Although we have reported on several mGlu4 in vitro and in vivo tool compounds, there still remains room for significant improvement in an acceptable probe for the mGlu4 receptor. To this end, we started from a functional high-throughput screening (HTS) initiated at Vanderbilt University, and discovered an interesting new lead compound, N-(4-(1,1,3,3-tetraoxidobenzo[d][1,3,2]dithiazol-2-yl)phenyl)furan-2-carboxamide, 8 (, Pubchem Compound ID (CID): 4066845; ChemDiv ID: K906-0087). Although compound 8 did not possess favorable calculated properties for a CNS probe compound (MW = 404.4, TPSA = 109.9), it represented a submicromolar starting point for our mGlu4 hit-to-lead program.
The 1,1,3,3-tetraoxidobenzo[d][1,3,2]dithiazol-2-yl)phenyl)carboxamide compounds were synthesized as outlined in . Commercially available mono-Boc protected 1,4-dianiline, 5, was reacted with 1,2-dibenzenedisulfonyl dichloride (Et3N, CH2Cl2) yielding 6. Next, the Boc group was removed under acidic conditions (4.0 M HCl/dioxane) and then the amides, 8a–l, were formed utilizing either the carboxylic acids (EDCI, HOBt, dioxane/DMF) or acid chlorides (DIEA, CH2Cl2).
Synthesis of 1,1,3,3-tetraoxidobenzo[d][1,3,2]dithiazol-2-yl)phenyl)carboxamides, 8a–l.
The structure-activity relationship (SAR) surrounding this set of compounds mirrored other series that we have reported previously ().18, 19, 24
Namely, the 2-pyridyl amide, 8d
, was the most potent heterocyclic substituent (hEC50
= 136 nM). Other aryl, cycloalkyl and 5-membered heterocycles also showed submicromolar potency at both the human and rat mGlu4
receptors. However, despite these modifications, the molecular weight (MW) and polar surface area (TPSA) remained suboptimal for the design of a CNS penetrant probe compound. To confirm this prediction, we profiled 8d
in both in vitro
and in vivo
rat pharmacokinetic (PK) experiments and assessed the extent of CNS penetration for this compound (). While the compound displayed selectivity against other mGlu subtypes (mGlu1–3,5,6–8
), the extent of brain penetration was poor (B/P = 0.085) following intraperitoneal (IP) administration to rats‥
Human mGlu4 potency and %GluMax response (as normalized to standard 1) for selected western amide analogs.
Characteristics of previously reported mGlu4 tool compounds from Vanderbilt University.
Next, we turned our attention to compounds that would have more favorable calculated properties for a CNS penetrant molecule (MW < 400, TPSA < 90) (). To this end we examined the effect of removing the carbonyl of the amide was evaluated. Subsequent replacement of the carbonyl with a methylene group (9b) or a cyclopropyl group (9c) effectively reduced the TPSA (113 to 96) of the scaffold, however with a cost of 10- to 40-fold reduction in mGlu4 potency. Likewise, the addition of a substituent ortho to the amide on the internal phenyl group (9a), or reversing the amide (9d), proved to alter the TPSA but, again, with a significant reduction in the potency for both compounds.
Human mGlu4 potency and efficacy (as normalized to standard 1) for bis-sulfonamide scaffold modifications.
One area of diversification where we have had success with other series is the 4-position of the internal phenyl group. Therefore, we evaluated the activity of a number of substitutions similar to the 1,1,3,3-tetraoxidobenzo[d][1,3,2]dithiazol-2-yl group, in an attempt to reduce TPSA and, ultimately, improve brain penetration. The inclusion of an imide group in lieu of the 1,1,3,3-tetraoxidobenzo[d][1,3,2]dithiazol-2-yl moiety represented a net lowering of the MW by 60 Da and the TPSA by approximately 30 Å2. The synthesis of the imide compounds is shown in .
