Tear Proteomic Analysis of Patients With Type 2 Diabetes and Dry Eye Syndrome by Two-Dimensional Nano-Liquid Chromatography Coupled With Tandem Mass Spectrometry

Abstract

Purpose.: Diabetesmellitus has been shown to be associated with and complicated by dry eye syndrome. We sought to examine and compare the tear film proteome of type 2 diabetic patients with or without dry eye syndrome and normal subjects using two-dimensional nano-liquid chromatography coupled with tandem mass spectrometry (MS)–based proteomics.

Methods.: Tears were collected from eight type 2 diabetes patients with dry eye syndrome, eight type 2 diabetes patients without dry eye syndrome, and eight normal subjects. Tear breakup time (BUT) was determined, and tear proteins were prepared and analyzed using two-dimensional strong cation-exchange/reversed-phase nano-scale liquid chromatography MS. All MS/MS spectra were identified by using SEQUEST against the human International Protein Index (IPI) database and the relative abundance of individual proteins was assessed by spectral counting.

Results.: Tear BUT was significantly lower in patients with diabetes and dry eye syndrome than in patients with diabetes only and normal subjects. Analysis of spectral counts of tear proteins showed that, compared to healthy controls, patients with diabetes and dry eye syndrome had increased expression of apoptosis-related proteins, like annexin A1, and immunity- and inflammation-related proteins, including neutrophil elastase 2 and clusterin, and glycometabolism-related proteins, like apolipoprotein A-II.

Conclusions.: Dry eye syndrome in diabetic patients is associated with aberrant expression of tear proteins, and the findings could lead to identification of novel pathways for therapeutic targeting and new diagnostic markers.

Introduction

Dry eye syndrome is a multifactorial condition characterized by eye irritation symptoms, blurred and fluctuating vision, tear film instability, increased tear osmolarity, and ocular surface epithelial syndrome. ^1–4 Diabetes mellitus has been shown to be associated with and complicated by dry eye syndrome. Diabetic patients frequently complain of symptoms typical of dry eye syndrome, such as burning and foreign body sensation. ^5,6 The manifestations of diabetes mellitus resultant from lacrimal gland and ocular surface dysfunctions appear to be related to dry eye syndrome. ^7,8 Manaviat et al. ⁹ showed that approximately half (54.3%, 108/199) of diabetic patients suffered from dry eye syndrome, and there was a significant association between dry eye syndrome and duration of diabetes. Jin et al. ¹⁰ also showed that patients with type 2 diabetes tend to have tear film dysfunction. Dry eye can lead to vision deficit, scarring and perforation of the cornea, and secondary bacterial infection. If this syndrome is diagnosed at the early stage and treated promptly, undesirable complications can be avoided or minimized. Therefore, early diagnosis and treatment of dry eye syndrome in type 2 diabetic patients is very important. 

Tears contain a variety of substances, including proteins, lipids, mucins, salts, and other organic molecules, and tear proteins are believed to have key roles in protecting the ocular surface from external insults, such as pathogenic infections, and promoting ocular wound healing. Recently, electrospray ionization (ESI) mass spectrometry (MS)/MS was used to identify some novel species in the tears. ^11–14 The number of proteins found in the tear film continues to grow, and Fung et al. ¹⁵ reported that approximately 500 proteins were detected and unambiguously identified by liquid chromatography (LC)/MS/MS However, there still is lack of agreement in the literature regarding the number of proteins in the tear film and the functions of individual proteins. Lactoperoxidase, lysozyme C, and elastase 2 are found to protect the ocular surface against antimicrobial infection or inflammation, and some other proteins, like cathepsin B, are related to ocular surface wound healing, or stability promoting through interaction with other ligands. ^16,17 In our study, we sought to examine and compare the tear film proteome of type 2 diabetic patients with or without dry eye syndrome and normal subjects using two-dimensional nano-LC coupled with tandem MS (2D-LC-nano-MS/MS)–based proteomics. 

Subjects and Methods

Subjects

We included in our study eight patients with type 2 diabetes with dry eye syndrome and eight patients with type 2 diabetes without dry eye syndrome. Type 2 diabetes was diagnosed according to the criteria of the World Health Organization (WHO). ¹⁸ Dry eye syndrome was diagnosed by a modified Dry Eye Workshop classification ¹⁹ and was considered present if Shirmer I test < 5 mm and tear breakup time (BUT) < 5 seconds, and absent if Shirmer I test > 10 mm and BUT > 10 seconds. A subject was included if he or she was aged between 50 and 60 years, and if he or she had type 2 diabetes of more than 5 years. A subject was excluded if he or she used eye drops within one month before the study, had a history of ocular surgery, had systemic immunologic syndromes, or had any ocular syndromes other than dry eye syndrome. In addition, eight age- and sex-matched healthy subjects were included as normal controls. The study protocol was approved by the Institutional Review Board of Shanghai Tenth People's Hospital and all study participants provided written informed consent. The study was carried out in accordance with the Declaration of Helsinki. 

