Discovery of a Potent and Selective TRPC5 Inhibitor, Efficacious in a Focal Segmental Glomerulosclerosis Model

Abstract

The nonselective Ca²⁺-permeable transient receptor potential (TRP) channels play important roles in diverse cellular processes, including actin remodeling and cell migration. TRP channel subfamily C, member 5 (TRPC5) helps regulate a tight balance of cytoskeletal dynamics in podocytes and is suggested to be involved in the pathogenesis of proteinuric kidney diseases, such as focal segmental glomerulosclerosis (FSGS). As such, protection of podocytes by inhibition of TRPC5 mediated Ca²⁺ signaling may provide a novel therapeutic approach for the treatment of proteinuric kidney diseases. Herein, we describe the identification of a novel TRPC5 inhibitor, GFB-8438, by systematic optimization of a high-throughput screening hit, pyridazinone 1. GFB-8438 protects mouse podocytes from injury induced by protamine sulfate (PS) in vitro. It is also efficacious in a hypertensive deoxycorticosterone acetate (DOCA)-salt rat model of FSGS, significantly reducing both total protein and albumin concentrations in urine.

Progressive chronic kidney disease (CKD) is a global health concern, affecting over 500 million people worldwide and about 26 million in the US alone.^1,2 CKD is regarded as an accelerator of cardiovascular disease (CVD), which, along with renal failure, is the primary cause of morbidity and mortality in this underserved population.³ As one of the leading causes of glomerular disease, focal segmental glomerulosclerosis (FSGS) in its most severe form is associated with the nephrotic syndrome characterized by symptomatic proteinuria (>3.5 g protein per day) in adults, scarring of the glomerulus, and loss of terminally differentiated podocytes.^4,5 Currently there are few effective treatments for FSGS, and many patients are treated chronically with steroids, which have undesirable side effects. Therefore, there is a significant unmet need for novel therapeutic approaches that can prevent or slow disease progression.

A hallmark of the pathogenesis of FSGS is podocyte injury and loss. Podocyte foot process effacement, one of the first morphologic alterations of the glomerular filtration barrier, is caused by the disruption of the podocyte actin cytoskeleton via activation of Rac1.⁶ TRPC5 has been identified as a key mediator of Rac1 activation in podocytes.^7,8 TRPC5 belongs to the mammalian superfamily of transient receptor potential (TRP) Ca²⁺-permeable nonselective cationic channels.⁹ Damage to podocytes activates Rac1, leading to translocation of TRPC5 to the cell membrane. Activation of the membrane-located TRPC5 channel leads to further Rac1 activation in a feed-forward loop, thereby driving cytoskeletal remodeling. The ensuing cytoskeletal changes lead to effacement of podocyte foot processes, podocyte loss, and proteinuria. Inhibition of TRPC5 by small molecules has been shown to protect podocytes in vitro and suppresses proteinuria and podocyte loss in vivo in a transgenic rat model of FSGS.^10,11 TRPC5-knockout (KO) mice were also protected from lipopolysaccharide (LPS)-induced albuminuria.¹⁰ Collectively, these results suggest that inhibition of  TRPC5 may provide a novel, effective, and safe approach for the treatment of FSGS and related proteinuric kidney diseases.

