Sample Western blot showing activity in healthy controls against aqueous-soluble (A) and detergent-soluble (B) retinal antigens. Each numbered lane corresponds to an individual sample (NC indicates negative control). Protein weights in kilodaltons are shown at the side of each figure. Bands corresponding to the reduced IgG heavy-chain band were not included in analysis (denoted by asterisk). Please note that the data from sample number 10 were omitted owing to poor quality. Bands included in analysis from panel A include the following: lane 6, 26.3 kDa; lane 7, 29.8 kDa; lane 9, 62.8 kDa; lane 12, 41.4 kDa; and lane 13, 40.7 kDa. Bands included in analysis from panel B include the following: lane 3, 33.2 kDa; lane 4, 112.4 kDa; lane 6, 27.8 kDa; lane 7, 31.0 kDa, 42.0 kDa; and lane 12, 44.6 kDa.
Ko AC, Brinton JP, Mahajan VB, Zimmerman B, Brinton GS, Stone EM, Folk JC, Mullins RF. Seroreactivity Against Aqueous-Soluble and Detergent-Soluble Retinal Proteins in Posterior Uveitis. Arch Ophthalmol. 2011;129(4):415-420. doi:10.1001/archophthalmol.2011.65
To characterize the seroreactivity against retinal proteins in patients with posterior uveitis, retinal disease of noninflammatory origin, and healthy controls.
Patients with posterior uveitis (n = 47), molecularly confirmed photoreceptor degenerations (n = 11), and healthy controls (n = 33) received dilated fundus examinations at the University of Iowa. Aqueous-soluble and detergent-soluble fractions of human retina were separated by gel electrophoresis and transferred to polyvinylidene fluoride membranes. Membranes were probed with patient serum samples to detect IgG, IgA, and IgM human antibodies that react with retinal antigens. The number of bands detected by Western blot was counted, and their molecular weights were determined.
Antibodies recognizing retinal proteins were found in healthy controls, in patients with posterior uveitis, and in patients with molecularly confirmed heritable retinal degenerations. In healthy controls, 42% of individuals had circulating autoantibodies that recognized retinal proteins. Healthy controls had a low odds ratio of serum reactivity to soluble antigens (0.7; 95% confidence interval [CI], 0.4-1.2). Patients with inflammatory retinal diseases and inherited retinal diseases had 4.89 (95% CI, 2.25-10.64; P < .001) and 2.71 (95% CI, 1.19-6.16; P = .02) times more activity against soluble retinal antigens compared with controls.
Healthy control patients exhibited a significantly higher level of background autoantibody activity against retinal proteins than previously reported. Antibody activity in healthy controls was primarily directed against membrane-bound retinal proteins, whereas in patients with pathologic retinal conditions, antibodies targeting nonmembrane-bound retinal proteins predominate.
Posterior uveitis encompasses a heterogeneous group of intraocular inflammatory conditions that can ultimately lead to blindness. While these diseases are mediated through inflammatory mechanisms, the exact cause of most posterior uveitides remains unknown. Although the eye is an immunologically privileged site, animal models have shown a breakdown of the blood retinal barrier allowing access of activated T cells into the eye.1 In experimental autoimmune uveitis (EAU), a murine model of uveitis, damage to the retina has been shown to begin with activation of antiretinal CD4+ T cells followed by infiltration into the retina and the recruitment of leukocytes that cause injury to the retina.2,3 While studies have suggested that T cells targeting retinal antigens may play a role in the progression of retinal degeneration in uveitis, relatively little is known about the role that the humoral immune response plays in patients with posterior uveitis.