Tear Sample Collection and Preparation

Tears were collected by placing a Schirmer strip over the lower lid approximately 6 mm from the lateral canthus. The lid was not anesthetized. The subjects were instructed to keep eyes closed during the 5-minute test. After wet length was recorded, the strip was placed in a 1-mL amber Eppendorf tube. The samples were placed immediately in ice transport tanks for transfer to the Chinese National Human Genome Center at Shanghai, where the samples were stored at −80°C until processed. Proteins were extracted from the Schirmer strip by incubation in 1 mL Tris buffer (pH = 8.3, 7 M urea, 5 mM ethylenediaminetetraacetic acid [EDTA], 1 mM phenylmethylsulfonyl fluoride [PMSF], and 0.5 mM dichlorodiphenyltrichloroethane [DTT]) at 4°C for 10 hours. After centrifugation at 12,000g for 10 minutes, the supernatants from all patients within each group were pooled and precipitated by acetone as described previously. ²⁰ Precipitated proteins were resuspended in 100 μL Tris buffer and quantitated using the BCA assay kit (Pierce, Rockford, IL). The normalized protein content in the tear samples was calculated using the following formula:  

 

 

Then, 12.5 μg protein from each group were incubated with 25 mM ammonium bicarbonate and 10 mM DTT for 60 minutes at 56°C followed by incubation with 25 mM indoleacetic acid (IAA) for 30 minutes in the dark at room temperature. Trypsin digestion was done at 37°C for 12 hours. The resultant peptides were dried and saved at −80°C. 

Two-Dimensional Strong Cation-Exchange/Reversed-Phase Nano-Scale LC MS

Extracted peptides were desalted using a 1.3 mL C18 solid phase extraction column (Sep-Pak Cartridge; Waters, Milford, MA) and dried using a vacuum centrifuge (Eppendorf, Westbury, NY), and then resuspended with loading buffer (5 mM ammonium formate containing 5% acetonitrile, pH 3.0), separated, and analyzed by 2D-LC-nano-MS/MS. The experiments were performed on a Nano Aquity UPLC system (Waters) connected to an LTQ Orbitrap XL mass spectrometer (Thermo Electron, Bremen, Germany) equipped with an online nano-electrospray ion source (Michrom Bioresources, Auburn, CA). A 180 μm × 2.4 cm SCX column (Waters), which was packed with a 5-μm PolySULFOETHYL Aspartamide (PolyLC, Columbia, MD) was used for the first dimension. For recovery of hydrophobic peptides retained on the SCX column after a conventional salt step gradient, an RP step gradient from 15% to 50% acetonitrile was applied to the SCX column. ¹⁹ A 9-μL plug was injected each time to form the step gradients (Table 1). The SCX column finally was cleaned with 1 M ammonium formate (NH₄FA) twice. The plugs were loaded onto the SCX column with a loading buffer at a 4 μL/min flow rate for 4 minutes. A 9-μL peptide sample was loaded onto the SCX column before the gradient plugs were injected. Eluted peptides were captured by a trap column (Waters), while salts were diverted to waste. The trap column (2 cm × 180 μm) was packed with a 5-μm Symmetry C18 material (Waters). The RP analytical column (20 cm × 75 μm) was packed with a 1.7-μm Bridged Ethyl Hybrid (BEH) C18 material (Waters), and was used for the second dimension separation. 

Table 1

 

Step Gradients for Peptide Recovery

SCX Gradients	NH4FA, mM, and ACN, %
1	5, 5% (sample loading)
2	100, 5%
3	200, 5%
4	250, 5%
5	300, 5%
6	400, 5%
7	500, 5%
8	500, 15%
9	500, 30%
10	500, 50%
11	1000
12	1000

 ACN, acetonitrile.

Peptides on the RP analytical column were eluted with a three-step linear gradient, starting from 5% B to 45% B in 40 minutes (A, water with 0.1% formic acid; B, acetonitrile with 0.1% formic acid), increased to 80% B in 3 minutes, and then to 5% B in 2 minutes. The column was reequilibrated at initial conditions for 15 minutes. The column flow rate was maintained at 300 nL/min and column temperature was maintained at 35°C. The electrospray voltage of 1.1 kV versus the inlet of the mass spectrometer was used. 

