This study protocol was reviewed and approved by the Institutional Review Board (IRB) of Seoul National University Bundang Hospital (IRB No: B-2105-682-105). In this study, a retrospective analysis was conducted on the collected clinical data and blood samples from the previous Bundang AMD Genetic Study⁹. The current study was conducted in accordance with the Declaration of Helsinki. A written informed consent was obtained from all participants in the secondary study.

Participants

Participants aged over 50 years who were newly diagnosed with intermediate AMD and nAMD between October 1, 2003, and December 31, 2016, were enrolled in the AMD group in our study. Participants aged over 50 years who were newly diagnosed with cataracts, without any retinal pathology between October 1, 2003, and December 31, 2016, were enrolled as the control group in our study. Intermediate AMD was defined as the presence of multiple large drusen (> 125 µm) on the macula with or without pigmentary changes, the absence of pre-existing central geographic atrophy (GA), and nAMD based on the international AMD classification¹². nAMD was defined as the evidence of choroidal neovascularization (CNV) associated with non-drusenoid retinal pigment epithelium detachment, serous sensory retinal detachment, subretinal hemorrhage, and/or subretinal exudation. We collected basic demographic information, including age and sex, at the first visit. Smoking status was classified as non-smoker or former/current smoker.

DNA-based genotyping of CFH SNPs

Patient DNA was obtained from peripheral blood using a DNA extraction kit (QIAamp DNA Maxi kit, Qiagen Inc.). Multiplex polymerase chain reaction using a single base extension technique was performed for SNP genotyping using the iPLEX Gold kit and MassARRAY Typer software version 4.0 (Sequenom, San Diego, CA). Two SNPs known to be major risk alleles for AMD in Koreans, namely CFH rs1061170 and rs800292, were analyzed⁹.

Selection of target tryptic peptides

For proteomic genotyping, it is necessary to determine how the SNPs of the CFH gene are translated into proteins. Two SNPs of the CFH gene (rs1061170 and rs800292) were retrieved from NCBI dsSNP (http;// www.ncbi.nlm.nih.gov/snp) and the amino acid sequence of CFH was obtained from Uniport knowledgebase (Uniprot ID: P08603) (http;//www.uniprot.org/). CFH SNPs rs1061170 and rs800292 result in a change in codon assignment from tyrosine (Y) to histidine (H) at amino acid positions 402 and from isoleucine (I) to valine (V) at position 62. For the selection of target peptides in the MRM assay, tryptic peptides unique to each protein isoform were identified from the UniProt database using the BLAST search tool (https://www.uniprot.org/blast/) (Table 1). Stable isotope-labeled standard (SIS) peptides with the same amino acid sequence as the target peptides, but incorporating C-terminal [¹³C₆, ¹⁵N₄] arginine (> 95% purity) were purchased from JPT Technologies (Berlin, Germany). The SIS peptides are used as quantitative standards in a mass spectrometer as they are physicochemically identical to the target peptides and have only a difference in mass. For each peptide, amino acid analysis was performed, and the absolute amount of peptide was provided by the vendor. The SIS peptides were re-solubilized in 20% acetonitrile with 0.1% formic acid to prepare a 1 mg/mL stock solution. Two other dilution series (50 and 100 ng/mL) were prepared to optimize the MRM assay. All solutions were stored at − 80 °C until use.

Optimization of MRM assay

We experimentally optimized most of the empirical parameters of the MRM assay by directly infusing the SIS peptides into a 5500 Qtrap mass spectrometer equipped with a turbo v ion source (SCIEX, Foster City, CA). First, the intensities of the doubly and triply charged ions (initial Q1) were scanned, and the optimal declustering potential (DP) voltage was determined by ramping the DP, and the dominant precursor ions (Q1) were then measured with the optimal DP voltage (final Q1). Second, at the optimized Q1 and DP parameters, we initially measured 10 highly intense product ions of a precursor ion using the base collision energy (CE) and then ramped the CE voltages to find the best CE values. Subsequently, we obtained the optimal collision cell exit potential (CXP) values of the selected 10 product ions. All optimized data were collected and compared to the theoretical spectra, and three highly intense y-ions were used for subsequent MRM assays (Table 1 and Supplementary Fig. S1).

