Cancer-retina antigens as potential paraneoplastic antigens

in melanoma-associated retinopathy

Alexandr V. Bazhin1*, Claudia Dalke2, Nadine Willner1, Oliver Absch€utz1, Hannes G. H. Wildberger3,Pavel P. Philippov4, Reinhard Dummer5, Jochen Graw2, Martin Hrab�e de Angelis6, Dirk Schadendorf1,Viktor Umansky

1and Stefan B. Eichm€uller1*

1Skin Cancer Unit, German Cancer Research Center, Heidelberg and University Hospital, Mannheim, Germany2Helmholtz Center Munich, German Research Center for Environmental Health, Institute of Developmental Genetics,Neuherberg, Germany3University Eye Clinic Zurich, University Hospital, Zurich, Switzerland4Department of Cell Signalling, A.N. Belozersky Institute of Physico-Chemical Biology,M.V. Lomonosov Moscow State University, Moscow, Russia5Department of Dermatology, University Hospital, Zurich, Switzerland6Helmholtz Center Munich, German Research Center for Environmental Health,Institute of Experimental Genetics, Neuherberg, Germany

Melanoma-associated retinopathy is a rare paraneoplastic neuro-logical syndrome characterized by retinopathy in melanomapatients. The main photoreceptor proteins have been found to beexpressed as cancer-retina antigens in melanoma. Here we presentevidence that these can function as paraneoplastic antigens in mel-anoma-associated retinopathy. Sera and one tumor cell line ofsuch patients were studied and ret-transgenic mice spontaneouslydeveloping melanoma were used as a murine model for mela-noma-associated retinopathy. Splenocytes and sera were used foradoptive transfer from tumor-bearing or control mice to wild-type mice. Retinopathy was investigated in mice by funduscopy,electroretinography and eye histology. Expression of photorecep-tor proteins and autoantibodies against arrestin and transducinwere detected in melanoma-associated retinopathy patients. In tu-mor-bearing ret-transgenic mice, retinopathy was frequently (13/15) detected by electroretinogram and eye histology. These patho-logical changes were manifested in degenerations of photorecep-tors, bipolar cells and pigment epithelium as well as retinaldetachment. Mostly these defects were combined. Cancer-retinaantigens were expressed in tumors of these mice, and autoantibod-ies against arrestin were revealed in some of their sera. Adoptivetransfer of splenocytes and sera from tumor-bearing into wild-type mice led to the induction of retinopathy in 4/16 animals. Wesuggest that melanoma-associated retinopathy can be mediated byhumoral and/or cellular immune responses against a number ofcancer-retina antigens which may function as paraneoplastic anti-gens in melanoma-associated retinopathy.' 2008 Wiley-Liss, Inc.

Key words: cancer-retina antigens; paraneoplastic antigen; mel-anoma; melanoma-associated retinopathy; ret transgene

To date, many authors have described the presence of autoanti-bodies (AA) against various neuronal proteins, paraneoplasticantigens, in sera of patients with different kinds of malignanttumors located outside the nervous system (for review see Ref. 1and 2). These AAs may cross-react with the corresponding para-neoplastic antigens or their epitopes, which are present in neurons,and thus initiate the development of a variety of neurological dis-orders and paraneoplastic syndromes, even though the primary tu-mor and its metastases have not invaded the nervous system.3–7

Melanoma-associated retinopathy (MAR) is a rare paraneoplasticsyndrome in patients with metastatic melanoma characterized byvisual disorders and the presence of serum AA against retinal pro-teins.8–14 MAR is closely related to cancer-associated retinopa-thy,15 which often occurs in lung cancer patients but differs inophthalmological symptoms and probably in the antigens recog-nized, which are unknown today (for review see Ref. 16).

The ophthalmological components of both cancer-associatedretinopathy and MAR are well understood. In case of MAR, thefull field flash electroretinogram (ERG) shows a reduction of theb-wave, indicating a dysfunction of depolarizing bipolar cells.

Reduced amplitudes of a-waves reflect degeneration in photore-ceptors cells.8–11 Sera from cancer-associated retinopathy patientsstain strongly photoreceptor cells but only weakly bipolar cells,while sera from MAR patients strongly react with bipolar cells8–11

suggesting that the MAR syndrome is induced by the degenerationof bipolar cells. Indeed, the histological examination of the retinafrom MAR patients indicates the degeneration of second-orderneurons—bipolar cells.12 This may explain why MAR patients of-ten suffer from sudden visual loss and night blindness, whereascolor vision is frequently normal.

