matrix compounds were observed at the retention times of JQ1 (1.77 min) and ISTD (1.41 min), thus indicating excellent selectivity for the developed LC-MS/MS methods. The LOD and LOQ obtained were 0.75 ng/mL (C.V.: 18.15%) and 5 ng/mL (C.V.: 4.50%) for the mouse plasma; 0.25 ng/mL (C.V.: 15.43%) and 0.5 ng/mL (C.V.: 9.84%) for the microdialysate samples, respectively.
Figure 4.
Representative LC-MS/MS chromatogram of JQ1 (left panel) and dabrafenib (right panel) generated for analysis of blank (analyte and ISTD free) mouse plasma (A) plasma spiked with 5 ng/mL JQ1 (B) and plasma sample obtained at 0.25h after administration of 30 mg/kg JQ1 intraperitoneally (C).
Figure 5.
Representative LC-MS/MS chromatogram of JQ1 (left panel) and dabrafenib (right panel) generated for analysis of blank (analyte and ISTD free) microdialysate (A) microdialysate spiked with 0.5 ng/mL JQ1 (B) and brain extracellular fluid sample obtained at 1h after administration of 30 mg/kg JQ1 intraperitoneally (C).
The calibration curves were found to be linear over the concentration ranges of 5–1000 ng/mL and 0.5–500 ng/mL for JQ1 in the mouse plasma and microdialysate samples, respectively. Table 1 summarizes the linearity parameters obtained for JQ1 from analyzing the mouse plasma and microdialysates calibration curves. Table 2 illustrates the results for inter-day and intra-day precision and accuracy at quality control levels for JQ1 in mouse plasma and microdialysate samples. The accuracy and precision values were within ± 10.59 % and ≤ 6.61% for the mouse plasma and microdialysate quality control samples, respectively, suggesting good reliability and reproducibility of the developed method.
Table 1.
Calibration curve parameters for JQ1 in mouse plasma and microdialysates (n=3).
| Sample Matrix | Calibration range (ng/mL) | Slope | Intercept | r2 |
|---|---|---|---|---|
| Mean ± S.D. | Mean ± S.D. | Mean ± S.D. | ||
| Mouse plasma | 5 – 1000 | (2.50 ± 0.1) X 10−3 | (1.30 ± 0.8) X 10−3 | 0.9955 ± 0.0021 |
| Microdialysates | 0.5 – 500 | (4.44 ± 0.2) X 10−3 | (−2.00 ± 1.0) X 10−4 | 0.9986 ± 0.0007 |
r2: Coefficient of determination
Table 2.
Inter-day and Intra-day precision and accuracy for JQ1 in mouse plasma and microdialysates (n=6 for Intra-day and n=18 for Inter-day).
| Sample Matrix | Nominal Concentration (ng/mL) | Mean Calculated concentration (ng/mL) | Accuracy (% R.E.) | Precision (% R.S.D.) |
|---|---|---|---|---|
| Mouse Plasma | Intra-day assay (n=6 replicates at each concentration) | |||
| 5 | 4.47 | −10.59 | 3.76 | |
| 15 | 14.89 | −0.76 | 5.24 | |
| 150 | 148.80 | −0.80 | 3.34 | |
| 800 | 720.29 | −9.96 | 2.22 | |
| Inter-day assay (n=6 replicates at each concentration, 3 days) | ||||
| 5 | 4.59 | −8.21 | 6.61 | |
| 15 | 14.64 | −2.42 | 5.26 | |
| 150 | 145.69 | −2.87 | 3.98 | |
| 800 | 724.33 | −9.46 | 3.03 | |
| Microdialysate | Intra-day assay (n=6 replicates at each concentration) | |||
| 0.5 | 0.52 | 4.28 | 3.46 | |
| 1.5 | 1.49 | −0.35 | 4.19 | |
| 100 | 95.89 | −4.11 | 2.36 | |
| 400 | 410.37 | 2.59 | 5.10 | |
| Inter-day assay (n=6 replicates at each concentration, 3 days) | ||||
| 0.5 | 0.53 | 5.79 | 5.96 | |
| 1.5 | 1.53 | 2.25 | 4.68 | |
| 100 | 96.44 | −3.56 | 2.87 | |
| 400 | 411.18 | 2.79 | 3.08 | |
%RSD: relative standard deviation; %RE: relative error
Matrix effect is a very well-known phenomenon affecting ESI mass spectrometric bioanalysis by altering the ionization of compounds either by inducing suppression or enhancement of ion response. In the current study, protein precipitated mouse plasma as well as microdialysate samples were subjected to SPE to reduce the matrix effect on LC-MS/MS analysis. As shown in Table 3, the post-extraction MF for JQ1 ranged from 0.95 to 1.01, with R.S.D. ≤ 5.77% for mouse plasma; 0.97 to 1.01, with R.S.D. ≤ 5.77% for the microdialysate samples, respectively. The MF values close to 1 signifies negligible ion suppression or enhancement for JQ1 in mouse plasma and microdialysate. Table 3 illustrates the extraction recovery for JQ1 from mouse plasma and microdialysate samples at LQC and HQC levels. The average extraction recoveries for JQ1 in CD1 and CD1 nude mouse plasma samples were 86.14% and 88.71%, respectively, with the R.S.D.≤ 6.50% for all the analyzed QC samples. Likewise, the average extraction recoveries for JQ1 in microdialysate samples was 100.18% with the R.S.D.≤ 4.33% for all the analyzed QC samples. The % R.S.D values of ≤ 15% for the recovery experiments suggest consistent recoveries for JQ1 at all the tested QC concentrations in mouse plasma and microdialysate samples. The average extraction recoveries for ISTD in mouse plasma and microdialysate samples were 76.58%, and 87.24%, respectively, with the R.S.D. value ≤ 11.01% for all the analyzed QC samples.
