Differential Proteomic Analyses of Cataracts From Rat Models of Type 1 and 2 Diabetes

Abstract

Purpose.: To identify differential changes in proteins and metabolites underlying “fast” type 1 (T1DC) and “slow” type 2 (T2DC) diabetic cataract (DC) formation in rat.

Methods.: Rat models of types 1 and 2 diabetes consisted of streptozotocin injection without and with high-fat diet, respectively. Cataract progression was examined weekly. At week 6, total protein changes were comparatively and quantitatively assessed by two-dimensional differential in-gel electrophoresis (2-D DIGE) coupled with mass spectrometry, and relevant metabolic changes were examined. Differences in high molecular weight (HMW) crystallin species between diabetic and control lenses were similarly identified.

Results.: Cataracts were morphologically different and progressed more slowly in T2DC versus T1DC. αA-crystallin, βB2-crystallin, and βA4-crystallin were significantly decreased in both DC types versus control. αB-crystallin was increased while βB1-crystallin was markedly decreased in T2DC. In T1DC, γB-crystallin and γS-crystallin fragmentation were increased. High-fat diet by itself had little impact, except for lowering γS-crystallin fragmentation. Despite significantly decreased opacity, a greater decrease in intermediate filaments (IFs) and more HMW crystallin species were observed in T2DC versus T1DC. However, aldose reductase expression and activity and sorbitol levels were increased to a greater extent in T1DC, while reduced glutathione (GSH) and reduced form of nicotinamide adenine dinucleotide phosphate (NADPH) levels were decreased to a greater extent and adenosine triphosphate (ATP) level was much lower in T1DC versus T2DC.

Conclusions.: The results suggest that osmotic damage, GSH loss, and decreased ATP production might be important pathological mechanisms in T1DC formation, whereas crystallin modification and cross-linking/aggregation as well as IF degradation may play more crucial roles in T2DC formation.

Introduction

A cataract is a highly frequent eye complication in type 1 and type 2 diabetic patients. A type 1 diabetic cataract (DC) is characterized by bilateral snowflake-type cortical deposits and/or a subcapsular cataract.¹ These opacities often rapidly evolve over a period of days. In contrast, type 2 DCs are morphologically similar to senile cataracts, and their opacities generally progress much more slowly than in type 1 DCs. Many studies have focused on type 1 DCs, and several mechanisms have been proposed.² However, most studies were performed in streptozotocin (STZ)-induced type 1 diabetic rats, in which a dramatic accumulation of sorbitol in the lenses caused “fast” lens opacification. The poor amenability of lens changes in this model led to the view that other pathways are relatively unimportant for DC formation, leading to decreased interest in studying the pathogenic mechanisms and developing new therapeutics.² However, considering the “slow” process of type 2 DCs, studies on such cataractogenesis might reveal novel mechanisms underlying DCs. 

Type 2 DCs have been studied in less detail for two major reasons. First, in humans, lens opacities in type 2 diabetes generally resemble senile cataracts, making it difficult to eliminate the impact of other factors, such as aging and cumulative radiation. Second, the use of an experimental model for type 2 diabetes is not optimal because the sequence of metabolic events associated with human type 2 diabetes does not occur spontaneously in rodents.³ However, Reed et al.⁴ recently generated an animal model (fat-fed rats injected with a relatively moderate amount of STZ) for type 2 diabetes that replicates the natural history and metabolic characteristics of the syndrome.^4,5 In the present study, we used this type 2 diabetic rat model and the conventional type 1 diabetic rat model (induced with relatively high doses of STZ) to perform proteomic and Western blot analyses and metabolic measurements to study the common and differential biochemical changes in both types of DC lenses compared with controls. 

Proteomic studies are invaluable for characterizing lenticular proteins and elucidating the complex factors involved in cataractogenesis.^6–8 However, these studies have predominantly focused on the abundant crystallin proteins (approximately 90% of the total protein), with little or no information on noncrystallin proteins, which typically have high molecular weights (HMWs; over 35 kDa) and are not usually visible on two-dimensional electrophoresis (2-DE) gels that have been optimized for the resolution of crystallins.⁹ However, the known noncrystallin proteins also play key roles, and thus understanding the precise roles of these proteins is critical to characterize the dysregulation of molecular pathways that leads to cataracts. 

Using two-dimensional differential in-gel electrophoresis (2-D DIGE), Guest et al.⁹ were the first to identify the HMW proteins in rat lenses that differ between the sexes. In the present study, we used 2-D DIGE with improved gel scanning to identify differentially expressed proteins in type 1 and type 2 DC lenses and control lenses. This is the first quantitative study of the abundant crystallins and HMW noncrystallin proteins using a 2-D DIGE gel. Several metabolic pathways associated with the differentially expressed proteins were subsequently examined. Furthermore, sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and liquid chromatography–tandem mass spectrometry (LC-MS/MS) analyses were performed to identify HMW crystallin species in type 1 and type 2 DC lenses and control lenses. These studies help to elucidate the shared and differential pathological mechanisms associated with the two types of DCs. 

Materials and Methods

Animals

The experiments were approved by the Regional Committee for Scientific Medical Ethics in Heilongjiang Province, China, and were performed according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

A total of 120 age-matched male Wistar rats (body weight 160–200 g, 7 weeks of age) were divided into four groups. Group N (normal control) rats were fed a standard diet (12% of calories as fat); in the group I rats (the type 1 diabetic group), type 1 diabetes was induced by an STZ (60 mg/kg) injection following a 4-week standard diet, after which the standard diet was continued; group II rats (the type 2 diabetic group) were administered a relatively moderate amount of STZ (50 mg/kg) following a 4-week high-fat diet (HFD; 40% of calories as fat), after which the HFD was continued, as described previously^4,5; and group H rats (the HFD control) were fed the HFD constantly, without diabetes induction. Blood glucose and insulin levels were determined weekly, and homeostasis model assessment–insulin resistance (HOMA-IR) was used to evaluate insulin resistance (HOMA-IR = fasting blood glucose level × fasting insulin level/22.5). The group I rats with nonfasting blood glucose levels > 13.8 mM were considered to have type 1 diabetes, and the group II rats with nonfasting blood glucose levels > 13.8 mM and significantly (P < 0.01) increased HOMA-IR values were considered to have type 2 diabetes. The induction of type 1 and type 2 diabetes was confirmed 2 days after STZ injection. 

Cataract Analysis and Hematoxylin-Eosin (HE) Staining

Cataract progression was examined weekly. The rats were anesthetized with diethyl ether, and the pupils were dilated with 1% atropine sulfate. Vertical lens sections were then observed by ophthalmologists using a slit-lamp microscope equipped with an imaging system (Canon, Inc., Tokyo, Japan). Images of the entire lens in the horizontal plane were captured using an operating microscope system (Moller-Wedel, Wedel, Germany) for quantitative analysis, as described previously.^10,11 In brief, the lens images were converted from full color to grayscale and analyzed using Adobe Photoshop CS5 software (Adobe Systems, San Jose, CA, USA). The opacity was calculated as the ratio of the number of pixels in the opaque regions to the total number of pixels in the whole lens and was expressed as a percentage. Based on opacity analysis, characteristic early-stage cataracts could be observed at week 6 after the induction of diabetes in the two types of diabetic rats. Therefore, at this time point, lenses were extracted from subsets (n = 16–20) of rats from each group and frozen in liquid nitrogen for subsequent biochemical analyses. The glucose levels and lens opacities in the remaining rats in each group were recorded for 4 more weeks. 