However, sometimes the a-wave in the ERG of MAR patientshas been reduced, and rare degenerative regions in neural retinaand pigment epithelium, as well as retinal detachment, have beendetected. Sera of MAR patients appear to show reactivity againstphotoreceptor cells.12 In addition, AA against two photoreceptorproteins, transducin (Tr)17 and rhodopsin (Rho),18 were found inthe sera of patients with MAR. These data allow us to suggest thatnot only bipolar cells but also photoreceptor cells may be involvedin the MAR syndrome (for review see Ref. 16). The paraneoplas-tic retina degeneration in cancer-associated retinopathy was shownto be linked to the apoptotic death of photoreceptors cells inducedby AA directed against the paraneoplastic antigen recoverin(Rec).19,20 Another hypothesis attributes cancer-associated reti-nopathy to the presence of Rec-specific cytotoxic T cells.21–23

However, there is still a lack of comparable investigations inMAR.

We have recently shown that photoreceptor proteins could func-tion as cancer-retina antigens (CRA) in melanoma.24,25 In addi-tion, by screening retina and melanoma cDNA libraries with sera

Additional Supporting Information may be found in the online versionof this article.

Abbreviations: AA, autoantibodies; Arr, arrestin; CRA, cancer-retinaantigen(s); ERG, electroretinogram; GC1, photoreceptor-specific guanylylcyclase 1; MAR, melanoma-associated retinopathy; PDE6, cGMP-phosphodiesterase 6; Rec, recoverin; Rho, rhodopsin; Tr, transducin.

Grant sponsor: DKFZ Intramural Program; Grant sponsor: RussianFoundation for Basic Research; Grant number: 06-04-48018; Grant spon-sor: Swiss National Foundation; Grant number: 310040-103671; Grantsponsor: Gottfried and Julia Bangerter Rhyner Stiftung; Grant sponsor:German National Genome Research Network (NGFN); Grant number:01GR0430; Grant sponsor: Dr. Mildred Scheel Foundation for CancerResearch; Grant number: 106096; Grant sponsor: Tumorzentrum Heidel-berg/Mannheim.

*Correspondence to: German Cancer Research Center, Skin CancerUnit (G300), Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany.E-mail: a.bazhin@dkfz.de or s.eichmueller@dkfz.de

Received 7 April 2008; Accepted after revision 8 July 2008DOI 10.1002/ijc.23909Published online 23 September 2008 inWiley InterScience (www.interscience.

wiley.com).

Int. J. Cancer: 124, 140–149 (2009)' 2008 Wiley-Liss, Inc.

Publication of the International Union Against Cancer

from melanoma patients with or without MAR syndrome, wecould isolate 20 different antigens recognized by patients’ anti-bodies, including visual Rho and arrestin (Arr).18 In the presentwork, we address the question of whether CRA can function asparaneoplastic antigens in MAR. On the basis of their frequentexpression, we focused on 6 CRA previously found to beexpressed in melanoma tissues24 such as Rho, Tr, cGMP-phospho-diesterase 6 (PDE6), Rec, photoreceptor-specific guanylyl cyclase1 (GC1), Arr. We report on AA in the sera of MAR patients, thefrequent occurrence of retinal degeneration in melanoma-bearing,ret transgenic mice and the induction of retinal degeneration innormal mice by transferring splenocytes and sera from tumor-bearing, transgenic mice.

Material and methods

Patients

The tumor material used was surplus material after surgical re-moval of lymph node metastases. Tissue and serum specimenswere investigated after informed consent of the patients and ap-proval by the local ethical committee.

