In conclusion, these data offer strong preclinical support for DF-003 as a potent and targeted inhibitor for the root cause of ROSAH syndrome. The investigational drug DF-003 has completed human clinical testing in a Phase 1 trial (NCT05997641) and is now entering a Phase 1b trial in adult ROSAH patients (NCT06395285). In this Phase 1b trial, outcome measures including safety, pharmacokinetics, pharmacodynamics, eye evaluations, headache, and quality of life will be assessed.
All animal studies were approved by the Institutional Animal Care and Use Committees (IACUC) at Shanghai Yao Yuan Biotechnology Ltd., Zhejiang Yao Yuan Biotechnology Ltd. and Oujiang Laboratory under protocol numbers DF00320231017-001 and DF00320231109-001. All animals received humane care in accordance with the national guidelines for housing and care of laboratory animals (Ministry of Health, Beijing, China).
Both male and female mice were used to conduct the experiments reported herein, and no specific or consistent differences were observed as a function of sex.
Adenosine-diphosphate-D-glycero-β-D-manno-heptose (ADP-D-Heptose), used in the TR-FRET kinase, radiolabeled kinase, and ADP-Glo kinase assays, as well as uridine-diphosphate-β-D-mannose (UDP-mannose) used in the radiolabeled kinase assay, were synthesized in-house following previously published synthesis routes. D-glycero-D-manno-6-fluoro-2,3,4,7-Ac-heptose-1β-S-ADP (DF-006) used in cell-based assays, was also generated in-house and has been characterized previously. Adenosine-diphosphate-L-glycero-β-D-manno-heptose (ADP-L-heptose) used in surface plasmon resonance (SPR) experiments was purchased from InvivoGen (tlrl-adph-l, Hong Kong, China).
Please see the Supplementary Notes, 1-2 in the Supplementary Information.
Recombinant ALPK1 proteins used for thermal shift assay (TSA), TR-FRET (time-resolved Förster resonance energy transfer)-based kinase assay, radiolabeled kinase assay, and SPR assay, as well as the recombinant ALPK1[T237M] protein used for the radiolabeled kinase assay were purified from Sf9 insect cells. DNA sequences coding for human ALPK1 (NP_079420.3) and ALPK1[T237M] with 3 × Flag-tag (MDYKDHDGDYKDHDIDYKDDDDK) sequences added to the C-terminus of the protein were codon optimized for optimal expression in insect cell lines. The DNA sequence was cloned into the pFAST-BAC vector for the generation of a baculovirus used to transduce insect Sf9 cells to overexpress these recombinant proteins. Recombinant ALPK1 used in a Western blotting-based kinase assay for verification of the TSA screen was purified from human 293-F cells. The same codon-optimized ALPK1 coding sequence, with 3 × Flag tags at the N-terminus, was cloned into the pcDNA3.1 vector. 293 F cells were transiently transfected for 48 h, while in the last 4 h 4% Salmonella typhimurium (s.t.) lysate was added to the media. Recombintant proteins used for site-directed mutagenesis analysis to validate homology model were purified from expi293 cells without s.t. lysate stimulation. Sf9, expi293, or 293-F cells were homogenized in 50 mM Tris, 0.5 M NaCl, 5% glycerol, pH 8.0; proteins were bound to anti-Flag affinity beads, washed, and eluted using 3 × Flag peptides in the same solution. The coding sequence for human TIFA was synthesized and cloned into the pET28a (+) vector, with 6×His-tag or HA-6×His-tag sequences having been added to the 5' end of the coding sequence. These vectors were used to transform Escherichia coli. After lysing the bacteria, recombinant His-tagged and HA-His-tagged TIFA were both purified with a Ni column and were finally dissolved in a buffer containing 50 mM Tris and 150 mM NaCl (pH 8.0) after desalting. His-TEV-tagged Mouse ALPK1 (NP_082084.1) kinase domain aa. 949-1231 (mALPK1-KD) and human ALPK1 N-terminal regulatory domain aa. 1-446 (ALPK1-ND) used for crystallography were purified from E. coli. using a similar approach except that the His-tag was cleaved with a TEV protease.