Sample preparation for MRM assay

The plasma (2 µL) sample was buffered with 50 µL of 8 M urea in 50 mM ammonium bicarbonate (ABC). The sample was treated using 7 µL of 100 mM tris (2-carboxyethyl) phosphine and 7 µL of 200 mM chloroacetamide at 37 °C for 1 h in the dark. Urea was diluted to 1 M with 50 mM ABC before trypsin (Promega, Madison, WI, USA) digestion at 1:50 enzyme: substrate ratio and incubated at 37 °C for 16 h with mixing on a shaker at 600 rpm. Formic acid was added at a final concentration of 0.5% to stop the digestion reaction. The mixture was desalted using a Sep-Pak tC18 96-well plate (Waters, Milford, MA, USA) and dried using a vacuum centrifuge (miVac Duo Concentrator; Genevac, Suffolk, UK). Dried samples were stored in a deep freezer at − 80 °C until use.

LC-MRM-MS analysis

Peptide separation was performed using an ACQUITY UPLC M-class (Waters). Mobile phase A was 0.1% formic acid in the water, and mobile phase B was 0.1% formic acid in acetonitrile. Samples were reconstituted with 40 µL of SIS peptide mixture in mobile phase A, injected with a full sample loop injection of 5 µL, and separated in an ACQUITY UPLC peptide CSH C18 column (1 mm i.d., 10 cm length, pore size 130 Å, particle size 1.7 µm; Waters). A gradient with a flow rate of 20 µL/min and 5% B for 3 min, 5–25% B for 22 min, 25–60% for 1 min, 60–60% for 1 min, 60–5% B for 1 min, and 5% B for 5 min, followed by column washing with 80% B for 5 min, 80% to 5% for 6 min, 5–80% for 6 min, and re-equilibration with 5% B for 13 min. MRM analysis was performed using a 5500 Qtrap mass spectrometer. The MS detection was carried out in positive mode with the following parameters: ion spray voltage of 5500 V, curtain gas at 20 psi, nebulizer gas at 25 psi, heating gas at 20 psi, resolution at 0.7 Da (unit) for Q1/Q3, interface temperature at 400 °C, and scan mass range 300–1250 m/z QQQ mode. Quantification experiments were performed using the scheduled MRM mode. The MRM detection window time was 120 s, and the cycle time was 1.5 s. The mass spectrometer was operated with Analyst software (Version 1.6.2, SCIEX), which generated MRM-MS data with the file name *.wiff.

Quantitative MRM assay

The concentration of the endogenous target peptide can be calculated by comparing the peak areas of the peptide and the spiked SIS peptide resulting from the LC-MRM-MS analysis. Peak areas were integrated into the Analyst software by setting a three points Gaussian smooth width, 30 s RT window, 500 cps minimum peak height, and one point peak splitting factor. For an accurate quantification, analytical validation was performed using calibration curves, specificity, precision and accuracy¹³. The calibration curves were used to predict the concentration of targets and to determine the limit of quantification (LOQ), the lowest concentration within a coefficient of variation (CV) of ≤ 20% with a coefficient of determination (R²) of 0.99 (Supplementary Fig. S2). Specificity was calculated by comparing endogenous analytes and internal standards in double blank (DB), specificity analytes (S.A), blank and low QC samples. Under optimized MRM conditions, no interference was observed for each sample in the target peptide (Supplementary Table S1). To evaluate precision and accuracy, we prepared four types of quality control (QC) samples of different concentrations. For intra- and inter-day assays, each concentration level was tested for three consecutive days in five replicates. Both intra- and inter-day QC samples at each concentration for all four analytes were acceptable (Supplementary Table S2 and S3).

Statistical analysis

The statistical differences in age, sex, smoking status, and the genotype of CFH SNP rs1061170 and rs800292 between the normal controls and AMD groups were evaluated using the Chi-square test and Mann–Whitney test. The correlation between the plasma concentration of tyrosine and histidine and the genotypes of CFH SNP rs1061170 was evaluated using Kruskal–Wallis test. The correlation between the plasma concentration of valine and isoleucine and the genotypes of CFH SNP rs800292 was also evaluated using Kruskal–Wallis test. Scatter plots were drawn to visualize the correlation between the plasma tyrosine/histidine (Y/H) ratio and the genotypes of CFH SNP rs1061170 and between plasma valine/isoleucine (V/I) ratio and the genotypes of CFH SNP rs800292. Cut-off values of plasma CFH variant ratios were set to best delineate the CFH genotypes, and the correlation between MRM-based and DNA-based genotypes of CFH was evaluated. Statistical significance was defined as a P value of < 0.05. All statistical analyses were performed using SPSS (version 24.0; SPSS Inc., Chicago, USA).