To observe the morphologic changes in DC lenses, equatorial sections of the lenses were stained with HE and subsequently imaged as described previously.¹² 

Two-Dimensional DIGE Analysis

Eight lenses from different rats from each group were used for proteomic and Western blotting experiments. Each frozen lens was ground in liquid nitrogen and subsequently dissolved in 1 mL lysis buffer (7 M urea, 2 M thiourea, 4% CHAPS, and 30 mM Tris, pH 8.5) containing a protease inhibitor cocktail. The lysis buffer contained powerful disaggregating reagents and chaotropes to solubilize the water-insoluble proteins, including urea-insoluble proteins and certain membrane proteins, as described previously.¹³ Next, the mixture was sonicated on ice and centrifuged (40,000g) at 4°C for 40 minutes. The supernatant was collected, and the protein concentration was determined by the Bradford method. Equal amounts of protein from each lens from the same group were then pooled together, and this homogeneous mixture was referred to as protein sample N, I, II, or H, depending on the group. An examination of the pellet (data not shown) revealed that over 99% of the proteins were solubilized in the lysis buffer, and this percentage remained the same over time as the rats became diabetic relative to the controls. 

A pooled sample consisting of equal aliquots of each sample (N, I, II, and H) was used as an internal standard for quantitative analysis. The inclusion of an internal standard in 2-D gels eliminates errors resulting from electrophoretic artifacts and enables accurate quantitative analysis.¹⁴ The internal standard was labeled with fluorescent Cy2 dye, and the individual protein samples (N, I, II, and H) were labeled with Cy3 or Cy5. The 2-D DIGE experimental design is shown in Supplementary Table S1. The labeling and 2-DE procedures were performed as described previously.¹⁵ 

Image Acquisition and Quantitative Analyses

After 2-DE, each gel was scanned (Typhoon 9410 scanner; GE Healthcare, Piscataway, NJ, USA), and the Cy2, Cy3, and Cy5 fluorescence intensities of each gel were then individually imaged according to the manufacturer's instructions. Briefly, excitation/emission wavelengths specific to Cy2 (488/520 nm), Cy3 (532/580 nm), and Cy5 (633/670 nm) were selected. The photomultiplier tube (PMT) voltage was then adjusted for each scan wavelength to confirm that the maximum pixel value (in the entire gel image) was between 50,000 and 80,000, which is suitable for such analyses. The maximum pixel value should not exceed 100,000 because this value indicates that the signal is saturated, which would prevent quantitative analyses. These conditions have generally been used in previous studies to produce high-resolution images of the abundant crystallins (<34 kDa). 

However, large numbers of HMW noncrystallin proteins were not visible under these scanning conditions. To discriminate these proteins, we selected the HMW gel region (>34 kDa) for another scan and performed a quantitative analysis. Compared with the first scan, the PMT voltage was increased in approximately 20% increments to enhance the resolution of the HMW proteins while ensuring that the maximum pixel value (in the HMW region of the image) was in a suitable range (30,000–50,000). 

The whole-gel and HMW-region images were then analyzed (DeCyder 6.5 software; GE Healthcare). Spot intensities were normalized to the internal standard, and the significance of protein changes was determined using one-way ANOVA (among the four groups) and Student's t-test (between two of the four groups) based on multiple gels. Protein spots with a 1.5-fold or greater difference (average ratio ≥ 1.5, P < 0.05) in the whole-gel images and spots with a 1.2-fold or greater difference (average ratio ≥ 1.2, P < 0.05) in the HMW-region images were considered to be significantly different or regulated. The thresholds for the whole-gel (1.5-fold) and HMW-region (1.2-fold) images were different because most proteins visible on the whole-gel images were long-lived crystallins, which are gradually modified after embryogenesis, even in the normal lens,¹⁶ whereas most HMW proteins are functional proteins, which are usually regulated dynamically. 

Protein Identification by MALDI-TOF Mass Spectrometry

Spot picking and digestion were performed using preparative gels. Immobilized nonlinear pH gradient strips (pH 3–10) were loaded with 1250 μg protein, and the gels were stained with Coomassie brilliant blue. The protein spots of interest were then excised from the gels and digested with trypsin. The tryptic peptides were identified by matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry and tandem TOF/TOF mass spectrometry (4800 Plus MALDI TOF/TOF Analyzer; Applied Biosystems, Foster City, CA, USA), as previously described.¹⁷ The combined mass and mass/mass spectra were used to search rat sequences in the International Protein Index database (IPI_Rat V3.87; http://www.ebi.ac.uk/IPI; provided in the public domain by the European Molecular Biology Laboratory-European Bioinformatics Institute [EMBL-EBI], Hinxton, UK) using the Mascot database (version 2.2.04; Matrixscience, London, UK) search algorithms. Confident protein identification had a significant (P < 0.05) protein score and the best ion score. 

SDS-PAGE and Image Analysis

Each protein sample (N, I, II, and H; 15 μg protein per lane) was resolved by discontinuous SDS-PAGE using a 4% stacking gel and a 12% resolving gel. The ratio of acrylamide to bis-acrylamide was 30:0.8. After electrophoresis, the gels were silver stained and then scanned, and image analysis was performed (Quantity One software; Bio-Rad, Hercules, CA, USA). These SDS-PAGE experiments were repeated three times. 

LC-MS/MS Analysis

Based on the image analysis, three bands and two regions (see Fig. 6) from each lane were manually excised from the gels for LC-MS/MS studies. Certain spots (picked from the 2-DE gels) that were not successfully identified by MALDI-TOF mass spectrometry were also included in the LC-MS/MS identification. In particular, the proteins in the gels were reduced in 10 mM dithiothreitol (DTT) at 56°C for 1 hour, alkylated in the dark with 55 mM iodoacetamide at room temperature for 1 hour, and then digested with trypsin. The tryptic peptides were then completely dried by centrifugal lyophilization. 

The LC-MS/MS experiments were performed (described in detail in the Supplementary Material) using an LTQ-FT mass spectrometer (Thermo Scientific, San Jose, CA, USA) equipped with a nanospray source and an Agilent 1100 high-performance liquid chromatography system (Agilent Technologies, Livermore, CA, USA). The data acquired from the mass spectrometer were used to query the SwissProt 57.15 database (Rat; http://www.uniprot.org/; provided in the public domain by EMBL-EBI, Hinxton, UK) using Mascot. Peptides with an ion score that exceeded the identity score (a primitive Mascot attempt to estimate the score of a random match) were accepted. Proteins were considered to be identified if at least two peptides (with a peptide ion score above the identity score) were explained by the spectra and if the protein ion score was greater than 90. 