It is plausible that the inflammatory pathway leading to uveitis includes a humoral component. Antibodies recognizing retinal proteins are present in the serum of patients affected by various uveitis syndromes including acute zonal occult outer retinopathy, toxoplasmosis, multiple evanescent white dot syndrome, and birdshot retinochoroidopathy.4- 6 Numerous other retinopathies have been associated with the production of autoantibodies recognizing retinal proteins, including cancer-associated retinopathy, melanoma-associated retinopathy, and nonneoplastic autoimmune retinopathy.4,7- 10 However, antibodies that react with retinal proteins are also present in the serum samples of patients affected with retinal degenerations not typically associated with an autoimmune response. For example, some patients with retinitis pigmentosa have antirecoverin and other antibodies recognizing retinal proteins circulating in their serum.4,11,12 A summary of retinal proteins for which antiretinal antibody activity is known to be associated with retinopathy is given in Table 1.
The exact role of autoantibodies that bind retinal proteins in posterior uveitis remains unclear. These autoantibodies may represent innocuous markers of retinal damage, secondary complications that aggravate the underlying disease, or a primary instigator of posterior uveitis. Further complicating the issue is the fact that autoantibodies that label retinal proteins have been found in pooled blood bank serum samples and in control subjects with no history of ocular disease.14,15 Mitigating influences, whether of a genetic or environmental source, may determine whether the presence of these antibodies leads to visual complications.
This study was performed to determine the seroreactivity against retinal proteins in patients affected with posterior uveitis syndromes and to compare them with those of healthy controls and patients with retinal disease of noninflammatory origin. Characterization of the autoantibody patterns of patients within this group of disorders will help further elucidate the role that humoral factors play in these diseases. The presence of consistent antiretinal reactivity profiles within the subtypes of posterior uveitis could be used as markers to support diagnoses, predict disease severity, or monitor disease progression. Identification of targeted retinal proteins will allow for further studies in disease mechanism that may lead to the addition of specific treatments targeting the humoral response to those currently used for uveitis treatment.16
This study was approved by the institutional review boards at the University of Iowa and was conducted in accordance with the Health Insurance Portability and Accountability Act and the Declaration of Helsinki. Written informed consent was obtained from all subjects or their parents or guardians. All patients and controls had a complete eye examination with dilated pupils. For each individual included in the group of healthy controls, a board-certified ophthalmologist reviewed a detailed medical history. Control serum samples were obtained only from individuals with no known systemic inflammatory or autoimmune disease and with no eye disease other than refractive error or age-related cataract. Color fundus photographs were taken of all control subjects for documentation purposes. The uveitis syndromes were diagnosed by a fellowship-trained vitreoretinal specialist. Molecular diagnoses of patients with inherited retinal degenerations were made by the John and Marcia Carver Nonprofit Genetic Testing Laboratory, Iowa City.
Human donor eyes were obtained from the Iowa Lions Eye Bank, Iowa City. Human retinas were collected from 3 donors without known ocular pathologic conditions within 6.5 to 17 hours after death. Each whole retina was homogenized in phosphate-buffered saline (PBS) containing protease inhibitors (Roche Complete Kit; Roche Diagnostics Corp, Indianapolis, Indiana). The solution was centrifuged at 16 300 g for 15 minutes. The soluble fractions of retina protein were collected in the form of supernatant and pooled into 1 protein sample. The remaining pellet was then rinsed 3 times via resuspension in 500 μL of the same solution and centrifuged at 16 300 g for 15 minutes, and the resulting supernatant was discarded each time. The final rinsed pellet was resuspended in 175-μL PBS + 1% Triton X-100, ground and vortexed, and centrifuged at 3000 g for 2 minutes. The insoluble fractions of retina protein were collected in the form of a supernatant and pooled into 1 protein sample. A colorimetric protein assay was used to determine the concentration of protein in each solution (Bio-Rad DC Protein Assay Kit; Bio-Rad Laboratories, Hercules, California).
Blood was drawn from patients and controls. Samples were centrifuged, and serum samples were fractionated within 12 hours of collection and stored at −80°C until further processing.
