Indoleamine 2,3-Dioxygenase Overexpression Causes Kynurenine-modification of Proteins, Fiber Cell Apoptosis and Cataract Formation in the Mouse Lens

Maneesh Mailankot; Magdalena M. Staniszewska; Heather Butler; Moonkyung H. Caprara; Scott Howell; Benlian Wang; Catherine Doller; Lixing W. Reneker; Ram H. Nagaraj


Lab Invest. 2009;89(5):498-512. 

In This Article


Transgenic Mouse Lines

One hemizygous-independent and two homozygous-independent lines were established. HomTg lines were characterized by smaller eyes (Figure 2a). All transgenic lines showed normal reproductive patterns and longevity. Both nontransgenic and transgenic animals received the same diet and developed normally. In addition, to control for possible effects of the promoter on morphological and developmental features in the hIDO overexpression animals, we examined eyes and lenses from C57BL/6 mice that overexpress glyoxalase-I. In this mouse, the chicken δ1-crystallin enhancer and αA-crystallin promoter construct were used to drive overexpression of glyoxalase-I in the lens (Gangadhariah et al., unpublished observation) but lens development and morphology were normal. We used ~3-month-old mice for the entire study, as at this age homTg animals had bilateral mature cataracts (see below).

Figure 2.

Lens diameter in the transgenic mouse. (a) Homozygous transgenic (homTg) mice (right) exhibited considerably smaller lenses. The eye size was normal in age-matched hemizygous transgenic (hemTg) (middle) and wild-type (Wt) mice (left). (b) Lens weight was significantly lower in homTg mice compared to Wt, glyoxalase-I-transgenic (TgGloI), and hemTg mice. (c) Lens diameter in homTg mice was significantly shorter than Wt, TgGloI, and hemTg mice. Results are mean ± s.d., n=5 mice (10 lenses). *P < 0.0001.

Morphological Changes

Lenses from the homTg lines failed to attain normal weight and diameter. Compared to Wt, in homTg lenses the weight was reduced by 91% and the diameter was reduced by 64% (Figure 2b and 2c). HomTg lens shape and histology also differed (Figure 3). The anterior and posterior segments lacked typical opposite-end curvature and clearly failed to overlap and form normal suture branches between and within successive growth shells (Figure 3a). The lens epithelium in homTg lenses had vacuoles (Figure 3b). The equatorial segments were also vacuolated in these lenses (Figure 3c). In homTg, a large number of nucleated cells were present throughout lens, indicating defects in denucleation of fiber cells (Figure 3d). In contrast, lenses from hemTg, Wt, and glyoxalase-I-overexpression mice contained no vacuoles or nucleated fiber cells.

Figure 3.

Morphological changes in the lens. Hematoxylin and eosin staining revealed differences in homTg lenses compared to Wt and other transgenic lines. (a) In the whole eye, the lens shape and size differed markedly in homTg. (b) The epithelial cells of homTg lenses had vacuoles (arrow). (c) At the lens equator, the homTg lens had vacuoles (arrow). (d) In the lens nucleus, the homTg lens had nucleated fiber cells (arrow). In contrast, all three features were normal in hemTg, TgGlo I, and Wt lenses (magnification: a, x 4; b, c, and d, x 10).

To investigate ultrastructural morphological differences in epithelial and fiber cells, we prepared homTg and Wt lenses for thin-section electron microscopy. In the Wt, the fiber cell architecture was regular (Figure 4a). Characteristic gap junctions were frequent between fiber cells. In contrast, in homTg lenses, the fiber cells were disorganized and swollen with many large vacuolar structures (Figure 4b). Gap junctions were scarce. The epithelial cells in Wt lenses formed a monolayer, and both epithelial and fiber cells showed tight connections in lenses (Figure 4c). In contrast in homTg lenses, the epithelial cells irregularly arranged, interdigitated with each other, and were loosely opposed to the fiber cells (Figure 4d).

Figure 4.

