MK-8245 – RO91978 chemical

Development of a Liver-Targeted Stearoyl-CoA Desaturase (SCD) Inhibitor (MK-8245) to Establish a Therapeutic Window for the Treatment of Diabetes and Dyslipidemia

■ INTRODUCTION

One of the fundamental goals for pharmaceutical intervention is to achieve the desired eﬃcacy with as few side eﬀects as possible. This can be particularly challenging in cases where the target enzyme or receptor is ubiquitously expressed and deemed essential for certain physiological processes. One such target is stearoyl-CoA desaturase-1 (SCD1). SCD1 is the key enzyme involved in the synthesis of monounsaturated fatty acids and catalyzes the installation of a cis-double bond at the Δ9 position of long chain saturated fatty acyl-coenzyme A esters.1 The monounsaturated lipid products (palmitoleoyl-CoA (C16:1) and oleoyl-CoA (C18:1)) are key building blocks in the synthesis of membrane phospholipids, cholesterol esters, and triglycerides. It has been reported that rodents deﬁcient in SCD1 either by gene deletion,2—4 antisense oligonucleotide (ASO) treatment,5—7 or pharmacological inhibition8—12 are resistant to diet-induced weight gain and have an improved insulin sensitivity, glucose tolerance, and lipid proﬁle. In humans, elevated SCD1 levels are positively correlated with higher plasma triglycerides13 as well as increased BMI and high insulin levels.14 On the basis of the human and rodent ﬁndings, SCD1 represents an attractive novel there are reports of skin and eye abnormalities in the SCD1—/— mice3,15 as well as in rodents treated with SCD1 inhibitors.8—10 These adverse events (AEs) consist of dry eye, squinting, and alopecia and are believed to be due to mechanism-based deple- tion of essential SCD-derived lubricating lipids. Thus, the main challenge with developing a small molecule SCD inhibitor is to achieve adequate therapeutic eﬃcacy without aﬀecting the hu- man skin and eye functions. Given the nonlife threatening nature of metabolic diseases, it is crucial to establish a suﬃcient therapeutic window for SCD inhibitors for treating metabolic syndromes before embarking on human clinical trials. One potential way to achieve a therapeutic window for SCD inhibi- tion is to target the SCD inhibitor to the organ believed to be responsible for the therapeutic eﬃcacy (liver) while minimizing its exposure in the tissues associated with mechanism-based SCD depletion of essential lubricating lipids (skin and eye).

Two pivotal SCD1 knockdown experiments provided evi- dence that it may be possible to develop a safe and eﬀective SCD inhibitor for therapeutic application. First, treatment of mice with an SCD1 ASO resulted in 60 70% reduction of SCD1 mRNA in liver and fat and prevented diet-induced obesity and improved therapeutic target for the treatment of type II diabetes, dyslipidemia, obesity, and metabolic diseases. However, in addition to the positive metabolic eﬀects associated with SCD1 inhibition, insulin sensitivity.5 Importantly, after 10 weeks of treatment with the SCD1 ASO, the mice did not demonstrate the hair, skin, and eye abnormalities observed either with SCD1 inhibitor-treat- ment or in the global SCD1—/— mice. Second, Cre-loX technol- ogy was used to generate mice with a liver-speciﬁc knockout of SCD1 (LKO),4 and these mice were protected from high- carbohydrate diet-induced adiposity and hepatic steatosis. Unlike the global SCD1—/— mice showing alopecia and closed eyes, the LKO mice were indistinguishable from both wild-type and LoX mice (18 weeks).

In drug discovery, tissue-targeting refers to a process which delivers a pharmaceutical agent (small molecule or biologic) to the desired speciﬁc tissue. Targeting a pharmaceutical agent to the liver (i.e., liver-targeting) can be highly eﬀective to treat liver- related diseases when it is essential to minimize exposure to other tissues where adverse events can occur as a result of mechanism- based or oﬀ target drug activity. To target a pharmaceutical agent to the liver, the following approaches have been studied: (i) nanoparticles to deliver incorporated therapeutic materials such as small molecules, proteins, genes, and siRNAs,16 (ii) HepDirect cytochrome P450-activated prodrugs of certain small molecules such as nucleotides,17 and (iii) utilization of liver-speciﬁc transport proteins.18 The approach to liver-target- ing outlined in this paper relies on the latter strategy. Described herein will be the strategy used to design liver-targeting SCD inhibitors and how this approach resulted in the discovery of the lead compound MK-8245 (7), which maintains preclinical pharmacological eﬃcacy while signiﬁcantly improving the ther- apeutic window compared to systemically distributed SCD inhibitors (i.e., 1).