Table 3.
Matrix effect and recovery evaluation for JQ1 in mouse plasma and microdialysates (n=3).
| Sample Matrix | Compound | Nominal Concentration (ng/mL) | Matrix Effect | Recovery | ||
|---|---|---|---|---|---|---|
| Mean calculated MF value | % R.S.D. | % Mean Recovery | % R.S.D. | |||
| Mouse Plasma CD1 | JQ1 | 15 | 0.95 | 4.91 | 82.18 | 6.50 |
| 800 | 1.01 | 0.26 | 90.09 | 6.48 | ||
| ISTD | 500 | 0.99 | 2.66 | 74.84 | 11.01 | |
| Mouse Plasma CD1 nude | JQ1 | 15 | 0.96 | 5.77 | 84.43 | 2.88 |
| 800 | 1.00 | 0.93 | 92.99 | 1.31 | ||
| ISTD | 500 | 0.99 | 2.72 | 78.31 | 7.49 | |
| Microdialysate | JQ1 | 1.5 | 0.97 | 8.66 | 104.35 | 4.43 |
| 400 | 1.01 | 0.21 | 96.00 | 1.55 | ||
| ISTD | 500 | 1.03 | 0.65 | 87.24 | 5.60 | |
%RSD: relative standard deviation
The results of the stability assessments performed for JQ1 spiked in mouse plasma and microdialysates under different experimental storage conditions are summarized in Table 4. JQ1 was stable for 24 h in spiked mouse plasma and microdialysate samples after short-term storage at 4°C and room temperature, indicating that the samples were stable under the laboratory handling conditions. The stability of JQ1 was confirmed in spiked mouse plasma and microdialysate samples after long-term storage at −80°C for 91 and 21 days, respectively, highlighting the reliability of the developed LC-MS/MS method to handle in vivo study samples. In addition, the extracted quality control samples for JQ1 in mouse plasma and microdialysates were stable in autosampler for 48 h, indicating good post-extractive stability for the analyte. It was observed that no-stability related issues for JQ1 were observed for mouse plasma and microdialysate samples undergoing three freeze-thaw cycles. When stored at −80°C for 88-days, the primary stock solution of JQ1 was found to be stable with R.S.D ≤ 1.16%.
Table 4.
Summary of stability evaluation for JQ1 in mouse plasma and microdialysates (n=3).