An LTQ Orbitrap XL mass spectrometer (Thermo Electron) was operated in the data-dependent mode to switch automatically between MS and MS/MS acquisition. Survey full-scan MS spectra with two microscans (m/z 300–1800) were acquired in the Obitrap with a mass resolution of 60,000 at 400 m/z, followed by 10 sequential LTQ-MS/MS scans. Dynamic exclusion was used with two repeat counts, 10-second repeat duration, and 60-second exclusion duration. For MS/MS, precursor ions were activated using 35% normalized collision energy at the default activation q of 0.25. 

Data Analyses

All MS/MS spectra were identified by using SEQUEST (v.28 [revision 12]; Thermo Electron) against the human International Protein Index (IPI) database (IPI human v3.45 fasta with 71983 entries). To reduce false-positive identification results, a decoy database containing the reverse sequences was appended to the database. The searching parameters were set up as follows: partial trypsin (KR) cleavage with two missed cleavages was considered, the variable modification was oxidation of methionine, peptide mass tolerance was 20 parts per million (ppm), and fragment ion tolerance was 1 Da. The Trans Proteomic Pipeline software (revision 4.0; Institute of Systems Biology, Seattle, WA) then was used to identify proteins based upon corresponding peptide sequences with ≥95% confidence. The peptides results were filtered by the Peptide Prophet ²¹ with a P value over 0.95 and a Protein Prophet ²² probability of 0.95 was used for the protein identification results. Spectral counts (SC) correlate with protein abundance. ²³ The relative abundance of individual proteins was assessed by spectral counting, in which we counted how many times the unlabeled version of a protein was identified by the fragmentation spectra of its peptides. Additionally, we normalized SC using the following formula:  

 

 

For analysis of the difference of abundance among the proteomes, proteins were divided into three groups according to SC: the high-abundance group (SC ≥ 50), the medium-abundance group (20 < SC < 49), and the low-abundance group (SC < 20). Significant upregulation was defined as an SC ratio ≥3 and SC distance ≥5; significant downregulation was defined as an SC ratio ≤0.33 and SC distance ≤−5. The SC ratio was calculated from the equation  

 

and SC distance was calculated using the following formula:  

 

 

Statistical Analysis

The χ² tests and Student's t-tests were performed using the SPSS statistical software (SPSS, Inc., Chicago, IL) and a P value of 0.05 or less was considered significant. 

Results

Demographic and Baseline Characteristics of the Study Subjects

Demographic and baseline characteristics of the study subjects are summarized in Table 2. The subjects of the three groups were matched in age and sex. Patients with diabetes and dry eye syndrome had a median duration of diabetes of 11.37 (range, 7–20) years, which was markedly longer than that of patients with diabetes only (median, 9.25 years; range, 5–15 years, P < 0.01). The best corrected visual acuity in patients with diabetes and dry eye syndrome (OD, 0.38 ± 0.25; OS, 0.53 ± 0.38) was markedly lower than that of patients with diabetes only (OD, 0.78 ± 0.32; OS, 0.78 ± 0.34, P < 0.01). Tear BUT was significantly lower in both eyes in patients with diabetes and dry eye syndrome (OD, 3.38 ± 3.07; OS, 2.75 ± 3.24) than that of patients with diabetes only (OD, 9.25 ± 5.52; OS, 8.25 ± 5.73, P < 0.01). Tear secretion as revealed by the Schirmer I test also was markedly reduced in patients with diabetes and dry eye syndrome (OD, 2.5 ± 2.39; OS, 4.0 ± 6.70) compared to that of patients with diabetes only (OD, 7.0 ± 5.24; OS, 0.63 ± 0.74, P < 0.01). 

Table 2

 

Demographic and Baseline Characteristics of the Study Subjects

	Healthy Subjects, n = 8	Patients With Diabetes, n = 8	Patients With Diabetes and Dry Eye Syndrome, n = 8
Age, y
Median	61.75 ± 6.61	59.75 ± 6.23	62.87 ± 5.72
Range	51–68	51–68	50–67
Sex
Male	4	4	4
Tear breakup time, s
OD	11.75 ± 3.85	9.25 ± 5.52*	3.38 ± 3.07†
OS	13.0 ± 3.93	8.25 ± 5.73*	2.75 ± 3.24†
Schirmer I test, mm/5 min
OD	15.0 ± 6.55	7.0 ± 5.24*	2.5 ± 2.39†
OS	13.88 ± 6.60	6.38 ± 3.54*	4.0 ± 6.70†
Fluorescent test
OD	0.00 ± 0.00	0.75 ± 0.87*	9.63 ± 4.01†
OS	0.00 ± 0.00	0.63 ± 0.74*	11.63 ± 4.24†
Best corrected visual acuity
OD	0.96 ± 0.17	0.78 ± 0.32*	0.38 ± 0.25†
OS	0.85 ± 0.28	0.78 ± 0.34*	0.53 ± 0.38†
Duration of diabetes, y
Median	0	9.25 ± 5.34	11.37 ± 6.05
Duration	0	5–15	7–20
Blood glucose, mmol/L	4.80 ± 0.31	7.88 ± 1.36*	8.69 ± 1.22†

*  Comparison of patients with diabetes only to healthy subjects (P < 0.01).

