Prior Art Analysis

The inhalational anesthetics are recognized as being exceedingly promiscuous and are associated with undesirable off-pathway effects, and having the lowest therapeutic ratio. Injectable anesthetics such as propofol and etomidate (Figure 1), are more specific but are still associated with significant side effects, as the recent death of pop-star Michael Jackson so tragically emphasizes. All of the currently used injectable anesthetics were discovered empirically, but have been useful in implicating a specific molecular target, the GABA_A receptor. This screening effort builds on this knowledge by continuing the search for the inherently more specific injectable agents and exploiting the GABA_AR mimicry of the apoferritin system. It is to be emphasized that this is the first rational, unbiased approach to discovering novel classes of general anesthetics, as current efforts in the field almost exclusively involve structural modifications of known anesthetics to modulate their pharmacokinetic profile and/or physiochemical properties.

Figure 1

Example general anesthetics propofol and etomidate.

2.1. Assays

qHTS assay using apoferritin and 1-AMA

The primary qHTS assay was performed in PBS, pH 7.2, supplemented with 0.1% PEG 400. Three μL of reagents (free 1-AMA; 50% saturated solution of 1-aminoanthracene in the assay buffer) in columns 3, 4 as negative control and 1-AMA/apoferritin mixture (10 μM apoferritin, final concentration; 50% 1-AMA, final concentration) in columns 1, 2, 5–48 was dispensed into 1536-well Greiner black assay plates. Compounds (23 nL) was transferred via Kalypsys pintool equipped with 1,536-pin array (10 nL slotted pins, V&P Scientific, San Diego, CA). Titration of the positive control, propofol, was delivered via pin transfer from separate control plates to the 2^nd column of each assay plate. The starting concentration of the control, dissolved in DMSO, was 160 mM, followed by two-fold dilution points in duplicate, for a total of 16 concentrations. The plates were incubated at room temperature for 10 min before being read on an EnVision HTS multilabel reader (Perkin-Elmer, Waltham, MA) (end point read, 3min/plate) using a Fura2 BFP excitation filter (380 nm, bandwidth 10 nm) and fluorescein emission filter (535 nm, bandwidth 25 nm) set (see Table 1 for protocol steps) coupled with a LANCE/DELFIA dichroic mirror with a cutoff wavelength at 400 nm. Two dispensers and two EnVisions were applied to facilitate the screen. Throughout the screen, reagent bottles and all liquid lines were made light-tight to minimize reagent degradation. All screening operations were performed on a fully integrated robotic system (Kalypsys, San Diego, CA) containing one RX-130 and two RX-90 anthropomorphic robotic arms (Staubli, Duncan, SC).

Table 1

Apoferritin qHTS assay protocol.

Library plates were screened starting from 6.9 to 153 μM, with the top concentration achieved through a “double-dipping” step of the highest concentration. This was done to achieve a higher compound concentration, and a total concentration series of 5 was performed against 1-AMA/apoferritin. Vehicle-only plates, with DMSO being pin-transferred to the entire 5–48 compound area, was inserted uniformly at the beginning and the end of each library in order to monitor for and record any shifts in the background. Activity was computed as the normalized fluorescence response relative to free 1-AMA and 1-AMA/apoferritin complex values. Concentration-effect relationships were derived by using publicly-available curve fitting algorithms developed in-house. A four parameter Hill equation was used to fit the concentration-response data by minimizing the residual error between the modeled and observed responses. (BioAssay AID 2323, AID 485281, AID 489008).

2.2. Probe Chemical Characterization

Probe Characterization of ML306

*Purity >95% as determined by LC/MS and ¹H NMR analyses.

2-(6-Oxo-3-p-tolylpyridazin-1(6H)-yl)-N-phenethylacetamide: LC-MS Retention Time: t₁ (Method 1) = 5.440 min and t₂ (Method 2) = 3.466 min; ¹H NMR (400 MHz, DMSO-d₆) δ; 2.36 (s, 3 H), 2.73 (t, J = 7.4 Hz, 2 H), 3.27 – 3.35 (m, 2 H), 4.72 (s, 2 H), 7.05 (d, J = 9.8 Hz, 1 H), 7.17 – 7.25 (m, 3 H), 7.29 (m, 4 H), 7.77 (d, J = 8.2 Hz, 2 H), 8.05 (d, J = 9.8 Hz, 1 H) and 8.24 (m, 1 H); ¹³C NMR (400 MHz, DMSO-d₆) δ; 20.8, 35.0, 54.3, 125.7, 126.1, 128.3, 128.6, 129.5, 129.6, 131.0, 131.5, 139.0, 139.3, 143.5, 158.9, 166.2; HRMS (ESI) m/z (M+H)⁺ calcd. for C₂₁H₂₂N₃O₃, 348.1707; found 348.1708.

Figure 2List and structures of probe and related analogs that have been submitted to the MLSMR

2.3. Probe Preparation

Preparation of 2-(3-oxo-6-p-tolylpyridazin-1(3H)-yl)-N-phenethylacetamide.TFA (ML306) as depicted in Scheme 1

Scheme 1Synthetic route to probe ML306

Scheme 2General synthetic routes to intermediates and analogs

A mixture of 2-(3-oxo-6-(p-tolyl)pyridazin-1(3H)-yl)acetic acid (1.2 g, 4.9 mmol, 1 eq.), EDC (1.88 g, 9.8 mmol, 1.5 eq.) and HOBT (0.75 g, 4.9 mmol, 1 eq.) in DMF (10 mL) was stirred at room temperature for 5 min. Diisopropylethylamine (0.86 mL, 4.9 mmol, 1 eq.) and 2-phenylethanamine (0.75 mL, 5.9 mmol, 1.2 eq.) were then added and stirred at room temperature for 1 hr. The reaction mixture was filtered and the filtrate containing the product was purified via reverse phase chromatography to give 2-(3-oxo-6-(p-tolyl)pyridazin-1(3H)-yl)-N-phenethyl-acetamide TFA (ML306) as a white solid.