■ CHEMISTRY

The synthesis of the thiadiazole tetrazole 4 was prepared as described in Scheme 1. Starting with the bromo thiadiazole amine 9, addition of the piperadine 8 under basic conditions resulted in the formation of the amine 10. The amine 10 was then converted to the nitrile 11 using standard conditions and, ﬁnally, the tetrazole 4 was generated by reacting nitrile 11 with sodium azide. The syntheses of tetrazole 5 and the tetrazole acetic acid 6 are described in Scheme 2. The bromo thiadiazole 12 was reacted under basic conditions with the piperadine 13 to generate the nitrile 14. Compound 5 was prepared by reacting nitrile 14 with sodium azide. Subsequent reaction of the tetrazole 5 with sodium hydride and ethyl bromoacetate generated two alkylated pro- ducts, the major one bearing the ethylacetate at the N-2 position of the tetrazole. Upon chromatographic separation of the two alkylated products, the desired major product was subjected to basic ester hydrolysis to generate the desired tetrazole acetic acid 6. The synthesis of the optimal liver-targeted tetrazole acetic acid 7 was prepared as outlined in Scheme 3. The key transformation is an unprecedented three-component 1,3-dipolar cycloaddition and nucleophilic substitution reaction between 13, 15, and 16 to give the isoXazole intermediate 17 in 20% yield. The mechan- ism of this transformation is still not clear. Because 3-bromoisoX- azole is quite resistant to nucleophilic substitution, it is reasonable to hypothesize that the piperidine 13 and the dipole precursor 16 may react prior to the cycloaddition reaction with propiolate 15. The ester group in compound 1719 was converted to the corresponding amide 18 by reaction with ammonium hydroXide. The amide 18 was then dehydrated to the nitrile 19 by reaction with triﬂuoroacetic anhydride. The nitrile 19 was converted to the tetrazole 20 and ﬁnally to the desired tetrazole acetic acid 7 in a similar manner as previously described for compound 6.

■ RESULTS AND DISCUSSION

Strategy to Design a Liver-Targeted SCD Inhibitor. Our goal was to increase drug exposure in liver (target organ) by engaging active transport into heptocytes via the liver-specific organic anion transporting polypeptides (OATPs). Our goal was to incorporate key transporting elements into SCD inhibitors to enable recognition by the OATP transport proteins while main- taining SCD potency. Moreover, we also sought to decrease the extent of unselective passive cell diffusion in order to minimize exposures in off-target tissues and cells (skin and eye). Fortu- nately, acidic moieties used to engage the active transporters generally impart reduced passive cell penetration. Nonetheless, we required a research operating procedure (ROP) that would allow us to funnel potential compounds and help us identify compounds which may exhibit a liver-targeting tissue distribu- tion profile. The ROP to select for liver-targeted SCD inhibitors is depicted in Figure 1 and can be summarized as follows:on the right-hand side produced compounds completely devoid of SCD inhibition (3, Figure 2). One of the ﬁrst acidic moieties to be appended was the tetrazole (4, Scheme 1). This resulted in a signiﬁcant loss of potency on the SCD1 rat enzyme ( 140-fold, 4 vs 1, Table 1) as well as being inactive in the HepG2 cell assay. However, compound 3 exhibited a modest SCD inhibition in the rat hepatocyte assay which contains active OATPs (IC50 = 3467 nM). This demonstrated that appendage of an acidic moiety could decrease passive cell penetration and engage in active transport into a hepatocyte (assumption based on HepG2 vs hepatocyte shift). To improve potency on the rat SCD enzyme assay, structure activity relationship (SAR) studies were per- formed on the right-hand side phenoXy portion and, as has been previously reported,22 2,5-dihalogen substitution resulted in a signiﬁcant 8-fold improvement in potency (5, rat enzyme IC50 = 34 nM, Scheme 2, Table 1). Compound 5 was signiﬁcantly shifted in the HepG2 assay (370-fold) and slightly more potent in the hepatocyte assay (4-fold vs HepG2). A further increase in potency could be obtained by appending an acetic acid side chain onto the tetrazole heterocycle (6, rat enzyme IC50 = 7 nM, Table 1). In this case, the acidic tetrazole moiety was eﬀectively replaced with a carboXylic acid. As with the tetrazoles 4 and 5, acid 6 was >100-fold shifted in potency in the HepG2 cellu- lar assay and more potent in the hepatocyte assay. Finally, there were no signiﬁcant diﬀerences in potencies between the rat, mouse, and human SCD1 (i.e., for 7, the rat, mouse, and human SCD1 enzyme IC50s were 3, 3, and 1 nM, respectively). More- over, all of the SCD inhibitors listed in Table 1 were highly selective over the Δ-5 and Δ-6 desaturases (i.e., >100000 μM vs rat and human Δ5D and Δ6D as assessed in the HepG assay21). Liver-Targeted Tissue Distribution Profile. To determine if our strategy to select compounds based on their shift in the HepG2 vs hepatocyte assays resulted in compounds which exhibit a liver-targeted tissue distribution profile, the inhibitors in Table 1 were dosed in mice (10 mg/kg, PO) and 6 h post dose the animals were euthanized and the concentration of compound in select tissues (plasma, liver, skin, and the eye-lubricating Harderian glands) was measured. To simplify the discussion, the liver-to-Harderian gland ratio was used as a measure of the degree of liver-targeting. Starting with what we suspected to be a systemically distributed compound (i.e., a compound that dis- tributes readily to most tissues), 1, which is actually more potent in the HepG2 assay vs hepatocyte assay, did show almost equivalent levels of compound in the Harderian gland vs liver (Table 1). We were gratified to observe that the tetrazole 3, which is at least 14-fold more potent in the hepatocyte vs HepG2 assay, did demonstrate a liver-targeted tissue distribution profile. As indicated in Table 1, the liver-to-Harderian gland ratio for 3 was >300:1, which is clearly superior to the 1.5:1 ratio obtained with the nonacidic compound 1. Finally, the more potent acidic SCD inhibitors (5, 6, and 7) also demonstrated a liver-targeted tissue distribution profile as judged by their liver-to-Harderian gland ratios.