†  Comparison of patients with diabetes and dry eye syndrome to healthy subjects (P < 0.01).

Patients With Diabetes and Dry Eye Syndrome Have Increased Normalized Protein Content in Tears

We determined further the normalized protein content in tear samples. The normalized tear protein content in patients with diabetes and dry eye syndrome was 1.19 ± 0.7, which was markedly higher than that of patients with diabetes only (0.27 ± 0.13, P = 0.001) and normal healthy subjects (0.20 ± 0.14, P = 0.001, Fig. 1). 

Figure 1

 

Mean normalized protein concentration in tears in healthy subjects (group A), patients with diabetes (group B), and patients with diabetes and dry eye syndrome (group C).

Tear Proteomic Characteristics of Patients With Diabetes and Dry Eye Syndrome

The 2D-nano-LC-MS/MS identified a total of 357 unique proteins from 4004 peptides from the tear samples, including 210 (58.8%) proteins in healthy subjects, 180 (50.4%) proteins in patients with diabetes, and 219 (61.4%) proteins in patients with diabetes and dry eye syndrome (Fig. 2A). The MS/MS spectra of complement C3 by 2D-nano-LC-MS/MS are shown in Figure 3. Of the proteins, 91 (25.5%) were shared among healthy subjects, patients with diabetes, and patients with diabetes and dry eye syndrome. In addition, 6 (1.7%) proteins were found in healthy subjects and patients with diabetes only, 12 (3.4%) proteins were present in patients with diabetes only, and patients with diabetes and dry eye syndrome, including β2-microglobin and superoxide dismutase (Table 3), and 50 (14.0%) proteins were shared by healthy subjects and patients with diabetes and dry eye syndrome. There were 63 (17.7%) proteins unique to healthy subjects, 71 (19.9%) proteins to patients with diabetes, and 66 (18.5%) to patients with diabetes and dry eye syndrome. The percentage of high and low abundance proteins was comparable among healthy subjects, patients with diabetes, and patients with diabetes and dry eye syndrome (P > 0.05, Fig. 2B). Meanwhile, the percentage of medium abundance proteins in patients with diabetes and dry eye syndrome (12%) was markedly higher than that of healthy subjects (7%, P < 0.05). 

Figure 2

 

Integrated analyses of tear proteome data from groups A to C. (A) The diagram shows the proteins identified from the groups A to C tear samples. The numbers of proteins identified in the three samples are shown in the circles. (B) The distribution of high-, medium-, and low-abundance proteins in tear film. The medium-abundance proteins in group C samples obviously increased than those in group A samples (*P < 0.05). (C) Molecule function categorizations of tear proteins identified. CK, cytoskeleton; EI, enzyme inhibitor.

Figure 3

 

The MS/MS spectrum of peptide KVLLDGVQNPR from complement C3 in tear samples.

Table 3

 

Shared Proteins in the Tear Samples of Patients With Diabetes Only and Patients With Diabetes Plus Dry Eye Syndrome

Description	SC			Normalized SC
	A	B	C	A	B	C
Isoform 1 of protein wiz	0	45	19	0	0.0088635	0.0031799
Putative uncharacterized protein DKFZP686C15213	0	25	18	0	0.0049242	0.0030126
β-2-microglobulin	0	7	12	0	0.0013788	0.0020084
Fibrinogen β chain precursor	0	3	11	0	0.0005909	0.001841
IGHM protein	0	3	22	0	0.0005909	0.003682
Fatty acid-binding protein, epidermal	0	3	3	0	0.0005909	0.0005021
Isoform A 1 of ACYL-COA-binding protein	0	2	7	0	0.0003939	0.0011715
PDZ and LIM domain protein 1	0	2	3	0	0.0003939	0.0005021
Myosin regulatory light chain	0	2	2	0	0.0003939	0.0003347
IGL@ protein	0	2	37	0	0.0003939	0.0061925
Superoxide dismutase	0	1	1	0	0.000197	0.0001674
Isoform 2 of calpastatin	0	1	2	0	0.000197	0.0003347

 A, healthy subjects; B, patients with diabetes only; C, patients with diabetes plus dry eye syndrome.