3.1. Summary of Screening Results

Initially, the assay was validated in an offline pilot screen using the library of pharmacologically active compound (LOPAC¹²⁸⁰) collection[21] where fluorescence was monitored on a ViewLux CCD camera-based microplate imager (PerkinElmer, Waltham, MA). Samples were excited at 340 nm and emission was collected at 530 nm. The excitation filter that comes with the ViewLux imager (340 nm) does not match the maximum excitation wavelength of 1-AMA (380 nm) very well. The lower excitation wavelength provided by the ViewLux imager resulted in an enhanced compound interference.[22, 23] In order to further alleviate compound interference, the assay was further optimized by using an Envision multilabel reader, which was equipped with an excitation filter centered at 380 nm and an emission filter centered at 535 nm, better matching the spectral property of our assay reporter, 1-AMA. To further validate this approach, the top concentration of the LOPAC¹²⁸⁰ collection was tested using both instruments. It was found that red shifting the excitation wavelength by as little as 40 nm was accompanied by a significant reduction in the number of interfering compounds, identified as those whose response was above the positive controls (see Figure 4A for Envision data and B for Viewlux data).

Figure 4

Optimization of the apoferritin/1-AMA fluorescence assay. A) and B) are scatter plots of the top LOPAC concentration plate run on Envision (ex: 380 nm/em: 535 nm) and on ViewLux (ex: 340 nm/em: 530 nm), respectively; C) Effect of 0.1% PEG 400 in an AmpC (more...)

One downside of using the EnVision reader for end point fluorescence assays is that it takes longer per plate (3 min on Envision vs. <1 min on ViewLux). For the large collection qHTS screen, this was partially compensated through the utilization of two identical EnVision readers, whose performance will be addressed in section (3.2).

Another aspect that we investigated during assay optimization was the need to minimize the effect of compound aggregation and to reduce assay false positives caused by such mechanism. Tween-20, a common detergent used for this purpose, was tested, but it showed interference with our assay readout. Therefore, PEG400 was further evaluated over a few concentrations to examine its effect on assay signal window and assay Z’ [24]. It was found that at 0.1% concentration, PEG400 had minimal effect on the assay S:B or Z’, but a 1% concentration led to a decrease of S:B by more than 30% (data not shown). Additionally, the effect of 0.1% PEG400 was validated in an AmpC β-lactamase absorbance assay [25], using a known aggregator, TIPP [26]. It can be seen from Figure 4C that the addition of 0.1% PEG 400 exerted a similar effect as that of 0.01% Triton in the assay in that the apparent inhibitory effect observed in the very left figure in panel C was unmasked and TIPP was revealed as an inactive.

The above assay was used in a fully-integrated robotic screen of a library containing 351, 367 compounds tested at five concentrations ranging from 6.9 to 153 μM in a qHTS mode.[20] The flexibility of choosing a biologically relevant compound concentration range for screening purpose is offered by the qHTS paradigm,[27] and in this particular case, this range was selected based on the sensitivity of the assay. In other words, the relatively high concentration of apoferritin applied in the assay (10 μM) suggested that the lowest detection limit of assay would be in that range, rendering the screen of lower concentration plates (low single digit micromolar to submicromolar) unnecessary.

The screen was completed within 4 days. An average Z’ of 0.76 was achieved across the 1,509 plates screened, with those read on Envision reader 1 giving slightly higher Z’ than Envision reader 2 (Figure 5A). Low signal window plates were rejected, and in total, 1,228 plates passed quality control and were included in the final dataset. Data were further corrected against DMSO-containing plates and normalized against positive and negative controls for the establishment of dose-dependent concentration curves for each compound. EC₅₀ of each compound was directly determined from the primary screen. Figure 5B shows a 3D activity plot of the control titration (in green) along with potential inhibitors of apoferritin. The control propofol yielded a highly reproducible EC₅₀ with an average value of 23 μM. Samples that produced full or partial curves and had maximum response greater than six sigma activity (~60%) were declared as top actives. Additionally, top actives from normalized and uncorrected curve fits were included. All other compounds that resulted in signal decrease were grouped as weak inhibitors or inconclusives. Out of all the compounds screened, 0.64% were classified as top actives (Figure 5B, dark blue) and 6.7% were weak inhibitors or inconclusive (Figure 5B, light blue). Total actives shown in the 3D plot represent 7.3% of the entire chemical library screened.

Figure 5

Summary of apoferritin/1-AMA screening results. A) scatter plot of Z’ values of the apoferritin/1-AMA assay across all screened plates (average Z’: 0.76); B) 3D plot of qHTS data with control titrations (shown in green), high confidence (more...)

A total of 2,563 top actives were clustered using Leadscope structural fingerprints. A tanimoto coefficient of 0.7 was used for clustering, and 261 clusters were identified. Additionally, 176 structurally diverse singletons were found. Compounds were chosen for follow-up based on structural diversity, structure activity relationships and property distributions such as MW, logP, and ligand efficiency. A collection of top 700 cherrypicked compounds were retested in the primary qHTS assay for activity confirmation. Ligand efficiency,[28] a measure of binding affinity per heavy atom in a compound, was computed for all 700 compounds, and top 80 compounds were selected for confirmation in ITC assay for orthogonal assay (see Figure 6 for assay work flow chart).