Western Blot Analysis

The proteins were separated by SDS-PAGE and transferred onto polyvinylidene fluoride membranes at 200 mA at room temperature for 1 hour (2 hours for HMW proteins). The membranes were then incubated with primary monoclonal antibodies (Abcam, Cambridge, UK) at room temperature for 3 hours, followed by incubation with an alkaline phosphatase-conjugated secondary antibody (Santa Cruz Biotechnology, Paso Robles, CA, USA) at room temperature for 1 hour. The resulting bands on the immunoblots were visualized using a BCIP/NBT kit (Sigma-Aldrich Corp., St. Louis, MO, USA) and then analyzed (Quantity One software; Bio-Rad). 

Enzymatic Activity Assay

Lenses from different individuals (n = 6) from each group were homogenized in 10 mM phosphate-buffered saline (PBS; pH 6.8) containing 2 mM DTT. These homogenates were centrifuged, and the supernatants were collected for aldose reductase (AR) analysis. The reaction was performed in the assay mixture (100 mM PBS at pH 6.8, 100 mM DL-glyceraldehyde, 0.1 mM reduced form of nicotinamide adenine dinucleotide phosphate (NADPH), and 0.1 mL supernatant). Enzymatic activity was measured spectrophotometrically, as described previously.⁹ 

Four lenses from each group were used for glucose-6-phosphate dehydrogenase (G6PD) analysis. Each lens was homogenized in 50 mM PBS (pH 7.0) containing a protease inhibitor cocktail and subsequently centrifuged. The supernatant was then collected. The reaction mixture contained 3.3 mM MgCl₂, 55 mM Tris buffer (pH 7.8), 100 mM glucose-6-phosphate, and 6 mM nicotinamide adenine dinucleotide phosphate (NADP). The G6PD activity was measured as previously described.¹⁸ 

Lens Metabolite Measurements

Lenses from different individuals (n = 6) from each group were weighed and deproteinized by homogenization in perchloric acid. The levels of reduced glutathione (GSH), oxidized glutathione (GSSG), glucose, sorbitol, adenosine triphosphate (ATP), adenosine diphosphate (ADP), and adenosine monophosphate (AMP) were then measured in concentrated K₂CO₃-neutralized perchloric acid extracts using a spectrofluorometer and enzymatic procedures, as described previously.^19–21 For the estimation of NADPH, the lenses (n = 4/group) were weighed and homogenized in 0.2 M KOH. Next, an equal volume of 0.23 M KH₂PO₄ was added for neutralization, and the mixtures were centrifuged. The NADPH levels were measured in the supernatant following the method of Giblin and Reddy.²² 

Statistical Analysis

Statistical analysis was performed by Student's t-test and one-way ANOVA using DeCyder 6.5 software (GE Healthcare) for 2-D DIGE analysis and SPSS version 17.0 software (SPSS, Inc., Chicago, IL, USA) for other experiments. P < 0.05 was considered to represent a significant difference. The band intensities in the SDS-PAGE and Western blotting experiments were quantified using image analysis software (Quantity One software; Bio-Rad) before statistical analysis. 

Results

Experimental Models

Before induction of diabetes (STZ injection), the rats given the HFD (groups II and H) had significantly higher HOMA-IR values compared with the rats fed the standard diet (groups N and I; 4.87 ± 1.06 vs. 2.21 ± 0.48, P < 0.01). This IR induced by the HFD persisted throughout the 10-week period in which the rats from groups II and H had diabetes (data not shown). The prediabetic state, which is characterized by IR, revealed the natural history of type 2 diabetes,⁴ and the transition from prediabetes to type 2 diabetes occurred when pancreatic beta cells were partially destroyed by the STZ (group II) and the secretory capacity was no longer able to compensate for the IR. 

Nonfasting blood glucose concentrations in normal and HFD controls were similar in the experimental period (Supplementary Table S2). After induction of diabetes, the nonfasting blood glucose concentrations in the type 1 and type 2 diabetic groups both increased (~3-fold) significantly (P < 0.01), and the blood glucose levels in the two groups were similar until week 8 (Supplementary Table S2). The similarly high blood glucose in the two types of diabetic rats and the presence of IR in the type 2 diabetic rats indicated both shared and different metabolic characteristics, which might lead to biochemical changes in the respective lenses at the early stages of the two DC types. 

Changes in Lens Transparency

All lenses of normal and HFD control rats were clear throughout the experimental period. Cataracts were observed in type 1 diabetic rats after a 3-week duration of diabetes on average, whereas in type 2 diabetic rats, cataracts appeared after 2 more weeks. These lens opacities were analyzed quantitatively (Fig. 1A). Cataracts in type 2 diabetic rats progressed much more slowly than those in type 1 diabetic rats. Most opacities observed in type 1 diabetic lenses began with vacuole and cleft formation in the peripheral lens fibers, which then diffused as cortical opacities and rapidly evolved into mature cataracts. In contrast, most opacities in type 2 diabetic lenses began as punctate opacities in the cortex and/or nucleus that slowly increased and diffused. This morphologic difference indicated potentially different pathological mechanisms underlying the generation of the two types of DCs. Characteristic lens and HE staining images from each group at the end of week 6 after diabetes induction are shown in Figures 1B and 1C; both the type 1 and the type 2 diabetic rat lenses shown were considered to have early-stage cataracts (opacity < 50%, Fig. 1A). Numerous studies have confirmed that any changes affecting the structure or amount of specific crystallins can lead to a cataract; therefore, 2-D DIGE was performed to investigate the crystallin abundance changes in the two types of DC lenses. 

Figure 1

 

Analysis of lens opacity. (A) Quantitative analysis of lens opacity in normal control (N), type 1 diabetic (I), type 2 diabetic (II), and high-fat diet control (H) rats. Images of the entire lens (dilated pupil) in the horizontal plane were obtained weekly. The opacity was calculated as the ratio of the number of pixels in the opaque regions to the total number of pixels in the entire lens, and this value is expressed as a percentage. Overall, the values are expressed as the mean ± SEM, n = 8 to 14/group. (B) Vertical sectional images of rat lenses 6 weeks after diabetes induction. Lenses were observed weekly using a slit-lamp microscope equipped with an imaging system. All lenses in the normal (N) and high-fat diet (H) control rats were clear. The characteristic opacities observed in type 1 diabetic (I) lenses were cortical cataracts diffused from vacuoles and clefts in the peripheral lens fibers, whereas type 2 diabetic (II) lenses had punctate opacities in the cortex and nucleus. (C) Morphologic changes in the peripheral lens fibers 6 weeks after diabetes induction. Equatorial sections of the lenses were stained with HE. Characteristic changes in type 1 diabetic lenses were extensive vacuoles and clefts, while in type 2 diabetic lenses only a few small vacuoles were found. These changes in diabetic lenses relative to controls are indicated by solid arrows. Original magnification: ×40.

