In Vitro Binding Affinity vs. In Vivo Site Occupancy: a PET Study of Four Diastereomers of Dihydrotetrabenazine (DTBZ) in Monkey Brain

Abstract

Introduction.

In vivo imaging methods such as Positron Emission Tomography (PET) can be used to examine the relationship between in vitro binding affinity and in vivo occupancy of binding sites in the brain for new drug candidates. In this study, PET imaging in monkey brain was used to evaluate that correlation for a set of four diastereomers of the compound dihydrotetrabenazine (DTBZ), the pharmacologically active metabolite of the drug tetrabenazine.

Methods.

PET studies of DTBZ diastereomers were completed in a single monkey brain. In vivo occupancies (ED₅₀) were estimated using multiple drug doses and the vesicular monoamine transporter 2 specific radioligand (+)-α-[¹¹C]DTBZ, employing a test-retest sequence of control PET scan, drug administration and a second PET scan completed on a single day.

Results.

DTBZ has three chiral carbon centers and eight possible stereoisomers, and in vivo occupancy of the target site VMAT2 was observed only for the four diastereomers of DTBZ having the 11bR absolute configuration. The estimated in vivo occupancies (ED₅₀ values from 0.023 to >3.15 mg/kg) correlated well (R² = 0.95) with the in vitro binding affinities (K_i values of 4 to 600 nM for the VMAT2), and an even better correlation (R² = 0.99) was found for the three isomers with in vitro binding affinities less than 100 nM.

Conclusions.

If the physiochemical (MW, log P, pKa) or physiological (metabolism, transport, protein binding) properties of a set of drug stereoisomers are considered similar, the binding affinities determined from in vitro assays may predict the in vivo occupancies of the target binding site in the monkey brain.

1. Introduction

The development of a targeted pharmaceutical for therapeutic or diagnostic imaging purposes, where a drug molecule is intended to occupy a high affinity binding site on a protein macromolecule (enzyme, receptor, transporter or ion channel), often begins with the identification of a molecular structure with significant in vitro affinity. Lead molecules might be identified based on natural products, through de novo molecular design, modification of existing drug structures, or via high throughput screening of molecular libraries. Medicinal chemists then often synthesize a series of derivatives of that lead structure, with the goal of increasing the in vitro affinity to improve specificity or selectivity, together with enhancing the ADME (absorption, distribution, metabolism and excretion) profile of the drug candidate.

An important question in the design of CNS drugs or brain imaging radiopharmaceuticals is whether the in vitro binding affinity necessarily predicts in vivo target occupancy. In the “free drug hypothesis”, the binding site occupancy is dependent on unbound drug concentrations, which in turn are only a fraction of the total brain concentration that also includes non-specific binding of drug to proteins and phospholipids. The total concentration of drug in the brain is dependent on multiple factors including blood-brain-barrier permeability, actions of membrane transporters involved in both uptake and extrusion, plasma protein binding, and metabolism. Studies have demonstrated that the unbound brain fractions correlate better with in vivo target site occupancy and likely pharmacological efficacy. The importance of the free drug concentrations in the brain makes it difficult to independently evaluate the impact of alterations of in vitro binding affinities since free fractions in the brain can vary widely between compounds of similar physiochemical properties (molecular weight, lipophilicity, pKa) and even within a set of compounds of similar chemical structure [1–5].

If binding site occupancy is postulated to be directly related to pharmacological effect of a therapeutic drug, understanding the relationships of in vitro binding affinities, brain uptake, and free fractions of drugs in the brain to in vivo site occupancy becomes valuable information for choosing optimum drug molecules for further development. The recent introduction of molecular imaging techniques that utilize radiotracers has provided methods to estimate the in vivo occupancy of a drug binding site, and has been of particular interest for the development of pharmaceuticals and radiopharmaceuticals that target specific sites in the human brain. Molecular imaging thus provides a non-invasive method of studying drug molecule uptake and binding in the absence of easy methods for tissue sampling and such studies have now become routine in the pharmaceutical industry [6–10].

We have considered that a study of the correlation of in vitro affinities to in vivo binding site occupancies might be best undertaken in a set of drug stereoisomers, where physiochemical (e.g., MW, logP, pKa, polar surface area) and physiological (membrane transport, metabolism) characteristics might be most similar. Many low molecular weight neuroactive drugs are chiral: the development of single isomers of new drug candidates has achieved greater emphasis in recent years, and older racemic drugs can be redesigned into preparations of a single active isomer (termed chiral switching [11]). In many cases, within an enantiomeric pair there is often a high (eutomer) and a low affinity molecule (distomer), and molecular imaging techniques can be employed to determine if there is any significant target engagement by the low affinity isomer. Many such studies have demonstrated that in vivo occupancy by enantiomers correlates with in vitro binding affinity in a simple binary fashion (one active, one inactive), but to examine the possibility of intermediate values for both affinity and occupancy, a larger set of diastereomers should be examined.