Several subtype-selective small molecule TRPC5 inhibitors have recently emerged in the literature.¹²⁻¹⁴ The structures of representative compounds are summarized in Figure 1. Chemical probes such as ML204,¹⁵ Clemizole,¹⁶ and AC1903¹⁷ have been instrumental in understanding the biological roles of TRPC5. In addition, this pioneering work helped to underscore the therapeutic potential associated with its inhibition. Although effective as molecular probes, these molecules generally suffer from low potency, off-target activities, or unknown selectivity and unfavorable pharmacokinetic properties. However, xanthine analogues such as HC-608 and HC-070, identified by Hydra/Boehringer-Ingelheim, are highly potent.¹⁸ While these compounds have further helped advance TRPC₄ and TRPC5 biology, especially in the CNS, this series of inhibitors suffers from poor physicochemical properties such as low solubility (kinetic solubility of HC-608 is 0.03 μM at pH 7.4 in phosphate buffer) and very high protein binding (>99.5% bound). In our search for novel TRPC5 inhibitors with an improved overall profile, we screened a structurally diverse library of ∼400,000 compounds against TRPC5, using a FLIPR assay, which resulted in the identification of several low micromolar hits. From the hit set, pyridazinone 1 (Figure 2) was selected as an attractive starting point for further exploration based on the encouraging potency, selectivity over TRPC6, good ligand efficiency, low cLogP, as well as the potential for rapid SAR exploration. Another important consideration was the ability of 1 to block the TRPC5 channel with similar potency, independent of the activators used (Rosiglitazone: IC₅₀ 5.0 μM, Riluzole: IC₅₀ 7.8 μM, (−)-Englerin A: IC₅₀ 4.0 μM, and Lanthanum: IC₅₀ 4.4 μM).

Figure 1.

Representative TRPC5 inhibitors from the literature.

Figure 2.

HTS hit, 1.

Our initial chemistry efforts focused on the improvement of the potency in blocking TRPC5 channel Ca²⁺ current. A survey of benzylic groups indicated dynamic SAR and suggested the possibility of improving potency. As exemplified by compounds 1, 2, and 3 (Table 1), aromatic and aliphatic R2 groups were tolerated. However, incorporation of heteroaromatic groups tended to lead to erosion of potency as represented by 6-1,6-2, 34, and 36. As shown by 5, addition of a p-fluoro group boosted potency 2.5× relative to compound 1, while removal of the o-methyl group led to a 3× loss in potency. Racemic methyl substitution on benzylic methylene was tolerated (4). Turning our attention to the pyridazinone ring, methylation and alkylation of the NH as in 7 and 8, respectively, greatly diminished the TRPC5 potency, suggesting that the NH may be involved in a productive H-bond. Investigation of modifications to the piperazine core proved informative. Introduction of a chiral Me group provided enantiomers with divergent activities as in 9 vs 10. However, methylation adjacent to the basic piperazine N, as in 11, led to ablation of all potency. As demonstrated with 13, replacing the basic N with carbon resulted in complete loss of potency. Homopiperazine (14) exhibited diminished activity as well. Attempts to introduce bridged and spiro features to the piperazine ring turned out to be unproductive (15, 16), except analogue 17, which has comparable activity to 1, but with much less favored physiochemical properties. Interestingly, introduction of the lactam carbonyl saw a 10-fold boost in potency (12). Contrasting to the basic nature of the piperazine core, the lactam displays very distinct electrostatic properties. The switch from sp³ to sp² also results in conformational and trajectory change for piperazine and the adjacent benzyl group (See Figure S1 for modeling study). This highly specific SAR suggests that piperazine might sit in a well-defined binding pocket, and the potency gain observed with the lactam is indicative of productive interactions with the protein.

Table 1. Summary of Early Dynamic SAR and Core Modification.

^aQpatch assay measuring the outward current at +80 mV in HEK 293 cells overexpressing human TRPC5 activated by 30 μM rosiglitazone. Results are mean of at least two experiments. Experimental errors within 30% of value.

^bThis denotes the end connecting to pyridazinone.

^cChloro analogue.

Owing to the potential instability, we turned our attention to the bromide functionality in this series. We speculated that it may play multiple roles: (a) halogen bonding with protein residue; (b) fine-tuning of the pK_a of the acidic proton in the warhead; and (c) steric and conformational constraint of the piperazinone scaffold. As such, we replaced the Br with groups of varying size and polarity. Replacing Br with a methyl (18) afforded measurable potency. Increasing the alkyl size (18 to 21) led to the increase in inhibitory activity with cyclopropyl as the most optimal. However, larger groups such as tetrahydropyran were not tolerated (22). Unsurprisingly, the Cl analogue (23) yielded similar potency to 12. Small electron-withdrawing substituents were also tolerated albeit with 4- to 20-fold reduction in potency (24, 25). Electron-donating substitution (26, 27) proved to be ineffective, hinting at the need to maintain the pK_a of the acidic NH proton. Through this effort, we identified Cl as a suitable replacement for Br, which is a common element in many approved medicines.¹⁹