The mouse kinase domain (mALPK1-KD; amino acids 949-1231) and human N-terminal regulatory domain (ALPK1-ND; amino acids 1-446) were mixed and allowed to form a complex and further purified using size exclusion chromatography and concentrated to 10 mg/ml. A co-crystal was obtained using a microtiter plate format from a well containing a solution of 12% PEG 3350 and 0.1 M cadmium bromide tartrate polyethylene glycol (CBTP), pH 8.2, with 0.1% seed crystal, by hanging drop vapor diffusion. The X-ray diffraction data was collected by Synchrotron (Shanghai Synchrotron Radiation FacilityBeamline BL17U) at the wavelength of 0.979 Å. The data was processed using the programs DENZO and SCALEPACK. The phase was determined by SAD (Single-wavelength Anomalous Dispersion), and automatic model building was performed in PHENIX. The rest of the model was manually built with Coot and refined in PHENIX.
Homology modeling of ALPK1 was performed using the Prime module in Schrödinger Maestro Suite 2021-02 following standard knowledge-based model building methods. From the crystal structure of ALPK1 and Dictyostelium myosin II heavy chain kinase A (3PDT), a prototypical member of the atypical alpha-kinase family of kinases with a conserved domain organization similar to ALPK1, the ATP binding pocket was built by identifying domain movement, building a chimera structure and remodeling the activation loop (under the OPLS4 force field). This model was optimized prior to docking using the protein preparation workflow in Schrodinger Maestro Suite 2021-02. Energy minimization of the protein structure was carried out using OPLS4. Molecular dynamics (MD) simulations for the generation of multiple protein conformations were used to address flexibility in the binding site. The model was then refined by induced fit docking (Induced Fit Docking Panel, standard protocol, using the OPLS_2005 force field) to generate grid-box coordinates for docking and validated by chemistry structure-activity relationship (SAR) analyses. The optimized grid was used for docking analysis (Glide module, Standard precision mode, using the OPLS_2005 force field and setting the hydrogen bond interaction of hinge binding as a constraint) of DF-003 and ATP.
Molecular dynamics (MD) simulation results can provide insights into protein-binding interactions between molecules and proteins. For this study, a 200 ns MD simulation was performed for the ALPK1-DF-003 complex using the Desmond module from Schrodinger Release 2021-02, the optimized potentials for liquid simulation (OPLS4) force field at pH 7.4. Before conducting this MD simulation, the complex was solvated in SPC solvent within an orthorhombic box and counterions (Na, Cl) were added to maintain a 0.15 M salt concentration. The simulation was completed under a bar pressure of 1.01325 and a constant temperature of 300 K, with a mainlining recording interval of 200 ps (Supplementary Notes 1, Table 8). The structural stability of DF-003 within the ALPK1-DF-003 complex was assessed using the root mean square deviation (RMSD) of protein and DF-003. The RMSD of the protein was calculated based on atom selection, and changes of the order of 1-3 Å were considered acceptable for a stable structure. The RMSD of the ligand was measured after the protein-ligand complex was first aligned to the protein backbone. If the observed value was not significantly larger than the RMSD of the protein, then the ligand had not diffused away from its initial binding site and the structure was considered stable. The simulation was performed three times under the same conditions to verify the consistency of the observed results.
A small molecule library containing ~160,000 compounds (HTS Diversity V2) was prepared by WuXi AppTec (Shanghai, China) in 384 well thin-wall qPCR plates, with 20 nL of 10 mM compound preparations in each well. On each screening plate, 2 columns of wells were loaded with 20 nL of pure DMSO to leave spaces for negative controls (no compounds) and the positive control (AS-252424). Using a Multidrop™ Combi (Thermo Fisher Scientific, Waltham, MA, USA), 3 mixtures were added to each well sequentially and mixed: 5 µl of compound solvent (50 mM Tris, 500 mM NaCl, 2% DMSO, pH 8.0) to compound wells and negative control wells, 3 μl of recombinant full-length ALPK1 kinase purified from Sf9 cells diluted in 50 mM Tris, 500 mM NaCl pH 8.0, and 2 µl of 5× Protein Thermal Shift™ Dye (Thermo Fisher Scientific, 4461146). The final concentration of ALPK1 was 150 µg/mL (~1.1 µM), and the compound concentration was 20 µM. The mixture was spun down at 1000 rpm for 1 min, then immediately applied to a protein melting reaction on a Roche LightCycler® 480 II Real-time PCR system, heating the mixture from 30 °C to 50 °C at the ramp rate of 0.02 °C/s while continuously acquiring the fluorescence signal. The melting temperature (T) of the ALPK1 protein with or without each compound was calculated using the Protein Thermal Shift™ Analysis Desktop Software v1.2 (Thermo Fisher Scientific). For positive control wells, 40 μM AS-252424 in 5 μl compound solvent was added manually and used as a positive control in the thermal shift screening. AS-252424 was discovered to cause a 3.4 °C increase in T in a pilot TSA screening using a kinase inhibitor library (EFEBIO, Shanghai, China). Binding of AS-252424 and 6 other hits from pilot screening to ALPK1 was confirmed in surface plasmon resonance (SPR) experiments (described below) to validate the TSA method. Compounds that caused an increase in T > 1 °C were considered to be hits and were tested in a Western blotting-based kinase assay and/or a TR-FRET kinase assay (described below) to investigate whether the binding leads to the inhibition of ALPK1 kinase activity and to measure the relative inhibitory potency of each inhibitory compound for the SAR study.