The functions of the proteins identified by the 2D-nano-LC-MS/MS technique were interpreted using the gene ontology tool available in the public domain at http://david.abcc.ncifcrf.gov/ (Fig. 2C). The 10 most upregulated and downregulated proteins in the tear samples from each group are shown in Table 4. Analysis of spectral counts of tear proteins showed that, compared to healthy controls, patients with diabetes and dry eye syndrome had increased expression of apoptosis-related proteins, immunity- and inflammation-related proteins, and glycometabolism-related proteins (Table 5). The interconnection of these proteins is shown in Figure 4. 

Figure 4

 

Interactive map of proteins in the tear samples of patients with diabetes and dry eye syndrome.

Table 4

 

The 10 Most Upregulated and 10 Most Downregulated Proteins in Tear Samples

Description	SC			Normalized SC			Up/Down Regulated
	A	B	C	A	B	C
Diabetes vs. controls, △B–A
Isoform 1 of serum albumin precursor	1304	177	1308	0.22	0.0349	0.2189	Down
Cystatin-S precursor	128	22	58	0.0216	0.0043	0.0097	Down
Putative uncharacterized protein	76	0	0	0.0128	0	0	Down
Proline-rich protein 1 precursor	67	21	24	0.0113	0.0041	0.004	Down
Proline rich 4, lacrimal isoform 1	52	2	7	0.0088	0.0004	0.0012	Down
Keratin, type I cytoskeletal 19	51	12	72	0.0086	0.0024	0.0121	Down
Putative uncharacterized protein	36	0	0	0.0061	0	0	Down
Isoform 1 of endothelin-1 receptor precursor	32	0	0	0.0054	0	0	Down
Isoform 1 of ATP-dependent RNA helicase DD×25	21	0	0	0.0035	0	0	Down
Isoform 3 of PNMA-like protein 2	18	0	0	0.003	0	0	Down
Lipocalin-1 precursor	796	1383	665	0.1343	0.2724	0.1113	Up
Keratin 4	79	159	168	0.0133	0.0313	0.0281	Up
Protein S100-A9	44	100	89	0.0074	0.0197	0.0149	Up
Isoform 1 of keratin, type I cytoskeletal 13	24	63	7	0.004	0.0124	0.0012	Up
Protein S100-A8	19	60	42	0.0032	0.0118	0.007	Up
Isoform α-enolase of α-enolase	12	30	34	0.002	0.0059	0.0057	Up
Isoform 1 of 14-3-3 protein σ	6	18	21	0.001	0.0035	0.0035	Up
Isoform er1 of ankyrin-1	0	57	0	0	0.0112	0	Up
Isoform 1 of protein wiz	0	45	19	0	0.0089	0.0032	Up
Histone H2A type 1-B	0	43	0	0	0.0085	0	Up
Diabetes plus dry eye syndrome vs. controls, △C–A
Putative uncharacterized protein	76	0	0	0.0128	0	0	Down
Proline-rich protein 1 precursor	67	21	24	0.0113	0.0041	0.004	Down
Proline rich 4, lacrimal isoform 1	52	2	7	0.0088	0.0004	0.0012	Down
Secretoglobin family 1D member 1 precursor	36	24	16	0.0061	0.0047	0.0027	Down
Putative uncharacterized protein	36	0	0	0.0061	0	0	Down
Isoform 1 of endothelin-1 receptor precursor	32	0	0	0.0054	0	0	Down
Isoform 1 of ATP-dependent RNA helicase DDX25	21	0	0	0.0035	0	0	Down
Isoform 3 of PNMA-like protein 2	18	0	0	0.003	0	0	Down
Protein S100-A6	13	0	4	0.0022	0	0.0007	Down
B2M protein	12	0	0	0.002	0	0	Down
β-2-microglobulin	0	7	12	0	0.0014	0.002	Up
Fibrinogen β chain precursor	0	3	11	0	0.0006	0.0018	Up
IGHM protein	0	3	22	0	0.0006	0.0037	Up
Isoform a 1 of acyl-CoA-binding protein	0	2	7	0	0.0004	0.0012	Up
IGL@ protein	0	2	37	0	0.0004	0.0062	Up
CDNA FLJ78387	0	0	72	0	0	0.0121	Up
Putative uncharacterized protein DKFZP686O16217	0	0	25	0	0	0.0042	Up
Keratin 13 isoform B	0	0	22	0	0	0.0037	Up
α-1-acid glycoprotein 2 precursor	0	0	17	0	0	0.0028	Up
Isoform 3 of shootin-1	0	0	10	0	0	0.0017	Up
Diabetes plus dry eye syndrome vs. diabetes only, △C–B
Lipocalin-1 precursor	796	1383	665	0.1343	0.2724	0.1113	Down
IGHA1 protein	177	111	25	0.0299	0.0219	0.0042	Down
α-2-glycoprotein 1, zinc	138	175	138	0.0233	0.0345	0.0231	Down
Isoform 1 of keratin, type I cytoskeletal 13	24	63	7	0.004	0.0124	0.0012	Down
Isoform ER1 of ankyrin-1	0	57	0	0	0.0112	0	Down
Histone H2A type 1-B	0	43	0	0	0.0085	0	Down
Isoform ER1 of ankyrin-1	0	57	0	0	0.0112	0	Down
Isoform 1 of protein wiz	0	45	19	0	0.0089	0.0032	Down
Histone H2A type 1-B	0	43	0	0	0.0085	0	Down
Isoform 4 of collagen α-1, XII chain precursor	0	29	0	0	0.0057	0	Down
Isoform 1 of serum albumin precursor	1304	177	1308	0.22	0.0349	0.2189	Up
Cystatin-S precursor	128	22	58	0.0216	0.0043	0.0097	Up
Proline-rich protein 4 precursor	109	70	126	0.0184	0.0138	0.0211	Up
Polymeric immunoglobulin receptor precursor	101	60	111	0.017	0.0118	0.0186	Up
IGL@ protein	0	2	37	0	0.0004	0.0062	Up
CDNA FLJ78387	0	0	72	0	0	0.0121	Up
Putative uncharacterized protein DKFZP686O16217	0	0	25	0	0	0.0042	Up
Leukocyte elastase inhibitor	14	1	25	0.0024	0.0002	0.0042	Up
Haptoglobin precursor	14	17	42	0.0024	0.0033	0.007	Up
Heat shock 70 kDa protein 1	7	9	24	0.0012	0.0018	0.004	Up