Having addressed several shortcomings related to the original hit, the potency for blocking the Ca²⁺ current still had room for further improvement (Table 2). Walking methyl around the phenyl ring established that the ortho monomethyl substitution is most effective (23, 28, 29, 30). Based on this result, we screened a range of ortho-substituted analogues in order to identify a group to further improve the potency and balance physicochemical properties and microsomal stability. Strcuture activity relationship (SAR) suggested lipophilic groups in this position were well tolerated with CH₃, Cl, or CF₃ (23, 31, 32, GFB-8438). While incorporating nitrogen next to CF₃ (34) improved the properties, it led to dramatic loss in activity. Substitution in the para-position with F to block the potential metabolic soft spot caused slight erosion in potency (35). Polar motifs (33, 36, 38) showed a potency loss trend as well, suggesting this part of the compound might bind in a lipophilic pocket. Indeed 4-Me-cyclohexyl (37) was equally potent as 23, but unfortunately with very unfavorable physiochemical properties.

Table 2. Summary of SAR in Br and Benzyl Region.

^aResults are mean of at least two experiments. Experimental errors within 30% of value. Activated (by 30 μM Rosiglitazone) TRPC5 overexpressed HEK 293 cells used for outward current measurement at +80 mV in Qpatch assay.

Several analogues identified in Table 2 have reasonable potencies and potential to serve as tool compounds for in vivo study. As such we determined the PK parameters for GFB-8438, 32, and 35 (Table 3). All three compounds demonstrated good metabolic stability when incubated with both human liver microsomes and hepatocytes, but with less stability in rat. The observed intrinsic microsomal clearance correlated well with in vivo rat clearance, and oral bioavailability in rats was modest and similar for each of the three compounds. When dosed subcutaneously, GFB-8438 exhibited significantly higher plasma exposure. With the subcutaneous route overcoming the limited oral bioavailability, GFB-8438 was deemed suitable for dosing in our efficacy model.

Table 3. DMPK Profiling of GFB-8438, 32, and 35^a.

	8438	32	35
HT-solubility (pH = 7.4) (μM)	316	114	12
rLM/hepatocyte CLint (mL/min/kg)	24/94	55/147	18/26
hLM/hepatocyte CL (mL/min/kg)	7.8/5.5	0.14/4.1	4.3/5.3
Pharmacokinetics in rat
CL (mL/min/kg)	31	56	27
T1/2 (h)	0.50	0.45	1.05
Vss (L/kg)	1.17	1.27	1.81
F (PO/SC) (%)	17/33	19/-	22/-

^aSee Supporting Information for experimental details.

Having achieved a desired pharmacokinetic profile in rat, we embarked on further characterizing GFB-8438. We first confirmed the inhibition of human TRPC5 by manual patch clamp. GFB-8438 inhibited the whole cell current with an IC₅₀ of 0.28 μM comparable to the IC₅₀ in Qpatch assay (0.18 μM) (Table 4). We then determined the potency of GFB-8438 against the TRPC5 rat ortholog and found it is equally potent in rodent TRPC5 Qpatch assay. Screening for activity against a panel of TRP channels, we also found that GFB-8438 was equipotent against TRPC4 and TRPC5, and showed excellent selectivity against TRPC6, other TRP family members (such as TRPC3/7, TRPA1, TRPV1/2/3/4/5, and TRPM2/3/4/8, data not shown), NaV 1.5, as well as limited activity against the hERG channel. In addition, GFB-8438 did not display significant off-target activity when screened against the safety panels of kinases (59 kinases) and receptors (87 targets).

