Human Melanoma-Associated Retinopathy (MAR) Antibodies Alter the Retinal ON-Response of the Monkey ERG In Vivo

Abstract

purpose. Melanoma-associated retinopathy (MAR) is a paraneoplastic condition that causes visual symptoms of night-blindness and photopsias. The electroretinogram (ERG) of MAR patients is characteristically abnormal in a way that implicates retinal depolarizing bipolar cell (DBC) dysfunction. Whether an injection of IgG from MAR patients into the vitreous of monkeys would alter the ERG acutely as a demonstration of a functional basis for patients’ visual symptoms was explored.

methods. MAR IgG was isolated from three visually symptomatic melanoma patients. Control IgG was from melanoma patients with no vision problems. The ERG was monitored after intravitreal injections into monkey eyes. One eye was injected with 2-amino-4-phosphonobutyric acid (APB), which is known to block DBC ON-pathway responses. Retinal immunocytochemistry was performed using fluorescein isothiocyanate–labeled goat anti-human IgG.

results. Within 1 to 3 hours after MAR IgG injection, the ERG photopic b-wave was diminished, with far less effect on the a- and d-waves. These changes are characteristic of DBC dysfunction and were similar to the effects of APB. The scotopic ERG b-wave, which reflects activity of rod-driven DBCs, showed a loss of amplitude and threshold sensitivity after MAR IgG. Retinal immunocytochemistry with anti-IgG antibody showed IgG penetration throughout the retinal layers, but staining was not specific for a single type of retinal neuron.

conclusions. Intravitreal injection of human MAR IgG altered the monkey ERG acutely in ways that implicate functional disruption of retinal DBC signaling. These results support the hypothesis that MAR IgG circulating antibodies are responsible for the reported visual symptoms. Bipolar cells in the ON-pathway appear to be affected more than OFF-pathway bipolar cells of the cone pathway in this acute preparation.

Some patients with metastatic cutaneous malignant melanoma experience visual symptoms of night blindness and shimmering photopsias that begin months to years after tumor diagnosis. ^{1 2 3} This visual paraneoplastic syndrome is termed melanoma-associated retinopathy (MAR). The electroretinogram (ERG) in MAR patients shows a reduction or loss of the photopic b-wave, with relative sparing of the a- and d-waves. ^{1 3} The ERG scotopic b-wave is also reduced. ¹ These ERG changes are characteristic of a functional defect in the retinal ON-pathway responses involving depolarizing bipolar cells (DBCs). ^{4 5}  

In addition to MAR, cancer-associated retinopathy (CAR) also produces vision symptoms. ^{6 7 8} Serum antibodies in MAR and CAR are reactive against retinal cells and proteins, ^{2 7 8} suggesting an autoimmune basis for both conditions. The proposed mechanism involves a B-lymphocyte response against tumor cell antigens, resulting in autoantibodies that cross-react with retinal cellular antigens. MAR antibodies bind to bipolar cells and their dendrites in the outer plexiform layer of human retinal sections. ² CAR antibodies react with recoverin, a 23-kDa calcium-binding protein found in photoreceptors and some types of cone bipolar cells. ^{7 8} This presumed mechanism of autoimmune disease is also known for the hearing disorders, in which some paraneoplastic inner ear diseases cause sensory neural hearing loss. ⁹ Treatment by immunosuppression can sometimes partially rescue the vision or hearing deficit, this is a further indication of immune system involvement in the sensory component of these conditions. ^{6 10}  

In the case of CAR, intravitreal injection of anti-recoverin antibodies induces death of recoverin-positive rat photoreceptor and bipolar cells in vitro ¹¹ and in vivo. ¹² We used this experimental approach and gave MAR IgG by intravitreal injection in monkey eyes to explore whether this could produce an acute functional change in the ERG. We found that within 3 hours after MAR IgG injection the ERG waveform changed in a way that roughly mimicked that reported for MAR patients. The ERG changes in monkey were characteristic of deficient signaling by depolarizing (ON-) bipolar cells and could be simulated by application of the glutamate agonist 2-amino-4-phosphonobutyric acid (APB), which blocks signal transmission to DBCs. ¹³ This demonstrates that MAR antibodies can affect the function of retinal DBCs relatively acutely and provides additional evidence for an autoimmune etiology for MAR disease. 