Figure 1. Strategy to identify liver-targeting SCD inhibitors. HepG2 cell SCD assay21 is a gauge of the extent of passive cell diﬀusion; rat hepatocyte SCD assay identiﬁes compounds which are actively trans- ported into the liver; mouse tissue distribution is a measure of the extent of liver-targeting.

In addition to generating OATP substrate aﬃnity, our liver- targeting strategy required the minimization of passive cell diﬀusion to prevent the penetration of any circulating SCD inhibitor to oﬀ-target tissues. The most potent SCD inhibitor 7 was signiﬁcantly shifted in the non-OATP HepG2 cellular assay ( 350-fold), indicating that this compound should possess the desired lack of cell penetration. To conﬁrm this, the diﬀusion rate of 7 was evaluated across monolayers of LLC-PK1 cells and the Papp was determined to be 3.8 10—6 cm/s, which is indicative of poor to moderate cell penetration24 (Supporting Information Method 2). It should be noted that we required the liver-
targeting compound to be absorbed in the GI tract in order to be delivered to the portal vein eﬀectively. Thus, there still had to be some degree of cell penetration (i.e., Papp > 1 10—6 cm/s and <10 10—6 cm/s) and it was determined that compound 7 was moderately bioavailable in multiple species (F = 12%, 28% and 40% in mice, rats and dogs, respectively). Moreover, it is known that OATPs are expressed in the intestinal wall25,26 and this may also be responsible for the absorption of compound 7. To determine if mouse liver-targeting via OATPs would translate to liver-targeting in other species,27 7 was dosed orally to mice, rats, dogs, and rhesus monkeys and at 6 h post dose, the animals were euthanized, and the concentration of compound in select tissues (plasma, liver, skin, and Harderian glands) was measured. Only rodents possess Harderian glands, and because the corresponding eye-lubricating meibomian glands in dogs and rhesus monkeys were too diﬃcult to extract, skin was used as the oﬀ-target tissue to measure liver-targeting in these higher species. As depicted in Figure 3, in all species examined, 7 was distributed mainly to the liver, with low exposure in tissues associated with potential adverse events (i.e., skin and eye lubricating glands). The liver-to-skin ratios were >30:1 in all four species.OATP Transport Profile. To demonstrate that 7 is actively transported via human OATPs, uptake studies were performed in butyrate-treated stably transfected OATP1B1 and 1B3 MDCKII cell lines. As shown in Figure 4, 7 was found to be a substrate of both OATP1B1 and 1B3 and the uptake was inhibited by the OATP inhibitor sulfobromophthalein28 (BSP). This data sug- gests OATP1B1 and OATP1B3 likely contribute toward the liver selectivity of 7 by mediating uptake into hepatocytes. The potency of 7 was measured in both freshly isolated and cryopre- served human hepatocytes and was found to be highly potent (IC50 = 5 nM, Supporting Information Method 3) compared to the non-OATP human HepG2 cell line (IC50 = 1066 nM), further indicating active uptake transport of 7 by human OATPs. Pharmacological Efficacy of the Liver-Targeted SCD Inhibitor 7. To assess the diabetic efficacy, lipid efficacy, and safety profile of a systemically distributed vs liver-targeted SCD inhibitor, the following models were used: (i) acute oral glucose tolerance in eDIO mice (oGTT) to assess glucose clearance and (ii) chronic administration in eDIO mice to assess body weight effects and lipid profile as well as evaluating the adverse event profiles in skin and eyes. Compounds 1 and 7 were given orally to eDIO mice 1 h before administration of an oral glucose challenge, and blood glucose levels were monitored for 2 h thereafter. As shown in Figure 5, both the systemically distributed 1 and the liver-targeted 7 improved glucose clearance dose-dependently to a similar extent, with ED50 values of 3 and 7 mg/kg, respectively. This was the first indication that selectively inhibiting liver SCD1 with a liver-targeted small molecule SCD inhibitor was effective at lowering glucose levels.