Figure 6

Critical path work flow for the identification of small molecule inhibitors of apoferritin/1-AMA interaction that possess general anesthetic properties.

Of the 80 compounds, the 6-phenylpyridazin-3(2H)-one chemotype, as exemplified by CID-3244374 (Figure 7), emerged as highest in ligand efficiency and had nascent SAR directly from the primary screen. The 6-phenylpyridazin-3(2H)-one chemotype and its derivatives in the primary screen exhibited high ligand efficiency and good physicochemical properties. Given the compounds’ profiles and the initial qHTS SAR, the series was chosen for resynthesis, reconfirmation of purified actives, and confirmation in tadpole in vivo assays.

Figure 7

Initial hit identified from the qHTS and validated using apoferritin and tadpole assays.

3.2. Dose Response Curves for Probe

Figure 8Isothermal Titration Calorimetry (ITC) of the HSAF interaction with ML306 (K_D = 8.2 μM)

3.3. Scaffold/Moiety Chemical Liabilities

One of the strengths of ML306 and its probe series is the general drug-like nature of the 6-phenylpyridazin-3(2H)-one chemotype. Like many anesthetics, the probe and analogs are small in molecular weight and overall size and possess good physicochemical properties such as low number of hydrogen bond acceptors and donors and low log P. ML306 does not contain any reactive functional groups and was found to be stable in both mouse plasma and PBS buffer conditions at various pH conditions. The probe has good in vitro ADME properties as well as good in vivo PK profile; microsomal stability with both human and mouse liver microsomes was assessed and ML306 found to have a short half-life that is ideal for anesthetics. Lastly, the probe compound was found to have good BBB penetration. Overall, ML306 represents an excellent starting point for further pre-clinical development of general anesthetics.

3.4. SAR Tables

Table 2General anesthetic profile in tadpoles and inhibition of 1-AMA/Apofferitin interaction

Representative Analogs (modification at 6-position of the pyridazin-3-one). All analogs were synthesized at NCGC

View in own window

Entry	CID	SID	NCGC IDs	R	IC50 (μM)a	% immobilityb	toxicityc
1	45548816	134419067	NCGC00262353	H	inactive	0	5
2	49704509	134419066	NCGC00262352	CH3	24.2	5	15
3	56643049	134419068	NCGC00262354	isopropyl	inactive	0	5
4	56643049	134419069	NCGC00262355	cyclopropyl	inactive	0	5
5	56643032	134419051	NCGC00262337	cyclohexyl	12.1	40	5
6	5053472	134419085	NCGC00262371	phenyl	5.4	20	5
7	5070704	13449032	NCGC00140268	4-CH3-phenyl	3.04	90	0
8	56643035	134419049	NCGC00262335	3-CH3-phenyl	3.4	20	5
9	7671962	134419050	NCGC00262336	2-CH3-phenyl	19.2	0	10
10	4993789	134419033	NCGC00140457	4-OCH3-phenyl	4.3	75	5
11	7665103	134419055	NCGC00262341	3-OCH3-phenyl	3.0	30	10
12	7665103	134419054	NCGC00262340	2-OCH3-phenyl	17.1	5	15
13	7664506	134419046	NCGC00262332	4-F-phenyl	4.8	5	0
14	56643067	134419087	NCGC00262373	4-CF3-phenyl	9.6	0	5
15	56643069	134419089	NCGC00262375	4-CN-phenyl	5.4	0	5
16	56643059	134419061	NCGC00262347	4-OCF3-phenyl	13.6	80	10
17	56643068	134419057	NCGC00262343	4-NMe2-phenyl	3.8	0	10
18	56643038	134419060	NCGC00262346	4-(2-PrO)-phenyl	4.8	95	5
19	56643051	134419062	NCGC00262348	4-isobutyloxyphenyl	13.6	100	0
20	56643060	134419088	NCGC00262374	4-n-Bu-phenyl	4.8	5	5
21	56643052	134419063	NCGC00262349	4-t-Bu-phenyl	6.8	90	10
22	7663799	134419047	NCGC00262333	3,4-methylenedioxyphenyl	4.3	0	0
23	56643026	134419053	NCGC00262339	4-CH3SO2-phenyl	21.5	5	5
24	7659957	134419056	NCGC00262342	3,4-(OCH3)2-phenyl	10.8	10	15
25	56643055	134419048	NCGC00262334	4-pyridyl	30.4	5	10
26	56643062	134419064	NCGC00262350	3-pyridyl	24.2	0	15
27	56643057	134419065	NCGC00262351	2-CH3-3-pyridyl	5.4	0	0

a

qhts IC₅₀;

b

Tadpole experiment immobility at 10 μM after 60 min;

c

% immobility after 24 h (10 μM)

Table 3General anesthetic profile in tadpoles and inhibition of 1-AMA/Apofferitin interaction

Representative Analogs (modification of the core). All analogs were synthesized at NCGC

View in own window

Entry	CID	SID	NCGC IDs	R	IC50 (μM)a	% immobilityb	toxicityc
28	56643042	134419074	NCGC00262360		4.3	0	0
29	56643028	134419071	NCGC00262357		3.0	0	0
30	56643041	134419073	NCGC00262359		12.1	0	0
31	56643047	134419076	NCGC00262362		24.1	0	5
32	56643030	134419075	NCGC00262361		inactive	0	0
33	56643065	134419042	NCGC00249440		4.8	95	45
34	135363127	56844190	NCGC00250074		14.5	10	0
35	56643039	134419045	NCGC00250075		6.1	0	0
36	135363124	56844193	NCGC00250071		14.5	0	0
37	135363126	56844195	NCGC00250073		inactive	0	0
38	135363128	56844194	NCGC00250077		inactive	10	0
39	135363123	56844191	NCGC00250070		inactive	0	0
40	135363121	56844197	NCGC00249762		9. 2	10	0
41	56643058	134419043	NCGC00249763		21.5	0	0
42	135363125	56844196	NCGC00250072		18.2	0	0

a

qhts IC₅₀;

b

Tadpole experiment immobility at 10 μM over 60 min;

c

% immobility at 24 h (10 μM)