Changes in the Abundance of Crystallins

The 2-D DIGE gels were scanned using a relatively low PMT voltage. Abundant proteins, most of which were crystallins with low molecular weights (≤28 kDa), were clearly observed (Figs. 2, 3A). A total of 20 spots, as indicated, showed significantly different levels (average ratio ≥ 1.5; P < 0.05) in the four groups. Among these spots, 19 were successfully identified. The protein identities and fold-change information are shown in Table 1. Notably, the abundance of α-crystallins, which function as chaperones in the lens, was altered in DC lenses compared with their controls. Compared with its levels in controls, the abundance of αA-crystallin (spots 7 and 8) was significantly decreased in both type 1 and type 2 DC lenses, whereas αB-crystallin (spots 9, 10, and 11) was expressed at a similar level in type 1 DC lenses but at a higher level in type 2 DC lenses (Figs. 3A, 3B; Table 1). Additionally, βB1-crystallin (spot 19) was significantly decreased in type 2 DC lenses. βA3-crystallin (spots 16 and 17), βA4-crystallin (spots 13 and 14), and βB2-crystallin (spot 18) were decreased in type 2 and type 1 DC lenses (Table 1); these decreases were greater in type 2 DC lenses, which might indicate increased modification and/or aggregation of these crystallins in type 2 DC lenses compared with type 1 DC lenses and controls. To confirm this possibility, SDS-PAGE and LC-MS/MS analyses were subsequently performed. 

Figure 2

 

Fluorescent images of 2-D DIGE gels. 2-D DIGE was performed to compare pooled (n = 8) lens extracts from normal control (N), type 1 diabetic (I), type 2 diabetic (II), and high-fat diet control (H) rats. (A) Proteins from normal control rats labeled with Cy3 (Gel 1, Supplementary Table S1). (B) Proteins from type 1 diabetic rats labeled with Cy5 (Gel 1). (C) Proteins from type 2 diabetic rats labeled with Cy3 (Gel 3). (D) Proteins from high-fat diet control rats labeled with Cy5 (Gel 3). The separations of the protein species in the lenses of the four groups were extremely similar. Protein spots with different abundances in the four groups are indicated by numbers. The approximate molecular weight (in kDa) and pH are indicated.

Figure 3

 

Identification of proteins with different abundances. Analytical 2-D DIGE analysis was performed to compare pooled (n = 8) lens extracts from normal control (N), type 1 diabetic (I), type 2 diabetic (II), and high-fat diet control (H) rats. (A) Whole-gel image of normal controls showing abundant proteins. Spots with differential protein levels in the four groups are indicated by numbers and subsequently identified (Table 1). (B) Three-dimensional image showing spots 7 (αA-crystallin), 10 (αB-crystallin), and 19 (βB1-crystallin), as indicated in (A), with different abundances in the four groups. (C) Image of 2-D DIGE analysis of normal controls revealing high molecular weight (HMW) proteins (in the region indicated by the solid rectangle in [A]). To visualize the HMW proteins, the HMW region (>34 kDa) of each gel was rescanned under optimized conditions, as described in the text. Several spots that were not visible in the whole-gel images were visible at high resolution after the second scan. Protein spots with different abundances in the four groups are indicated by numbers and subsequently identified (Table 2).

Table 1

 

Identification of Abundant Proteins by MALDI-TOF and Tandem TOF/TOF Mass Spectrometry*

Spot No.	Gene Name/Protein	Accession No.	Mascot Score†	Sequence Coverage, %	Peptide Count	Average Ratio‡
						I vs. N	II vs. H	H vs. N	II vs. N
1	CRYAA/αA-crystallin, fragment	IPI00188127	69	48.8	9	−1.63	−1.1	−1.01	−1.11
2	CRYGS/γS-crystallin, fragment	IPI00767786	185	27.5	3	1.75	1.11	−1.78	−1.61
3	NI§					1.78	1.51	−1.13	1.34
4	FABP5/fatty acid-binding protein	IPI00327830	116	34.8	4	1.21	1.76	−1.52	1.16
5	CRYAA/αA-crystallin, fragment	IPI00188127	167	35.3	7	−1.08	−2.62	1.26	−2.08
6	CRYAB/αB-crystallin, fragment	IPI00215465	130	50.9	7	−1.54	−1.09	1.02	−1.06
7	CRYAA/αA-crystallin	IPI00188963	238	45.2	8	−1.87	−1.87	−1.24	−2.32
8	CRYAA/αA-crystallin	IPI00188127	147	41	9	−1.51	−1.52	−1.16	−1.76
9	CRYAB/αB-crystallin	IPI00215465	102	49.1	7	−1.12	2.15	−1.22	1.76
10	CRYAB/αB-crystallin	IPI00215465	143	45.7	7	−1.01	1.76	−1.06	1.66
11	CRYAB/αB-crystallin	IPI00215465	125	38.3	5	−1.04	1.57	1.04	1.64
12	CRYGB/γB-crystallin	IPI00358037	142	36	6	1.52	1.15	1.31	1.5
13	CRYBA4/βA4-crystallin	IPI00231258	216	42.5	6	−1.34	−1.82	1.02	−1.78
14	CRYBA4/βA4-crystallin	IPI00231258	305	41.1	6	−1.6	−1.72	1.02	−1.68
15	CRYBB3/βB3-crystallin	IPI00189738	505	59.7	11	1.39	1.33	1.25	1.66
16	CRYBA1/βA3-crystallin	IPI00231371	105	30.2	6	−1.37	−1.96	−1.24	−2.43
17	CRYBA1/βA3-crystallin	IPI00231371	221	25.1	4	−1.17	−1.97	−1.2	−2.38
18	CRYBB2/βB2-crystallin	IPI00231628	268	57.6	11	−1.53	−1.58	−1.36	−2.15
19	CRYBB1/βB1-crystallin	IPI00189736	164	45.4	9	1.38	−4.32	1.06	−4.08
20	AKR1B1/aldose reductase	IPI00231737	235	38	13	1.57	1.14	1.11	1.26

*  The whole-gel images showing abundant proteins (Figs. 2A, 3A) were analyzed. Protein spots with a 1.5-fold or greater difference and an ANOVA (among four groups) or a t-test (between two groups) score < 0.05 were considered significantly changed or regulated (based on multiple gels). The differential protein spots, as indicated in Figures 2 and 3A, were then identified by MALDI-TOF and tandem TOF/TOF mass spectrometry.

†  Protein Mascot score obtained by mass spectrometry identification; confident protein identification had a significant (P < 0.05) protein score (>60) and the best ion score.

‡  The average spot volume ratio between the two groups is listed in this column. Average ratio ≥ 1.5, and P < 0.05 (t-test) was considered significant (shown in bold).

§  Not identified confidently.

In addition, one HMW protein (spot 20), identified as AR, was significantly more abundant in type 1 DC lenses, and another noncrystallin protein, fatty acid-binding protein (spot 4), was less abundant in HFD controls compared with normal controls (Fig. 3A; Table 1). However, under traditional gel scanning conditions, large numbers of HMW noncrystallin proteins were not visible, even though the HMW areas of the gels were specifically studied (data not shown); this prevented the performance of quantitative analyses. Changes in the abundance of these proteins might occur earlier and be more important than alterations in crystallin abundance during DC genesis. 