Figure 3. Plasma and tissue concentration ratios at 6 h post oral dosing of 7 in mouse, rat, dog, and rhesus monkey. (a) Concentration of 7 in tissues 6 h post dose. Compound 7 was dosed orally (formulated in 0.5% methylcellulose) to male C57BL6 mice (n = 2, 10 mg/kg), male Sprague Dawley rats (n = 3, 10 mg/kg), female rhesus monkeys (n = 2, 5 mg/kg), and male beagle dogs (n = 2, 5 mg/kg). (b) Ratio of tissue concentrations in liver vs tissues associated with potential adverse events. ND = not determined.

Figure 4. Uptake of 7 into (a) OATP1B1 stably transfected MDCK11 cells and (b) OATP1B3 stably transfected MDCK11 cells. OATP1B1 and 1B3 were found to mediate the BSP sensitive uptake of 7, suggesting these transporters may play a role in the uptake of 7 into human hepatocytes. All data are mean ( sem, n = 3.

Chronic eﬃcacy of the systemically distributed 1 and the liver- targeting SCD inhibitor 7 was evaluated in a 4-week oral dosing study in eDIO mice. Once-daily oral dosing was used for compound 1 based on its 6.4 h plasma half-life in mouse,9 whereas bid dosing was required for 7 (mouse plasma half-life = 0.7 h) to provide similar SCD inhibition over the course of the experiment. Table 2 compares the mouse pharmacokinetics of compounds 1 and 7 and also indicates that the terminal trough liver levels of both compounds in the chronic eDIO mouse study are similar and would provide near complete liver SCD inhibition (see Supporting Information Results 2). As shown in Figure 6a, compound 1 treatment resulted in a signiﬁcant, dose-dependent decrease in body weight ( 10% and 20% for the 1 and 5 mg/kg treatment groups, respectively), comparable to the eﬀect seen with the CB1 inverse agonist AM251.29 On the other hand, the liver-targeted inhibitor 7, dosed at 20 or 60 mg/kg bid, showed only a modest prevention of body weight gain ( 5% BW gain reduction, Figure 6b). The maximally eﬃcacious dose of 1, based on body weight eﬀects, was 5 mg/kg, qd. For compound 7, the maximally eﬃcacious dose was 20 mg/kg bid based on body weight and liver triglyceride reduction (60 mg/kg bid dose did not provide improved eﬃcacy results). Histological analysis showed that hepatic steatosis was signiﬁcantly reduced, which was conﬁrmed by a reduction of liver triglyceride levels (see Supporting Information Results 3, Figure 3). In these eDIO studies, there was no eﬀect on food consumption with either compound.

Evaluation of Skin and Eye Adverse Effects in Rodents Treated with Systemic vs Liver-Targeted SCD Inhibitors. In addition to evaluating efficacy end points in the chronic 4-week eDIO mouse studies, comparison of the skin and eye adverse effect profiles of a systemic vs liver-targeted SCD inhibitor was of paramount importance. Notably, would a liver-targeted SCD inhibitor impart a therapeutic window compared to a systemi- cally distributed inhibitor? As previously reported for systemi- cally distributed SCD inhibitors,8—10 both skin and eye abnormalities were observed in compound 1-treated eDIO mice (Figures 7a,b). In contrast to systemically distributed SCD inhibitors, which induce significant skin and eye adverse effects in the eDIO mouse within 5 10 days of treatment at therapeutic doses, the liver-targeted 7 did not induce meaningful skin or eye adverse events during the 4-week treatment period (Figures 7c, d). To confirm that the improved adverse event profile observed with 7 in the 4-week eDIO mouse study is due to lack of SCD inhibition in the off target tissues, the desaturation indices (DIs) in the off-target tissues were compared to that obtained in the liver. The DI is the ratio of SCD product-to-SCD substrate ([C18:1, oleic acid]/[C18:0, stearic acid]) and is a direct measure of chronic SCD activity.13 A reduction in DI indicates inhibition of the SCD enzyme and DI can be monitored in plasma and tissues. For the systemically distributed 1 at a maximally efficacious dose of 5 mg/kg qd, the liver SCD DI was reduced by 56% (Figure 8a, not significant due to high variation in vehicle-treated mice), while there was significant reduction of SCD activity in both the Harderian gland (92% V DI, Figure 8b) and skin (25% V DI, Figure 8c). These significant DI reductions are consistent with the high exposure of 1 in all of these tissues ([liver] = 10 μM, [Harderian gland] = 5 μM, [skin] = 4 μM).