SPR assays were conducted on a Biacore 8 K (Cytiva). A Series S NTA chip (Cytiva) was first conditioned with 350 mM EDTA for 60 s at 30 µL/min, activated with 0.5 mM NiCl for 60 s at 10 µL/min, and then activated with a 50/50 mixture of 0.04 M EDC and 0.1 M NHS (final concentrations 0.02 M and 0.05 M, respectively) for 420 s at 10 µL/min. Recombinant ALPK1 (purified from Sf9 cells, final concentration 120 μg/mL, 857 nM) in immobilization buffer (10 mM HEPES pH 7.4, 150 mM NaCl, 0.05% Tween-20, 0.5 mM TCEP) was injected at 5 µL/min with a contact time of 1000 s. The chip was then deactivated with 1 M ethanolamine-HCl pH 8.5 for 420 s at 10 µL/min and washed with 350 mM EDTA for 60 s at 30 µL/min to remove Ni. For analyte testing, flow rates were maintained at 30 µL/min. An ADP-L-heptose 500 nM solution was prepared in running buffer (10 mM HEPES pH 7.4, 150 mM NaCl, 0.05% Tween-20, 0.5 mM TCEP, 1% DMSO). Analyte was 2-fold serially diluted in the same running buffer containing 500 nM ADP-L-heptose, with a maximum final concentration of 1 μM for DF-003 and 50 μM for selected TSA chemical hits. Analyte binding was measured using an A-B-A method, beginning with a 50 s injection of 500 nM ADP-L-heptose ("A") followed by injection of analytes ("B") and dissociation of analytes in B. For K measurements, the association time was 80 s and the dissociation time was 120 s. The data collection rate was 10 Hz, and all data were double referenced and solvent corrected. Analysis was performed using Biacore Insight Evaluation software version 5. The 1:1 kinetic binding affinity models were used to fit the data, and all parameters were fitted globally.
Compounds were tested at a single final concentration of 20 μM in an in vitro kinase assay including 30 nM recombinant ALPK1 (purified from 293-F cells stimulated with 4% s.t. lysate 4 h before cell lysis) and 1.6 μM His-tagged TIFA in a 20 µl volume. Recombinant ALPK1 produced in this fashion exhibited kinase activity such that it was able to phosphorylate TIFA without any additional agonist. An in vitro kinase reaction was conducted at 27 °C for 40 min in the following buffer: 20 mM HEPES pH 7.0, 1 mM TCEP, 10 mM MgCl, 0.1 mM NaVO, 0.6 μM BSA, 5% DMSO and 25 μM ATP. Reactions without compound (replaced with pure DMSO) and reactions with neither compound nor ATP were regarded as negative control (0% inhibition) and positive control (100% inhibition), respectively. After these in vitro kinase reactions, the kinase-substrate mix was fractionated by electrophoresis on denaturing SDS-PAGE gels and transferred to Immobilon-P PVDF membranes (Millipore, MA, USA). Membranes were blocked with 5% milk in TBS with 0.1% Triton-X100. Phosphorylated TIFA were detected by incubating the membrane with an anti-phospho-threonine (#9386, Cell Signaling Technology, Danvers, MA, USA) antibody.