Materials and Methods

These studies were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Four rhesus monkeys (Macaca mulatta) were used. Animals were sedated with ketamine hydrochloride (10 mg/kg/h, IM) and xylazine (1 mg/kg/h, IM). ERGs were recorded simultaneously from both eyes using Burian-Alan bipolar corneal electrodes (Hansen Ophthalmic Development Laboratory, Iowa City, IA) after corneal anesthesia (topical proparacaine 0.5%) and full pupillary dilation (phenylephrine HCl 10% and atropine 1%). The indifferent electrode was a stainless steel needle placed subcutaneously on the back. Signals were amplified at 10,000-gain at 0.1 to 1000 Hz (−3dB points), and a notch filter was used to minimize contamination from 60 Hz line noise. Responses were digitized at 1 kHz rate, averaged, and analyzed off-line. The Ganzfeld stimulus had maximum intensity of 2.3 log cd/m², and intensity was controlled with neutral density filters. Animals were dark-adapted for 1 hour before scotopic ERG recordings, which were performed with 50-msec flashes. Photopic ERGs were recorded on a steady background of 34 cd/m², and photopic responses were elicited with 200-msec flashes to distinguish between the b-wave and d-wave responses. 

MAR IgG was from visually symptomatic MAR patients and prepared by column fractionation. ² We had only a limited number of monkey eyes available for study, consequently we pooled MAR sera from three visually symptomatic patients to enhance the possibility of seeing an ERG effect. All three patients had experienced photopsias and night-blindness and had ERG changes. The IgG serum fraction from all three MAR patients had given strong bipolar labeling by immunocytochemistry. Two of these MAR subjects were previously described. ^{3 14}  

The appropriate dose for MAR IgG was not known, and we used the same concentration of IgG sera, undiluted, as in the previous study ² and injected 0.1 ml volume. The dark-adapted ERG was monitored for indications of a sensitivity change. If none was observed after about 1 hour, a second injection was given, while maintaining dark-adaptation. Elapsed time given in the results for observing ERG changes is after the second injection. Intravitreal injections were made through the pars plana with a 30-gauge needle. Injection of 0.1 ml does not itself alter the ERG, ⁴ and we reconfirmed this in several of these eyes (Table 1) . None of these injections caused retinal hemorrhage or lens damage. 

Control-IgG was from serum of three patients with cutaneous melanoma who were not experiencing any visual symptoms. None of the three reported photopsias or night-blindness, which might have indicated that circulating antibodies had targeted retinal cells, but full ERG evaluations were not performed on these subjects to rule out a“ subclinical” state of disease. We again pooled the IgG fractionated sera from these three subjects to increase the likelihood of observing any ERG changes with minimal usage of monkey eyes. 

One eye (monkey 4) received a 0.1 ml injection of 32 mM APB (Sigma Chemical, St. Louis, MO) solution prepared in phosphate-buffered saline (PBS), with the pH adjusted to 7.3 to 7.4 with NaOH, to demonstrate the expected ERG waveform change associated with blocking light-evoked responses of DBCs under long-flash photopic conditions. ⁴  

Retinal immunocytochemistry was performed on the eyes of one animal euthanatized 2 hours after the ERG recording. After removal, eyes were slit at the pars plana and fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) for 6 hours at 4°C. The anterior segments were removed, and the eyecups were placed in 10% phosphate-buffered sucrose at 4°C. Retinas were cryoprotected in sucrose and cryosectioned at 12 μm thickness. Sections were incubated for 1 hour at room temperature in secondary antibody (goat anti-human IgG labeled with fluorescein isothiocyanate [FITC] at 1:50 in PBS with 0.3% Triton X-100), rinsed twice with PBS for 30 minutes at room temperature, and placed on coverslips in 90% glycerol in PBS containing 2% 1,4-diazabicyclo(2,2,2) octane. The sections were photographed with a Nikon microscope (Tokyo, Japan) equipped for epifluorescence. 