In contrast, for the liver-targeted SCD inhibitor 7, while there was very signiﬁcant reduction in liver DI (63%) following 4 weeks of dosing 20 mg/kg bid in eDIO mice, there was no signiﬁcant reduction in Harderian gland DI (Figures 8d,e, respectively; skin DI was not measured in the compound 7- treated mice). The reduction of DI (liver) or lack thereof (Harderian gland) was consistent with the exposure of 7 in these tissues ([liver] = 5 μM, [Harderian gland] = 0.9 μM). It should be noted that there was a small erosion of liver-targeting upon chronic dosing of 7 as assessed by liver-to-Harderian gland ratio (21:1 following single administration vs 5:1 after 4 weeks of dosing). The liver-to-Harderian gland ratio for 7 reached steady state (5:1) after 1 week of dosing, did not worsen upon chronic dosing (4 weeks) and was suﬃcient in providing a therapeutic window in the chronic eDIO mouse model (i.e., lack of harderian gland SCD inhibition (Figure 8e) and no signiﬁcant eye AEs (Figure 7d)).

One of most well-documented classes of liver-targeting small- molecule therapeutics are HMG-CoA reductase inhibitors (statins). As with SCD inhibitors, distribution of statins to tissues other than liver is linked to adverse events (myopathy in the case of statins).30 Statins are liver-targeted by virtue of their aﬃnity for the liver-speciﬁc organic anion transporting polypeptides (OATPs),and it was strategized that targeting the same OATP transporters would lead to the generation of a liver-targeted SCD inhibitor. The fact that statins were substrates for the OATPs was not known a priori and this liver-targeting feature was not knowingly designed into the compounds to improve the therapeutic index. In fact, realization that statins were OATP substrates occurred after the ﬁrst statin was already on the market. It was the goal of our research eﬀorts to implement an in vitro strategy to predict the liver-targeting capabilities of a compound.

Figure 6. Eﬀect on body weight following chronic dosing of (a) 1 vs (b) 7 in eDIO mice. eDIO mouse fed on HFD for at least 18 weeks prior to initiation of study were dosed orally with vehicle, 1 (1 or 5 mg/kg, qd) or 7 (20 or 60 mg/kg, bid). Doses were selected to target sustained liver SCD inhibition based on mouse liver PD results (Supporting Information Method 4). HFD, high fat diet; Veh, vehicle; n = 8—10 mice/group; mean ( sem.

Figure 7. Skin and eye adverse event proﬁle of 1 (a and b) and 7 (c and d) in a 4-week eDIO mouse study. Skin scoring system: 0, normal; 1, small patch (1 3 mm) of hair loss; 2, larger patch (3 0 mm); 3, extensive patch with lesions. Eye scoring system: 0, normal; 1, squinting eye; 2, partial eye closure, one eye; 3, both eyes closed. To lessen the severity and allow for completion of the study, eye lubrication (BNP: Neomycin and Polymyxin B Sulfates and Bacitracin Ophthalmic Ointment, USP) was applied to the eyes of compound 1-treated mice starting on day 12. Veh = vehicle.

Transporters can aﬀect the tissue distribution of drugs and thereby contribute to the selective distribution of drugs to speciﬁc tissues. Two of the OATP transporters (1B1 and 1B3) are speciﬁcally expressed in the liver, are localized to the basolateral membrane of the hepatocytes, and facilitate the uptake of drugs from the blood into hepatocytes.31 Drug sub- strates of OATPs 1B1 and 1B3 include a large number of structurally diverse compounds, such as statins (i.e., atorvastatin, rosuvastatin, simvastatin acid, etc.), endothelin receptor antago- nists (i.e., atrasentan), certain antibiotics (i.e., benzylpenicillin), and angiotensin II receptor antagonists (i.e., olmesartan, valsartan), just to name a few.31 One of the key structural features that these drugs have in common is an acidic moiety, either a carboXylic acid or tetrazole moiety. It was determined that adding an acidic moiety to the left-hand side of systemically distributed SCD inhibitors allowed for retention of SCD potency (compounds 4, 5, 6, and 7, Table 1). However, simply adding an acidic moiety on the left-hand side of the molecule did not automatically impart recognition by the liver-speciﬁc OATPs, as judged by the shift in potency in HepG2 vs hepatocyte assays as well as in tissue distribution studies. Detailed SAR studies regarding the correct positioning of the acidic moiety to maximize both SCD potency and liver-targeting will be the subject of subsequent publications.

