Discovery of potent small-molecule inhibitors of lipoprotein(a) formation

Compounds

Chemical synthesis and characterization of the compounds were performed as indicated in the Supplementary Information.

Recombinant protein expression

The nucleotide sequence encoding the human LPA gene (NP_005568.2) was inserted into a mammalian expression vector containing a cytomegalovirus promoter. Protein expression was performed by transient transfection of human embryonic kidney 293 (HEK293) cells cultured in serum-free medium. The culture medium was collected five days after transfection and human apo(a) protein was purified using l-lysine affinity chromatography. Nucleotide sequences encoding various K_IV domains of human apo(a) and human and rat plasminogen were inserted into a pET21a Escherichia coli expression vector. Bacterial BL21(DE3) was used as the expression host, and the induction of protein expression was performed in 2× TY medium with 1 mM isopropyl β-d-1-thiogalactopyranoside at 37 °C for 5 h. Cells were collected and stored at –80 °C for subsequent protein purification and refolding. The frozen cell pellets were lysed and inclusion bodies were purified. The protein refolding was performed using the rapid-dilution method, and the refolded proteins were further purified by size-exclusion chromatography. All protein concentrations were determined by A280.

ITC

Experiments were performed on a MicroCal Auto-iTC200 titration calorimeter (Malvern Panalytical), with a cell volume of 200 μl and a 40-μl microsyringe. Experiments were performed at 25 °C while stirring at 750 rpm in ITC buffer (50 mM phosphate buffer at pH 8.0). The microsyringe was loaded with a solution of the respective compounds, with the concentration measured by quantitative nuclear magnetic resonance, and was inserted into the calorimetric cell prefilled with the respective recombinant proteins of known concentration in ITC buffer. The system was equilibrated to 25 °C and an extra delay of 60 s was applied. The first titrations were performed using an initial control injection of 0.5 μl, followed by 18 identical injections of 2 μl, with a duration of 4 s per injection and 150-s intervals between injections. The second titration was performed over the first titration without cleaning the cell and was carried out with continuous injections as an automation method in a MicroCal Auto-iTC200. The second titrations were performed using 19 identical injections of 2 μl, with a duration of 4 s per injection and 150-s injection intervals²⁹. The data were corrected for ligand heats of dilution and were deconvoluted using MicroCal PEAQ-ITC Analysis Software to yield the enthalpy of binding (ΔH) and the binding constant (K). Thermodynamic parameters were calculated using the basic equation of thermodynamics (ΔG = ΔH – TΔS = –RTlnK_d), where ΔG, ΔH and ΔS are the changes in free energy, enthalpy and entropy of binding, respectively; K_d is the equilibrium dissociation constant; T is the absolute temperature (kelvins); and R is the gas constant (1.987 cal mol⁻¹ K⁻¹). A single-binding-site model was used in MicroCal PEAQ-ITC Analysis Software v.0.9.

Lp(a) assembly assay

HepG2 cells (HB-8065, ATCC) were cultured in Dulbecco’s modified Eagle’s medium, supplemented with 10% fetal bovine serum, 20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) and 100 U ml⁻¹ penicillin–streptomycin. apoB-conditioned media were collected after 24 h of culture at 37 °C and 5% CO₂ from confluent wild-type HepG2 cells that endogenously express and secrete apoB. Apo(a)-conditioned media were collected after 24 h of culture at 37 °C and 5% CO₂ from HEK293 cells (CRL-1573, ATCC) that stably express and secrete recombinant human 16K apo(a) protein containing one K_IV type 1, six K_IV type 2 repeats, one each of K_IV types 3–10 and one K_V.