In each well of a 384-well plate, three mixtures were added sequentially: 2 µL of blank control or serially diluted compounds (inhibitory binders identified from the TSA screen or compounds following SAR optimization) in 5% DMSO, 4 µL of substrate/kinase mix (final concentrations of 260 nM recombinant HA-His-tagged TIFA, 2 nM recombinant ALPK1, 50 mM HEPES, 10 mM MgCl, 1 mM EGTA, 2 mM DTT, 0.01% (v/v) Tween-20, pH 7.5), and 4 µL of ATP/ADP-D-heptose mix (final concentrations of 6.25 µM ATP and 20 nM ADP-D-heptose). The 10 µL kinase reactions were allowed to proceed for 1.5 h at 27 °C. Kinase reactions were terminated by adding 10 µL of the HTRF KinEASE detection kit buffer (Cisbio, Bedford, MA, USA; 62SDBRDF) containing monoclonal anti-HA-Eu cryptate (Cisbio; 610HAKLB) and monoclonal Anti HA-d2 to each well. TIFA phosphorylation levels were monitored through the detection of oligomerization signals generated upon TIFAsome assembly. Both donor and acceptor antibodies recognize TIFA and will generate a TR-FRET signal when bound to the same TIFA oligomer formed upon TIFA phosphorylation. After incubation at 26 °C for 1 h, fluorescent signals at 620 nm and 665 nm, for europium donor and d2 acceptor fluorescence, respectively, were detected using a Tecan INFINITE NANO+ microplate reader (Tecan, Männedorf, Switzerland). The TR-FRET signals were calculated as the ratio of the acceptor and donor emission signals with the formula below:
Ratio = Signal at 665 nm/ Signal at 620 nm × 10,000
ALPK1 kinase activity was calculated as follows: % Activity = 100% × (TR-FRET signal of test compound - mean TR-FRET signal of no ATP 1% DMSO control) / (mean TR-FRET signal of 1% DMSO control - mean TR-FRET signal of no ATP 1% DMSO control).
In multi-dose assays, compound activity was calculated via logarithmic interpolation. Concentration-response curves were fitted with a four-parameter logistic nonlinear regression model and the IC values for these compounds were calculated in GraphPad Prism 6. In early studies, eltrombopag and tolcapone were identified as hits from a pilot TSA screen of an FDA-approved drug library and were found to inhibit ALPK1 activity in the Western blotting-based assay described above. Both compounds also abolished the TR-FRET signal with IC of 7.6 μM and 18 μM, respectively, validating the TR-FRET assay. In one case, the inhibitory potency of DF-003 was tested at increasing ATP concentrations ranging from 0.15625 μM to 25 μM.
To identify additional ALPK1 inhibitor pharmacophores, a second high-throughput screen was conducted using a small molecule library containing ~200,000 compounds (HTS Diversity V3) prepared by WuXi AppTec in 384-well plates, with 10 nL of 10 mM compound preparations in each well. The TR-FRET kinase assay was conducted as described above, except that 4 µL of 2.5% DMSO was first added to each well to dilute the compounds therein, followed by 4 µL of substrate/kinase mix (final concentrations of 260 nM recombinant HA-His-tagged TIFA, 2 nM recombinant ALPK1) and 2 µL of ATP/ADP-D-heptose mix (final concentrations of 6.25 µM ATP and 20 nM ADP-D-heptose). Sixteen repeats of negative control [vehicle (1% DMSO) with ATP], 16 repeats of positive control (vehicle without ATP) and 16 repeats of reactions containing 20 µM compound 23 (Supplementary Notes, 1 Table 3) used as positive control were conducted on each screening plate. Compounds that inhibited >40% of ALPK1 kinase activity at a final concentration of 10 µM were considered as hits and applied to subsequent dose-dependent potency analyses and SAR optimization.
Assays were conducted by Reaction Biology (Malvern, PA, USA). To measure DF-003 IC value for ALPK1, recombinant human ALPK1 (0.5 nM), human TIFA (10 µM), and ADP-D-heptose (5 nM) were mixed in a kinase reaction buffer (20 mM HEPES (pH 7.5), 10 mM MgCl, 1 mM EGTA, 0.01% Brij35, 0.02 mg/mL BSA, 0.1 mM NaVO, 2 mM DTT, 1% DMSO) to which 3-fold dilutions of DF-003 (1 µM - 50.8 pM; prepared in DMSO) were added. After the addition of [P]-ATP (20 µM; specific activity: 0.01 µCi/µL final), the kinase reaction was allowed to proceed at room temperature for 60 min, and samples were then spotted onto P81 ion exchange paper (#3698-915, Whatman). After extensive washing with 0.75% phosphoric acid, the radioactive phosphorylated substrate remaining on the filter paper was measured to quantify ALPK1 kinase activity, which was expressed as the percentage of remaining ALPK1 kinase activity relative to vehicle (DMSO, no DF-003) conditions. Curve fitting and IC value calculations were performed using GraphPad Prism 4. The same kinase assay was used to measure DF-003 IC values for ALPK1[T237M], except that ALPK1 was replaced by ALPK1[T237M] and ADP-D-heptose was replaced by 10 μM UDP-mannose. The time-course of TIFA phosphorylation by ALPK1[T237M] agonized by 5 nM ADP-D-heptose or 10 μM UDP-mannose was measured by detecting radioactive phosphorylated substrate on the filter paper at 20, 40, and 60 min after reaction initiation in the same kinase reaction condition without addition of DF-003.