Scheme 2 Synthesis of N-4-(2,5-dioxopyrrolidin-1-yl)-phenylpicolinamides.’a
Evaluation of the compound containing a sultam group revealed a complete loss of activity (16) (). This core phenyl group does not contain a halogen as do the other compounds; however, the compound containing an internal halogen was not attainable by synthesis. Next, a series of imide compounds were investigated. The initial five-membered imide compound was active (15a, hEC50 = 6200 nM), however the compound lost ~20-fold activity compared to 8d. Potency could be recovered by increasing the bulk of the right-hand side by utilizing phthalimide (and substituted phthalimides) (15b, hEC50 = 59.4 nM; 15c, hEC50 = 42 nM). However, an initial metabolic stability assessment revealed these compounds suffered from extensive oxidative metabolism (results not shown). Having demonstrated the phenyl moiety of the phthalimide as the predominant metabolic soft spot, we turned to saturated versions of the phthalimide (15d – 15l). The initial compound was the direct saturated comparator to 15b (15d, hEC50 = 370 nM). Although there was a ~10-fold loss of potency, 15d still showed <500 nM potency against hmGlu4. The trans-isomer was also evaluated with similar potency to the cis-15d (results not shown). Increasing the bulk of the saturated six-membered ring resulted in interesting compounds. The [2.2.1]-bridged oxo compound (15e, hEC50 = 1300 nM) exhibited a 4-fold loss in potency; however the unsaturated [2.2.1]-bridged carbon analog (15f, hEC50 = 291 nM) and unsaturated [2.2.2]-analog (15g, hEC50 = 287 nM) were more potent than 15d. Reducing the ring size from 6-membered to a substituted 3-membered ring resulted in an equipotent compound (15h, hEC50 = 435 nM). Substitution of the 5-membered imide led to a loss of activity (15i, hEC50 = 1400 nM), but remained comparable to the initial 5-membered imide (15a). However, when the size of the substituent was increased from the gem-dimethyl to cyclohexyl (15j, hEC50 = 158 nM) there was a 10-fold increase in activity. Other bulky substituents were evaluated (15k, hEC50 = 674 nM; 15l, hEC50 = 771 nM); however, these compounds lost activity when compared to 15f.
Human mGlu4 potency and efficacy (as normalized to standard 1) for selected bis-sulfonamide replacements.
With our best compounds evaluated for human potency, we evaluated the in vitro potency of the compounds on the rat mGlu4 receptor. Additionally, we compared compounds in terms of their relative efficacy which we evaluated by assessing the ability of a 30 µM concentration of each compound to shift a glutamate concentration-response curve to the left (represented as fold shift, FS) (). For examination of compound activity at the rat receptor, we employ an assay in which we measure the ability of the Gi/o
receptor to couple to G Protein Inwardly Rectifying Potassium (GIRK) channels.25
In contrast to the calcium mobilization assay, we generally do not observe large increases above the response elicited by saturating concentrations of glutamate alone when GIRK activation is used to assess activity; rather, compounds shift the glutamate concentration-response curve to the left only. Compounds from the 1,1,3,3-tetraoxidobenzo[d][1,3,2]dithiazol-2-yl)phenyl)carboxamide series (8d–8k
) all possessed GluMax values comparable to that of 1
and their efficacies in shifting the glutamate curve were modest, between 3.3 and 7.3. However, moving from the 1,1,3,3-tetraoxidobenzo[d][1,3,2]dithiazol-2-yl)phenyl)carboxamide series to the imide series produced significant divergence in compound efficacies. In particular, two phthalimide compounds (15b
) exhibited large differences in efficacy (leftward fold shift values of 23.6 and 83.0, respectively). Compound 15c
produced the largest leftward shift of the glutamate response curve reported to date. It is also noted that moving away from the phthalimide structures to the saturated imides led to a significant reduction in efficacy and the largest fold shift value for the saturated compounds was 16.7 (15l
), with most of the compounds tested exhibiting fold shifts of <10 (See Supplemental Figure 1
for graphs of human and rat potency and fold shift experiments). The exceptions were 15f
(12.1) and 15l
Rat mGlu4 receptor potency and efficacy (fold shift) data for selected compounds.