Results

The general strategy was to inject MAR IgG into the vitreous and monitor the ERG for waveform or sensitivity changes. MAR IgG produced ERG changes in all three animals that we studied. Typical photopic ERG changes are illustrated in Figure 1 from monkey 1 (Table 1) . MAR IgG suppressed the photopic b-wave but had minimal effect on the a-wave, and it only smoothed the d-wave without affecting the initial time course of the OFF-response. These MAR IgG effects are consistent with a primary site of action on DBCs, with relative sparing of cone photoreceptors and hyperpolarizing bipolar cells. ^{4 15} The MAR effect compares well with that of APB (Fig. 1 , bottom trace). APB blocks light-evoked activity of the DBCs. ¹³ Note that APB, MAR IgG, and non-MAR control sera all altered the a-wave slope minimally and to a similar degree. We confirmed the MAR IgG findings in two additional monkeys (Fig. 2 , with results from monkey 1 shown for comparison). Although some waveform variability was observed in the three monkeys after MAR IgG, all three cases showed suppression of the b-wave to a greater extent than the d-wave. 

Non-MAR control sera in the fellow eye had the opposite effect and enhanced the photopic b-wave (Fig. 1 , middle trace). The primate photopic b-wave is a balance between a positive-going DBC component and a negative-going component from hyperpolarizing bipolar cells (HBCs), which normally masks the full extent of the DBC b-wave. ⁴ Blocking HBC activity with piperidine dicarboxylic acid causes the b-wave to become larger. ^{4 15} Thus, the enhanced b-wave after non-MAR control sera need not be an effect on DBCs. We do not know why the control sera enhanced the photopic ERG b-wave, but this does not undermine the finding that the MAR IgG appears to target DBC activity. 

MAR IgG also affected the scotopic b-wave (Fig. 3) . The scotopic b-wave derives directly or indirectly from activity of DBCs in the rod pathway. ⁵ The b-wave suppression was quantitatively greater than that caused by the control non-MAR IgG sera (Fig. 3B) . For criterion responses of 50 μV, MAR IgG caused a 1.8 log unit sensitivity loss compared with only 0.6 log unit loss for control sera. We removed these eyes for immunohistochemistry and did not track the ERG changes over a longer period. However, in the two other animals, the MAR IgG effect was reversible with time. As shown in Figure 4 (monkey 3, Table 1 ), MAR IgG acutely caused a 1.2 log unit sensitivity loss of the b-wave (50 μV criterion), and this recovered to within 0.2 log units of baseline when retested at 4 months. The photopic b-wave also recovered in this animal. This recovery suggests that the MAR IgG did not permanently damage retinal neuronal function. We did not study interim time points and do not have a more precise estimation of the recovery time course. 

Eyes of the first monkey (monkey 1) were enucleated and processed for immunofluorescence study about 2 hours after we observed the acute ERG changes. Unlike the previous study in which MAR IgG on human retinal sections showed staining of bipolar soma, ² this was not observed in the retina of monkey 1. However, immunofluorescence was observed throughout the retinal thickness, indicating that the human MAR IgG had penetrated the retina after the intravitreal injection and could account for the effect on the ERG. All sections were treated with FITC-labeled secondary antibody against MAR IgG, not with primary antibody because MAR IgG (i.e., the primary antibody) had been given in vivo by intravitreal injection during ERG recording. A saline-injected control eye showed no retinal labeling, indicating that the secondary antibodies did not react with any retinal structures. The anti-human IgG did cross-react with the monkey IgG in blood vessels in either the experimental or control eyes. 

Discussion

These results provide the first direct experimental evidence that MAR IgG has retinal reactive immunoglobulins that affect retinal function acutely and provide further indication that the vision abnormalities experienced by human MAR patients result from circulating antibodies. The main effect appears to be on DBCs of the ON-pathway of both the cone and rod systems. The photopic ERG d-wave, attributable to the OFF-pathway responses of hyperpolarizing bipolar cells, ¹⁶ was less affected. The suppression of rod ERG b-wave responses in monkey is consistent with MAR patients’ night-blindness symptoms and reduced scotopic b-wave. The rod ERG b-wave recovered to normal after an extended time in these monkey experiments, indicating that the effect is functional and not from cellular toxicity. Reversal has been reported with experimental CAR disease in which antibodies were detected in the photoreceptor layer within 24 hours after intravitreal injection and caused intense staining by 48 hours but faded to a nondetectable level by 7 days. ^{11 12}  