For assessing kinase activity of ALPK1 and its single amino acid substitution mutants, the ADP-Glo (Promega, V9101) reagent was used following the manufacturer's instructions, with recombinant human ALPK1, ALPK1-GLU1137A, ALPK1-GLU1137K, or ALPK1-TYR1133A (all used at 2.5 nM), human TIFA (1.7 µM), and ADP-D-heptose (20 nM). The Km of ATP for each enzyme was first determined by measuring the velocity of ADP production under various ATP concentrations ranging from 2 μM-500 μM. The IC of DF-003 for each enzyme was determined by adding half-log10 serial dilutions of DF-003 (5 µM - 158 pM; prepared in DMSO) with ATP added to the kinase reaction at the Km concentration (16 μM for ALPK1, 113 μM for ALPK1-GLU1137A, 33 μM for ALPK1-GLU1137K, and 49 μM for ALPK1-TYR1133A). Kinase reactions were conducted at 28 °C for 40 min. Curve fitting and Km and IC value calculations were performed using GraphPad Prism 4.
The inhibitory effects of DF-003 on the activity of 394 human kinases were measured by Reaction Biology using the previously described HotSpot™ Kinase Assay. The general experimental approach was the same as the in vitro kinase assay detailed above, with an extended 120-min reaction time, an ATP concentration of 10 µM, and a single tested DF-003 dose (10 µM). The percentage of kinase activity was calculated for each kinase with DMSO and DF-003 treatment. The dose-dependent inhibitory activity of DF-003 against the top 3 non-ALPK1 human kinases that were most strongly inhibited by 10 µM DF-003 was further determined in kinase reaction assays using a 10-dose dilution series of DF-003 (20 µM-0.763 nM), with all other experimental conditions being the same as above. Additionally, a 10-dose dilution series of staurosporine and other enzyme-appropriate compounds were included as positive control inhibitors.
THP-1 cells (Cell Bank of The Chinese Academy of Sciences, Beijing, China) were routinely cultured in RPMI-1640 (Hyclone, Logan, UT, USA) containing 10% heat-inactivated fetal bovine serum (FBS, Hyclone), 0.05 mM 2-mercaptoethanol, 1% penicillin (100 U/mL), and streptomycin (100 µg/mL) (Gibco, Waltham, MA, USA) in a humidified 5% CO incubator at 37 °C. To induce macrophage-like differentiation, these cells were seeded in 24-well plates (4×10/well) and treated for 48 h with phorbol myristate acetate (PMA; 50 ng/mL). Media was then exchanged for fresh complete medium without PMA and pretreated with a range of DF-003 concentrations for 2 h (0.3 nM - 1000 nM), after which they were stimulated for an additional 4 h using 5 nM DF-006. Cells were then collected for qPCR to analyze the expression of TNF and CXCL8 (See below). Samples were analyzed in quadruplicate, and IC values for the inhibition of ALPK1 agonist-induced cytokine upregulation were calculated by plotting relative mRNA expression against DF-003 concentration for each gene of interest using GraphPad Prism 6.
HEK-293 cells were purchased from the Cell Bank of The Chinese Academy of Sciences and routinely cultured in DMEM/High-Glucose (Hyclone). For TIFAsome analyses, the coding sequence for the GFP-TIFA fusion protein was cloned into the pcDNA3.1 plasmid. HEK-293 cells cultured on coverslips were transiently transfected with the construct, treated with either 1 µM or 200 nM DF-003 (or DMSO as a vehicle control), followed by 1 μM DF-006 stimulation for 1 h. Each condition was conducted in sextuplicate. Cells were fixed with 4% paraformaldehyde and stained with DAPI (Shanghai Zhenghuang, Shanghai, China; C1005). Fluorescent images were acquired using a Motic upright fluorescent microscope (PA53 BIO FS6). For each coverslip, five 2560 μm × 1760 μm fields were imaged and analyzed by a researcher blinded to treatment conditions, calculating the percentage of TIFAsome-positive cells among all positive cells exhibiting green fluorescence.