In this study, we have applied molecular imaging techniques (Positron Emission Tomography (PET)) to studies of dihydrotetrabenazine (DTBZ: 1,3,4,6,7,11b-hexahydro-9,10-methoxy-3-(2-methylpropyl)-2H-benzo[a]quinolizin2-ol), the metabolite of the clinically used drug tetrabenazine (TBZ: 1,3,4,6,7,11b-hexahydro-9,10-dimethoxy-3-(2-methylpropyl)-2H-benzo[a]quinolizin-2-one). In vivo, TBZ is rapidly and essentially completed reduced to DTBZ through the action of liver carbonyl reductase, and thus the pharmacological activity of TBZ is entirely due to its metabolites. DTBZ has three chiral carbon centers and thus eight potential stereoisomers (Fig. 1). The isomers of DTBZ have a wide range of in vitro binding affinities (from 1 to >20000 nM) for the vesicular monoamine transporter-2 (VMAT2) site in the mammalian brain, and the availability of the VMAT2 radioligand (+)-α- [¹¹C]DTBZ provided an opportunity to examine the relationship between in vitro binding affinities and in vivo occupancies.

Fig. 1.

Structures and in vitro binding affinities (from ref. 12) for the eight stereoisomers of dihydrotetrabenazine (DTBZ).

2. Materials and Methods

2.1. Chemicals

Racemic tetrabenazine (3RS,11bRS) was obtained from Centaur Pharmaceuticals PVT, LTD (Mumbai, India). The 2R,3R,11bR (92% ee), 2S,3R,11bR (98% ee), 2S,3S,11bS (100% ee) and 2R,3S,11bS (100% ee) isomers of DTBZ were prepared from racemic TBZ by minor modifications of literature procedures [12], with identities verified by ¹H-NMR, ¹³C-NMR and high-resolution mass spec analyses. Isomeric purity (enantiomeric excess) for each isomer was determined by chiral HPLC analyses (Chiralpak column, Phenomenex, USA; 50:50 acetonitrile:water, 1.6 ml/min). 9-O-Desmethyl-(+)-α-DTBZ (2R,3R,11bR isomer) was obtained from MonomerChem Inc (Durham, USA).

2.2. Radiochemical Synthesis

(+)-α-[¹¹C]DTBZ was prepared by reaction of 9-O-desmethyl-(+)-α-DTBZ with no-carrier-added [¹¹C]methyl iodide, using established procedures [13]. A new radiochemical synthesis of (+)-α-[¹¹C]DTBZ was completed for each PET scan. The final product was purified by high-performance liquid chromatography and formulated in a sterile solution of isotonic saline. Radiochemical purities exceeded 95% with an average molar activity of 135 +/− 96 TBq/mmol at the time of injection. The injected dose of radioactivity was equivalent for first and second scans (177 +/− 20 and 183 +/− 13 mBq), as were the mass amounts of cold compound present in the radioligand injections (0.59 +/− 0.32 and 0.65 +/− 0.45 microgram).

2.3. PET Studies

All procedures were approved by the University of Michigan Institutional Animal Use and Care Committee (IACUC). Thirty-two individual [¹¹C]DTBZ PET imaging studies were completed, all performed in a single adult female rhesus monkey (Macaca mulatta, 8 yr old at initiation of study, weight 6.5–7.2 kg during period of study). The minimum time between study sessions was one week. All studies were done as part of test-retest pairs involving a control scan (saline injection only) done first, a timed delay, administration of the test drug article, and a second PET study. Drug doses were administered in random order. To test reproducibility of the PET imaging procedure, two sessions were done as control:control protocols, with an intervening saline injection only. All of the studies were completed using the same imaging equipment (Concorde MicroPET P4, Siemens/Concorde Microsystems, Inc., Knoxville, TN, USA).

For all studies, the monkey was initially anesthetized with ketamine (15 mg/kg i.m) and glycopyrrolate (0.004 mg/kg i.m.), then intubated and placed on continuous inhalation of 2% isoflurane in oxygen, and a percutaneous catheter was inserted into one hindlimb for administration of radiotracers and test drug solutions. The monkey was placed supine with the head positioned and secured in the MicroPET P4 scanner. The body temperature (37°C) was maintained by an electronic heating pad, and respiratory rate, heart rate, and blood oxygenation (SPO2) were continuously monitored during imaging. Fluid balance was maintained using saline injections. Following a measured transmission scan, a dynamic emission PET scan was performed for 60 minutes. At the start of the dynamic sequence [¹¹C]DTBZ in 0.2– 1.0 ml was injected intravenously followed by a 1 ml flush of saline. For each dual study, 50 min after completion of the first emission scan, the pharmacological test article was injected intravenously (0.1–0.5 ml in saline) followed 10 min later by injection of [¹¹C]DTBZ and initiation of a second 60 min emission PET scan.