Table 4General anesthetic profile in tadpoles and APO inhibition

Representative Analogs (modification of the side chain). All analogs were synthesized at NCGC

View in own window

Entry	CID	SID	NCGC IDs	R	IC50 (μM)a	% immobilityb	toxicityc
43	56643048	134419034	NCGC00189159		5.0	90	0
44	56643036	134419035	NCGC00249422		3.0	95	0
45	49852991	104224352	NCGC00189154		30.4	85	0
46	56643061	134419037	NCGC00249428		3.4	85	0
47	56643031	134419038	NCGC00249429		3.4	85	0
48	7662444	134419082	NCGC00262368		3.4	10	5
49	50925844	134419077	NCGC00262363		3.4	65	0
50	56643046	134419052	NCGC00262338		3. 0	5	20
51	5099141	134419084	NCGC00262370		17.1	0	5
52	49852775	104224356	NCGC00189158		15.2	0	0
53	56643034	134419039	NCGC00249431		10.2	100	100
54	4582591	104224364	NCGC00189166		31. 6	0	-
55	49853286	104224370	NCGC00189172		29.2	0	-
56	3241729	104223755	NCGC00050102		35.5	0	-
57	56643066	134419078	NCGC00262364		15.2	0	0
58	56643054	134419083	NCGC00262369		8.6	5	0
59	56643063	134419086	NCGC00262372		inactive	0	0
60	56643037	134419081	NCGC00262367		3.0	0	5
61	56643027	134419040	NCGC00249436		10.8	95	0
62	56643040	134419041	NCGC00249437		12.1	0	10

a

qhts IC₅₀;

b

Tadpole experiment immobility at 10 μM over 60 min;

c

% immobility at 24 h (10 μM)

Table 5General anesthetic profile in tadpoles and inhibition of 1-AMA/Apofferitin interaction

Representative Analogs (modification of the side chain). All analogs were synthesized at NCGC

View in own window

Entry	CID	SID	NCGC IDs	R	I C50 (mM)a	% immobilityb	toxicityc
63	135363119	56844198	NCGC00249760		inactive	0	0
64	3245279	4251618	NCGC00065315		18.2	0	0
65	135363122	51663603	NCGC00249765		inactive	0	10
66	135363120	56844192	NCGC00249761		18.2	0	0
67	56643045	134419044	NCGC00249766		21.5	0	0

a

qhts IC₅₀;

b

Tadpole experiment immobility at 10 μM over 60 min;

c

% immobility at 24 h (10 μM)

3.5. In Vivo Activity, mouse model

ML306 produced loss of righting reflex (LORR) in less than 20 seconds in mice, with durations that depended on dose. An 80 mg/kg injection iv produced an average time of 73 ± 45 seconds until mice were able to right themselves (n = 6). The response to this dose varied between 0 seconds (misplaced injection) and 256 seconds. Conversely, 100 mg/kg ivintravenous (IV) administration of ML306 produced LORR in all animals (n = 3), with LORR ranging from 140 to 362 seconds. 40 mg/kg IV administration was determined to be ineffective in all mice tested (n = 4). Control injections with the vehicle solution ranged from 1304 mg HP-β-CD/kg body mass to 2557 mg/kg and produced no observable effects (n = 4). The latter dose approximates published data that suggests decreased activity and breathing irregularities following an intravenous injection of 2250 mg/kg in rats.[29]

Upon recovery of righting reflex, the mice were subdued, exhibited slow exploratory behavior, and did not attempt to avoid the handler. Apparently normal behavior was observed within ten minutes following the injection. No apparent toxicity or residual effect of the injection was noted out to two weeks. Two weeks post injection, each mouse had gained a minimum of 0.4 grams of body mass, not different from the vehicle injected controls.

The molecular weight of the ML306 (461 g/mol) salt is ~2.6 times that of propofol (178 g/mol). For comparison, a 20 mg/kg injection of propofol suspended in intralipid induces LORR for 216 ± 25 seconds. 100 mg/kg ML306 is equal to 0.217 mmol/kg, whereas 20 mg/kg propofol is equal to 0.112 mmol/kg.

3.6. Profiling Assays

Pharmacokinetic Profiling of Probe

Table 7ADME profile for ML306

All experiments were conducted at Pharmaron Inc

View in own window

Compound	Solubilitya	Caco-2 Permeability (10−6 cm/s)	Efflux Ratio	Microsomal Stability T1/2 (min)b		% CYP Inhibition @ 3 μM		Plasma Stability (mouse) % remaining (2 hr)	PBS Buffer Stability % remaining (2 hr)
ML306	57.9 μM	8.2	1.6	11 (human)	5.2 (mouse)	24 (2D6)c	12 (3A4)d	98.9	100%

a

represents the kinetic solubility in PBS buffer (pH = 7.4).

b

represents the stability in the presence of NADPH. The probe compound showed no degradation without NADPH present over a 1 hr period.

c

dextromethorphan was used as the substrate.

d

midazolam was used as the substrate.