Patient MAR1 suffered from breast cancer at the age of 55years, which was treated successfully by surgery and axillarlymphadenectomy. In 1994, 7 years later, a malignant melanoma(SSM-type, Breslow 1.0 mm, Clark level III; initial Stage IA,pT1a; S-100, HMB-45, and Melan A positive) was diagnosed onher back. Ophthalmological symptoms appeared since January1999: acquired night blindness in the left eye, 1 month later alsoin the right eye. The conventional ERG from April 1999 demon-strated a diffuse loss of rod function in both eyes, while cone func-tion was better preserved. The pattern ERG demonstrated a pro-gressive loss of retinal function in the most central retina of botheyes. Palpable lymph nodes were found in the right axilla followedby radical lymphadenectomy, with 7 of 16 lymph nodes positive.In July 1999 subjective and psychophysical visual functions wereunchanged, but a further deterioration of rod and cone responseswas found by conventional ERG. After positive PET/CT findings(12/1999), lymphadenectomy of the right axilla showed malignantmelanoma metastases in 4 of 7 lymph nodes (2/2000). Subjectiveor psychophysical visual functions remained unchanged. ERG2001 unchanged. In 8/2001, numerous lymph node metastasiswere detected (under the kidney and in retroperitoneal, axillar,infraclavicular and subpectoral lymph nodes). Patient receivedDTIC (decarbazine) chemotherapy in 2001/2002, but showed stillprogressing lymph node metastases. In 2002, distant metastaseswere detected and the patient died in 3/2002 with stage IV(T1aN3M1c).

Additional sera were obtained from previously described MARcases.17,26 Patients MAR-2 and MAR-3 were male and sufferedfrom night blindness, photopsias and problems with contrast sensi-tivity. Both patients had abnormal ERG and tritan defects on theFM 100-Hue test.26 Patient MAR-4 was a woman with bilateralnight blindness and decreased visual acuity. Bilateral centrocecalscotomas with enlarged blind spots were noted.17

Cell culture

A melanoma cell line (MM990428) was established from alymph node metastasis of patient MAR-1 as described earlier.27

Cells were cultivated in RPMI-1640 medium, supplemented with10% fetal calf serum, L-glutamine (2 mM) and penicillin/strepto-mycin solution (5 U/mL) at 37�C and 5% CO2.

Mice

Mice (C57Bl/6 background) transgenic for the human ret geneunder the control of metallothionein-I promoter expressing ret inmelanocytes28 were kindly provided by Dr. Nakashima (Japan).All mice were crossed and kept under specific pathogen-free con-ditions in the animal facility of German Cancer Research Center(Heidelberg). Experiments were performed in accordance with the

institution’s guidelines. Spontaneous tumor development wasassessed twice a week. After a short latency (20–90 days), around30% of mice develop skin tumors on the face (nose, ears, eyes andneck), back or on the tail. Tumor-bearing mice developed metasta-ses in lymph nodes and some distant organs like liver, lungs andbrain. Histological analysis of primary tumors and metastasesrevealed the morphology of malignant melanoma. In addition, itwas found that these tumors expressed melanoma-associated anti-gens tyrosinase, tyrosinase-related protein-1 and -2, as well asgp100 (Umansky, unpublished observations).

Transfer of splenocytes and serum

Single-cell suspensions from spleens of tumor-bearing ret trans-genic (experimental group) or nontransgenic wild-type mice (con-trol group) were prepared by mechanical dissociation. After thelysis of erythrocytes by a short NH4Cl hypotonic solution treat-ment, residual cells were washed twice, resuspended in PBS andused as a source of T- and B lymphocytes. In addition, serum sam-ples were prepared as a source of immunoglobulins. Spleen cells(2 3 107 cells/200 lL PBS) and serum (100 lL) were transferredi.v. into wild-type mice. Three weeks later, mice were killed andeyes were isolated, and the retinal histology was examined.

Funduscopy

The posterior parts of both eyes were examined after pupil dila-tion with 1 drop of atropine (1%). The mouse is grasped firmly inone hand and clinically evaluated using a head-worn indirect oph-thalmoscope (Sigma 150 K, Heine Optotechnik, Herrsching, Ger-many) in conjunction with a condensing lens (90D lens, Volk,Fronh€auser, Unterhaching, Germany) mounted between the oph-thalmoscope and the eye.