Table 4. In Vitro Profile of GFB-8438^a.

hTRPC5 Qpatch IC50 (μM)	0.18
hTRPC5 manual patch clamp IC50 (μM)	0.28
rTRPC5 Qpatch IC50 (μM)	0.18
hTRPC4 Qpatch IC50 (μM)	0.29
hTRPC6 Qpatch IC50 (μM)	>30
Nav1.5 manual patch clamp IC50 (μM)	>30
hERG manual patch IC50 (μM)	8.7
Eurofins kinase panel (%ctrl @10 μM)	>50 (LOK 33%)
Eurofins receptor panel (%inh @10 μM)	All <50%
Equilibrium Solubility (pH = 7.4) (μM)	48
Human plasma protein binding (% bound)	69
MDCK-MDR1 PA-B/ER (10–6 cm/s)	20/1.6

^aSee Supporting Information for experimental details.

Based on the observed potency, selectivity, the overall favorable physicochemical properties, and PK profile, GFB-8438 was selected as an attractive tool compound for further testing in in vitro and in vivo efficacy studies.

Thus, we sought to assess the ability of GFB-8438 to elicit a protective effect on podocytes following injury induced by protamine sulfate (PS).^10,20,21 Conditionally immortalized mouse podocytes were incubated with PS, an indirect activator of TRPC5. Extensive actin disruption and loss of stress fibers was observed as demonstrated by the loss of synaptopodin and a reorganization of the phalloidin stained actin cytoskeleton (Figure 3B). In contrast, pretreatment of mouse podocyte with compound GFB-8438, followed by incubation with PS, effectively blocked synaptopodin loss and cytoskeletal remodeling (Figure 3C). This pharmacological effect is consistent with the prior study using tool TRPC5 inhibitor ML204,¹⁰ further validating the critical role of the TRPC5 channel in maintaining podocyte cytoskeletal integrity.

Figure 3.

In vitro efficacy study for GFB-8438 in PS-induced mouse podocytes injury assay. (A) Control mouse podocytes without PS and GFB-8438; (B) mouse podocytes after treatment with PS (300 μM for 15 min); (C) mouse podocytes pretreated with GFB-8438 (1 μM for 30 min) and then insulted with PS. The tiled images were acquired using a Zeiss LSM880 Airyscan super resolution confocal microscope. Green color: Synaptopodin; Red color: Phalloidin; Blue color: Dapi.

Next, GFB-8438 was evaluated in an in vivo deoxycorticosterone acetate (DOCA)-rat model of FSGS. The DOCA-salt hypertensive rat is a well-established model of mineralocorticoid hypertension with renal dysfunction, characterized by increased concentrations of urinary protein and albumin excretion.²² Based on the plasma exposure, we selected the highest tolerable subcutaneous dose of 30 mg/kg QD for this study. As shown in Figure 4, GFB-8438 demonstrated robust benefits in this model. Statistically significant reductions in urine protein concentrations were seen after 3 weeks when compared to the DOCA control group. This trend was even more robust when analyzing the urine albumin concentrations, which were significantly lower at both two- and three-week time points and similar to the eplerenone control group. There was no significant difference in body weight in all treated groups. Animals receiving DOCA-GFB-8438 did not exhibit significant differences in mean arterial blood pressure (BP), diastolic and systolic BP compared to the DOCA group, while significant reduction of BP was observed at both week 2 and 3 in eplerenone treated group. The effect of GFB-8438 on proteinuria without affecting blood pressure contrasts with mechanisms of action of angiotensin-converting enzyme (ACE) inhibitor or an angiotensin II receptor blocker (ARB) frequently used for the treatment of FSGS. These data are consistent with TRPC5 inhibitors such as GFB-8438 having disease-modifying activity, addressing the pathogenesis of FSGS by protecting the integrity of the podocyte cytoskeleton.

Figure 4.

In vivo efficacy study for GFB-8438 in DOCA rat model (n = 8 in sham group, n = 15 in all other groups). Sprague–Dawley rats were unilaterally nephrectomized; after one-week recovery, rats were implanted with a DOCA pellet (45 mg) and provided with high salts-containing tap water for 3 weeks of treatment. DOCA-salt rats received once daily subcutaneous of GFB-8438 at 30 mg/kg or twice daily oral dosing of eplerenone at 50 mg/kg for 3 weeks. Hemodynamic parameters and urinary albumin and protein were measured weekly. (A) total urinary protein excretion at week 0, 1, 2 and 3; (B) total urinary albumin excretion at week 0, 1, 2 and 3; (C) mean arterial blood pressure measured at week 0, 1, 2, and 3. No difference in creatinine clearance across all groups. Data are presented as mean ± SEM.