Emission data were collected using one of two framing sequences: sequence A (5 × 2 min, 4 × 5 min and 3 × 10 min) or B (12 × 5 min). The same frame sequence was always used within a single test-retest session. The PET emission scans were corrected for radioactive decay, dead-time, random and scattered coincidences, and attenuation, then reconstructed using two iterations of standard OSEM (ordered subset expectation maximization) for rapid initial convergence, followed by 6 iterations of a maximum a posteriori (MAP) algorithm that accounts for the point spread function of the detectors, yielding a reconstructed resolution of approximately 1.4 mm. Images from all scans were co-registered to the initial scan using SPM12 [14].

For quantitative analyses, three-dimensional volumes-of-interest (VOIs) for the striatum and the occipital cortex were defined on the average image of all scans on the monkey. The VOIs were defined using in-house software (IDA; Image Display & Analysis) by first defining a boundary surrounding each structure, and then setting threshold to exclude all voxels below the threshold value. As these VOIs were defined on the average of all of the co-registered scan images, this approach assures that the exact same VOIs were used for each scan. Striatal and occipital cortex tissue time-activity curves were then generated.

The image-derived tissue time-radioactivity data was analyzed using the Logan graphical method with the occipital cortex as the reference region [15] to calculate striatal distribution volume ratios (DVR), from which binding potentials were calculated as BP_ND-Logan = DVR – 1. The occupancy values were calculated as

$$\%\text{Occupancy}=100\left[\left({\text{BP}}^{\text{control}}–{\text{BP}}^{\text{test}}\right)/{\text{BP}}^{\text{control}}\right].$$

3. Results and Discussion

Tetrabenazine (TBZ) is a potent and reversible inhibitor of the vesicular monoamine transporter type 2 (VMAT2), the protein responsible for the movement of monoamines from the cytosol into storage vesicles in neurons [16]. Developed in the late 1950s/early 1960s as an antipsychotic, TBZ has been more recently approved in Europe, Japan and the United States for treatment of various hyperkinetic disorders [17,18]. Tetrabenazine has two chiral carbon centers and has always been and is currently marketed as a racemic mixture of the 3R,11bR and 3S,11bS enantiomers based on the expectations of these being the thermodynamically more stable isomers arising from the synthetic approach: the 3S,11bR and 3R,11bS isomers are not contained in commercial samples of TBZ. Studies of the metabolism of TBZ in rats and humans have conclusively demonstrated that it is rapidly and essentially completely reduced by carbonyl reductase in the liver [19] to form the pharmacologically active metabolite dihydrotetrabenazine (DTBZ: also called HTBZ or DHTBZ; Fig. 1). The reduction of the ketone to the hydroxyl produces a third chiral carbon center, and thus four probable in vivo DTBZ species (2R,3R,11bR, 2S,3R,11bR, 2R,3S,11bS and 2S,3S,11bS) arise from a dose of racemic (3RS,11bRS)TBZ.

The absolute configurations of any of the chiral centers in the DTBZ isomers remained undefined for decades [20], but following the first chiral resolution of DTBZ in 1997 [21] and X-ray crystallography, the assignment of the absolute configuration of the three chiral carbon centers for 2R,3R,11bR-DTBZ was made. Subsequently, all of the eight potential stereoisomers of DTBZ have been synthesized [12,22,23] and the in vitro binding affinities of each to the VMAT2 have been determined (Fig. 1). The two highest affinity isomers, 2R,3R,11bR-DTBZ and 2S,3R,11bR-DTBZ (termed α- and β-isomers) are most likely responsible for the pharmacological effect of racemic TBZ [24]. The 2R,3R,11bR-isomer formed the basis for the development of carbon-11 and fluorine-18 labeled VMAT2 imaging agents for human PET studies [25], and that single isomer was also used in the development of the recently approved new VMAT2 inhibitor for clinical use, valbenazine [24]. Several of the other lower affinity isomers have garnered interest for therapeutic applications not clearly related to their binding to the VMAT2 [21].

In this study, the relationship of in vitro binding affinity to in vivo occupation of the VMAT2 binding site has been examined in the rhesus monkey brain using PET. To minimize variability, all studies (N = 32 total) were performed in the same monkey, using the same microPET imaging equipment, with all image data analyses done using an identical protocol. The studies were done in the test-retest fashion utilizing a control [¹¹C]DTBZ scan, a one-hour interval for radioactive decay, and intravenous administration of the test article 10 min before a second [¹¹C]DTBZ scan. For two sessions using only administration of saline before the second scan, changes in BP_ND-Logan from the first to the second scan were −3% and −7%, indicating acceptable stability of the measure through the full 3-hour experimental period. The reproducibility of specific binding measures (binding potential, BP_ND-Logan) for the control [¹¹C]DTBZ scans was also very good over the entire project period [26].