Electroretinography

Whole-field ERGs were recorded simultaneously from botheyes to examine the retinal function as described.29 In brief, micewere dark-adapted for at least 12 hr and anesthetized. After pupildilation (1% atropine), individual mice were fixed on a sled, andgold wires (as active electrodes) were placed on the cornea. Theground electrode was a subcutaneous needle near the tail; a refer-ence electrode was placed subcutaneously between the eyes. ERGwas performed at 2 luminance levels, at 500 cd/m2 and 12,500 cd/m2. The mice were introduced into an ESPION ColorBurst Hand-held Ganzfeld LED stimulator (Diagnosys LLC, Littleton, MA) ona rail to guide the sled (High-Throughput Mouse-ERG, STZ forBiomedical Optics and Function Testing, T€ubingen, Germany).Responses were recorded with an ESPION Console (DiagnosysLLC) and stored. To determine the cutoff value for the ERGinvestigation,29 10 control mice were subjected to ERG by a lumi-nance of 500 cd/m2 and 12,500 cd/m2. The cutoffs for the a-wavewere set to 22.5 at 500 cd/m2 and 223.8 at 12,500 cd/m2. Trans-genic mice were measured for 10 times at both illumination inten-sities. Mean values of the a-waves higher than cutoff were inter-preted as a result of degenerated photoreceptor cells. The cutoffsfor the b-wave were set to 109 (at 500 cd/m2) and 121 (at 12,500cd/m2); lower mean values were interpreted as degeneration ofbipolar cells.16

Histology of eyes

Eye balls were fixed for 24 hr in Davidson solution (mixture of31.4% ethanol, 8.3% formaldehyde and 11.1% acetic acid inwater), dehydrated and embedded in plastic medium (JB4-Plus;Polysciences, Inc., Eppelheim, Germany). Transverse 2-lm sec-tions were cut with an ultramicrotome (Ultratom OMU3; Reichert,Walldorf, Germany) and stained with methylene blue and basicfuchsin. Sections were evaluated by light microscopy and imageswere taken. Absence of tumor cells in eyes was determined by 2independent researchers.

141CANCER-RETINA ANTIGENS AND MAR

Immunohistology

Snap-frozen eye balls were embedded in Tissue-Tek (SakuraFinetek, Zoeterwoude, The Netherlands). Sections were first fixedin ice-cold absolute acetone and were then preabsorbed using theAvidin/Biotin Blocking solution (Vector Laboratories, Burlin-game, CA) and subsequently incubated with 5% normal goat se-rum. The antibodies against Ret were applied at a concentration2 lg/mL overnight. After incubation with a goat anti-rabbit IgGcoupled to biotin for 1 hr and subsequent incubation with a strep-tavidin–biotin–phosphatase complex, specific staining was visual-ized by alkaline phosphatase substrate (all components from theAlkaline Phosphatase Rabbit IgG ABC Kit and AP substrate Kit;Vector Laboratories) as described by the manufacturer. Hemalaunsolution was used for counterstaining.

Melanoma tissue specimen were fixed in 4% formaldehyde,dehydrated and embedded in paraffin. Sections were dewaxed,rehydrated and preabsorbed using the Avidin/Biotin Blocking so-lution (Vector Laboratories) and subsequently incubated with 5%normal goat serum. The application of the antibodies against Rho(0.3 lg/mL), Tr (0.3 lg/mL), Rec (4.5 lg/mL) and Arr (0.3 lg/mL) and further procedures were done as described for the cryo-sections mentioned earlier.

RNA and proteins

RNA and proteins were extracted with TriPure Isolation Rea-gent (Roche Diagnostics, Mannheim, Germany) for RNA andlysate buffer (25 mM Tris-HCl, pH 7.5 with 0.05% Triton X-100)for protein, as described by the manufacturers. Normal human ret-ina RNA was obtained from a commercial source (BD BiosciencesClontech, USA). Bovine photoreceptor proteins were obtained asdescribed.24 Polyclonal, monospecific rabbit antibodies againstRho, Tr, Rec and Arr were obtained as described.24 Polyclonal,monospecific rabbit antibodies against PDE6 and anti-human c-Ret were purchased from Affinity BioReagents (Golden, CO) andIBL (Gunma, Japan), respectively. Polyclonal, monospecific rab-bit antibodies against GC1 were kindly provided by Prof. K.-W.Koch (Oldenburg, Germany).

RT-PCR analysis

One microgram of the total RNA from cell line MM990428 orfrom murine tumor tissues was reverse-transcribed by using the1st strand cDNA synthesis kit (Roche Diagnostics) at 42�C for50 min as described by the manufacturer. PCR-amplification wasperformed using 1 lL from the RT-reaction mixture in 25 lL ofthe PCR-mixture containing 50 pmol of sense and antisense pri-mers. After the initial incubation at 94�C for 2 min, 33 cycles ofamplification were carried out for genes of interest, and 25 cyclesfor GAPDH. Primer sequences for visual genes and GAPDH,annealing temperatures and sizes of PCR products are listed inSupporting Table S1. A sample was judged as being positive, if aclear band was visible in an ethidium bromide stained gel in atleast 2/3 of independent PCR experiments. The amplified PCRproducts were electrophoresed on 1.5–3% agarose gels and subse-quently stained with ethidium bromide.