The assembly assay was conducted in 96-well polypropylene plates by combining HepG2 and apo(a)-HEK293-conditioned media in the presence of a range of compound concentrations. After a two-hour incubation period at 37 °C, the reaction was stopped by the addition of 6-aminocaproic acid. Lp(a) was detected using a sandwich enzyme-linked immunosorbent assay (ELISA), with a polyclonal goat anti-apo(a) capture antibody (Abcam ab31675) and a biotin-conjugated, polyclonal goat anti-apoB detection antibody (Abcam ab20898). In brief, ELISA plates were coated with 100 μl of the capture antibody at a 1:12,500 dilution (around 4 μg ml⁻¹) into HEPES-buffered saline (HBS), incubated overnight at 4 °C, washed and blocked by the addition of 200 μl of 1% casein. After removal of the blocking buffer, 20 μl of each of the assembly reactions or of a dilution series of purified human Lp(a) (Athens Research, 12-16-121601), diluted 1:5 into HBS supplemented with 0.5% casein, were added to the ELISA plate. After a 2-h incubation period at room temperature, plates were washed five times with HBS + 0.1% Tween-20. To each well, 100 μl of biotin-conjugated detection antibody was added and allowed to bind for one hour at room temperature. Plates were washed five times with HBS + 0.1% Tween-20, and 100 µl of streptavidin–horseradish peroxidase (HRP) was added to each well, followed by another one-hour incubation at room temperature. Unbound streptavidin–HRP was removed by washing plates five times with HBS + 0.1% Tween-20, and the ELISA was developed using the HRP substrate 3,3’,5,5’-tetramethylbenzidine (TMB). The reaction was stopped after 15 min by the addition of 1N sulfuric acid. Absorbance at 450 nm was measured on an Envision plate reader.

The concentration of Lp(a) formed in each test condition was determined on the basis of the standard curve constructed from the purified human Lp(a) used as a reference standard, and was normalized to high (buffer) and low (no apoB) controls. Data were fitted to a standard Hill equation. Data were analysed in GraphPad Prism v.9.5.1.

Protein crystallography

The K_IV7 and K_IV8 domains (1263–1357 and 1377–1470, respectively, on the basis of GenBank NP_005568.2) were expressed in E. coli BL21(DE3) cells; inclusion bodies were purified by repeated centrifugation and resuspension, then unfolded in 6 M guanidine hydrochloride, diluted in Tris, NaCl, 0.5 M urea and 1.25 mM oxidized and reduced glutathione. After dialysis with Tris pH 8, the protein was purified and concentrated to 22 mg ml⁻¹ (final buffer 50 mM Tris pH 9, 150 mM NaCl and 10% glycerol). For the monovalent compound, tagless K_IV8 was set up in vapour diffusion sitting drops at 21 °C at a ratio of 1.5:1, with a well solution of 100 mM HEPES pH 6 and 1 M tri-sodium citrate pH 7. Crystals appeared within one day and were transferred on the 30th day to a soaking drop containing 10 mM LSN3353871, 100 mM HEPES pH 6 and 1 M tri-sodium citrate pH 7. They were sealed and left at 21 °C for two hours, then transferred to a drop containing 10 mM LSN3353871, 100 mM HEPES pH 6, 1 M tri-sodium citrate pH 7 and 22% glycerol, then collected and flash frozen in liquid nitrogen. For the bivalent compound LSN3441732, K_IV7 was set up with a well solution of 100 mM MES pH 6.5 and 1.6 M magnesium sulfate. After 11 days, the crystal was collected in similar conditions plus 22% glycerol and flash frozen in liquid nitrogen. The trimeric compound, LY3473329, was set up at 2 mM with tagless protein in vapour diffusion sitting drops at 21 °C with a well solution of 100 mM sodium acetate pH 4.5 and 30% PEG 300. After two weeks, a plate-shaped crystal was collected in similar conditions with 22% ethylene glycol and was flash frozen in liquid nitrogen.