| Sample Matrix | Stability Study | Nominal Concentration (ng/mL) | Mean ± S.D. Calculated Concentration (ng/mL) | Precision (% R.S.D.) | Accuracy (% R.E.) | % Mean deviation |
|---|---|---|---|---|---|---|
| Mouse Plasma | Process a | 15 | 14.93 ± 1.10 | 7.34 | −0.46 | −3.80 |
| 800 | 686.39 ± 43.52 | 6.34 | −14.20 | −7.24 | ||
| Bench-top RT b | 15 | 16.27 ± 1.80 | 11.07 | 8.47 | 4.84 | |
| 800 | 767.80 ± 29.98 | 3.90 | −4.02 | 3.76 | ||
| Bench-top 4C c | 15 | 15.57 ± 0.98 | 6.30 | 3.80 | 0.33 | |
| 800 | 767.45 ± 16.91 | 2.20 | −4.07 | 3.71 | ||
| Freeze-thaw d | 15 | 14.40 ± 0.51 | 3.52 | −3.98 | −7.20 | |
| 800 | 793.88 ± 23.31 | 2.94 | −0.77 | 7.29 | ||
| Long-term e | 15 | 14.70 ± 0.69 | 4.69 | −2.03 | −5.31 | |
| 800 | 788.56 ± 19.49 | 2.47 | −1.43 | 6.57 | ||
| Microdialysate | Process a | 1.5 | 1.45 ± 0.10 | 6.88 | −3.09 | 0.46 |
| 400 | 388.38 ± 2.23 | 0.57 | −2.91 | 1.39 | ||
| Bench-top RT b | 1.5 | 1.49 ± 0.06 | 3.89 | −0.42 | 3.23 | |
| 400 | 403.22 ± 4.61 | 1.14 | 0.81 | 5.26 | ||
| Bench-top 4C c | 1.5 | 1.53 ± 0.03 | 1.84 | 2.22 | 5.96 | |
| 400 | 400.64 ± 13.48 | 3.37 | 0.16 | 4.59 | ||
| Freeze-thaw d | 1.5 | 1.62 ± 0.017 | 1.03 | 7.83 | 11.77 | |
| 400 | 408.69 ± 6.31 | 1.54 | 2.17 | 6.69 | ||
| Long-term f | 1.5 | 1.63 ± 0.15 | 8.88 | 8.85 | 12.83 | |
| 400 | 404.60 ± 6.46 | 1.60 | 1.15 | 5.62 |
%R.S.D.: relative standard deviation; %R.E.: relative error
Stability assessed after 48 h in auto sampler at 4°C
Short-term stability after 24h at room temperature
Short-term stability after 24h at 4°C
Stability evaluated after three freeze-thaw cycles
91 days long-term stability in mouse plasma at −80°C
21 days long-term stability in microdialysate at −80°C
The precision (R.S.D.) and accuracy (Bias) for the 20-fold dilution of mouse plasma samples were 11.10% and −3.36%, respectively, while the precision and accuracy values for the 10-fold dilution of microdialysate samples were 1.47% and 0.53%, correspondingly. The data suggested that the dilution integrity was ensured up to 20- and 10-fold for the mouse plasma and microdialysate samples, respectively. Finally, no analyte or ISTD carryover was observed in blank plasma and microdialysate samples following the injection of ULOQ samples, indicating efficient clean-up of the LC system between sample injections.
3.6. Non-specific binding studies
The first non-specific binding study showed that JQ1 in Ringer’s solution was highly bound to both plastic, glass, and lo-bind tubes (Table 5). Using an additive (i.e., 10% BCD) and transferring the samples to a tube containing methanol before freezing, successfully reduced the non-specific binding (Table 5). All the samples containing 10% BCD and that were transferred to methanol were within 14% of the nominal concentration.
Table 5.
JQ1 non-specific binding to reconstitution and storage vessels.
| Ringer’s solution | 10% BCD in Ringer’s solution, transferred to MeOH | |||||
|---|---|---|---|---|---|---|
| Reconstitution Vessel | Collection Vial | Sample | Measured Concentration (ng/ml) | Accuracy (%) | Measured Concentration (ng/ml) | Accuracy (%) |
| 1 - Plastic | A - Glass | 1-A | 1.23 | 2.45 | 45.512 | 91.02 |
| 1 - Plastic | B - Plastic | 1-B | 0.27 | 0.53 | 44.973 | 89.95 |
| 1 - Plastic | C - Lo-bind | 1-C | 1.02 | 2.03 | 49.077 | 98.15 |
| 2 - Plastic | A - Glass | 2-A | 1.82 | 3.64 | 45.512 | 91.02 |
| 2 - Plastic | B - Plastic | 2-B | 0.33 | 0.65 | 44.973 | 89.95 |
| 2 - Plastic | C - Lo-bind | 2-C | 1.48 | 2.96 | 49.077 | 98.15 |
| 3 - Glass | A - Glass | 3-A | 18.52 | 37.04 | 44.137 | 88.27 |
| 3 - Glass | B - Plastic | 3-B | 3.90 | 7.79 | 43.074 | 86.15 |
| 3 - Glass | C - Lo-bind | 3-C | 30.55 | 61.11 | 47.239 | 94.48 |
| 4 - Glass | A - Glass | 4-A | 31.25 | 62.50 | 45.868 | 91.74 |
| 4 - Glass | B - Plastic | 4-B | 7.22 | 14.45 | 43.179 | 86.36 |
| 4 - Glass | C - Lo-bind | 4-C | 34.51 | 69.02 | 47.883 | 95.77 |
Nominal concentration 50 ng/ml
The second non-specific binding study showed that JQ1 did not bind to the tubing used in the microdialysis system. The percent binding of the drug to the tubing was less than 3% (Table 6), and no drug was able to be measured in the system when flushed after being perfused with a spiked solution.