Abundance Changes in HMW Noncrystallin Proteins

To optimize the resolution of HMW proteins in 2-D DIGE gel images, we selected the HMW (>34 kDa) region and elevated the PMT voltage for a subsequent scan (Supplementary Fig. S1). Consequently, the overall intensity of spots in this region was significantly higher (Figs. 3A, 3C). Based on this optimization, a total of 34 spots, as indicated (Fig. 3C), showed significantly different (average ratio ≥ 1.2; P < 0.05) protein levels in the four groups. Among these spots, 25 were successfully identified (Table 2). Notably, glutathione synthetase (GS; spot 27) and lengsin (a lens-specific member of the GS superfamily; spot 31) expression was significantly lower in type 1 but not type 2 DC lenses compared with normal controls. Additionally, three glycolic enzymes were significantly differentially expressed among the four groups. Expression of phosphoglycerate kinase 1 (PGK1; spot 20), which catalyzes the seventh step of glycolysis for ATP production, was lower (1.27-fold) in type 1 DC lenses but higher (1.27-fold) in type 2 DC lenses compared with normal controls. Expression of α-enolase (spot 19), which catalyzes the ninth glycolytic step, was higher (1.22-fold) in type 2 DC lenses compared with normal controls but similar between controls and type 1 DC lenses. Expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH; spots 5 and 6), which catalyzes the sixth glycolytic step, was increased in HFD-fed rats compared with normal controls. Moreover, expression of the key enzyme in the sorbitol pathway, AR (spot 4), was increased by 1.59- and 1.27-fold in type 1 and type 2 DC lenses, respectively, compared with normal controls (Fig. 3C; Table 2), consistent with the whole-gel analysis (Fig. 3A; Table 1). In contrast, compared with normal controls, expression of annexin A2 (spot 18) was decreased (1.39-fold) in type 2 DC lenses, and expression of annexin A5 (spot 1) was decreased (1.52-fold) in type 1 DC lenses. In Figure 4, standardized log abundances are shown for GS, lengsin, PGK1, α-enolase, annexin A2, and annexin A5 in the four groups for every 2-D DIGE gel, which were determined using DeCyder software. Metabolic studies were then performed to further validate the importance of the observed changes in the enzymes. 

Figure 4

 

Standardized log abundance graphs for differentially expressed proteins. Spot numbers (31, 27, 20, 19, 1, and 18) are indicated in Figure 3C; the corresponding proteins were identified (lengsin, glutathione synthetase, phosphoglycerate kinase 1, α-enolase, annexin A5, and annexin A2, respectively), and fold-change information is shown in Table 2. The thick black line represents the median of two gels containing the same sample. N, normal control; I, type 1 diabetic; II, type 2 diabetic; H, high-fat diet control group.

Table 2

 

Identification of HMW Proteins by MALDI-TOF Mass Spectrometry* and LC-MS/MS†

Spot No.	Gene Name/Protein	Accession No.‡	Mascot Score	Seq. Cov., %	Peptide Count	Average Ratio§
						I vs. N	II vs. H	H vs. N	II vs. N
1	ANXA5/annexin A5	P14668	115	9.7	2	−1.52	1.05	−1.17	−1.11
2	NI||					−1.61	1.18	1.12	1.33
3	NI					−1.39	1.25	1.13	1.42
4	AKR1B1/aldose reductase	IPI00231737	235	38	13	1.59	1.13	1.13	1.27
5	GAPDH/glyceraldehyde-3-phosphate dehydrogenase	IPI00567177	95	16.8	5	−1.09	−1.09	1.2	1.1
6	GAPDH/glyceraldehyde-3-phosphate dehydrogenase	IPI00567177	116	24	6	−1.12	−1.22	1.27	1.04
7	BFSP2/phakinin, fragment	IPI00372464	152	20.2	8	−1.13	−1.14	−1.07	−1.22
8	BFSP2/phakinin, fragment	IPI00372464	165	18.3	7	−1.09	−1.18	−1.06	−1.25
9	CRYAA/αA-crystallin, dimer	IPI00188127	90	21.4	4	−1.31	−1.08	1.19	1.1
10	NI					−1.18	1.04	1.08	1.12
11	NI					1.23	1.08	1.06	1.15
12	ACTG1/γ-actin	IPI00896224	97	13.6	4	1.23	1.11	−1.08	1.03
13	ACTG1/γ-actin	IPI00896224	86	14.4	4	1.03	−1.08	−1.12	−1.21
14	ACTG1/γ-actin	IPI00896224	70	8.3	3	−1.04	1.29	−1.09	1.18
15	NI					−2.3	−1.22	1.02	−1.2
16	NI					−1.69	−1.68	−1.16	−1.95
17	NI					1.12	−1.12	−1.06	−1.19
18	ANXA2/annexin A2	Q07936	216	20.9	5	−1.05	−1.21	−1.15	−1.39
19	ENO1/α-enolase	IPI00464815	163	27	9	−1.01	1.21	1.01	1.22
20	PGK1/phosphoglycerate kinase 1	P16617	169	18.5	4	−1.27	1.07	1.2	1.27
21	BFSP2/phakinin	IPI00372464	195	25.5	9	−2.12	−3.75	−1.08	−4.05
22	BFSP2/phakinin	IPI00372464	185	29.3	10	−1.68	−3.62	−1.17	−4.24
23	BFSP2/phakinin	IPI00372464	211	29.8	10	−1.3	−3.22	−1.09	−3.5
24	BFSP1/filensin	IPI00232002	98	12.6	7	−1.21	−1.4	−1.06	−1.48
25	BFSP1/filensin	IPI00232002	114	35.2	13	−1.26	−1.6	−1.18	−1.88
26	BFSP1/filensin	IPI00232002	65	8.3	5	−1.29	−2.25	−1.05	−2.36
27	GSS/glutathione synthetase	P46413	102	6.3	2	−1.44	1.14	−1.15	−1.01
28	TUBB2B/tubulin beta-2B chain	IPI00655259	115	15.5	6	−1.13	−1.22	−1.11	−1.35
29	VIM/vimentin	IPI00230941	131	14.2	6	1.03	−1.32	−1.15	−1.5
30	VIM/vimentin	IPI00230941	153	16.1	7	1.1	−1.32	−1.22	−1.61
31	LGSN/lengsin	IPI00327094	110	17.3	9	−1.61	−1.07	1.04	−1.03
32	NI					1.32	1.09	1.04	1.13
33	CCT5/T-complex protein 1 subunit epsilon	Q68FQ0	124	4.8	2	−1.24	1.34	−1.11	1.21
34	NI					1.26	−1.44	−1.27	−1.83

*  The HMW-region gel images showing noncrystallin proteins (Fig. 3C) were analyzed. Protein spots with a 1.2-fold or greater difference and an ANOVA (among four groups) or a t-test (between two groups) score < 0.05 were considered significantly changed or regulated (based on multiple gels). The differential protein spots, as indicated in Figure 3C, were then identified by MALDI-TOF and tandem TOF/TOF mass spectrometry. Confident protein identification had a significant (P < 0.05) protein score (>60) and the best ion score.