Figure 8. Desaturation indices in tissues of eDIO mice treated with 1 or 7. eDIO mice fed on HFD for at least 18 weeks prior to initiation of study were dosed orally with vehicle, 1 (5 mg/kg, qd, 3 days) or 7 (20 mg/kg, bid, 28 days). Following sacriﬁce, tissues were harvested and DI measurements were obtained. (a and d) represent liver DI measurements in vehicle vs 1 or 7-treated mice, respectively; (b and e) represent Harderian gland (HG) DI measurements in vehicle vs 1 or 7-treated mice, respectively; (c) represents skin DI measurements in vehicle vs 1-treated mice (skin DI for 7-treated mice was not determined). n = 8—10 mice/group; mean ( sem.

The optimization of liver-targeted SCD inhibitors depicted in Table 1 involved the important use of the HepG2 and hepatocyte SCD cellular assays. The principle reason for using cellular SCD assays to gauge OATP substrate aﬃnity is that they are very high throughput. Assays to directly measure OATP substrate aﬃnity certainly exist (see EXperimental Section) but are not suﬃciently high throughput to be able to test hundreds of compounds per month. The goal of the HepG2 assay is to gauge the extent of passive diﬀusion and compounds which are not potent in this assay are desired to minimize diﬀusion and distribution to non- liver tissues. Conversely, in the OATP hepatocyte cell assay, compounds which are recognized by the OATPs will be actively transported into the hepatocyte and should be more potent at inhibiting intracellular SCD compared to the HepG2 assay. It should be noted that there is only a loose correlation between HepG2-to-hepatocyte potency shift and liver-targeting as mea- sured by compound levels in mouse liver vs Harderian gland. However, as depicted in Table 1, HepG2/hepatocyte ratios of g4 do translate to signiﬁcant and likely suﬃcient liver-targeting as determined from the mouse tissue distribution analyses. In the tissue distribution analysis, particular attention is focused on the drug concentration in tissues where adverse events have been observed in the SCD1—/— mice (the eye lubricating Harderian gland and skin tissue). Focusing only on the plasma-to-liver ratio can be misleading because many nontransported drugs reach high concentrations in the liver due to rapid exchange of blood constituents with the liver tissue. Drugs absorbed in the gastro- intestinal tract are completely transported to the liver via the portal vein, and after steady state has been reached, most compounds do not demonstrate cell speciﬁcity (liver levels of compound vs that in other tissues). For example, the liver-to- plasma ratios for the systemically distributed 1 and the liver- targeted compound 4 were very similar (14:1 for 1 vs 15:1 for 4) and could not be used to discriminate these two compounds. However, when comparing the liver-to-Harderian gland ratios, there is clearly a diﬀerence (1.5:1 for 1 and >300:1 for 4), thereby demonstrating the need to examine drug concentrations in the oﬀ-target tissues to properly assess liver targeting. Our target tissue distribution proﬁle was determined to be a concentration of drug in liver that fully inhibits SCD and that is at least 20-fold higher than levels found in the oﬀ-target tissues (Harderian gland and skin). The most potent liver-targeted SCD inhibitor 7 met these criteria not only in mouse but also in rat, dog, and rhesus monkey. Furthermore, compound 7 is likely to be liver-targeted in humans due to the fact that it was found to be a substrate for the human OATP1B1 and 1B3 (Figure 4) and was also very potent in a human hepatocyte assay (IC50 = 5 nM). It has been reported that there is a strong association between simvastatin- induced myopathy and two tightly linked variants of the OATP1B1 gene.32 This is likely due to impaired liver uptake of simvastatin in those subjects with the OATP1B1 variants which, in turn, leads to increased systemic exposure of the statin and adverse events in the oﬀ target muscle tissue. The fact that 7 is actively taken up in the liver by both OATP1B1 and OATP1B3 may mitigate the risk of skin and eye adverse events in patients with the OATP1B1 variants. Nonetheless, the eﬀect of OATP1B1 polymorphisms and tolerability of compound 7 will be followed up in the future with in vitro uptake tools and in clinical studies.