Table 6.
JQ1 adsorption to microdialysis tubing.
| Syringe 1 | Syringe 2 | |
|---|---|---|
| Stock Average (1S & 2S) | 48.38 | 48.16 |
| Perfusate Average (1P, 2P, 3P) | 46.86 | 47.04 |
| Percent binding (%) | 3.14 | 2.32 |
Accordingly, the microdialysis studies were performed using glass collection tubes, 10% BCD in Ringer’s solution as the perfusate, and 20 μL of the samples would immediately be transferred to lo-bind tubes containing 200 μL of methanol in which 20 μL of the samples were immediately transferred.
3.7. Application to plasma pharmacokinetic and cerebral microdialysis studies
A plasma pharmacokinetic study of JQ1 was performed in non-tumor bearing mice dosed with 30 mg/kg JQ1 intraperitoneally. All JQ1 concentrations collected after 6 h post-dose fell under the LOQ. JQ1 total plasma concentration-time data were well captured with a linear one-compartment model, in which the intraperitoneal absorption was assumed immediate and total (Figure 6). The model parameter estimates are reported in Table 7. The limited sampling model determined the following informative time-points for plasma sampling during the cerebral microdialysis study: 15 min, 1 h, and 6 h post-dose.
Figure 6.
JQ1 plasma and extracellular fluid pharmacokinetic model (A) and concentration-time profiles (B) in CD-1 nude mice dosed with 30 mg/kg JQ1 intraperitoneally. Observed and model-predicted plasma data are shown by circles and solid line, respectively. Observed and model-predicted extracellular fluid data are shown by squares et dashed line, respectively.
Table 7.
JQ1 pharmacokinetic parameter estimates.
| Parameter (Unit) | Definition | Estimate (RSE%) | IIV (RSE%) |
|---|---|---|---|
| Plasma model | |||
| CL (L/h/kg) | Central clearance | 4.3 (11) | 0.32 (25) |
| V (L/kg) | Volume of distribution | 7.0 (11) | 0.33 (24) |
| Extracellular fluid (ECF) model | |||
| CLin (L/h/kg) | Plasma to ECF influx clearance | 4.01·10−5 (6) | 0.11 (47) |
| CLef (L/h/kg) | ECF to plasma efflux clearance | 1.24·10−3 (1) | - |
RSE%: relative standard error; IIV: interindividual variability reported as standard deviation
Cerebral microdialysis was performed in non-tumor bearing mice dosed with 30 mg/kg JQ1 intraperitoneally to sample brain unbound ECF JQ1 concentrations. The mean (SD) in vitro recovery of microdialysis probes determined by the zero-flow rate method was 65.9 ± 10.8%. JQ1 unbound ECF concentration-time data were well captured with a one-compartment model linked to the plasma model (Figure 6). Final parameter estimates are reported in Table 7. All the model parameters were well estimated with relative standard errors (RSE%) values below than 30%.
Mean (SD) JQ1 total plasma and unbound ECF AUC0–6h were 6975 ± 1895 h·ng/mL and 156.8 ± 54.2 h·ng/mL, respectively. As shown in Figure 6, JQ1 unbound ECF concentrations were above the IC50 values of 35.2/15.1 ng/mL (i.e., 77/33 nM) for BRD4(1/2) determined in cell-free assays, for at least 2 h post-dose [10].
4. Conclusions
A precise, sensitive, and selective LC-MS/MS method was developed and validated for quantification of JQ1 in mouse plasma and cerebral microdialysates in accordance with the US FDA bioanalytical guidelines. A low volume mouse plasma (20 μL) and microdialysates (20 μL) SPE sample preparation protocol was developed which resulted in consistent and reproducible recoveries for JQ1 and ISTD with minimal matrix interferences. Noteworthy advantages of the LC-MS/MS method include a relatively short chromatographic runtime of 5 min and a wide dynamic linear range for the developed assays (5–1000 ng/mL for the mouse plasma curve and 0.5–500 ng/mL for the microdialysate curve). Furthermore, in the current work, we have successfully demonstrated an application of the LC-MS/MS methods to JQ1 pharmacokinetic and cerebral microdialysis studies. Lastly, the developed mouse plasma and microdialysate LC-MS/MS methods could be further explored in future JQ1 related bioanalytical studies.