Having identified several key compounds based on human and rat potency as well as efficacy at the rat receptor, we evaluated these compounds in a battery of pharmacokinetic assays including an assessment of intrinsic clearance (CLint
) in hepatic microsomes. In addition to intrinsic clearance allowing for the prediction of pertinent rat and human PK parameters (CL and t1/2
), intrinsic clearance assessments allow for a rank ordering of compounds with respect to oxidative lability and a subsequent predicted stability in our in vivo
PK and efficacy models (). Although two compounds based on the phthalimide scaffold (15b
) showed excellent in vitro
potency on mGlu4
, these compounds have been previously reported by Merck and thus were not further evaluated.26
The three compounds based on the saturated imide scaffold (15f, 15j–k
) displayed elevated clearance in human liver microsomes, each exceeding hepatic blood flow in humans (QH
, 21 mL/min/kg). Conversely, an assessment in RLM indicated the compounds to possess moderate-to-high predicted hepatic clearance (CLHEP
), with compound 15k
predicting the best stability in rat (CLHEP
= 30 mL/min/kg). Assessment of the compound’s affinity for plasma proteins was also estimated (percent free, unbound; %fu) in vitro in cryopreserved plasma (). While compounds 15j
were highly protein bound in both human and rat plasma (%fu <1), 15f
displayed lower protein binding under the same conditions (%fu >3). In addition, 15f
was also investigated for nonspecific binding (NSB) in freshly prepared rat brain homogenate employing rapid equilibrium dialysis and found to display a decrease in NSB (%fu >7%) relative to binding to plasma proteins. As a first-tier screen for potential drug-drug interaction liability, the compounds were evaluated for their inhibition of the cytochrome P450 (CYP450 or CYP) enzymes utilizing a cocktail approach in human liver microsomes designed to inform on the inhibition of the five major drug-metabolizing and/or polymorphic CYPs (CYP2C9, 2D6, 3A4, 2C19, and 1A2). All compounds tested showed no significant activity against CYP2D6 and CYP3A4 (>30 µM); and only 15k
showed moderate inhibition of 2C9, 2C19 and 1A2 (8.7 µM, 17.6 µM, and 8.9 µM, respectively).
Predicted hepatic clearance, plasma protein binding, CYP450 inhibition, and mGlu selectivity values for selected compounds
Lastly, these compounds were evaluated for their selectivity against the other mGlu receptors. For initial studies, a 10 µM concentration of compound was applied prior to a complete agonist concentration-response curve (CRC) appropriate for each receptor. This technique allows for the detection of potentiator (leftward shift) or antagonist (rightward shift and/or decrease in the maximal response) activity within a single experiment. Using this assay, all compounds were selective versus the Group II mGlus (2 and 3) and only 15f showed weak PAM activity against a group I receptor (mGlu5, 2.1 FS). However, these compounds did show varying selectivity against the other Group III receptors. Whereas 15j only showed weak activity against mGlu6 (PAM, 2.1 leftward FS of the agonist CRC), 15f was weakly active against both mGlu6 (PAM, 3.1 FS) and mGlu7 (PAM, 2.9 FS), and 15k showed activity against all 3 receptors (mGlu6, PAM, 5.8 FS; mGlu7, mixed agonist/PAM (Ago-PAM) activity, 12.9 FS; mGlu8, PAM, 2.1 FS); the allosteric agonist activity against mGlu7 was equivalent to that observed for mGlu4 (12.9 FS vs. 12.1 FS).