Diffraction data were collected at the Lilly Research Laboratories Collaborative Access Team (LRL-CAT) beamline at Sector 31 of the Advanced Photon Source. Crystals stored in liquid nitrogen were mounted on a goniometer equipped with an Oxford Cryosystems cryostream maintained at a temperature of 100 K. The wavelength used was 0.9793 Å, collecting 900 diffraction images at a 0.2° oscillation angle and a 0.12-s exposure time on a Pilatus3 S 6M detector at a distance of 392 mm. The diffraction data were indexed and integrated using MOSFLM 7.0.5 and merged and scaled with Scala 3.3 and Truncate 6.5 from the CCP4 6.5 suite³⁰.

Non-isomorphous data readily yielded the initial structure by molecular replacement using an internal proprietary crystal structure. The initial structure coordinates for the dataset were further refined using REFMAC v.5.8 (CCP4), applying anisotropic temperature factors. Model building was performed with Coot v.0.8 (CCP4) and final structure validation with MolProbity v.4.02 (ref. ³¹) and CCP4 validation tools. See Supplementary Table 1 for crystallographic data statistics.

Protein coordinates and structure factors have been deposited with the Protein Data Bank (https://www.wwpdb.org/) under the access codes 8TCE (LSN3353871), 8V9B (bivalent LSN3441732) and 8V8Z (trimeric LY3473329).

In vivo studies

All procedures were conducted in compliance with the Animal Welfare Act, the Guide for the Care and Use of Laboratory Animals and the Office of Laboratory Animal Welfare (Covance Laboratories for cynomolgus monkey studies and Eli Lilly and Company for mouse and rat studies).

LPA transgenic mouse studies

The LPA gene (GenBank accession code NM_005577) encoding the human apo(a) protein with a signal peptide, one K_IV type 1, six K_IV type 2 repeats, one each of K_IV types 3–10, one K_V and one protease domain was subcloned into a transgenic vector containing a mouse albumin promoter cassette and a human growth hormone polyadenylation signal. The transgenic mouse line was generated by pronucleus injection of the LPA transgenic vector. Eight individual transgenic founder mice were assessed for germline transmission of LPA by genotyping and assessment of the levels of apo(a) in the plasma. Positive founder mice were crossed with ApoB100 transgenic mice (Taconic model 1004), and the resulting mouse line nomenclature (B6.SJL-Tg(APOB)1102Sgy Tg(Alb-LPA)32Arte) was selected on the basis of the confirmation of germline transmission, the identification of LPA and huApoB100 transgenes and the levels of Lp(a) in the plasma. Mice hemizygous for both transgenes were used for pharmacology studies. The studies were conducted in female Lp(a) double transgenic mice (age around 7–17 months). Mice were housed in cages with a standard light cycle (12-h light, 12-h dark), at room temperature (22 ± 4 °C), with a relative humidity of 30–70%. Mice were identified by numbers on the cage cards. After arrival, mice were fed on a normal diet (Harlan Teklad diet 2014). Mice were randomized to treatment groups (n = 5 per group, unless otherwise noted) by body weight and baseline plasma Lp(a) concentration using a block randomized allocation tool (BRAT) for the study. In an example Lp(a) transgenic mouse study in which 66 mice were prescreened for baseline Lp(a) levels, the average Lp(a) level was 45 ± 1.8 µg ml⁻¹ (range 20–78 µg ml⁻¹). Mice were dosed with various compounds in vehicle (10 ml kg⁻¹, 1% hydroxyethyl cellulose (HEC), 0.25% Tween-80, Antifoam) or with vehicle alone as a control orally BID for five days. Blood samples were collected by tail bleeds at designated times, and Lp(a) levels were determined by ELISA by an investigator who was blinded to group allocation. In brief, lipoprotein particles were captured by a goat anti-Lp(a)-antibody-coated plate (Abcam ab31675, diluted 1:12,500 in HEPES-buffered saline), the plates were washed and samples were detected using an HRP-conjugated goat anti-apoB antibody (Abcam ab27622, diluted 1:3,000). Colorimetric peroxidase substrate 3,3’,5,5’-TMB was added, and the reaction was stopped using 1 N sulfuric acid. Absorbance at 450 nm was read on a Molecular Devices SpectraMax plate reader. Blood samples were collected at four hours after dosing on day 5 of dosing to confirm exposure. Full pharmacokinetic (PK) profiles were not evaluated. Concentrations of compounds were evaluated from dried blood spots through a non-Good Laboratory Practice (GLP) liquid chromatography with tandem mass spectrometry (LC–MS/MS) assay. Data were analysed in GraphPad Prism v.9.5.1