To obtain HEK-293 cells stably overexpressing ALPK1 or ALPK1[T237M], codon-optimized DNA coding sequencing for Flag-tagged human ALPK1 and ALPK1[T237M] were inserted into the pcDNA3.1 vector. Plasmids were transfected into HEK-293 cells using Lipofectamine 2000 as per the manufacturer's protocol. At 48 h post-transfection, G418 (#A1720, Sigma-Aldrich, St. Louis, MO, USA) was used to select resistant transformed cells, and single-cell clones were prepared through a limiting dilution-based approach. Western blotting (See Below) was used to assess Flag-tagged ALPK1 and ALPK1[T237M] overexpression. Cells were treated in triplicate with a range of DF-003 concentrations (0.64 nM - 20 µM) for a total of 30 h, during which media was refreshed with equivalent DF-003 doses after 24 h. Cells were then harvested to analyze the expression of CXCL10, TNF, and CXCL8 by qPCR (See Below). CXCL8 concentrations in supernatants collected from these cells were analyzed with a Human CXCL8 ELISA Set (#555244, BD Biosciences, Franklin Lakes, NJ, USA) based on the manufacturer's instructions, and results were analyzed with a microplate reader (Multiskan FC, Thermo Fisher Scientific).
To prepare stable NF-κB reporter cells, HEK-293 cells were transfected with an NF-κB dual reporter construct (a kind gift from Dr. Lei Sun of Fudan University). The construct contains an NF-κB responsive element (5'-GGGAATTTCCGGGAATTTCCGGGAATTTCCGGGAATTTCCGGGAATTTCCGGGAATTTCCGGGAATTTCCGGGAATTTCC-3'), a minimized CMV promoter, followed by coding sequences for secreted alkaline phosphatase (SEAP) and firefly luciferase linked by a 2 A peptide. G418 (#A1720, Sigma-Aldrich) was used to select resistant transformed cells, and single-cell clones were prepared through a limiting dilution-based approach. Positive clones were confirmed by increased alkaline phosphatase activity and luciferase activity after stimulating cells with DF-006. To measure increased NF-κB signal caused by ALPK1[T237M] and DF-003's inhibitory effect, reporter cell clones were transfected with ALPK1 or ALPK1[T237M] expression vectors. Four hours after transfection, cells were treated with serially diluted DF-003 for an additional 48 h Intracellular luciferase activity was measured using Luciferase Reporter Gene Assay Kit (Beyotime, Jiangsu, China; RG027) per the manufacturer's instructions.
RNA was extracted from harvested tissue and cell samples using the TRI reagent (T9424, Sigma-Aldrich). The HiScript Q-RT SuperMix (Vazyme, Nanjing, China) was then used to synthesize cDNA. Real-time PCR was carried out using a QuantStudio™ 5 Real-Time PCR System (Applied Biosystems) with the AceQ qPCR SYBR Green Master Mix Kit (Vazyme). All qPCR assays were run in 384-well thin-wall plates with a total reaction volume of 10 µL, including 2 µL of cDNA (20 ng), 5 µL of 2x AceQ qPCR SYBR Green Master Mix, 0.2 µL of ROX Reference dye (50x), 0.1 µL of each primer (F + R, 10 µM of each), and 2.6 µL of ddHO. Thermocycler settings included an initial 5 min at 95 °C followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s. Relative gene expression was analyzed using the 2 method, and GAPDH served as a normalization control for human cell lines and Rpl13 for murine tissues. Data are presented as the fold-change in expression relative to vehicle control unless otherwise noted. Primers used for these analyses are presented in Supplementary Table 3.
Mice were housed in a specific pathogen-free (SPF) animal facility at Drug Farm (Shanghai, China) in individually ventilated cages on a 12 h light/dark cycle with a bedding of wood shavings and ad libitum access to rodent chow. The facility was maintained at a temperature of 20 °C to 26 °C with humidity between 40% and 70%. Six 8 week-old male C57BL/6J mice from Zhejiang Vital River Laboratory Animal Technology Co., Ltd. (Zhejiang, China) were randomly assigned to 2 groups (3 animals/group). Animals were administered DF-003 by oral gavage once daily at 3 and 5 mg/kg for 10 consecutive days. DF-003 were dissolved in 0.1% (w/v) methyl cellulose (MC, 1500 cP) + 0.2% (v/v) Tween 80 in purified water. The oral dosing volume for all animals was 10 mL/kg body weight. Drinking water and certified rodent diet were available to animals ad libitum, except that animals were fasted 16 h prior to the last administration until 4 h after the last dosing. Blood samples at 0 (right before the last dosing), 0.5, 1, 2, 4, 6, 8, 12, and 24 h after the 10 dosing were collected from the tail vein and transferred into tubes containing K-EDTA. The tubes were gently inverted several times to ensure mixing and immediately placed on wet ice. Plasma was obtained by centrifugation at 3200 × g at 4 °C for 10 min and stored at ≤ -60 °C until subsequent analysis. Twenty-four hours after the 10th dosing, the animals were euthanized by CO inhalation. The eyes were harvested, and the retinas and optic nerves were separated. In addition, a piece of cerebral cortex tissue ( ~ 50 μg) from each mouse was snap-frozen in liquid nitrogen. The concentrations of DF-003 in the plasma, retina, and cerebral cortex of each mouse were measured by Wuhan Haipu Biomed Inc. (Wuhan, China) using an LC-MS/MS approach. The concentration of DF-003 in the optic nerve was also measured using this same strategy, combining all 6 optic nerves from the 3 mice in each dosing group as a single sample.