Table 5

 

Comparison of Proteins Related to Apoptosis, Immunity, Inflammation, and Oxidative Stress in the Tear Samples of Study Subjects

Accession No.	Protein Name	SC			Normalized SC			Change
		A	B	C	A	B	C
Apoptosis-related proteins
P04792	Heat shock protein 1, 27 kDa	17	28	35	0.2868	0.5515	0.5858	Up
P08107	Heat shock protein 1A, 70 kDa	7	9	24	0.1181	0.1773	0.4017	Up
O43707	α 4-actinin,	6	5	17	0.1012	0.0985	0.2845	Up
P63104	Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein	10	8	21	0.1687	0.1576	0.3515	Up
P04083	Annexin A1	35	52	44	0.5905	1.0242	0.7364	Up
P09211	Glutathione S-transferase PI	8	7	16	0.135	0.1379	0.2678	Up
P23528	Cofilin 1, nonmuscle	7	4	14	0.1181	0.0788	0.2343	Up
P68371	β 2C-tubulin,	0	0	7	0	0	0.1171	Up
Q96BY2	Modulator of apoptosis 1	0	0	6	0	0	0.1004	Up
P30101	Protein disulfide isomerase family A, member 3	0	0	5	0	0	0.0836	Up
P27797	Calreticulin	0	0	4	0	0	0.0669	Up
P07858	Cathepsin B	3	0	6	0.0506	0	0.1004	Up
P00747	Plasminogen	0	0	3	0	0	0.0502	Up
P55072	Valosin-containing protein	0	0	2	0	0	0.0334	Up
P52565	Rho Gdp dissociation inhibitor, Gdi, α	0	0	2	0	0	0.0334	Up
P32119	Peroxiredoxin 2	6	2	1	0.1012	0.0394	0.0167	Down
P31944	Caspase 14, apoptosis-related cysteine peptidase	2	1	1	0.0337	0.0197	0.0167	–
P08758	Annexin A5	6	2	5	0.1012	0.0394	0.0837	–
Immunity and inflammation-related proteins
P02787	Transferrin	73	91	101	1.2317	1.7924	1.6904	Up
Q8WUK1	Immunoglobulin heavy locus	0	3	22	0	0.0590	0.3682	Up
P01024	Complement component 3	21	27	39	0.3543	0.5318	0.6527	Up
P19652	Orosomucoid 2	0	0	17	0	0	0.2845	Up
P02763	Orosomucoid 1	11	8	18	0.1856	0.1576	0.3013	Up
P0C0L4	Complement component 4A	3	2	6	0.0506	0.0394	0.1004	Up
P05155	Serpin peptidase inhibitor, clade G, C1 inhibitor	1	0	4	0.0168	0	0.0669	Up
P08246	Elastase 2, neutrophil	0	0	1	0	0	0.0167	Up
P10909	Clusterin	11	4	2	0.1856	0.0788	0.0334	Down
P04264	Keratin 1, epidermolytic hyperkeratosis	28	22	28	0.4724	0.4333	0.4686	–
Oxidative stress-related proteins
P06727	Apolipoprotein A-IV	6	7	9	0.101232	0.1379	0.150628	Up
P22079	Lactoperoxidase	3	0	1	0.050616	0	0.016736	Down
P00441	Superoxide dismutase 1, soluble	3	1	1	0.050616	0.019697	0.016736	Down
Q99497	Parkinson syndrome, autosomal recessive, early onset 7	6	1	4	0.1012	0.0197	0.0669	–
Glycometabolism-related proteins
P02765	α-2-HS-glycoprotein	0	0	2	0	0	0.0334	Up
P02652	Apolipoprotein A-II	12	13	20	0.2025	0.2561	0.3347	Up
P52209	Phosphogluconate dehydrogenase	0	0	4	0	0	0.066946	Up
P40925	Malate dehydrogenase 1, Nad, Soluble	2	0	5	0.033744	0	0.083682	Up
P09467	Fructose-1,6-bisphosphatase 1	6	0	3	0.101232	0	0.050209	–
P00558	Phosphoglycerate kinase 1	7	7	7	0.1181	0.1379	0.1172	–
P13929	Enolase 1	1	0	1	0.016872	0	0.016736	–