Morphological changes in the lens revealed by electron microscopy. The Wt lens had normal fiber cell architecture (a) and flat epithelial cells (c). The fiber cells in homTg (b) were disorganized and had large vacuole-like structures. The epithelial cells in homTg (d) were irregularly arranged. Arrowhead indicates space between the epithelial and fiber cells and between fiber cells. Circles indicate gap junctions. Magnification, x 10 000. FC, fiber cell; V, vacuole.

Lenses from homTg lines had a strikingly dense nuclear cataract (Figure 5a) and opacity (Figure 5b); the cataract developed in both homozygous lines after about 3 months of age. The hemTg lens had no evidence of cataract formation and was similar to the Wt and glyoxalase-I-overexpression mouse lenses at this age.

Figure 5.

Lens opacity in transgenic mice. (a) Slit lamp images revealed that lenses from homTg mice were cataractous by 3 months of age. (b) Three-month-old homTg lenses had marked opacity. Lenses from age-matched hemTg, TgGlo I, and Wt mice were normal.

IDO Overexpression

IDO activity in the lens was quantified by estimating the amount of KYN formed upon incubation of water-soluble lens proteins with 1 mM L-tryptophan. Reverse-phase HPLC revealed a peak at 12-14 min that corresponded to KYN (Figure 6a). Both homTg and hemTg lenses showed IDO activity whereas the Wt showed negligible activity (Figure 6b). Enzyme activity increased approximately eightfold in homTg compared to hemTg lenses. Heat treatment (65°C, 15 min) abolished enzyme activity, and incubation with the IDO inhibitor, 1-methyl tryptophan resulted in a 40% reduction in the enzyme activity in homTg lens proteins confirming that the measured activity was from IDO.

Figure 6.

Indoleamine 2,3-dioxygenase (IDO) activity in the lens. IDO activity was estimated by measuring its product, kynurenine, by reverse-phase HPLC. (a) The elution profile of kynurenine showed that the peak between 12 and 14 min was kynurenine. Top panel, kynurenine standard and bottom panel, kynurenine in homTg lens extract. (b) homTg lenses had eightfold higher IDO activity than hemTg lenses. Methyltryptophan (MT)-treated homTg lens extract (homTg MT) had 40% less activity compared to the untreated sample, whereas the activity in Wt lens extract and heat-treated homTg lens extract (homTg heat) was negligible. Results shown are mean ± s.d. of five lenses from five mice. *P < 0.0001. (c) Immunohistochemical localization of IDO in the lens. Prominent DAB staining (dark brown) was observed throughout the homTg lens. In hemTg lens, staining occurred only in the anterior epithelium and to some extent in the bow and outer posterior regions. Wt lens did not show appreciable DAB staining (magnification x 4). The negative control (labeled-control), where the primary was omitted did not show DAB staining. The inset shows western blotting for IDO (45 kDa) in the water-soluble proteins using the IDO monoclonal antibody (1:5000 diluted). Lane 1, Wt; lane 2, hemTg; lane 3, homTg. GAPDH loading control (36 kDa) is also shown. (d) Quantitative real-time PCR analysis for hIDO mRNA showed that hIDO mRNA was 4.5-fold higher in homTg lens than in hemTg lens. In the Wt lens no hIDO mRNA was present. Results shown are mean ± s.d. of five lenses from five mice. *P < 0.0001.

IDO localization in the lens was determined by immunohistochemistry (Figure 6c). In homTg lens, prominent DAB staining was observed throughout the lens. In hemTg lens, staining occurred only in the anterior epithelium and to some extent in the bow region and outer posterior regions. The negative control where the primary antibody was omitted did not show immunoreaction. The Wt lens did not show appreciable DAB staining. Western blotting using an mAb for hIDO showed a protein band corresponding to the molecular weight of IDO (45 kDa).

The hIDO mRNA level was assessed with real-time PCR. The mRNA level was 4.5-fold higher in homTg than in hemTg lenses. Wt lenses showed no hIDO mRNA (Figure 6d). These data suggest that both the IDO transcript and protein are elevated in homTg and hemTg lenses and resulted in higher enzyme activity.