For each Western blot, 154.5 μg of soluble or insoluble retina protein was suspended in a solution containing 1× NuPAGE LDS (lithium dodecyl sulfate) sample buffer and 1× NuPAGE reducing agent (Invitrogen, Carlsbad, California). Proteins were separated with a 1.5-mm × 2D-well NuPAGE 4% to 12% Bis-Tris gel at 200 V. After electrophoresis, the proteins were transferred from the gel to a polyvinylidene fluoride membrane at 30 V for 2 hours. The membrane was dried overnight and cut into 5-mm wide strips and wetted in methanol. Mass processing of individual samples allowing the use of uniform reagents was achieved using a device invented by one of the authors (A.C.K.). Membrane strips were blocked for 1 hour in 5% nonfat dry milk (NFDM) in PBS. The strips were then each incubated with human serum at a dilution of 1:500 in 2% NFDM for 1 hour and rinsed with 1× Tris-buffered saline + 0.1% Tween 20 (TBST). The strips were washed twice for 10 minutes in TBST and probed with a 1:30 000 dilution of horseradish peroxidase conjugate of goat-antihuman IgG/A/M antibody (Pierce Protein Research Products, Rockford, Illinois) and washed 3 times for 10 minutes each in TBST. The membranes were then developed using uniform exposure times with the ECL Plus Western Blotting Detection System (GE Healthcare, Amersham, England). Membrane strips that were only probed with secondary antibody served as negative controls; membrane strips that were probed with a serum sample that had previously demonstrated antienolase antibody activity served as a positive control.13 The number of bands on Western blot was counted in a masked fashion for each patient sample. Bands corresponding to the reduced IgG heavy-chain band were not included in analysis; visualized bands were omitted if the band signal intensity was less than 1.5 times that of the background for that sample as quantified using Adobe Photoshop (version 10.0; Adobe Systems Inc, San Jose, California). To determine the protein mass of each visualized band on Western blot, protein standards were used to derive a linear relationship between the logarithm of protein masses and its Rf value. Rf values were determined for each retinal protein to which serum showed reactivity, and their masses were calculated. For each serum sample, the number of positive bands and their molecular weights were determined for both retinal fractions.
The prevalence of antibodies recognizing retinal proteins in healthy controls and in the 2 categories of ocular conditions was computed and then compared using the Fisher exact test. Of the visible bands, the ratio of aqueous-soluble to detergent-soluble bands (or odds of aqueous-soluble to detergent-soluble bands) within each of the ocular conditions and healthy controls were compared using logistic regression analysis with the logit model fitted by the method of generalized estimating equations. Because a subject can have more than 1 band, this method was used to account for the correlation of outcomes (ie, bands representing activity against aqueous-soluble or detergent-soluble fractions of retina protein) of the bands from the same subject. From this analysis, the comparison of the ratio of aqueous-soluble to detergent-soluble activity between those with ocular conditions and healthy controls was expressed and tested in terms of the odds ratio (OR) of a band being soluble with ocular condition relative to normal. This same logistic regression analysis was performed on the bands from the healthy controls with age group as the independent variable to test for the effect of age on the aqueous-soluble to detergent-soluble band ratio.
Samples were obtained from 47 patients with ocular inflammatory diseases (7 acute zonal occult outer retinopathy, 14 multifocal choroiditis, and 26 ocular histoplasmosis); 11 patients with molecularly confirmed photoreceptor degenerations (9 cases of autosomal dominant retinitis pigmentosa and 2 cases of Leber congenital amaurosis); and 33 healthy control aged patients (22 women and 11 men). Thirteen controls were aged between 18 and 30 years, 14 controls were aged between 31 and 60 years, and 6 controls were older than 60 years.
Autoantibodies that bind to retinal antigens were found in all disease groups and in healthy controls. The Figure shows a sample Western blot with associated positive bands. A summary of Western blot findings is listed by disease in Table 2, and listings of detected bands by calculated molecular weight are given in Table 3 and Table 4. The proportion of patients with antiretinal autoantibody activity in healthy controls; inflammatory retinal diseases (acute zonal occult outer retinopathy, multifocal choroiditis, and ocular histoplasmosis); and inherited retinal degeneration groups (retinitis pigmentosa and Leber congenital amaurosis) were 42.4%, 48.9%, and 45.4%, respectively. Within the healthy control group, the proportion of patients with antiretinal autoantibody activity in the 18- to 30-year, 31- to 60-year, and older-than-60-year groups were 61.5%, 42.8%, and 0%, respectively.