Cynomolgus monkey studies

Female or male Chinese Macaca fascicularis cynomolgus monkeys (as described in the figure legends) of unspecified ages with body weight ranging from 2.0 kg to 5.0 kg were housed in cages with a standard light cycle (12-h light, 12-h dark) at room temperature (20–26 °C); the relative humidity ranged from 30% to 70%. Monkeys were given fruits, vegetables or dietary enrichment as a form of environmental enrichment, and various cage enrichment devices. Purina Lab diet 5048C was provided BID and individual monkeys were identified by cage card. The monkeys were randomized to treatment groups by body weight and baseline plasma Lp(a) concentration, as determined by a commercially available Randox assay (Randox LP2757), using a BRAT for the study. In an example cynomolgus monkey study in which 39 monkeys were prescreened for baseline Lp(a) levels, the average Lp(a) level was 350 ± 57 µg ml⁻¹ (range 53–1,667 µg ml⁻¹). Monkeys were dosed orally with vehicle (purified water) or compounds QD or BID for 14 days, at dose levels and frequencies as previously described for each compound. Plasma samples were collected on days before dosing to establish baseline Lp(a) and after the morning dose during the study. Lp(a) levels were measured by an investigator who was blinded to group allocation using the Randox assay, and were quantified relative to the calibrator series (Randox LP3404). Data were analysed in GraphPad Prism v.9.5.1. The Lp(a) percentage change from before dosing in non-human primates was analysed by repeated measures ANOVA in SAS v.9.4 (SAS). At each time point, the Lp(a) percentage change for each treated group was compared with vehicle by the Bonferroni method.

Plasma samples were collected from all monkeys before dosing and at four-hour time points only on days 1, 3, 5, 9 and 14. A full PK profile was obtained in the first three monkeys of each dose group on day 15, collecting samples before dosing and 1, 2, 4, 8, 12, 24, 48 and 96 h after dosing. Concentrations of compounds were evaluated in plasma samples using a non-GLP LC–MS/MS assay. The non-compartmental plasma PK parameters were calculated using Watson (v.7.5).

Radioligand-binding assays

Full-length apo(a) and various shorter constructs were prepared as described. Human and rat plasminogen, purified from plasma by affinity chromatography, were obtained from Innovative Research (IHGPG and IRPLG, respectively).

All reagents used in the binding assays were prepared in assay buffer containing 50 mM Tris-HCl, pH 7.4, and 0.1% bovine serum albumin. All apo(a) proteins were diluted to a final concentration of 200 pM. Human and rat plasminogen were diluted into assay buffer, and 250 ng was added to each well.

Saturation binding studies of ³H-LSN3441732 or ³H-LSN3374443 were conducted by adding to each well of a white-wall, clear-bottom 96-well plate, 50 µl each of (1) dimethyl sulfoxide (DMSO) or a saturating concentration (40 μM) of an unlabelled ligand; (2) protein diluted into assay buffer; (3) resuspended wheat germ agglutinin polyvinyltoluene scintillation proximity assay beads (2 mg ml⁻¹); and (4) a dilution series of radioligand. Plates were incubated for one to two hours at room temperature and counted on a Wallac Trilux 1450 liquid scintillation counter. Non-specific binding, defined as binding in the presence of the unlabelled ligand, was subtracted to determine specific binding. Specific binding was plotted as a function of ligand concentration, and a single-site binding equation of the form y = B_max × x/(K_d + x) was fitted to the data, where B_max is the maximum saturable binding and K_d is the equilibrium dissociation constant.