†  Spots that were not identified successfully by MALDI-TOF mass spectrometry were then identified by LC-MS/MS. Proteins were considered to be identified if at least two peptides (with a peptide ion score above the identity score) were explained by the spectra and if the protein ion score was higher than 90. Five such spots (spots 1, 18, 20, 27, and 33) were confidently identified by LC-MS/MS.

‡  The data acquired from MALDI-TOF mass spectrometry were used to query the International Protein Index database (IPI_Rat V3.87) using Mascot, and the corresponding accession numbers (beginning with “IPI”) are listed. The data acquired from LC-MS/MS were used to query the SwissProt database (Rat V57.15) using Mascot, and the corresponding accession numbers (beginning with “P” or “Q”) are listed.

§  The average spot volume ratio between the two groups is listed in this column. Average ratio ≥ 1.2, and P < 0.05 (t-test) was considered significant (shown in bold).

||  Not identified confidently by MALDI-TOF mass spectrometry and LC-MS/MS.

In addition, two beaded filament structural proteins (BFSPs), filensin and phakinin, were decreased in DCs, and especially in type 2 DC lenses, compared with normal controls. Moreover, compared with normal controls, vimentin, another intermediate filament (IF) protein, showed a similar level in type 1 DC lenses but was decreased in type 2 DC lenses. Western blot analysis was used to verify changes in filensin (Fig. 5A), phakinin (Fig. 5B), and vimentin (Fig. 5C) levels in DC lenses compared with controls. 

Figure 5

 

Comparison of filensin (A), phakinin (B), and vimentin (C) levels. N, I, II, and H represent individual lens extracts from normal control, type 1 diabetic, type 2 diabetic, and high-fat diet control rats at the end of week 6 (after induction of diabetes). The upper graphs show the immunoreactive bands identified by Western blotting. The lower graphs show the corresponding band intensities normalized to β-actin. Mean ± SEM, n = 5/group. *P < 0.05 and **P < 0.01 versus normal controls; #P < 0.05 and ##P < 0.01 type 2 diabetic group versus type 1 diabetic group.

Increase in HMW Crystallin Species Levels in DC Lenses

High molecular weight crystallin species are usually modified or aggregated crystallins and are present in aged and/or cataract lenses.¹³ The formation of these species might have contributed to the observed decrease in the levels of most crystallins (Table 1) in DC lenses, especially in type 2 DC lenses. Sodium dodecyl sulfate–PAGE analysis was performed to enhance the detection of these HMW crystallins. Based on image analysis, several differences were evident among the four groups in the 26- to 43-kDa region of the SDS-PAGE gel (Fig. 6A). A comparison of the intensities in this gel region in each lane is shown in Figures 6B and 6C. The intensities of regions a (36–42 kDa) and b (28–32 kDa) were higher (P < 0.05) in the lanes containing type 1 and type 2 diabetic protein samples compared with the lanes containing normal control samples, but the HFD controls had intensities that were similar to those of the normal controls. The main components of regions a and b were various crystallins (Table 3). Considering that crystallin monomers are theoretically ≤28 kDa in size, the proteins with highly increased levels in the two regions were HMW crystallin species. Compared with type 1 DC lenses, there was a greater number (P < 0.05) of such HMW species in type 2 DC lenses (Fig. 6B), although these lenses had lower opacity. 

Figure 6

 

SDS-PAGE gel image and intensity analysis. (A) SDS-PAGE analysis was performed to enhance the detection of high molecular weight (HMW) crystallin species. Proteins (15 μg per lane) were resolved by SDS-PAGE, and the gels were silver stained and subsequently scanned. Under these conditions, the gel images were suitable for quantitative analysis of low-abundant proteins with relatively HMW (>26 kDa), as their intensities were midrange (not too low and not saturated). The first lane on the left contains a molecular weight marker. Lanes N, I, II, and H represent a mixture of eight extracts from normal control, type 1 diabetic, type 2 diabetic, and high-fat diet control rat lenses, respectively, at the end of week 6 (after induction of diabetes). The images (>26 kDa) were analyzed using the Quantity One software, and several differences among the four groups were observed in the 26- to 43-kDa region (indicated by a black box). Regions a and b and bands c, d, and e were identified by LC-MS/MS analysis (Table 3). (B) Comparison of the intensities of the 26- to 43-kDa regions in lanes N, I, and II. The intensities of bands c and d and the average intensities of regions a and b were normalized to that of band e (the primary fraction identified as β-actin; Table 3) and then compared among the three groups. Band f was not included in the analysis because of its low intensity and continuity with region b. *P < 0.05 versus normal controls and #P < 0.05 versus type 1 diabetic group (based on analyses of three individual SDS-PAGE gels). (C) Comparison of the intensities of the 26- to 43-kDa regions in lanes N and H. *P < 0.05 versus normal controls.

Table 3

 

Main Proteins Identified in Each Region/Band by LC-MS/MS*

Band/Region	Main Composition of Each Band/Region in Each Lane of SDS-PAGE Gels†
	N	I	II	H
a	βB1(96)-, βB3(23)-, βA4(17)-, βA3(16)-, αA(16)-crystallin	βB1(177)-, βA3(43)-, βA4(34)-, αA(26)-, βB2(17)-crystallin	βB1(221)-, βA3(61)-, βA4(47)-, αA(38)-, βB3(33)-crystallin	βB1(125)-, βA3(27)-, βA4(18)-, βB3(16)-, αA(16)-crystallin
b	βB1(70)-, αA(21)-, βA3(10)-, βB3(9)-, βA4(9)-crystallin	βB1(87)-, αA(39)-, βB3(16)-, βA4(12)-, βA3(12)-crystallin	βB1(99)-, αA(58)-, βA3(29)-, αB(21)-, βA4(10)-crystallin	βB1(76)-, αA(21)-, βB3(14)-, βA4(11)-, βA3(7)-crystallin
c	Aldose reductase (66), L-lactate dehydrogenase (10), βB1-crystallin (9)	Aldose reductase (93), βB1-crystallin (20), L-lactate dehydrogenase (11)	Aldose reductase (79), βB1-crystallin (31), L-lactate dehydrogenase (14)	Aldose reductase (57), βB1-crystallin (16), L-lactate dehydrogenase (10)
d	GAPDH (71), aldose reductase (16), βB1-crystallin (12)	GAPDH (64), aldose reductase (19), βB1-crystallin (17)	GAPDH (75), βB1-crystallin (24), aldose reductase (15)	GAPDH (87), aldose reductase (14), βB1-crystallin (14)
e	β-actin (109), βB1-crystallin (6)	β-actin (111), βB1-crystallin (9)	β-actin (106), βB1-crystallin (9)	β-actin (101), βB1-crystallin (8)

*  Proteins in each band/region from SDS-PAGE gels, as indicated in Figure 6, were identified by LC-MS/MS. Only peptides that had an ion score that exceeded the identity score (a primitive Mascot attempt to estimate the score for a potential random match) were accepted.

†  Main protein compositions in each band/region are listed, and the corresponding accepted peptide numbers are shown in parentheses.