These data reveal statistically significant differences in autoimmune activity against soluble vs insoluble retinal antigens. Healthy controls had a low OR of serum reactivity to aqueous-soluble antigens (OR, 0.7; 95% confidence interval [CI], 0.4-1.2). However, both disease groups had high odds ratios of serum reactivity to aqueous-soluble antigens. As a group, the inflammatory retinal diseases showed predominant activity against aqueous-soluble antigens (OR, 3.6; 95% CI, 2.0-6.6). The inherited retinopathy group (OR, 2.0; 95% CI, 1.0-3.8) also showed more activity against soluble retinal antigens. Compared with healthy controls, patients with inflammatory retinal diseases and inherited retinal diseases had 4.89 (95% CI, 2.25-10.64; P < .001) and 2.71 (95% CI, 1.19-6.16; P = .02) times more bands on Western blot representing activity against soluble retinal antigens. There was an increasing trend in the ratio of aqueous-soluble to detergent-soluble bands with older age that was not found to be statistically significant.
There is compelling evidence that antibodies reactive to specific retinal antigens play a role in autoimmune retinopathy. In addition, autoimmune retinopathy may mimic inherited retinal disease in some cases, as we noted recently in a patient with anti–γ-enolase activity.13 The role of circulating antibodies reactive to retinal proteins in the pathogenesis of other retinal diseases is still unclear and was discussed in a recent letter and a response to that letter in this journal.17- 19 Previous studies looking at healthy populations have reported varying levels of antiretinal antibody activity. The primary limitation that has faced the field has been in the definition of a normal population. One group of investigators used a commercially available set of human serum samples from random blood bank donors,14 whereas another group used serum samples from patients who answered a questionnaire that they had no ocular disease or vision-related problems.15 An earlier study tested reactivity in normal subjects against bovine rather than human retina.20 Jampol and Fishman stated in an editorial17 and in a reply to a subsequent letter to the editor18,19 that a more rigorously defined control group is needed.
In the present study, we used a control group for which a board-certified ophthalmologist reviewed a detailed medical history and performed a comprehensive dilated eye evaluation on every control subject. Subjects were eliminated if they had any history of systemic inflammatory or autoimmune disease or even 1 chorioretinal scar in the fundus. Even with these strict criteria, 42% of the controls possessed antibodies that reacted with some retinal proteins. The usefulness of testing serum samples for antibodies recognizing retinal proteins as a cause for disease depends on their prevalence in a population without the disease. This finding calls into question the significance of laboratory results that do not limit test positivity to the finding of specific, identifiable, pathogenic antibodies such as antirecoverin and antienolase.
To our knowledge, this is the first human antiretinal autoantibody study to use aqueous-soluble (nonmembrane-bound protein) and detergent-soluble (membrane-associated protein) fractions of retina. The fractioning of retina proteins into these 2 groups offers additional information about the location of targeted retinal antigens and study of less complex mixtures of proteins. Furthermore, because the concentration of protein in the detergent-soluble fraction is less than that of the aqueous-soluble fraction in an unfractionated retina, the separation of these 2 components allows for standardization of the amount of protein exposed to patients' serum samples. The separation of these 2 fractions of retina conferred greater sensitivity to our laboratory testing process.
Curiously, the healthy control group had more autoantibody activity directed against detergent-soluble retinal antigens, whereas patients with posterior uveitis and molecularly confirmed retinal degenerations had more activity directed against aqueous-soluble retinal antigens. This was statistically significant and may point to a mechanism by which the antibodies were produced. The higher prevalence of serum reactivity to the aqueous fraction of retina protein in patients with ocular conditions may be caused by exposure to internal cell proteins from dying or injured retinal cells. It is unclear whether these antibodies are pathogenic or simply represent epiphenomena of disease. In healthy controls, the presence of autoantibodies recognizing retinal antigens despite the lack of pathologic ocular conditions may indicate multiple subclinical disruptions of the blood-retinal barrier and subsequent exposure of the immune system to cell surfaces in the retina. The higher proportion of activity against detergent-soluble retinal antigens may be evidence of immune exposure to mainly surface-bound proteins on undamaged retinal cells.