Kynurenine Content and Protein Modification

Lenses from transgenic animals showed KYN, whereas it was not detected in Wt lenses (Figure 7a). KYN levels were approximately threefold higher in lenses from homTg compared to those from hemTg. To determine if KYN formation occurred in younger than 3-month-old lenses, we determined KYN content in 1-month-old Wt and homTg lenses. At this age, homTg lenses were transparent, but much smaller than Wt lenses of same age. The KYN levels at this age were already elevated (Supplementary Figure S1a).

Figure 7.

Lens kynurenine content and protein modification. (a) Kynurenine content was estimated by reverse-phase HPLC. Kynurenine content was threefold greater in homTg than in hemTg but was not detected in Wt lens. (b) ELISA for kynurenine-induced modifications of lens proteins. Modifications were ~2.6 times greater in homTg lens than in hemTg or Wt lens. Results shown are mean ± s.d. of five independent experiments of five lenses from five mice. *P < 0.0001; **P=0.002. (c) Immunostaining for kynurenine modification. Marked kynurenine modifications (red) were found in homTg lens (panel C3) especially in the nucleus but not in Wt (panel C1) or hemTg (panel C2) lenses. Panel C4 presents regions at higher magnification to show kynurenine modification of nuclei (top) and fiber cells (bottom). The immune reactivity in homTg lens was markedly reduced in the absence of anti-kynurenine mAb (panel C5) and when the antibody was preincubated with kynurenine-modified RNase A (panel C6). Nuclei were stained with DAPI (blue) (magnification x 40). Figures are representative of images obtained from three independent experiments.

KYN-modified proteins were measured by a direct ELISA. The mAb used in this assay was developed by us and it recognizes mainly the Michael adduct formed from the reaction of histidine with deaminated KYN.[25] The homTg lenses exhibited ~2.7 times greater KYN modifications compared to hemTg lenses. The lenses from Wt showed little modifications (Figure 7b). No immunoreactivity was present when the primary antibody was preincubated with KYN-modified RNase A (data not shown). These data suggest that overexpression of hIDO resulted in greater KYN and KYN-modified proteins in the transgenic animal lenses.

To localize lens protein modifications produced by KYN, we performed immunohistochemistry on paraffin-embedded sections. Immunoreactivity was present throughout the homTg lens, but predominated in the central nuclear region (Figure 7c). The hemTg (C2) lenses showed much weaker immunoreactivity compared to the homTg (C3). Interestingly, immunoreactivity in homTg lenses coincided with the nucleated fiber cells. The immunoreactivity was markedly reduced when the antibody was preincubated with KYN-modified RNase A (C6). Together these results demonstrate that in transgenic lenses, enhanced IDO activity leads to higher KYN content and KYN-modified proteins. These results also suggest that KYN modifications may be responsible for nucleated undifferentiated fiber cells.

To determine if KYN modification preceded cataract development, we determined KYN modification by immunohistochemistry in 1-month-old homTg and Wt lenses. Unlike 3-month-old lenses (Figure 7, C3), 1-month-old lenses showed KYN modified proteins in the outer cortical region, more densely at the bow regions (Supplementary Figure S1b). These observations together with results in Supplementary Figure S1a suggest that KYN formation and KYN modification of proteins occur before development of mature cataracts in homTg animals.

Incubation of Lens Homogenate and Lens Organ Culture with L-Tryptophan

Incubation of water-soluble lens proteins from transgenic mice with L-tryptophan resulted in KYN formation (Figure 8a). KYN formation increased approximately sevenfold in proteins from homTg lens compared with hemTg lenses. The Wt lens proteins did not produce KYN. These experiments also revealed that ~12% of added L-tryptophan was converted to KYN within 1 h in homTg lens protein extract, whereas only ~1.5% was converted in hemTg lens proteins.

Figure 8.

Kynurenine and GSH in the lens. (a) Incubation of lens proteins with L-tryptophan revealed an approximately sevenfold increase in kynurenine formation in homTg lenses relative to hemTg lenses. The Wt lens proteins did not produce kynurenine. (b) Lens organ culture with (black bar) and without (gray bar) L-tryptophan. Kynurenine content increased in transgenic lenses but not in Wt lens. (c) Lens GSH content decreased by ~36% in homTg lenses relative to Wt and hemTg lenses. Results shown are mean ± s.d. of five lenses from five mice. *P < 0.0001.