MAR IgG altered the ERG of all three monkeys in similar fashion but to a different extent. It is not surprising that intravitreal injection of a mixture of complex proteins might yield differences in waveforms in different animals. We have previously noted differences in photopic ERG waveforms even after applying glutamate analogs (e.g., see waveforms after APB ^{4 15 16} ). Furthermore, even though all 20 human MAR patients reported to date have night-blindness and photopsias, they show different degrees of ERG alteration. ^{1 2 3} Some MAR patients are reported to show only partial b-wave reduction in early disease stages and initially have ERG asymmetry between the two eyes that later evolves to severe bilateral ERG involvement over the course of several weeks (Kenneth Alexander, personal communication; see also the patient course reported in ^{Ref. 2)} . In MAR patients circulating antibodies would present chronically to the retina. By comparison, our experimental monkey manipulation provided only a single acute application, which may explain why we did not observe a fully electronegative scotopic response as has been reported for some patients. The consistency of the MAR IgG effect on the photopic ERG in producing an electronegative a-/b-wave complex in two of three monkeys (Fig. 2 , animals 1 and 3) seems remarkable in this context. The effect of MAR IgG on the photopic b-wave was quite evident in the waveform when the OFF-response was separated using the long-flash ERG. In all cases, however, both the rod and cone systems showed a deficit of the ON-pathway ERG attributable to an effect on DBCs. These findings provide further evidence of an autoimmune basis for the visual symptoms experienced by MAR patients. 

In an effort to find an alternative non-primate animal model in which to test these ideas further, we conducted a limited set of experiments with MAR IgG in the rod-dominated rat and in the guinea pig, which has a usable photopic ERG. The human MAR IgG decreased the scotopic and photopic ERG responses nonselectively and to a degree similar to that of control serum from normal subjects without melanoma. These studies provided no basis for further experiments with human MAR IgG in these species. 

Intravitreal injection of APB produced results that were similar to the MAR IgG effect on the ERG. Although APB specifically targets the mGluR6 receptor on DBCs, ^{17 18} this does not necessarily mean that MAR IgG targets this receptor, because bipolar cell function could also be impaired at the level of a membrane channel or at other sites. Also, the immunostaining pattern of human retinal sections with MAR patient serum showed binding to the bipolar cell soma and was not limited to the synaptic region where mGluR6 receptors are located. ² The retinal antigen recognized by human MAR antibody may be a 33-kDa protein, ¹⁹ and further information about the specific bipolar cell component that is involved will become clearer once this subcellular target is identified. 

 Supported by Grants EY01311 and EY06094 and Vision CORE Grants EY01730 and EY07003, National Institutes of Health, Bethesda, Maryland; by The Foundation Fighting Blindness, Hunt Valley, Maryland; by The Mackall Foundation Trust, New York, New York; and by Research to Prevent Blindness, Inc., New York, New York.

 Submitted for publication March 10, 1999; revised July 8, 1999; accepted August 2, 1999.

 Commercial relationships policy: N.

 

Table 1.

 

Ocular Injections

Monkey	Right Eye	Left Eye
1	MAR sera	Control non-MAR sera
2	MAR sera	Saline
3	MAR sera	Not available for study*
4	APB	Saline

*  Left eye of this animal was sighted but not usable as a control.

Figure 1.

 

Monkey photopic ERG to 200-msec stimuli that elicit a- and b-waves at light-onset and the d-wave at stimulus termination. Top: Response before (thin trace) and after MAR IgG injection (thick trace). Middle: Control non-MAR IgG sera from melanoma patients without visual symptoms: thin trace, before injection; thick trace, after non-MAR IgG. Bottom: The effect of 1 mM APB (thick trace) versus PBS vehicle control solution (thin trace) from monkey 4 (Table 1) .

Figure 2.

 

Photopic ERG of three monkeys before (thin trace) and after (thick trace) MAR IgG intravitreal injection. Waveform from monkey 1 is repeated from Figure 1 for comparison.

Figure 3.

 

(A) Scotopic ERG intensity series of monkey 1 before and after MAR IgG intravitreal injection. (B) Amplitude versus intensity plot of these scotopic b-waves shows a 1.8 log unit threshold elevation after MAR IgG versus only 0.6 log unit in the fellow control eye (for 50-μV threshold criterion).

Figure 4.

 

ERG recovery study of the scotopic b-wave of monkey 3. MAR IgG reduced the b-wave sensitivity by 1.2 log unit within 2 hours after injection. When tested 4 months later, the b-wave amplitude had recovered nearly completely.

The authors thank Samuel G. Jacobson, Kenneth R. Alexander, Donald C. Falgoust, Lynn G. Feun, Gerald A. Fishman, Eliot L. Berson, and Alan C. Bird for providing sera to Ann Milam.