The predicted in vitro ADME properties of these compounds were consistent with advancement for in vivo studies. Thus, the hydrochlorides of these compounds (15f, 15j, 15k) were synthesized and dosed intraveneously (1 mg/kg) and orally (10 mg/kg) to determine pertinent PK parameters such as clearance (CL), volume-of-distribution at steady state (Vss), half-life (t1/2) as well as bioavailability (%F) (). Although all three compounds displayed modest stability (CLHEP) when assessed in rat liver microsomes; 15k proved to be a lower clearance compound in vivo (5.4 mL/min/kg). Both 15f and 15j demonstrated moderate clearance values (37 mL/min/kg and 48 mL/min/kg, respectively), values that correlated well with their in vitro CLHEP values. Due to the exceptional stability of 15k following parenteral administration to the rat, we evaluated this compound in both beagle dog and rhesus monkey employing an IV cassette dosing paradigm. Compound 15k showed excellent stability in both dogs (CL, 7 mL/min/kg) and monkeys (3 mL/min/kg) with each species diplaying a t1/2 of ≥ 3 hrs. Although 15k showed superior stability comparatively, it unfortunately displayed an oral bioavailability less than 10%; Conversely, compound 15f demonstrated an excellent oral bioavailability of 97% whereas 15j displayed a modest bioavailability of 25%.
Rat in vivo pharmacokinetic values for selected compounds
The combination of enhanced bioavailability, oral PK, in addition to an exceptionally low protein binding value, indicated that compound 15f possessed the characteristics to be a potential tool compound which could be evaluated for brain exposure after oral dosing (). The hydrochloride salt of 15f was dosed orally in a microsuspension (10% Tween-80 in 0.5% methylcellulose) and plasma (systemic and hepatic portal) and brain samples were collected for bioanalysis of parent compound up to 6 hours following administration of the compound. The compound showed similar levels in the systemic and HPV plasma, indicating neglible hepatic first pass metabolism (AUCSYS/AUCHPV = 1.01). Compound 15f also displayed exceptional CNS penetration, with compound exposures reaching 800 nM*h (AUC0–6) and a corresponding brain:plasma ratio that reached unity (0.93).
PO pharmacokinetic parameters of 15f, (10 mg/kg, 10% Tween 80 in 0.5% methylcellulose).
Considering the promising PK results for 15f
, we evaluated the compound for selectivity against the other mGlu subtypes. In these studies, a 10 point concentration-response of 15f
was applied prior to an EC20
concentration of agonist appropriate for each receptor. As with our original fold shift-type studies, 15f
was shown to be inactive against mGlu 1, 2, 3, and 8 and was active against mGlu5
= 1.7±0.8 µM, 64.3±0.8% GluMax), mGlu6
>10 µM, 45.8±3.0% GluMax) and mGlu7
(PAM, 2.9±0.7 µM, 64.0±1.4% GluMax) (See Supplemental Table 1
). In addition, 15f
was tested in Ricerca’s Lead Profiling Screen (binding assay panel of 68 GPCRs, ion channels and transporters screened at 10 µM), and was found to not significantly with all the 68 assays conducted (no inhibition of radioligand binding > 50% at 10 µM).
Based on the previous PK studies, 15f
was chosen as a potential tool compound for an in vivo
proof-of-concept study and was examined in a rat model of neuroleptic-induced catalepsy. Here we evaluated the ability of 15f
to reverse haloperidol-induced catalepsy, a preclinical rodent model of motor impairments associated with Parkinson’s disease. The D2
receptor antagonist haloperidol induced robust catalepsy (). As shown in , 15f
(0.1 – 56.6 mg/kg, p.o.) produced a reversal of haloperidol-induced catalepsy (F(8,89)
= 5.8697, p<0.0001), significant at doses of 0.1 – 56.6 mg/kg, as compared with the vehicle control. The effects of 15f
were equally efficacious to those observed using an adenosine A2A
antagonist from Neurocrine Bioscience (compound 23
as a positive control, administered at a dose of 56.6 mg/kg, po.
Figure 10 Reversal of haloperidol-induced catalepsy in rats by compounds 15f after oral dosing. Catalepsy was measured in haloperidol treated (0.75 mg/kg) rats after oral administration of compound (0.1, 0.3, 1, 3, 10, 30 and 56.6 mg/kg) after 30 mins. An adenosine (more ...)