Human and rat in vitro clot dissolution assay

Platelet-poor plasma was generated from whole blood donated by healthy volunteers, and Sprague Dawley rat citrate plasma was purchased from Innovative Research. One hundred microlitres of plasma was added to a 96-well plate (Corning) and the following were added to the indicated concentrations: 7.5 mM CaCl₂, 117 mM NaCl and 5 µl of test compound in 50% DMSO. Wells were mixed and warmed to 37 °C. A mixture of thrombin (Sigma) and tissue plasminogen activator (Calbiochem) was prepared. Final concentrations in the reaction were 0.05 NIH U ml⁻¹ and 0.5 nM, respectively. The mixture was added and the plate was sealed and placed in a prewarmed plate reader, shaken and read every 2 min for 15 h. The maximum slope for each curve was determined and the percentage inhibition was calculated relative to untreated controls³². Data were analysed in GraphPad Prism v.9.5.1.

Rat Plg mRNA, plasminogen activity assay and plasminogen ELISA

Male Sprague Dawley rats (Charles River, average body weight 310 g, aged eight to nine weeks old) were housed in cages with a standard light cycle (12-h light, 12-h dark), at room temperature (22 ± 4 °C), with a relative humidity of 30–70%. Rats were identified by numbers on the cage cards. After arrival, rats were fed on a normal diet (Harlan Teklad diet 2014) and randomized to treatment groups (n = 5 per group) by baseline plasma plasminogen concentration and activity using a BRAT. LSN3441732 and LY3473329 were administered QD orally at three dose levels (3.7, 12.3 and 37 mg kg⁻¹; 3.2, 10.5 and 31.6 mg kg⁻¹ per dose, respectively) or vehicle (10 ml kg⁻¹, 1% HEC, 0.25% Tween-80, Antifoam) for four days. Terminal blood was drawn under general anaesthesia 24 h after the day-4 oral dose. Plasma samples were collected via the abdominal aorta into a premeasured anticoagulant BD Vacutainer used for blood collection (2.7 ml, buffered sodium citrate). The liver was flash frozen in liquid nitrogen.

RNA was isolated from the liver using the PureLink RNA Mini Kit (Invitrogen) and quantified using the NanoDrop spectrophotometer (Thermo Fisher Scientific). Complementary DNA (cDNA) was prepared using the High Capacity cDNA Reverse Transcriptase Kit (Applied Biosystems). Quantitative PCR was conducted using TaqMan reagents and primers (Plg Rn00585167_m1, Gapdh Rn99999916_s1). Relative quantitation was calculated using the comparative Ct method. Statistical significance was evaluated using JMP v.16.1.0 software (SAS).

To measure plasminogen activity, 100 µl plasma per well was added to a 96-well plate (Corning) and the following were added to the indicated concentrations: 7.5 mM CaCl₂ and 117 mM NaCl. Wells were mixed and warmed to 37 °C. Urokinase (Millipore) was added at 125 U per well and the plate was incubated at 37 °C for 4 min. The plasminogen substrate Chromogenix S-2251 (DiaPharma) was added to a concentration of 3.3 mM. The mixture was added and the plate was sealed and placed in a prewarmed plate reader, shaken, and the absorbance read every 30 s for 2 h. The maximum slope for each curve was determined and the percentage inhibition was calculated relative to untreated controls. Plasminogen protein was measured using ELISA kits for rat (Abcam ab157740) according to the manufacturer’s instructions. Plasminogen activity (percentage) and mass (µg ml⁻¹) in rat were analysed by ANOVA in JMP 16.1.0. The percentage reduction of plasminogen activity and mass for each treated group was compared with vehicle by Dunnett’s method.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.