Table 8In vivo PK (mouse) at 40 mpk IV and 5 mpk IV

All experiments were conducted at Pharmaron Inc. with male CD1 mice (6–8 weeks of age). Data was collected in triplicate at 8 time points over a 24 hr period

View in own window

Compound	Routea	T1/2 (h)	Co (ng/mL)	AUCinf (h*ng/mL)	Cmax (brain) (ng/g)	Tmax (brain) (hr)	MRTb (hr)	Cl (mL/min/kg)	P/B Ratioc
ML306	IV (5 mpk)	0.28	7878	1508	ND	ND	0.27	56.5	ND
ML306	IV (40 mpk)	0.21	65162	11596	22033	0.08	0.17	58.1	2.4

a

Both formulated as a solution [HP-β-CD in saline (1 g/mL)].

b

Mean residence time (the time for elimination of 63.2% of the IV dose).

c

Plasma to brain ratio [AUC_last(plasma)/AUC_last(brain)].

Figure 9

(A) PK plasma profile for ML306. CD1 Mice IV (40 mpk (blue) and 5 mpk (red)) dosed as a solution with n = 3 for each time point. (b) PK brain profile for ML306. CD1 Mice IV (40 mpk) dosed as a solution with n = 3 for each time point.

See Section 4.2 “Mechanism of Action Studies”.

Structure Activity Relationships of ML306

A high-throughput screening campaign of a library containing 397,939 compounds in the 1-AMA/apoferritin assay and subsequent application of rigorous chemoinformatics analysis (see above for details) resulted in the selection of the 6-substituted pyridazin-3-one chemotype as our lead series (e.g. CID 3244374, Figure 7). Following confirmation of these pyridazin-3-one derivatives via re-synthesis in the qHTS assay, and synthesis of a small set of analogs (~10) we then profiled these compounds for anesthetic properties and toxicity in tadpole assay. In the tadpole assay, anesthetic profile was measured as % of immobility after 60 minutes at 10 μM (100% immobility is desired) and the toxicity was measured as % of immobility after 24 hr (0% immobility is desired). With a good correlation between 1-AMA/apoferritin inhibition and anesthetic profile for majority of the compounds in tadpoles, we planned more extensive systematic structural modifications to the 6-substituted pyridazin-3-ones in an effort to establish a full SAR profile. It should be noted that given the limitations of the qHTS assay sensitivity (~3 μM) and the fact that improving potency in this surrogate assay is not our ultimate goal, we chose to use the tadpole assay for SAR analysis. Though this clearly presents challenges for iterative medicinal chemistry optimization, we felt it was the most efficient way to discovery novel general anesthetics.

As shown in Table 2, our initial exploration at 6-position of the pyridazin-3-one showed that smaller groups or rings are not well tolerated. For example, entries 1–3 (CID 45548816, CID 49704509, and CID 56643049) showed no activity in the HTS assay and had little or no effect on immobilizing the tadpoles. However, these modifications caused some toxicity (immobilization at 24 h). Similarly, replacing with more hydrophilic groups such as pyridine analogs 25–27 (CID 56643055, CID 56643062, and CID 56643057) and N,N-dimethylphenyl analog 17 (CID 56643068) resulted in diminished apoferritin activity and complete loss of activity in tadpoles. Phenyl or para-substituted phenyl groups improved the potency that is exemplified by the % of immobility of tadpoles at 10 μM and the HTS IC₅₀ values of compounds 7 (CID 5040404, R = 4-Me-phenyl), 10 (CID 4993789: R = 4-OMe-phenyl), 18–19 (CID 56643038 and CID 56643051) and 21 (CID 56643052: R = 4-t-butyl-phenyl). Further, these compounds showed minimal or no toxicity against tadpoles at 10 μM. Electron-donating lipophilic groups at the para position of the phenyl ring is preferred over ortho and meta-substitution. As shown in Table 2 compounds 7 & 10 (CID 5070704 and CID 4993789) displayed better potency both in tadpole assay and surrogate assay as compared to entries 8–9 (CID 56643035: R = meta-Me; and CID 7671962: R = ortho-Me, respectively) & 11–12 (CID 7665103: R = meta-OMe; CID 7665103: R = ortho-OMe, respectively) with ortho being the least potent. Further exploration with di-substituted phenyl analogs such as compounds 22 & 23 (CID 7663799 and CID 56643026) resulted in no activity in the tadpole assays despite good activity in the enzymatic assay. The 6-substitution pattern of the pyridazin-3-one appears to be essential for the modulation of anesthetic activity as evidenced by complete loss of potency for 5 and 4 substituted compounds 36–37 (CID 135363124 and CID 135363126). Given that the p-tolyl group at 6-position of the pyridazin-3-one appeared to provide maximal potency and minimal toxicity, it was held constant while other regions of the molecule were explored for SAR.

Although attempts to replace the pyridazin-3-one core with smaller or similar heterocycles provided several compounds with moderate activity as measured in the surrogate assay, these were shown to be inactive in the orthogonal tadpole assay (entries in Table 3). This observation with these and several other analogs might be due to lack of necessary pharmacological properties essential to penetrate the target in the in vivo system. On the other hand, incorporation of an OMe-substituted phenyl ring, entry 33 (CID 56643065) retained the anesthetic profile but exhibited severe toxicity. Moreover, reduced activity for compound 40 (CID 135363121: 5-position = Me), suggests that further substitution in the pyridazin-3-one core is not tolerated.