The Western blot analysis verified the SDS-PAGE analyses of the HMW crystallin species and the 2-D DIGE analyses of the crystallin monomers. Compared with normal controls, αA-crystallin (20 kDa) was significantly decreased (P < 0.01) in type 1 and type 2 DC lenses, whereas its HMW species (25–40 kDa) were present at higher levels (P < 0.01), especially in type 2 DC lenses (Fig. 7A). However, compared with type 1 DC lenses and normal control lenses, αB-crystallin (21 kDa) was expressed at a higher level (P < 0.05) in type 2 DC lenses, and its HMW species (26–40 kDa) were also present at higher levels (P < 0.01) (Fig. 7B). In contrast, the level of βB1-crystallin (28 kDa) was lower (P < 0.01) in type 2 DC lenses compared with normal controls, whereas its HMW species (30–40 kDa) were present at higher levels in type 1 and type 2 DC and HFD control lenses (Fig. 7C). Except for the HMW βB1-crystallin species, there were no significant differences in the crystallins between normal and HFD controls. 

Figure 7

 

Comparison of αA-crystallin (A), αB-crystallin (B), and βB1-crystallin (C) levels. N, I, II, and H represent individual lens extracts from normal control, type 1 diabetic, type 2 diabetic, and high-fat diet control rats, respectively, at the end of week 6 (after induction of diabetes). The upper graphs show the immunoreactive bands observed by Western blotting. For each crystallin, both the monomer and the high molecular weight (HMW) species were detected by the same antibody. Molecular weights are indicated to the left of each image. The lower graphs show the corresponding band intensities normalized to β-actin. Mean ± SEM, n = 5/group. *P < 0.05 and **P < 0.01 versus normal controls; #P < 0.05 and ##P < 0.01 type 2 diabetic group versus type 1 diabetic group.

In addition, by SDS-PAGE (Fig. 6), band c, primarily containing AR (Table 3), showed higher (P < 0.05) protein levels in type 1 and type 2 DC lenses compared with normal controls, whereas band d, primarily containing GAPDH, showed higher (P < 0.05) protein levels in HFD controls compared with normal control lenses (Fig. 6; Table 3), supporting the 2-D DIGE results. 

Activation of AR in DC Lenses

Changes in AR expression (2-D DIGE results) indicated activation of AR in DC lenses, and especially in type 1 DC lenses. To confirm this finding, AR activity and the levels of lens glucose (substrate) and sorbitol (product) were measured. Lens glucose and sorbitol concentrations were significantly higher in rats after 6 weeks of type 1 or type 2 diabetes compared with the respective nondiabetic groups (Figs. 8A, 8B). Additionally, the glucose level in type 1 DC lenses was slightly higher (~1.2-fold, P = 0.03) than the level in type 2 DC lenses. However, the sorbitol level in type 1 DC lenses was also significantly higher (~2-fold, P < 0.01) than the level in type 2 DC lenses, which indicated more severe osmotic stress in type 1 DC lenses compared with type 2 DC lenses. Aldose reductase activity followed the same trend (Fig. 8C) as did the lens sorbitol concentration in the four groups. The lens sorbitol and AR activity analyses were consistent with the proteomic data indicating increased AR protein levels (Fig. 3C; Table 2). 

Figure 8

 

Comparison of lens glucose (A) and sorbitol (B) concentrations and AR activities (C). Lenses from each group were extracted and assessed at the end of week 6 (after induction of diabetes). N, normal control; I, type 1 diabetic; II, type 2 diabetic; H, high-fat diet control group. Mean ± SEM, n = 6/group. **P < 0.01 versus normal controls; #P < 0.05 and ##P < 0.01 type 2 diabetic group versus type 1 diabetic group.

Changes in Glutathione Level

The downregulated expression of GS (2-D DIGE results, Fig. 4) in type 1 DC lenses indicated a potential decrease in lens glutathione. To better understand its antioxidant status in DC and control lenses, the levels of GSH and GSSG in lenses were measured. Lens GSH levels in type 1 diabetic rats were decreased 4.4-fold compared with levels in normal control rats, whereas GSSG levels were similar between the two groups (Table 4). In type 2 DC lenses, GSH levels were decreased 2.1-fold, whereas GSSG levels were increased (P < 0.05) compared with levels in normal controls. As a result, the lens GSH/GSSG ratio was 4- and 3-fold lower in type 1 and type 2 DC lenses, respectively, compared with normal controls, and GSH and GSSG levels and the GSH/GSSG ratio were higher in type 2 DC lenses compared with type 1 DC lenses. In contrast, there were no significant differences in the three parameters between normal and HFD controls. Notably, the lower levels of GSH in the type 1 DC lenses revealed more severe oxidative damage. 

Table 4

 

Glutathione Redox Status in Lenses

	N	I	II	H
GSH	4.22 ± 0.72	0.97 ± 0.21*	2.05 ± 0.36*†	3.91 ± 0.64
GSSG	0.146 ± 0.019	0.135 ± 0.033	0.208 ± 0.041‡§	0.151 ± 0.037
GSH/GSSG	28.97 ± 0.856	7.22 ± 0.921*	9.97 ± 1.17*§	26 ± 1.99

 Lens GSH and GSSG levels were measured at the end of week 6 (after induction of diabetes). The data are expressed (μmol/g lens) as the mean ± SEM, n = 6/group.

*  P < 0.01 versus N.

†  P < 0.01 group II versus group I.

‡  P < 0.05 versus N.

§  P < 0.05 group II versus group I.

Decreases in the NADPH Level and G6PD Activity in DC Lenses

The changes in glutathione levels could also be related to the hexose monophosphate shunt (HMPS), which produces NADPH to regenerate GSH from GSSG. Lens NADPH levels were measured and showed 2.4- and 1.4-fold decreases in type 1 and 2 diabetic rats, respectively, compared with normal controls (Fig. 9A). Moreover, similar changes were found in the activity of G6PD (Fig. 9B), which is the rate-limiting enzyme of the HMPS that catalyzes the first step in NADPH production. 

Figure 9

 

Comparison of lens NADPH (A) and G6PD activities (B). Lenses from each group were extracted and assessed at the end of week 6 (after induction of diabetes). N, normal control; I, type 1 diabetic; II, type 2 diabetic; H, high-fat diet control group. Mean ± SEM, n = 4/group. *P < 0.05 and **P < 0.01 versus normal controls; ##P < 0.01 type 2 diabetic group versus type 1 diabetic group.

Changes in Energy Status

Expression changes in glycolic enzymes (2-D DIGE results, Table 2) might alter the energy status of the two types of DC lenses. To detect this possible phenomenon, adenine nucleotide levels were measured. The ATP levels in type 1 DC lenses were lower (~1.4-fold) compared with levels in normal controls, whereas ADP and AMP levels were higher (Table 5). In type 2 DC lenses, ADP and AMP levels were also higher compared with levels in normal controls, whereas ATP levels were similar to those in normal controls, indicating an increase in ATP production and consumption in type 2 DC lenses. Additionally, the total adenine nucleotide levels were lower in type 1 DC lenses but higher in type 2 DC lenses compared with normal controls. 

