Study design
The research protocol and data collection were performed in accordance with the guidelines set out in the Declaration of Helsinki of the World Medical Association. The Medical Ethics Committee of Zhujiang Hospital, Southern Medical University (approval number: 2022-KY-219-01), reviewed and approved all procedures. All participants read and understood the study’s nature, purpose, risks, and procedures and signed an informed consent form. The following data were collected: age, sex, medication history within the previous 6 months, medication allergies, ocular medication history within the previous month, diabetes duration, ocular disease, surgical history, and other systemic disease histories. Ophthalmic examination included basic examination of the eye, ocular surface disease index (OSDI) score, comprehensive analysis of the ocular surface, tear ferning test, and conjunctival impression cytology. All participating researchers were qualified ophthalmologists.
Study participants
We prospectively included 33 eyes with diabetic retinopathy treated in the Department of Endocrinology and Department of Ophthalmology of Zhujiang Hospital of Southern Medical University between October 2022 and February 2023. The case group included 22 eyes with DK, and the control group included 11 eyes without DK. The inclusion criteria for the case group were as follows: diabetic retinopathy and a ≥ 5-year history of diabetes; abnormal corneal perception and corneal epithelium in one or both eyes; and positive corneal fluorescein staining, with more than five spots in the examined eye. The inclusion criteria for the control group were as follows: diabetic retinopathy and a ≥ 5-year history of diabetes; no abnormalities in corneal perception or corneal epithelium; and negative corneal fluorescein staining, with no more than five fluorescein staining spots. Additionally, we excluded patients with hypersensitivity to sodium fluorescein; skin mucinosis, autoimmune diseases, and systemic infectious diseases; eye surgery or eye trauma within 3 months; lacrimal punctum embolisation, ocular infection, and use of atropine, antihistamines, β-blockers, systemic oral contraceptives, and diuretics within 1 month; severe meibomian gland dysfunction and severe dry eyes; and use of artificial tears within 2 h.
OSDI score and corneal sensory examination
All patients completed the OSDI questionnaire, and the scores were calculated. Corneal sensation was then assessed using a disposable cotton swab to make a cotton thread for touching the cornea between the patient’s lower pupillary border and corneal limbus thrice, while avoiding the patient’s line of sight during the inspection. Sensation was recorded as ‘normal’, with obvious tactile sensation and blinking action in all three instances; ‘decreased’, with tactile sensation in only one or two instances but no blinking action; or ‘disappeared’, with no tactile sensation in all three instances and no blinking action. An assessment of ‘decreased’ or ‘disappeared’ sensation indicated abnormal corneal sensitivity.
Tear sample extraction and analysis
Tear samples for mass spectrometry analysis were collected using Schirmer strips, which were placed on the middle and outer one-third of the lower eyelid for 5 min. Samples for the tear ferning test were collected using a disposable microcapillary tube, with 1 µL of tear fluid obtained from the lower eyelid tear river or conjunctival sac. All tear samples were added to the bottom of an Eppendorf tube, immediately frozen, and stored at − 80°C until analysis.
Tear ferning test
For analysis, the tear fluid was thawed and centrifuged at 20°C. The tears were evenly spread on a clean glass slide, dried at 20℃ for 15 min, observed under a microscope, and graded according to the Rolando grading method.17
Ocular surface analysis
Ocular surface analysis was performed using a corneal topography instrument (Oculus Company, Germany), and the data were recorded. The observed lipid layer of the tear film was scored based on the following features: colour (rich [0 points] or single [1 point]), lipid layer distribution (uniform [0 points] or uneven [1 point]), and lipid layer thickness (normal [0 points], thin [1 point], or thick [2 points]). The meibomian glands were analysed for deformation or absence. The absence of upper and lower meibomian glands was classified as follows: normal (0 points), missing < 1/3 (1 point), > 1/3 missing < 2/3 (2 points), or missing > 2/3 (3 points). Meibomian gland deformations were defined as straight (0 points) or tortuous (1 point). Meibomian gland opening was described as without blockage (0 points) or with blockage (1 point). After ocular surface analysis, corneal fluorescein staining and first tear breakup time examinations were performed.
Conjunctival imprint cytology
Before specimen collection, proparacaine hydrochloride eye drops were instilled in the eye. The rough surface of an acetate filter paper was then placed on the bulbar conjunctiva at the corneal edge of the superior temporal quadrant and evenly pressed with a glass rod for 10 s. The specimens were carefully stripped using edentulous forceps and immediately fixed in 4% paraformaldehyde tissue fixation solution after slight drying at 20°C. The fixed specimens were stored at 4℃ and stained with periodic acid-Schiff within 24 h. The periodic acid-Schiff staining results were graded according to the Nelson method.18
Protein sample extraction
The Schirmer strips containing tear samples were retrieved from the − 80°C freezer, a lysis buffer (8 M urea, 1% protease inhibitor) was added to the samples, and ultrasonic lysis and centrifugation were performed at 12 000 rpm for 10 min at 4°C to remove cell debris. After centrifugation, the supernatant was transferred to a new centrifuge tube, and the protein concentration was determined using a BCA kit (KeyGEN BioTECH Company, China). Then, the tear samples were mixed randomly into five samples with similar concentrations according to the case group and control group.