Discussion

Schirmer's strip has been used for collecting tears by investigators as it could well retain proteins, especially for tear samples of scant volume, for protein identification and quantification. One drawback is that the samples on the strip may become contaminated by unintentional contact with the conjunctival sac. Despite this drawback, Schirmer's strip still is considered by many investigators as an appropriate approach for collecting tear samples and could capture subtle changes in protein contents in the tear samples. ^24,25 In our study, we carried out a systematic comparison of tear proteomic properties of patients with type 2 diabetes and dry eye syndrome, patients with diabetes only, and normal subjects. Our D-LC-MS/MS identified 357 distinct proteins from tear samples of all subjects, and these proteins include proteins identified in a previous study, such aslactoferrin and lipocalin, ²⁶ indicating that 2D-nano-LC-MS/MS is a suitable approach for obtaining rich protein information from biofluid, especially from samples of limited volumes. Approximately one-quarter (25.5%, 91/357) of these proteins were found in all subjects. Patients with diabetes, and patients with diabetes and dry eye syndrome shared an additional 3.4% (12/357) of these proteins. Furthermore, 18.5% (66/357) of the proteins were found only in patients with diabetes and dry eye syndrome. A similar percentage of the proteins (17.7%, 63/357) was present only in normal subjects, suggesting that some proteins present in normal tears are lost with development of diabetes or diabetes with dry eye syndrome. We further found that the expression of some proteins, such as complements, heat shock proteins, and annexins, is increased, while that of some other proteins, such as peroxiredoxin and superoxide proteins, is reduced, implying that changes in the proteome occur in diabetic patients before overt onset of dry eye syndrome. We also observed that patients with diabetes and dry eye syndrome exhibited increased expression of apoptosis-related proteins, immunity- and inflammation-related proteins, and glycometabolism-related proteins. These findings indicated different proteomic profiles in healthy subjects, patients with diabetes only, and patients with diabetes and dry eye syndrome. 

Srinivasan et al. ²⁵ showed that certain proteins, such as ezrin, Ig gamma-3 chain C region, and peroxiredoxin, were upregulated, while other proteins, including cystatin-S, lacritin, lipocalin-1, proline-rich protein-4, lysozyme, polymeric immunoglobulin receptor, and lactotransferrin, were downregulated in patients with moderate to severe dry eye disease. Consistently, we also found that cystatin-S, proline-rich protein-4, and polymeric immunoglobulin receptor were downregulated in dry eye patients without diabetes compared to normal subjects. However, the polymeric immunoglobulin receptor was upregulated in diabetic patients with dry eye disease. Zhou et al. ²⁷ reported that S100A8 and S100A9 upregulation correlated with severity of dry eye symptoms. In our study, we also found that S100A8 and S100A9 in tears of diabetic patients were upregulated compared to normal subjects. It has been reported that lipocalin-1 was downregulated in dry eye disease, ²⁸ while in our study we found apparent upregulation of lipocalin-1 in diabetic patients without dry eye disease, but marked downregulation of the protein was observed in diabetic dry eye patients. We speculated lipocalin-1 downregulation may be an underlying cause of dry eye disease. Meanwhile, β2 microglobulin was downregulated in diabetic patients with or without dry eye, which is consistent with the report by Kim et al., ²⁸ suggesting that B2M may be related to development of diabetes, but not dry eye. We also found IGL@protein and heat shock protein70 (HSP70) were increased in diabetic patients compared to normal subjects, which was even more significantly increased in diabetic patients with dry eye disease. We speculated that IGL@protein and HSP70 may be related to the development of dry eye disease in diabetic patients. However, whether they can serve as markers for diabetic dry eye disease needs further investigation. 