Highlights.
Developed and validated novel LC-MS/MS method for JQ1 in mouse plasma in dialysate
Method utilized HLB uElution plate-based extraction method for sample preparation
Method sensitive for JQ1 with LOQ of 5 and 0.5 ng/ml in mouse plasma and dialysate
Method applied to cerebral microdialysis study to estimate JQ1 CNS penetration
Acknowledgements –
This work was supported by a Cancer Center Support (CORE) Grant CA 21765, 2R01CA194057, and the American Lebanese Syrian Associated Charities (ALSAC). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health..
Footnotes
Conflict of interest
The authors have no conflict of interest to declare.
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- [1].Hovestadt V, Smith KS, Bihannic L, Filbin MG, Shaw ML, Baumgartner A, DeWitt JC, Groves A, Mayr L, Weisman HR, Richman AR, Shore ME, Goumnerova L, Rosencrance C, Carter RA, Phoenix TN, Hadley JL, Tong Y, Houston J, Ashmun RA, DeCuypere M, Sharma T, Flasch D, Silkov A, Ligon KL, Pomeroy SL, Rivera MN, Rozenblatt-Rosen O, Rusert JM, Wechsler-Reya RJ, Li XN, Peyrl A, Gojo J, Kirchhofer D, Lotsch D, Czech T, Dorfer C, Haberler C, Geyeregger R, Halfmann A, Gawad C, Easton J, Pfister SM, Regev A, Gajjar A, Orr BA, Slavc I, Robinson GW, Bernstein BE, Suva ML, Northcott PA, Resolving medulloblastoma cellular architecture by single-cell genomics, Nature 572(7767) (2019) 74–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Northcott PA, Robinson GW, Kratz CP, Mabbott DJ, Pomeroy SL, Clifford SC, Rutkowski S, Ellison DW, Malkin D, Taylor MD, Gajjar A, Pfister SM, Medulloblastoma, Nat Rev Dis Primers 5(1) (2019) 11. [DOI] [PubMed] [Google Scholar]
- [3].Matzuk MM, McKeown MR, Filippakopoulos P, Li Q, Ma L, Agno JE, Lemieux ME, Picaud S, Yu RN, Qi J, Knapp S, Bradner JE, Small-molecule inhibition of BRDT for male contraception, Cell 150(4) (2012) 673–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Guidance for Industry, Bioanalytical Method Validation, in: F. US Department of Health and Human Services, CDER, CVM; (Ed.) May 2018. [Google Scholar]
- [5].Matuszewski BK, Constanzer ML, Chavez-Eng CM, Strategies for the assessment of matrix effect in quantitative bioanalytical methods based on HPLC-MS/MS, Anal Chem 75(13) (2003) 3019–30. [DOI] [PubMed] [Google Scholar]
- [6].Jacus MO, Throm SL, Turner DC, Patel YT, Freeman BB 3rd, Morfouace M, Boulos N, Stewart CF, Deriving therapies for children with primary CNS tumors using pharmacokinetic modeling and simulation of cerebral microdialysis data, Eur J Pharm Sci 57 (2014) 41–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Campagne O, Davis A, Maharaj AR, Zhong B, Stripay J, Farmer D, Roussel MF, Stewart CF, CNS penetration and pharmacodynamics of the CHK1 inhibitor prexasertib in a mouse Group 3 medulloblastoma model, Eur J Pharm Sci 142 (2020) 105106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Elmeliegy MA, Carcaboso AM, Tagen M, Bai F, Stewart CF, Role of ATP-binding cassette and solute carrier transporters in erlotinib CNS penetration and intracellular accumulation, Clin Cancer Res 17(1) (2011) 89–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Krens SD, van der Meulen E, Jansman FGA, Burger DM, van Erp NP, Quantification of cobimetinib, cabozantinib, dabrafenib, niraparib, olaparib, vemurafenib, regorafenib and its metabolite regorafenib M2 in human plasma by UPLC-MS/MS, Biomed Chromatogr 34(3) (2020) e4758. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Filippakopoulos P, Qi J, Picaud S, Shen Y, Smith WB, Fedorov O, Morse EM, Keates T, Hickman TT, Felletar I, Philpott M, Munro S, McKeown MR, Wang Y, Christie AL, West N, Cameron MJ, Schwartz B, Heightman TD, La Thangue N, French CA, Wiest O, Kung AL, Knapp S, Bradner JE, Selective inhibition of BET bromodomains, Nature 468(7327) (2010) 1067–73. [DOI] [PMC free article] [PubMed] [Google Scholar]