Our next focus of SAR exploration was modification of the side chain at the 2-position as shown in Table 4 and Table 5 (only representative analogs are shown). We quickly learned that the amide linkage is critical for the anesthetic activity as evidenced by reduced potencies for compounds 63–67 in Table 5. Introduction of other liphophilic groups such as phenyl, furan, thiophene, cycloalkyl (5,6 or 7-membered rings) greatly facilitated the potency in the 1-AMA/apoferritin assay and in tadpoles with diminished toxicity. Compounds 43–47 and 61 in table 5 displayed over 85% immobility at 10 μM without any toxicity. In contrast, a decreased activity for compounds 50 (CID 56643046) and 51 (CID 5099141), suggests that substitution of the phenyl ring is not well tolerated. Replacing the phenyl group with smaller groups such as a cyclopropyl moiety [compound 60 (CID 56643037)] or smaller alkyl group [compound 62 (CID 56643040)] also negates tadpole activity despite moderate activity in the HTS assay. Surprisingly, polar groups and/or heterocycles led to complete loss of anesthetic profile as evidenced by the % immobility and IC₅₀ values for compounds 54–59. Finally, we turned our attention to optimization of the linker between the amide nitrogen and the phenyl ring. A two carbon linker is needed to manifest optimal anesthetic activity whereas a one carbon or ≥3 carbon linkers decreased the potency as shown for compounds 48 (CID 7662444) and 49 (CID 50925844), respectively. Therefore, our SAR exploration resulted in para-substituted phenyl group at 6-position with phenylethyl amide moiety side chain provided best anesthetic compounds with good potency and minimal toxicity. Figure 10 provides a summary of the SAR assessed during the medicinal chemistry optimization. Compounds 7 (CID 5070704), 19 (CID 56643051), 43 (CID 56643048) and 44 (CID 56643036) are highlighted as some of the best candidates for further studies. However, we focused on compound 7 (ML306) because it exhibited better aqueous solubility than the other analogs.

Figure 10

SAR Summary.

Having established compound 7 as a lead compound which exhibits favorable activity in the tadpole immobilization assay without any obvious adverse effects on the tadpoles viability (e.g. no acute toxicity) we were eager to progress this compound in vivo (mouse model, Table 6). However, before doing so we investigated the in vitro ADME and in vivo PK properties of the molecule as shown in Tables 7 and 8. The probe compound exhibits favorable aqueous solubility (PBS buffer) of ~58 μM which is well above the concentration required for tadpole immobilization. This compound also displays favorable Caco-2 permeability of 8.2 and a minimal efflux ratio of 1.6 where anything <2 is considered desirable. Given that Caco-2 cells express Pgp, and Pgp efflux plays a well-known role in BBB permeability, this data can be used to predict BBB penetration of discovery compounds. We then wanted to examine whether the probe compound inhibits specific cytochrome P450 enzymes 2D6 and 3A4 as these two isoforms account for the metabolism of approximately 80% of drugs. To assess the potential for CYP inhibition and drug-drug interactions (DDI), we looked at the effect of co-incubation of our compound at 3 μM with known CYP substrates (2D6: dextromethorphan and 3A4:midazolam). We found that ML306 exhibits modest CYP inhibition of 24% and 12% for 2D6 and 3A4 respectively. To further elucidate the potential for CYP liabilities, follow-up IC₅₀ values and testing of a more expansive representation of CYP isoforms will be required. ML306 was also profiled for its microsomal stability with both human and mouse liver microsomes and was found to have a short half-life of 11 and 5.2 min, respectively. In most drug discovery programs such a result would be considered a liability, ideal general anesthetics are ones that are cleared rapidly as drugs with a slow metabolism are associated with a more prolonged “hangover effect” (e.g. fatigue, weakness, pain, nausea and vomiting). However, it should be noted that there is also a market for longer acting or slower metabolized compounds (i.e. sedation in the ICU). Lastly, ML306 was profiled for its stability in both mouse plasma and PBS buffer and it was found to be stable in both media and at various pH’s (2, 7.4 and 9) as shown in Figure 3. With this in vitro ADME data in-hand we were eager to determine whether these results translated to appropriate in vivo exposure, namely T_1/2 and BBB penetration.

Table 6

ML306 In vivo loss of righting reflex (LORR) studies. Standard error shown for LORR time

Figure 3

Stability of ML306 in DPBS (pH 7.4) buffer at room temperature over 48 hr (Panel A). Stability of ML306 at pH 9 (Panel B) and pH 2 at room temperature over 24 hr (Panel C).

As fast-acting general anesthetics are typically administered via IV (and inhalation for gases) we aimed to determine the in vivo PK profile of the probe compound at two different doses (5 mpk and 40 mpk) to ensure that a linear PK profile was observed. Our initial study used 5 mpk because we did not yet know the dose which would provide a general anesthetic effect to the mouse. After subsequent in vivo PD studies (vide infra) we chose a dose closer (40 mpk) to what is required for LORR (Loss of Righting Reflex) in the mouse model (Table 6). These results are tabulated in Table 8 and closely mirror what we observed in the in vitro ADME studies. The T_1/2 in plasma was 17 min and 13 min for 5 and 40 mpk respectively, whereas the MRT in plamsa (mean residence time) was 16 min and 10 min, respectively. The T_max in the brain was determined to be ~5 minutes. (40 mpk) with 30% of the compound remaining in the brain tissue after 15 min. Interestingly, others have tested propofol in mouse models (32 mpk) doses IV and found the T_max to be ~3 minutes and the T_1/2 to be 9.6 minutes, which suggests that these compounds have similar residence in the brain tissue.[30] ML306 was found to have a linear PK profile in the brain that mimic that found in the plasma (Figure 9A & B) and had good BBB penetration with a P/B ratio of 2.4 and a C_max of 22033 ng/g @ 40 mpk dose. We were encouraged to initiate preliminary in vivo PD studies with this compound by its favorable PK profile (short half-life, adequate brain exposure) and the fact that this brain profile closely resembles propofol.