It has been suggested that some initial events, such as chronic hyperglycemia, corneal nerve damage, and impairment of insulin action, may lead to alterations in tear films and the ocular surface of diabetic patients. Those events contribute to tissue injury and may create an environment for inflammation, as nonspecific response could increase and perpetuate ocular injury. Our study also revealed changes in proteins involved in inflammation and immune response. Complement C3, which was upregulated in patients with diabetes and dry eye syndrome, has a central role in the activation of the complement system, and IL-1α, IL-6, and IFN-γ significantly enhance C3 secretion. ²⁹ The C3b interferes with IL-12 and IL-10 production via an ERK MAPK-dependent mechanism. ³⁰ Elastase 2, which was found only in patients with diabetes and dry eye syndrome in our study, belongs to the peptidase S1 family and induces IL-8 expression via an IL-1 receptor-associated kinase signaling pathway; these events involve cell surface membrane–bound toll-like receptor 4 (TLR4). ³¹ However, elastase 2 activates p38 MAP kinase, which upregulates NF-κB and AP-1 activities, thus, inducing IL-8 mRNA expression and protein synthesis. ³² Secreted orosomucoid proteins, which were upregulated only in tears of patients with diabetes and dry eye syndrome, function in modulating the activity of the immune system during the acute phase reaction. These findings suggest that inflammatory mediators may inhibit neural signals to the lacrimal gland, ³³ thus, depriving the gland of trophic stimulation needed for its maintenance and resulting in its progressive destruction. The inflammatory alterations may impair biochemical events, culminating in reduced tear secretion and dry eye syndrome in diabetic patients. 

Recently, increased awareness of oxidative stress damage and its relation with ocular surface damage have prompted researchers to unravel mechanisms in the development of dry eye syndrome. The toxic effect of reactive oxygen species and free radicals can be eliminated by enzymes, such as superoxide dismutase (SOD), which eliminates O₂₋ to produce H₂O₂. ³⁴ Behndig et al. ³⁵ reported the tears contain little SOD activity. We found that SOD level was reduced in the tear of dry eye syndromes in diabetic patients. We also identified 18 apoptosis-related proteins. Among them, 15 proteins, including heat shock proteins, annexins, modulator of apoptosis 1, were markedly upregulated in diabetic patients with dry eye syndrome compared to healthy controls. 

We also were interested in whether lipid metabolism, which is important for stabilizing tear films, was dysregulated in dry eye syndrome in diabetic patients. We found that lipid metabolism–related proteins, α-2-HS-glycoprotein, apolipoprotein A II, and A IV, were increased significantly in diabetic patients with dry eye syndrome. The α-2-HS-glycoprotein belongs to the fetuin family, and is associated with insulin-mediated inhibition of lipolysis and stimulation of lipogenesis. ³⁶ Apolipoprotein A IV is a major component of high density lipoprotein (HDL) and chylomicrons, and has a role in chylomicron and VLDL secretion and catabolism. Apolipoprotein A II may be a positional candidate gene for familial type 2 diabetes, ³⁷ and apolipoprotein A II may stabilize HDL structure by associating with lipids and affect HDL metabolism. The findings suggest that lipid metabolism becomes dysregulated in tear films of diabetic patients. 

In conclusion, we have demonstrated that dry eye syndrome in diabetic patients is associated with an altered proteomic profile with dysregulated expression of proteins involved in a variety of important cellular process, including inflammation, immunity, oxidative stress, and lipid and glucose metabolism. These findings suggested broad derangement in tear proteins in diabetic patients with dry eye syndrome. Further characterization of these proteins could provide potential diagnostic markers and therapeutic targets that may lead to better outcomes for these patients. Further studies are needed to elucidate changes in the signaling pathways whose proteins are expressed aberrantly in tear contents in diabetic patients with dry eye syndrome. 

Acknowledgments

Supported by the Shanghai Health Hospital Development Center Foundation Project (SHDC12007104), the Natural Science Foundation of Shanghai (No.11ZR1427900), the Youth Research Project of Shanghai Municipal Health Bureau (No.2010Y164), the Young Talent Training Plan of Tongji University (No.2010KJ018), and the Young Talent Training Plan of Shanghai Tenth People's Hospital (No.11RQ108). 

Disclosure: B. Li, None; M. Sheng, None; L. Xie, None; F. Liu, None; G. Yan, None; W. Wang, None; A. Lin, None; F. Zhao, None; Y. Chen, None 

Footnotes

 BL and MS contributed equally to the work presented here and should therefore be regarded as equivalent authors.