As a single published study [12] has reported the in vitro binding affinities of all eight DTBZ stereoisomers (Fig. 1), those values were used for the correlations with in vivo occupancies; binding affinities reported by other investigators may be different but have maintained the rank order of stereoisomers. Of the eight isomers, only the four with 11bR stereochemistry exhibited in vitro binding affinities below one micromolar. Prior in vivo PET studies with carbon-11 labeled (−)-α-DTBZ (2S,3S,11bS, K_i > 2 μM) had shown no VMAT2 dependent localization in rodent [27] or human brain [15]. The other three 11bS-isomers also have in vitro binding affinities greater than 1 micromolar, and thus efforts to determine the ED₅₀ values for the four 11bS isomers were not performed based on the conclusion that the 11bR configuration was necessary for high affinity VMAT2 binding and significant in vivo VMAT2 occupancy.

The dose vs. occupancy curves and estimated in vivo ED₅₀ values for the four 11bR isomers are shown in Fig. 2. A minimum of three doses was used for the ED₅₀ estimations: for purposes of safety to the animal no attempt was made to reach 100% VMAT2 occupancy with any of the isomers as potential effects (both on- and off-target) of high doses of any of the compounds are unknown, and we have previously discussed the difficulties in choosing safe doses of test drugs for primate studies [28]. The lowest ED₅₀ of 0.055 mg/kg was seen for the highest affinity 2R,3R,11bR isomer (K_i = 4 nM). The ED₅₀ for the lowest affinity 2S,3S,11bR isomer (K_i = 593 nM) was estimated at approximately 3 mg/kg using an extrapolation of the data to the 50% occupancy point. The two remaining isomers, with in vitro binding affinities of 13.4 nM (2S,3R,11bR) and 71 nM (2R,3S,11bR) exhibited intermediate ED₅₀ values (0.12 and 1.0 mg/kg respectively). The correlation of in vitro binding affinities (K_i, from ref. 12) and in vivo occupancies (ED₅₀) for the four 11bR isomers was very good (R² = 0.95, Fig. 3A) but is highly dependent on the value for the lowest affinity (2S,3S,11bR) isomer, where the ED₅₀ value could only be estimated by extrapolation. However, the in vitro-in vivo correlation was even better (R² = 0.99, Fig. 3B) when considering only the three isomers with in vitro binding affinities less than 100 nM.

Fig. 2.

Injected dose vs. occupancy graphs for the four 11bR isomers of DTBZ.

Fig 3.

Correlations of in vitro binding affinities (K_i) with in vivo ED₅₀ values (mg/kg) for all four 11bR isomers of DTBZ (panel A) and for only the three high affinity isomers (K_i < 100 nM, panel B).

The very good correlation between in vitro binding affinity and in vivo occupancy for the four high-affinity DTBZ stereoisomers with the 11bR configuration supports the proposition that, all other things being assumed equal (e.g., physiochemical parameters, uptake and clearance, protein binding, metabolism), the extent of initial in vivo occupancy (within the first hour) of the VMAT2 target site is primarily dependent on the binding affinity. Is it then reasonable to expect similar results for other drugs that have multiple chiral centers and thus a large set of possible diastereomers? The answer to that question is likely no: differences in logP and pK_a [29], protein binding [30], transport [31,32], metabolism [33] and target selectivity [34] have been reported for stereoisomers of drug molecules. One limitation of this study of DTBZ isomers in the monkey brain is that none of these parameters were directly determined for all four of the 11bR-isomers, and differences in logP [35] and metabolism in humans [36] of some DTBZ isomers have been reported. However, the excellent correlation of in vitro binding affinities to the estimated in vivo site occupancies achieved in these monkey studies suggests that any differences that do exist did not significantly affect the early in vivo distribution (drug uptake into brain) and target engagement for the DTBZ isomers.

Summary.

Using in vitro binding affinities for selecting the appropriate candidate molecule for further development is an ongoing challenge in medicinal chemistry. The in vitro testing of drug-target interactions is an important first step in the development of therapeutic drugs or diagnostic radiopharmaceuticals, but attempts to use in vitro affinities to predict in vivo occupancy of the target site for a set of candidate compounds with disparate structures can be complicated by differences in physiochemical properties and physiological disposition. In this study we have found that a good correlation between in vitro binding affinities and in vivo target occupancies can be demonstrated using a set of stereoisomers of a single drug molecule, in this case the VMAT2 inhibitor dihydrotetrabenazine. The results support that initial occupancy of the binding site was highly dependent on the binding affinity: the questions of whether higher binding affinities result in persistent occupation of the binding site (a longer residence time) or greater pharmacological effect were not answered. Those questions are still to be addressed in future experiments.