To determine whether KYN formation occurs in intact lenses, we performed organ culture experiments, in which we incubated whole lenses with 1 mM L-tryptophan. Similar to results for isolated proteins, KYN increased by 5-fold in homTg lenses and 3.2-fold in hemTg lenses (Figure 8b). The formation of KYN from lens proteins and whole lenses of transgenic animals also suggests that IDO overexpression causes higher production of KYN in the lens.

Glutathione Level

GSH levels were reduced by 36% in homTg lenses in comparison to Wt and hemTg lenses (Figure 8c). In 1-month-old homTg lenses, GSH levels were reduced by ~13% in comparison to 1-month-old Wt lenses (Supplementary Figure S2). This suggests that the loss of GSH started to occur (along with KYN formation) before the onset of mature cataracts in homTg animals.


TUNEL staining revealed apoptosis in the homTg lens nucleus (Figure 9a). Staining of the same sections for KYN modifications revealed that many cells that were TUNEL positive were also positive for KYN modifications (Figure 9b), and that KYN modifications appeared to be present in both the cytoplasm and the nucleus. These findings suggest IDO overexpression causes apoptosis of fiber cells and could be due to KYN modifications of cytosolic and nuclear proteins and DNA. Apoptosis of fiber cells were seen, albeit to a lesser extent, in 1-month-old homTg lenses (Supplementary Figure S3). This, together with data in Supplementary Figures S1a and b, suggests that apoptosis, probably as a consequence of KYN formation, started to occur in precataractous lenses.

Figure 9.

Apoptosis and kynurenine modification in the lens. (a) TUNEL-positive cells (green) were detected in homTg lens but not in Wt and hemTg lenses or in homTg lens incubated without terminal transferase. Nuclei were stained with DAPI (blue) (magnification x 40). (b) Cells that were positive for TUNEL (green) were also positive for kynurenine modifications (red) (magnification x 40). Images in the lower panels are sections magnified from the panels above. Figures are representative of images obtained from three independent experiments.

To further establish apoptosis in transgenic lenses, we measured caspase-3 and caspase-9 in lens homogenate. Caspase-3 activity was fivefold higher in homTg lenses when compared to hemTg and Wt lenses (Figure 10a). Likewise, caspase-9 activity was fivefold higher in homTg animals compared to lenses from hemTg and Wt lenses (Figure 10b). These results suggest that IDO overexpression activates apoptosis in lens fiber cells.

Identification of KYN-Modified Proteins

Proteins from homTg lenses were immunoprecipitated using the KYN mAb. The immunoprecipitated proteins were separated on SDS-PAGE. Several protein bands were observed (Figure 11). The major protein band was subjected to mass spectrometry analysis. Several proteins, mostly ß-crystallin subtypes, were identified, in addition to αA- and αB-crystallins (Table 1). Our initial attempts to identify specific sites of KYN modification on ß-crystallins were not successful, possibly because of structural diversity of modifications and/or low abundance of modification relative to protein concentration, but we hope to identify such sites with more sophisticated approaches in the future. Our attempts to identify proteins by western blotting also failed. The KYN antibody did not react with proteins in western blotting (these proteins had relatively large quantities of KYN modification as judged by ELISA, see Figure 7b). We believe that the antigenic epitope gets masked or destroyed during SDS-PAGE and/or western blotting (possibly by the reduction of the ketone group on the Michael adduct). Further studies are needed to verify this possibility.

Figure 11.

Identification of KYN-modified proteins. Water-soluble lens proteins were incubated with KYN antibody, Protein A/G sepharose gel was pelleted and subjected to SDS-PAGE followed by western blotting using KYN antibody. The major protein band (thick arrow) was analyzed by mass spectrometry and the results are presented in Table 1. Lane 1, Wt lens extract; lane 2, homTg lens extract; lane 3, Gel+homTg lens extract; lane 4, Gel+Ab. The heavy and light chains of the antibody are indicated by thin arrows.