Table 5

 

Energy Status of Lenses

	N	I	II	H
ATP	2.06 ± 0.33	1.43 ± 0.25*	2.15 ± 0.26†	1.97 ± 0.29
ADP	0.386 ± 0.089	0.537 ± 0.08‡	0.597 ± 0.112*	0.413 ± 0.075
AMP	0.108 ± 0.013	0.138 ± 0.018*	0.166 ± 0.023*§	0.111 ± 0.11
ATP/ADP	5.39 ± 0.55	2.73 ± 0.37*	3.62 ± 0.44*†	4.82 ± 0.54
ATP+ADP+AMP	2.55 ± 0.25	2.11 ± 0.31‡	2.92 ± 0.32‡†	2.49 ± 0.27

 Lens ATP, ADP, and AMP levels were measured at the end of week 6 (after induction of diabetes). The data are expressed (μmol/g lens) as the mean ± SEM, n = 6/group.

*  P < 0.01 versus N.

†  P < 0.01 group II versus group I.

‡  P < 0.05 versus N.

§  P < 0.05 group II versus group I.

Discussion

This is the first study examining the common and differential changes in proteins and metabolites underlying “fast” type 1 and “slow” type 2 DC formation in rat models. Notably, the marked improvement in the resolution of noncrystallin proteins in the 2-D DIGE ensured the accuracy of the quantitative analysis and directed subsequent metabolic analyses. 

The significant increase in lens glucose and sorbitol levels in type 1 diabetic rats is consistent with previous studies.^20,23,24 Elevated AR expression (Table 2) and activity (Fig. 8C) might lead to sorbitol accumulation and altered membrane permeability in type 1 diabetic lenses, resulting in cell lesions, lens swelling, and vacuole and cleft formation (Fig. 1C).^2,25 In contrast, in type 2 diabetic rats, lens AR expression and activity and sorbitol levels were significantly lower than in type 1 diabetic rats but higher than in controls. These findings indicated more moderate osmotic stress in type 2 DC lenses than in type 1 DC lenses, which explained the morphologic differences in opacity between the two groups and the slower cataract progression in type 2 diabetic rats. Additionally, in the current study, the glucose level in type 2 DC lenses was slightly lower than in type 1 DC lenses (Fig. 8A), which might promote moderate AR activation. However, considering the slight decline in glucose levels but greatly decreased AR activity in type 2 DC lenses compared with type 1 DC lenses, other mechanisms might also contribute to the moderate AR activation. 

Oxidative stress occurs in the lens early during the course of diabetes and is manifested by depletion of the main biological antioxidant, GSH, and an increase in the GSSG/GSH ratio.^26,27 The significant decrease in GSH levels is not solely due to utilization of glutathione^28,29; another factor causing decreased GSH levels might be decreased GSH synthesis.^30,31 In the present study, the 2-D DIGE results revealed that GS and lengsin levels were significantly decreased in type 1 but not in type 2 DC lenses compared with normal controls (Fig. 4), which might have contributed to the significantly lower levels of GSH in type 1 compared with type 2 DC lenses. Furthermore, the decrease in GSH levels in diabetic (type 1) rats may be affected by the decreased level of ATP (which is required for the de novo synthesis of GSH).^28,32 This hypothesis is consistent with the energy status of type 1 DC lenses observed in this study (Table 5). In addition, the accumulation of GSSG relative to GSH in type 1 and 2 diabetic rats could also be related to the inability of lens cells to produce sufficient NADPH via the HMPS, which was supported by the decreased levels of lens NADPH and G6PD activities observed in the current study (Fig. 9). 

Two-dimensional DIGE analysis also revealed that the level of PGK1 in type 1 DC lenses was lower than that in normal controls, whereas α-enolase and PGK1 expression in type 2 DC lenses was increased compared with that in normal controls (Fig. 4). These changes in the glycolytic enzymes might have contributed to the changes in ATP levels (Table 5) in diabetic lenses. Because the lens energy status is also related to the tricarboxylic acid cycle in epithelial cells, glycolytic enzymes may contribute partially but not solely to the changes in ATP. Notably, the energy status analysis also suggested an increase in ATP generation and consumption in type 2 DC lenses compared with normal controls. This compensatory mechanism might have been partially due to upregulated synthesis of the abundant αB-crystallin chaperone (requiring ATP) in type 2 diabetic rats (Fig. 7B). In addition, this compensatory mechanism might have resulted in the lower glucose level in type 2 DC lenses compared with type 1 DC lenses (Fig. 8A), leading to moderate AR activation. 

The levels of HMW crystallin species were significantly increased in both type 1 and type 2 DC lenses. These HMW species are most likely glycated and/or cross-linked crystallins in DC lenses, given that accumulation of glycated and advanced glycation end product (AGE)-modified β- and γ-crystallins was previously observed in STZ-diabetic (type 1) rats.^33–35 Glycation leads to protein cross-linking, which may act as a nucleation site, triggering further aggregation.³⁶ Notably, despite the significantly lower opacity, a greater number of HMW species were observed in type 2 DC lenses compared with type 1 DC lenses. Interestingly, a similar change was observed in the IF protein (filensin, phakinin, and vimentin) levels, which were decreased to a greater extent in type 2 DC lenses than in type 1 DC lenses (Fig. 5). Intermediate filaments can promote precise cellular organization in the lens,³⁷ and a decrease in IF levels might be due to their modification and subsequent degradation by proteolytic enzymes.³⁸ Therefore, it is reasonable to suggest that crystallin modification and cross-linking/aggregation along with IF modification and degradation might play more crucial roles in type 2 DC formation. Additionally, this increased protein modification in type 2 diabetic lenses may induce increased expression of the αB-crystallin chaperone (Fig. 7B), which may bind to and protect unfolded and/or modified crystallins from further denaturation.³⁹ 

To our knowledge, this is the first observation of decreased annexin A2 and A5 levels in type 2 and type 1 DC lenses, respectively (Fig. 4). Recombinant annexin A2 has been reported to inhibit the progression of diabetic nephropathy via recovery of hypercoagulability.⁴⁰ Annexin A5 also has a protective role, preventing the development of diabetic neurodegenerative diseases.⁴¹ However, the role of annexins in the lens remains unknown, and their potential DC-associated mechanisms require further investigation. 

In conclusion, comparably high blood glucose levels in type 1 and type 2 diabetic rat models might be the leading cause for the common changes in lenses, whereas type 2 diabetes-specific metabolic characteristics (such as IR) might initiate the differential changes. Identification of these similarities and differences might increase our understanding of diabetes-associated cataracts. However, in this study, the results were obtained in experimental animal models, and it is uncertain if the findings can be directly translated to humans. Additional studies will be required to elucidate the metabolic mechanisms underlying DCs in humans and to identify effective therapeutic agents for patients with DCs. 

Acknowledgments

The authors thank Cheng Ma and Bin Fu for their excellent technical assistance. 

Supported by grants from the Scientific Research Fund of the First Affiliated Hospital of Harbin Medical University (20131304), Specialized Research Fund for the Doctoral Program of Higher Education (20112307110013), and the National Natural Science Foundation of China (30973275). 

Disclosure: S. Su, None; F. Leng, None; L. Guan, None; L. Zhang, None; J. Ge, None; C. Wang, None; S. Chen, None; P. Liu, None