Alpk1 mice (with removal of exon 13 to abolish the kinase activity of endogenous mouse ALPK1) were generated previously. To prepare ROSAH model mice using CRISPR, a guide RNA (gRNA) (5'-AGTGAGGACCAGCGGTGCAGAGG-3') targeting exon 2 of the mouse Alpk1 gene (NCBI Gene ID: 71481) was generated. Targeting recombination vectors containing the following elements were prepared: (1) a 1.5 kb 5' arm homologous to the genomic sequence upstream of the start codon of mouse Alpk1 (including ATG); (2) the human ALPK1 coding sequence (encoding wild-type ALPK1 for the mAlpk1 allele or ALPK1[T237M] for the mAlpk1 allele, the latter of which contains the same C > G nucleotide substitution as in human ROSAH syndrome patients) with a poly-A tail; (3) a 1.4 kb 3' arm homologous to the genomic sequence downstream of the start codon for mouse Alpk1 (excluding ATG). In vitro transcribed gRNA and Cas9 mRNA, together with targeting recombination vectors, were co-microinjected into C57BL/6N Alpk1 mice-derived fertilized eggs. Founders with the hALPK1 CDS inserted into the mouse Alpk1 gene were screened via Southern blotting and further confirmed by PCR and Sanger sequencing and were backcrossed with Alpk1 mice.
Mice bearing the abovementioned alleles were intercrossed to generate founder mAlpk1 mice on the C57BL/6N background (Crb1). These mice were backcrossed with wild-type C57BL/6J (Crb1) mice (purchased from Zhejiang Vital River Laboratory Animal Technology Co., Ltd.) twice to respectively obtain mAlpk1; Crb1 and mAlpk1; Crb1 mice. These lines were intercrossed to generate mAlpk1; Crb1 (designated as hALPK1-KI) and mAlpk1; Crb1 (designated as hALPK1[T237M]-KI) mouse lines. Both lines exhibit loss-of-function for mouse Alpk1. Sixteen to seventeen-week-old hALPK1-KI and hALPK1[T237M]-KI littermates from the mating between these two lines were used for testing DF-003 efficacy in vivo. Our mice included the introduction of the wild-type Crb1 gene from C57BL/6J mice to address the rd8 mutation in this gene present in the C57BL/6N subline that can confound efforts to analyze ocular phenotypes.
To test the in vivo efficacy of DF-003 in ROSAH model mice, female hALPK1-KI and hALPK1[T237M]-KI mice (16-17 weeks of age; n = 24/genotype) were each randomized into vehicle (n = 8/genotype), DF-003 3 mg/kg (n = 8/genotype), and DF-003 5 mg/kg (n = 8/genotype) treatment groups, ensuring that there were no differences in starting body weight among groups. Mice were dosed orally with DF-003 (3 or 5 mg/kg) or vehicle control (0.1% MC) once per day for 10 total doses at a dosage volume of 10 mL/kg body weight. The status and body weights of all mice were monitored daily throughout this study. Twenty-four hours after the final dose, one retina, one optic nerve, and brain cortex samples were harvested from each mouse and were immediately homogenized in the TRI reagent (Sigma-Aldrich) and stored at -80 °C for analyses of gene expression. One eyeball from each mouse was fixed with 4% paraformaldehyde, and the retina was dissected and embedded in O.C.T. compound (Sakura Finnetek, Torrance, CA, USA; 4583) followed by storage at -80 °C.
Mouse retina samples were fixed in 4% paraformaldehyde (PFA), mounted in O.C.T embedding compound, and frozen at -20 °C to -80 °C, after which they were cut into 10 μm transverse sections near the center of the retina (through the optic papilla).