The pyridazinone GFB-8438 was efficiently prepared in a 3-step synthetic sequence described in Scheme 1 from commercially available starting materials. Standard alkylation of bromide with piperazinone under high temperature, followed by Boc deprotection and S_NAr reaction with 4,5-dichloropyridazinone furnished the synthesis of GFB-8438 in overall 17% yield after recrystallization from acetic acid and water.

Scheme 1. Synthesis of GFB-8438.

Reagents and conditions: (a) NaH (2 equiv), DMF, 0 to 110 °C; (b) TFA (7 equiv), DCM, 0 °C to rt; (c) DIEA (1.5 equiv), DMA, rt to 100 °C.

In summary, starting from HTS screening hit 1, a series of systematic medicinal chemistry SAR explorations led to the identification of GFB-8438, a novel, potent, and subtype selective TRPC5 inhibitor with favorable overall in vitro and in vivo properties. GFB-8438 was shown to protect mouse podocytes from injury induced by protamine sulfate in vitro. More importantly, GFB-8438 demonstrated robust efficacy in the hypertensive DOCA-salt rat model without affecting blood pressure, further validating the critical role of the TRPC5 channel in the pathogenesis of kidney disease.

Progress toward the identification of inhibitors suitable for testing in clinical trials will be reported in due course.

Glossary

Abbreviations

FSGS

focal segmental glomerulosclerosis

CKD

chronic kidney disease

TRP

transient receptor potential

TRPC5

TRP channel subfamily C, member 5

TRPA

TRP channel subfamily A

TRPV

TRP channel subfamily V

TRPM

TRP channel subfamily M

DOCA

deoxycorticosterone acetate

PS

protamine sulfate

CVD

cardiovascular disease

KO

knockout

LPS

lipopolysaccharide

ROS

reactive oxygen species

HTS

high throughput screening

SAR

structure activity relationship

CL

clearance

Vss

volume of distribution

F

bioavailability

T_1/2

half-life

MDCK-MDR1

Madin–Darby canine kidney-multidrug resistance protein 1

ER

efflux ratio

hERG

human ether-a-go-go-related gene

HEK

human embryonic kidney

BP

blood pressure

S_NAr

nucleophilic aromatic substitution

SC

subcutaneous dose

PO

oral dose

DMPK

drug metabolism and pharmacokinetics

hLM

human liver microsomes

rLM

rat liver microsomes

DMF

dimethylformamide

TFA

trifluoroacetic acid

DCM

dichloromethane

DIEA

diisopropylethylamine

DMA

dimethylacetamide.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.9b00430.

• General chemistry methods; general synthetic protocols; synthetic procedures for the preparation of selected compounds;  hTRPC5 fluorescence-based assay (FLIPR) protocols; hTRPC5 automated patch clamp assay (Qpatch) protocol; hTRPC5 manual patch clamp assay protocol; automated patch clamp assay (Qpatch) protocol; hTRPC6 automated patch clamp assay (Qpatch) protocol; Table S1. DMPK profiling of GFB-8438, 32, and 35; Table S2. In vitro activity profile of GFB-8438; PS-induced mouse podocytes injury protection assay; Efficacy study in DOCA-salt rat model; Figure S1. Modeling overlay for compounds 1 and 12; References (PDF)

Author Present Address

^§ (M.B.) Disc Medicine, 400 Tech Square, 10th floor, Cambridge, MA 02139, United States.

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors declare the following competing financial interest(s): M.Y., M.W.L., M.D., G.M., T.T.T., M.C-L.G., X-R.P-Z., A.W.-B., M.B., J.F.R., P.M., and J.-C.H. are current or past employees of Goldfinch Bio Inc.