The PK profile of ML306 compares more favorably to that of propofol than it does to the general population of drugs but this should not be viewed as a deficiency, particularly given the early-stage status of this chemotype. The activity of many drugs and compounds do not translate well from mouse to human; however, general anesthetics are remarkable in that the concentrations necessary to produce immobility are conserved within a factor of 2–4 fold across all animal species. Even the very complex neurosteroid anesthetics produce immobility in tadpoles at concentrations comparable to that for both rodents and people. The concentrations of inhalational anesthetics required for mouse immobility are within 20% of those required in humans. This is also true for intravenous anesthetics, with the exception that bolus doses are larger due to pharmacokinetic considerations. Thus, the bolus dose of propofol in a human to produce unconsciousness is about 2 mg/kg, whereas that for a mouse is 20 mg/kg. The calculated concentrations at effect sites, however, are comparable. This is not to say that off-pathway effects are the same. In many cases, lack of lethality in the tadpole cannot be reproduced in the mouse. In light of this, our results to date indicate that ML306 carries a sufficient promise to merit further development directed towards eventual human testing.

4.1. Comparison to Existing Art and How the New Probe is an Improvement

ML306 and related analogs have reduced activity compared to propofol, in both the tadpole assay and the in vivo mouse model. However, the true goal of this project was to validate a novel approach to discover new general anesthetics. The field of general anesthetics has not seen a rationally designed approach or one that relying on HTS, but rather serendipitous discovery or structural modifications to known compounds. Thus, the probe compound presented herein represents validation of the first approach of this kind with demonstrated in vivo efficacy and a PK profile comparable to propofol.

4.2. Mechanism of Action Studies

Propofol is, in part, thought to produce its effect by interacting with inhibitory cys-loop ligand-gated ion channels, although there exists evidence that it also has effects on other ion channels, such as the voltage gated cation channels. Although our surrogate screen is based an ability of apoferritin to mimic GABAergic allosteric binding, apoferritin clearly cannot mimic the ion channel activity of the GABA_A receptor. This activity, however, is simply not amenable to high throughput screening as it requires an electrophysiology rig. One assay that more specifically evaluates GABA_AR allosteric binding activity is the ability of compounds to enhance the binding of radiolabeled flunitrazepam. Preliminary results from Eckenhoff Lab show GABAergic activity for several members of the probe series. It is conceivable that a portion of the sedative action of members of the probe series is due to enhancement of GABAergic activity, as the original surrogate was designed to reveal. Although our preliminary results showed that the probe ML306 was not active in the radiolabeled flunitrazepam assay, the fact that at the same time a number of its analogues did show activity comparable to propofol (data not shown), indicates that additional studies using the radiolabeled flunitrazepam, as well as alternative assay platforms to interrogate GABAergic activity, are warranted.

ITC against apoferritin was used only to confirm the qHTS results, since the latter may be contaminated by various optical effects, such as inner filter, FRET and so forth. ITC proved very difficult as the compounds were not available in sufficient mass in powder form to conduct several runs (which, given the expected micromolar K_D, required the application of very high compound concentrations, beyond those supported by their aqueous solubilities) in the absence of DMSO. Cosolvents such as DMSO produce large heats of dilution which can obscure the often smaller heats of bimolecular binding interactions. Thus, after several attempts, the ITC confirmation of qHTS, even of the subsets, was abandoned in favor of the phenotype assays. However, ML306 and the original hit CID-3244374 did prduce a binding isotherm in apoferritin ITC experiments, yielding K_D’s of 8.2 μM (Figure 8) and 4.2 μM (Figure 9), respectively.

Pharmacological Profiling of Probe

ML306 was submitted to the National Institute of Mental Health’s Psychoactive Drug Screening Program (PDSP) G protein-coupled receptor (GPCR) Panel, and the results were represented as a heatmap (Table 9). ML306 along with propofol were tested at 1 μM against a panel of 149 GPCRs. Compounds with <4-fold stimulation at 1 μM were considered uninteresting for further potency determination. Internal positive controls were used in the panel (data not shown). Surprisingly, both ML306 and propofol were very clean with no significant activity against any of the tested GPCRs. Profiling against ligand gated voltage cannels and other selected receptors are ongoing. We were encouraged by this profile as it seems to indicate that the in vivo effect is not simply a result of promiscuous GPCR activity, however it also means that additional profiling must be done to help identify the target(s). It should be noted that such a task is a lofty goal given that researchers do not yet fully understand how known anesthetics work despite being studied extensively for decades.

Table 9

PDSP GPCR Profiling of ML306 and propofol tested at 1 μM concentration. White squares represent inactivity while blue squares represent fold stimulation at 1 μM