For immunofluorescent and immunohistochemical labeling, frozen sections were air-dried and then blocked using goat serum diluted in PBS containing 0.1% Triton X-100. The microglia were detected by immunofluorescence using primary rabbit anti-IBA1 (Wako, Richmond, VA, USA; 019-19741) and secondary goat anti-rabbit IgG (H + L) Cross-Adsorbed, Alexa Fluor™ 488 (Invitrogen, Carlsbad, CA, USA; A-11008). Nuclei were counterstained with DAPI (Shanghai Zhenghuang, Shanghai, China; C1005). The astrocyte activation marker GFAP was detected by immunohistochemistry with rabbit anti-GFAP (Abcam; ab68428) and secondary donkey anti-rabbit IgG H&L (HRP) (Abcam; ab6802). ImmPACT® DAB Peroxidase (HRP) Substrate (Vector Labs, Burlingame, CA, USA) was used for staining, followed by diaminobenzidine (DAB) coloration according to routine immunohistochemistry procedures, yielding a brown signal corresponding to GFAP positivity. Cell nuclei were counterstained with hematoxylin (Baso, Zhuhai, China) and dyed blue.
For hematoxylin and eosin (H&E) staining, cells were initially stained with hematoxylin, followed by differentiation and eosin staining. Images were acquired using a Motic upright fluorescent microscope (PA53 BIO FS6) and analyzed in the CaseViewer software (Version 3.3; 3DHISTECH Ltd., Budapest, Hungary).
For quantification, the Lasso tool was used to draw a curve aligned with the nerve fiber layer of the retina and measured as the length of the retina. The GFAP-positive nerve fiber layer length was measured, summed, and normalized to the overall length of the retina. Total IBA1-positive cells were counted in the whole outer nuclear layer and inner nuclear layer areas of analyzed retinal sections and also normalized to the length of the retina.
Testing for reduced sweating in experimental mice was performed as described in a previous study. A hind paw of a male hALPK1-KI or hALPK1[T237M]-KI mouse was painted with 2% (w/v) iodine/alcohol solution and allowed to dry. The surface was then painted again with 1:1 starch-castor oil 1:1 (w:v). Purple spots formed by sweating were observed 3-10 min thereafter.
The optomotor response (OMR) test was conducted to assess the visual acuity of unrestrained mice, following the protocol described previously. Mice were placed on a central platform within the qOMR system (PhenoSys, Berlin, Germany), which was enclosed by four screens that were positioned on each wall of a square box. During the test, visual stimulation was presented on the screens. The spatial frequency of the visual pattern varied across a range of frequencies from 0.1 to 0.5 cycles per degree, with increments of 0.05 cycles per degree. For each trial, the visual pattern moved at a constant speed of 12 degrees per second for a duration of 60 s. Throughout this process, the qOMR system automatically tracked and analyzed the head movements of the mice as they attempted to follow the movement patterns. After the test, the system evaluated the accuracy of the mice's tracking behaviors, categorizing them as either correct or incorrect. These data were then comprehensively analyzed to derive the crucial OMR index and determine the visual threshold, as described previously. Initially, we extracted the peak value of the OMR index from the fitted curves. Subsequently, we subtracted 1 from this maximum value and calculated one-quarter of the resulting difference to establish a threshold point. Finally, we aligned this calculated threshold point with the OMR index curve obtained from the experiment to identify the corresponding spatial frequency value. This spatial frequency value represented the visual threshold of the mice, indicating the maximum spatial frequency stimulus that they could accurately track.
OCT images were obtained using the OPTOPROBE system (Optoprobe, Glamorgan, UK). Mice were anesthetized by intraperitoneal injection of 1% pentobarbital sodium (0.0075 mL/g), and their pupils were dilated with compound tropicamide eye drops. Cross-sectional images of the retina, centering around the optic nerve head, were then captured.
Mice were anesthetized by intraperitoneal injection of 1% pentobarbital sodium and their pupils dilated as described above. The fundus was imaged using the Eyemera fundus camera (IIScience, Busan, South Korea).
Female hALPK1-KI and hALPK1[T237M]-KI mice were treated orally with vehicle or DF-003 (5 mg/kg) once per day for 10 days. Twenty-four hours after the last dose, the indicated plasma cytokine and chemokine levels were measured using the ProcartaPlex Kit (eBioscience, Thermo Fisher Scientific) as directed.
Unless otherwise noted, data are presented as means ± standard error of the mean (SEM). Statistical outliers were detected using a Grubb's test and excluded from further analysis and clearly indicated in Source Data files. Data were compared using independent sample t-tests or one-way ANOVAs, as appropriate, and the specific statistical tests used are indicated in the corresponding figure legends. Animal numbers are as noted, and each point in the individual figures corresponds to one animal. A p-value < 0.05 served as the significance threshold.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.