This study has limitations. First, stringent inclusion criteria for controls made it difficult to amass a large control group, particularly of healthy patients older than 60 years. The data show a trend in decreasing prevalence of antiretinal autoantibody activity with increasing age, but more control subjects are needed to make any statistically significant conclusion. There was also difficulty in amassing a large group of molecularly confirmed retinal degenerations owing to the rarity of the disease. Serum samples of patients with retinal degenerations without molecular confirmation were not included in this study in order to exclude possible phenocopies of inherited eye disease.13 Second, despite thorough ophthalmologic examination and review of medical charts, individuals within the control group may still have inflammatory disease that is not yet clinically evident. Third, it is possible that some of the bands on this or any similar Western blot are representative of autoantibodies cross-reacting with retinal protein epitopes rather than autoantibodies specifically raised against retinal proteins. Whether their primary target is the retina, however, does not change their clinical significance because antibodies have been shown to play a role in retinal pathologic changes if they are allowed to gain access to the retina (as with antirecovern autoantibodies in cancer-associated retinopathy related to a small-cell lung carcinoma7). Reactive bands on Western blot may also result from molecular mimicry. Among patients with autoantibodies recognizing epitopes on retinal proteins, whether disease develops may be related to normal variation between individuals in the integrity of the blood-retina barrier or ocular immunity. Fourth, for any band on Western blot, there is the possibility of it containing multiple proteins of the same or very similar molecular weights. The use of Western blots also limits the detection of different antibodies reactive to different epitopes of the same protein. Finally, our data analysis is based on the number of bands found on Western blot. We were unable to quantify and compare the signal intensity of each individual band because our exposure-based detection method resulted in saturation of high-intensity bands. This limits our ability to separate out subtle gradations in antiretinal activity.
Because many of the control patients had antibodies recognizing retinal proteins in their serum samples, their mere presence in a patient cannot be definitively considered the cause of an unusual retinopathy. However there is good evidence that antibodies to enolase and especially to recoverin cause retinal degeneration.7,11,13,20 Further studies are needed to determine the identity of pathogenic autoantibodies as well as the significance of reactivity against aqueous- and detergent-soluble retinal proteins. Patients with additional types of posterior uveitis should be tested, and serum samples should also be assessed during different stages of disease to correlate autoantibody activity with appearance of retinal inflammation. Additional control patients would be helpful to compare against the results found in patients with inflammation. Although the controls in this study had many bands, there were differences between controls and patients with inflammatory disease. Therefore, closer examination of the bound antigens—through mass spectrometry or other methods—may unmask antibodies found in the patients that are pathogenic and are not found in control patients. Finally, the high-intensity bands on Western blots should also be studied to find clues to the enigma of posterior uveitis.
Correspondence: Robert F. Mullins, PhD, Department of Ophthalmology and Visual Sciences, University of Iowa, 375 Newton Rd, Iowa City, IA 52242 (Robert-Mullins@uiowa.edu).
Submitted for Publication: April 16, 2010; final revision received July 6, 2010; accepted July 11, 2010.
Author Contributions: Ms Ko and Dr Brinton contributed equally to this work.
Financial Disclosure: None reported.
Funding/Support: This study was supported in part by the Doris Duke Research Foundation (Ms Ko), Research to Prevent Blindness (Dr Folk), Howard Hughes Medical Institute (Dr Stone), and NEI grants EY017451 (Dr Mullins) and EY016822 (Dr Stone).
Additional Contributions: Thomas A. Oetting, MD, A. Tim Johnson, MD, Andrew G. Lee, MD, and Hilary A. Beaver, MD, helped in the recruitment of patients for this study.