4.3. Planned Future Studies

Future studies will be multi-pronged. First, it is likely that only a subset of the original apoferritin qHTS actives have anesthetic activity, and these may not be the very top hits in the primary screen (as primarily determined by the hits’ displacement IC₅₀ values in the apoferritin assay). We now know, for example, that the apoferritin “anesthetic” site is likely to be a fatty acid binding site.[31] Thus, the overlap between anesthetics and fatty acid chemotypes is finite and is not likely to be large. Therefore, the original list of top actives needs to be revisited in an unbiased phenotype screen, such as the stereotypical photoresponses in the zebra fish embryo. We have initiated a collaboration with the Peterson Lab at Massachusetts General Hospital and Harvard Medical School to validate this approach using his medium throughput zebrafish screening operation. A second direction will be to focus further on the medicinal chemistry of the ML306 series. It is likely that this modest sized compound can be further modified to produce greater activity. Our goal would be to exceed the potency of propofol in the tadpole assay by at least 10-fold. Moreover, we plan to test several of the other compounds with promising tadpole activity in the in vivo model to see if improved activity is observed (e.g. compounds 18, 44, 46, 47 and 61). A third direction is to use ML306 as a probe to identify the mechanism of action for this compound and the members of its series. We will explore the mechanism through our own studies as well as collaborating with the Roth Laboratory at the University of North Carolina, Chapel Hill. These efforts include pull-down studies using light-activated analogs of ML306, similar to what the Eckenhoff lab has done for propofol (with m-Azipropofol) and/or additional profiling of potential targets. The SAR profile we have established through this project thus far should allow for such derivatives to be prepared without affecting the anesthetic properties.

ML306 Probe Characterization

Figure

Figure A1. LC-MS purity analysis for ML306

Figure A2. ¹H NMR spectrum of ML306

Figure A3. ¹³C NMR spectrum of ML306

Analog Characterization

CID 56643048: LC-MS Retention Time: t₁ (Method 1) = 5.542 min and t₂ (Method 2) = x min; ¹H NMR (400 MHz, CDCl₃) δ 1.48 (d, J = 6.3 Hz, 2 H), 1.65 (d, J = 5.9 Hz, 1 H), 2.39 (s, 3 H), 2.76 (s, 1 H), 2.89 (s, 2 H), 4.65 – 4.82 (m, 1 H), 5.03 (d, J = 15.3 Hz, 1 H), 5.23 – 5.41 (m, 1 H), 5.93 (q, J = 6.4 Hz, 1 H), 6.24 – 6.40 (m, 2 H), 6.77 (d, J = 9.4 Hz, 1 H), 7.23 (d, J = 7.0 Hz, 2 H), 7.32 – 7.42 (m, 2 H), 7.54 – 7.61 (m, 1 H), 7.70 (d, J = 9.4 Hz, 1 H); HRMS (ESI) m/z (M+H)⁺ calcd. for C₂₀H₂₂N₃O₃, 352.1656; found 352.1656.

CID 56643039: LC-MS Retention Time: t₁ (Method 1) = 5.438 min and t₂ (Method 2) = 3.430 min; ¹H NMR (400 MHz, CDCl₃) δ 1.48 (d, J = 7.0 Hz, 2 H), 1.65 (d, J = 6.7 Hz, 1 H), 2.40 (s, 3 H), 2.77 (s, 1 H), 2.84 (s, 2 H), 4.98 – 5.33 (m, 2 H), 5.95 (q, J =7.0 Hz, 1 H), 6.21 – 6.40 (m, 2 H), 7.05 (d, J = 9.4 Hz, 1 H), 7.25 (s, 1 H), 7.36 – 7.47 (m, 1 H), 7.66 – 7.71 (m, 3 H); HRMS (ESI) m/z (M+H)⁺ calcd. for C₂₁H₂₃N₂O₃, 351.1703; found 351.1709.

CID 56643058: LC-MS Retention Time: t₁ (Method 1) = 5.455 min and t₂ (Method 2) = 3.334 min; ¹H NMR (400 MHz, CDCl₃) δ 1.70 (dd, J = 15.5 and 6.9 Hz, 2 H), 2.47 (s, 3 H), 2.93 – 3.06 (m, 2 H), 5.50 (q, J = 6.3 Hz, 1 H), 6.16 – 6.21 (m, 1 H), 6.27 – 6.31 (m, 1 H), 6.39 – 6.42 (m, 1 H), 7.23 – 7.47 (m, 4 H), 7.81 – 7.91 (m, 1 H), 7.93 – 8.00 (m, 2 H), 8.45 (d, J = 8.6 Hz, 1 H); HRMS (ESI) m/z (M+H)⁺ calcd. for C₂₀H₂₀N₅O₂, 362.1612; found 362.1616.

CID 56643034: LC-MS Retention Time: t₁ (Method 1) = 6.125 min and t₂ (Method 2) = 3.643 min; ¹H NMR (400 MHz, CDCl₃) δ 1.32 – 1.84 (m, 12 H), 2.35 (s, 3 H), 2.72 (s, 1 H), 2.92 (s, 2 H), 3.79 – 3.84 (m, 1 H), 4.25 – 4.39 (m, 1 H), 4.94 – 5.08 (m, 2 H), 7.04 (d, J = 9.8 Hz, 1 H), 7.30 (d, J = 7.8 Hz, 2 H), 7.76 (d, J = 7.8 Hz, 2 H), 7.99 – 8.11 (m, 1 H); HRMS (ESI) m/z (M+H)⁺ calcd. for C₂₁H₂₈N₃O₂, 354.2176; found 354.2179.

CID 56643061: LC-MS Retention Time: t₁ (Method 1) = 5.843 min and t₂ (Method 2) = 3.503 min; ¹H NMR (400 MHz, CDCl₃) δ 1.54 (d, J = 7.4 Hz, 2 H), 1.72 (d, J = 6.7 Hz, 2 H), 2.41 (s, 3 H), 2.76 (s, 3 H), 5.02 – 5.28 (m, 2 H), 6.06 (q, J= 6.9 Hz, 1 H), 7.00 – 7.10 (m, 1 H), 7.24 – 7.43 (m, 7 H), 7.65 – 7.74 (m, 3 H); HRMS (ESI) m/z (M+H)⁺ calcd. for C₂₂H₂₄N₃O₂, 362.1863; found 362.1866.