Trypsinization
Protein (200 µg) was obtained from each sample for enzymatic hydrolysis. The volume was adjusted to the same using lysis solution, and dithiothreitol was added to achieve a final concentration of 5 mM and reduced at 56℃ for 30 min. Next, iodoacetamide was added to achieve a final concentration of 11 mM and incubated at 20℃ in the dark for 15 min. The alkylated sample was transferred to an ultrafiltration tube and centrifuged at 12 000 g for 20 min at 20℃. Then, 8 M urea was used to replace urea thrice; next, a replacement buffer was used to replace urea thrice. Trypsin was added in a 1:50 ratio (protease:protein, m/m) and hydrolysed overnight at 37℃. Centrifugation (12 000 ×g) was performed for 10 min at 20℃ to recover the peptide segment. The peptide segment was recovered once with double distilled water, and the two peptide segment solutions were merged for later use. Finally, the combined peptides were desalted using the Strata X SPE column.
Four-dimensional mass spectrometry
The peptide segment was dissolved in liquid chromatography mobile phase A and separated using the NanoElute ultra-high-performance liquid chromatography system. Mobile phase A constituted an aqueous solution containing 0.1% formic acid and 2% acetonitrile, and mobile phase B was an acetonitrile aqueous solution containing 0.1% formic acid. The liquid phase gradient setting was as follows: 0–70 min, 6–24% B; 70–84 min, 24–32% B; 84–87 min, 32–80% B; and 87–90 min, 80% B; the flow rate was maintained at 450 nL/min. The peptide segments were separated using an ultra-high performance liquid chromatography system and injected into a capillary ion spray for ionisation, followed by timsTOF Pro mass spectrometry analysis. The ion source voltage was set to 1.7 kV, and the peptide parent ions and their secondary fragments were detected and analysed using high-resolution TOF. The scanning range of the secondary mass spectrometry was set to 100–1700. Data were collected in the parallel accumulative serial fragmentation mode. After collecting primary mass spectrometry data, the primary mass spectrometer was operated in the parallel accumulative serial fragmentation mode 10 times to collect secondary spectra with the charge number of the parent ion in the range of 0–5. The dynamic exclusion time for tandem mass spectrometry scanning was set to 30 s to avoid repeated scanning of the parent ion.
The resulting tandem mass spectrometry data were processed using the MaxQuant search engine (v.1.6.15.0). Tandem mass spectra were searched against the human SwissProt database (20 389 entries) and concatenated with a reverse-decoy database. The enzyme digestion method was set to Trypsin/P, the number of missed cut positions to 2, the minimum length of the peptide segment to 7 amino acid residues, the maximum number of peptide modifications to 5, and the mass tolerance for the precursor ions to 20 ppm and 20 ppm in the first and main search, respectively. The mass tolerance for the fragment ions was set to 0.02 Da. Carbamidomethylation on cysteine was specified as a fixed modification, while acetylation at the protein N-terminus and oxidation on methionine were specified as variable modifications. The false discovery rate was adjusted to < 1%. Identifying proteins requires at least one unique peptide segment.
Differential protein screening
Based on the correction intensity of each protein obtained from the mass spectrometry database search in different samples, the relative quantitative values of the sample in different samples can be obtained by converting the correction intensity of each protein in different samples through a formula (Formula 1).
Formula 1: Relative quantitative conversion formula
Rij = Iij/mean (Ij)
Note
i, j, I, and R represent the sample, protein to be converted, correction intensity of the protein obtained after searching the library, and relative quantitative value, respectively.
Log2 conversion was performed on the relative quantitative values of proteins to ensure that the test data conformed to the normal distribution. A t-test was then performed on the relative quantitative values of each protein in various groups to calculate the P-value and determine the significance of the difference (Formula 2). A P-value of < .05 indicated that the difference was significant.
Formula 2: T-test for relative quantitative values of proteins in each group
Pk = t-test (Log2 (Rik, i∈1), Log (Rik, i∈0))
Note
P, R, i, I, k, 1, and 0 represent the P-value, relative quantitative value, sample, protein to be calculated, case group, and control group, respectively.
Next, the following formula was used to calculate the ratio of the mean relative quantitative values of each protein in the sample in different groups as the difference multiple. The difference multiple of a certain protein between the case (1) and control groups (0) was calculated (Formula 3).
Formula 3: Difference multiple calculation formula
FC1/0, k = Mean (Rik, i∈1) /Mean (Rik, i∈0)
Note
FC, R, i, k, 1, and 0 represent the multiple differences, relative quantitative value, sample, protein to be calculated, case group, and control group, respectively.
A P-value of < .05 and a difference multiple of > 1.5 were considered a significant increase, and a difference multiple of < 1/1.5 was considered a significant decrease.
Enrichment and protein interaction network analyses
The identified differential proteins were annotated using the Eggnog-mapper (v2.1.6) software with Gene Ontology (GO) terms for biological processes, cellular components, and molecular functions. Fisher’s exact test was used to analyse the significance of GO enrichment for differentially expressed proteins, and the difference was considered significant at P-values of < .05. The identified proteins were enriched in the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway.
We compared the screened differential proteins with a difference multiple of > 1.5 using the STRING (v11.5) protein interaction network database, extracted protein interactions with a confidence score of > 0.7, selected the top 50 most closely interacting proteins, and visualised them using R language software. The predicted proteins were screened to generate protein interaction graphs.
Statistical analyses
All statistical analyses were performed using SPSS 25.0. Measurement data are presented as mean ± standard deviation and were compared using the independent sample t-test. Categorical data are presented as proportions and were compared using Fisher’s exact test. The absence of a meibomian gland, conjunctival imprinting cytology, and lacrimal fern tests were examined using the Mann–Whitney U test. Differences were considered significant at P-values of < .05.