It was shown in Figure 5 that the liver-targeted inhibitor 7 was as eﬃcacious in the oGTT eDIO mouse model as the systemi- cally distributed 1. In an attempt to understand the mechanism by which hepatic SCD1 inhibition results in an improved oral glucose tolerance in eDIO mice, a hyperinsulinemic euglycemic glucose clamp experiment in overfed rats was carried out (see Supporting Information Results 1). As shown in Supporting Information Figure 1a, under hyperinsulinemic euglycemic conditions, compound 7 signiﬁcantly increased the glucose infusion rate (GIR) required to maintain euglycemia, indicative of improvement in whole body insulin sensitivity. This was accompanied by a nonsigniﬁcant suppression in hepatic glucose production (Supporting Information Figure 1b). These results indicate that a liver-targeted small molecule SCD inhibitor (7) improves whole body insulin sensitivity, consistent with the literature report with a liver-targeted ASO against SCD1.6 In a chronic eDIO mouse model, there were diﬀerences in eﬃcacy and tolerability noted between the systemically distributed SCD inhibitor 1 and the liver-targeted 7. While eﬃcacy on liver triglyceride and liver steatosis was similar for 1 and 7, the eﬀect on body weight reduction was signiﬁcantly diﬀerent (20% vs 5%, respectively, Figure 6). It is clear that the antiobesity eﬀect with liver-selective SCD inhibitors is very modest compared to that obtained with systemically distributed inhibitors. In an attempt to understand the obesity eﬃcacy diﬀerences, the liver-targeted SCD1 inhibitor 6 and the systemically distributed SCD1 inhi- bitor 1 were compared head-to-head in the eDIO mouse model and the following parameters were measured: (i) body weight and (ii) inhibition of SCD activity (DI) in fat vs liver tissues. As previously observed in Figure 6, the systemically distributed SCD inhibitor 1 had a signiﬁcant body weight loss ( 15%) vs vehicle- treated mice whereas the liver-targeted SCD inhibitor 6 did not promote signiﬁcant body weight loss in eDIO mice (see Sup- porting Information Results 4, Figure 4a). Both 1 and 6 had similar inhibition of liver desaturation index (DI = [C18:1]/ [C18:0], direct measure of chronic SCD activity) at termination of the 2-week study, indicating that similar levels of liver SCD inhibition had been obtained (Supporting Information Figure 4b). Conversely, while there was signiﬁcant inhibition of eWAT DI for compound 1, there was no inhibition observed in eWAT for the liver-targeted 6 (Supporting Information Figure 4c), which conﬁrms that liver-targeting prevents exposure of SCD compounds to other tissues. This data suggests that robust obesity eﬃcacy likely requires inhibition of adipose SCD and is in agreement with previously reported data.4

With regards to tolerability and adverse event proﬁle in the chronic eDIO mouse studies, there were important diﬀerences between the systemically distributed SCD inhibitor 1 and the liver-targeted inhibitor 7. As reported in parts a and b of Figure 7, compound 1 induced eye adverse events, followed a few days later by skin adverse events. At the maximally eﬃcacious dose of 7 (20 mg/kg bid), there was observed a liver-targeted tissue distribution proﬁle which correlated with liver TG eﬃcacy (Supporting Information Figure 3a), modest body weight loss and no signiﬁcant eye or skin adverse events during the course of the 4-week study (Figures 7c,d). At a 3-fold higher dose of 7 (60 mg/kg bid), a clean adverse event proﬁle was also observed, which indicates that a therapeutic window could be obtained by targeting the SCD inhibitor to the liver, consistent with reported tissue as assessed by monitoring the DI in the Harderian gland following chronic dosing in the eDIO mouse model (Figure 8e). The level of DI reduction in the liver and Harderian gland in this chronic study correlated with the high drug exposures of 7 in the liver and the lower drug exposure in skin and eye. This indicated that the liver-targeting SCD inhibitor 7 was successful at provid- ing good SCD inhibition in liver while sparing inhibition in tissues associated with adverse events.

■ CONCLUSION

In this report, a liver-targeting strategy was successfully implemented which resulted in the discovery of 7, a potent, liver-targeted SCD inhibitor with excellent in vitro and preclini- cal in vivo eﬃcacy. This approach utilized high throughput cellular assays, which predicted the extent of liver-speciﬁc OATP substrate aﬃnity. Following conﬁrmation of liver-targeting by tissue distribution studies, the lower throughput OATP assays could be used to conﬁrm human OATP substrate aﬃnity. While liver-targeting OATP approaches have been used to improve the therapeutic window of pharmaceutical agents such as statins, to the best of our knowledge, this is the ﬁrst report of how to intentionally design a liver-targeting agent via OATP substrate aﬃnity. Liver-targeting allowed compound 7 to demonstrate maximal liver SCD inhibition while sparing inhibition in the skin and eye tissues associated with adverse events. This ultimately resulted in an improved safety proﬁle for the liver-targeted SCD inhibitor 7 relative to the previously reported systemically distributed inhibitor 1. Future studies involve the clinical evalua- tion of 7 for the treatment of diabetes and dyslipidemia.

Method for Tissue Desaturation Index Determination. Liver, fat, Harderian gland, eWAT, and skin were collected for desatura- tion index determination. Liver, fat, eWAT, and Harderian glands were weighed and collected in strips of 12 attached microtubes in microracks (National Scientific, Claremont, California). Enough cold Dulbecco’s Phosphate Buffer Saline (DPBS; Cellgro, Manassas, Virginia) was added to liver and fat to obtain a final concentration of 25 mg/100 μL. A fiXed amount of 100 μL of cold DPBS was added to the Harderian glands. Three medium size (5/3200) stainless steel beads (MetalFini, St-Laurent, Qu´ebec) were added for homogenization. All tissues were mechanically milled using the GenoGrinder 2000 (ATS Scientific, Burlington, Ontario). Tough tissues like skin could not be properly homogenized using a GenoGrinder. Pieces of weighed shaved skin were collected in microcentrifuge tubes. The fresh or cryopreserved skin tissues were then transferred to 8 mL glass tubes. Enough 10N NaOH (J.T. Baker, Phillipsburg, New Jersey) was added to obtain a final concentration of 25 mg/500 μL. The skin samples were then hydrolyzed overnight at 100 °C. Upon cooling, 500 μL of hydrolyzed skin were transferred to new 8 mL glass tubes. Aliquots (100 μL) of fresh or cryopreserved liver, fat, eWAT, and Harderian gland tissue homogenates were also hydro- lyzed by transferring them to 8 mL glass tubes (VWR), adding 500 μL of 10N NaOH, and submitting them to 100 °C for 1 h. Once cooled to room temperature, 1 mL of isooctane (Sigma-Aldrich, Milwaukee, Wisconsin) was added to all samples to conduct the first extraction.

The uptake of [14C]-7 (10 μM) into MDCKII cells stably transfected with the human uptake transporters OATP1B1 or OATP1B3 was discarded. Samples were acidified to pH 2.0 by adding 700 μL of formic acid (Acros Organics, Geel, Belgium) and miXed. The extraction of the fatty acids was carried on by adding 500 μL of isooctane and thoroughly miXing. The solvent layers of isooctane containing the fatty acids were collected after centrifugation and transferred to new 8 mL glass tubes containing 1 mL of saturated NaCl solution with 1% formic acid. The purpose of this step is to remove all residual aqueous solution from the isooctane before gas chromatography (GC) analysis. Once miXed and centrifuged, an aliquot of 100 μL was transferred to GC vials (Chromatographic Specialties, Brockville, Ontario), containing 300 μL conical inserts with springs (Fisher Scientific, Ottawa, Ontario). The fatty acids were derivatized to their corresponding methyl esters with 20 μL of diazomethane solution prepared in-house and analyzed on an Agilent Technologies (Mississauga, Ontario) 6890N GC system equipped with a flame ionization detector (FID) using a HP-88 column (100 m length, 0.25 mm ID, 0.2 μM film), also from Agilent Technol- ogies. Samples (2 μL) were injected in split mode using ratio up to 25:1. Helium was used as carrier gas at a flow rate of 2 mL/min. Hydrogen, provided by a Parker hydrogen generator (Cleveland, Ohio), was used as auXiliary gas for the flame ionization detector. The injector and detector temperatures were set at 250 and 280 °C, respectively. The initial oven temperature was set at 120 °C, increased to 165 °C at a rate of 10 °C/min, increased to 210 °C at a rate of 5 °C/min, and increased to 230 °C at a rate of 5 °C/min.

Data analysis of the fatty acid methyl esters (FAMEs) was done using the Agilent ChemStation integration software. FAMEs identiﬁcation was achieved by comparison of retention times with those obtained from standard miXtures (NU-CHEK-PREP, Elysian, Minnesota). The Desa- turation Index was determined by calculating the SCD-1 product peak area/SCD-1 substrate peak area.Statistical Method. The P values reported in all figures were calculated using Merck CMG (comparing multiple groups) StatServer version 2.93, all pairwise comparisons, confidence intervals at 95%.