Western blot analysis

After SDS-PAGE in a 12.5% polyacrylamide gel, separatedproteins were electrotransferred to nitrocellulose membranes(Hybond-C; Amersham Bioscience, Buckinghamshire, UK) inTris-glycine-methanol buffer (pH 8.3). After blocking with 10%(w/v) delipidated dry milk in PBS containing TWEEN-20, themembranes were incubated with antibodies against photoreceptorproteins. Membranes were then incubated with an anti-rabbit per-oxidase conjugated secondary antibody (Santa Cruz Biotechnol-ogy, Santa Cruz, CA). Immunoreactive bands were visualized byan enhanced chemiluminescence system (Amersham Bioscience)according to the method described by the manufacturer.

Results

CRA are expressed in melanoma cells of a MAR patient and AAagainst CRA are detected in sera of MAR patients

Since MAR is extremely rare, we only had access to sera of 4patients with clinical symptoms of MAR. From one of thesepatients (MAR1), a stable tumor cell line (MM990428) could beestablished, and paraffin-embedded tumor tissue was available.

The cell line MM990428 was tested positive for Tr, Rec andArr at mRNA and protein levels, while PDE6 and GC1 geneswere found to be transcribed, but no protein could be detected byWestern blotting, Rho, CGC (cGMP-dependent channels) and rho-dopsin kinase were neither found at mRNA nor at protein levels(Fig. 1). Immunohistological analysis of paraffin tumor sectionsconfirmed the presence of Tr, Rec and Arr proteins in the histolog-ical sections, and demonstrated that not all tumor cells were posi-tive, but instead a rather spot-like staining pattern was observed(Fig. 2). Specificity was proven by omission or preabsorption ofthe primary antibody. An immunohistological staining with anantibody against Rho did not reveal any Rho-positive cells in thesections. The expression of PDE6 and GC1 in the sections couldnot be investigated because of the lack of antibodies against PDE6and GC1 working in immunohistology on paraffin sections. Inter-estingly, while the staining for Tr and Rec was restricted to mela-noma cells, the anti-Arr antibody stained, in addition, some stro-mal components. The expression of CRA in the tumors of theother 3 patients with MAR could not be assessed because of thelack of tumor material.

The presence or absence of AA against CRA was examined inall 4 sera by Western blotting, using corresponding photoreceptorproteins as antigens. We detected AA (at a serum dilution of1:100) against Tr in the sera of patients MAR1 and MAR4, andagainst Arr in sera of patients MAR2 and MAR4. AA against Rho,PDE6 and Rec were not detected in any of the sera. Thus, at leastAA against Tr and Arr may play a role in MAR development.

Tumor-bearing ret transgenic mice, but not wild-type mice showsigns of retinal degeneration

In order to investigate the potential involvement of CRA inMAR pathogenesis, we used ret transgenic mice spontaneouslydeveloping cutaneous malignant melanoma28 as a model systemfor MAR.

For ophthalmological investigations, we used the following pa-rameters: (i) funduscopy of optic media and of fundus oculi, (ii)ERG, and (iii) histological retina examination. First, we performedfunduscopy of 10 C57Bl/6 wild-type mice that did not show anychanges in optic media and fundus oculi. Cutoff values for ERGwere determined using these mice, as described.29 Histological ex-amination of eye balls obtained from wild-type mice did not showany pathological changes: the retinal structure was intact, alllayers were well preserved, and photoreceptor cells, bipolar cellsand pigment epithelium were clearly distinguishable (Fig. 3a).

Next, we tested 10 ret transgenic mice without any visibletumors using the same ophthalmological approaches (Table I).Five mice did not show any difference from wild-type mice, nei-ther in funduscopy nor in ERG (Fig. 4b) and retinal histology(Fig. 3b, mouse TgN#02). Four mice (TgN#03, 08, 09 and 10) hadslight reduction of the a-wave (mouse TgN#10) and the b-wave(mice TgN#03 and #08) or both (mouse TgN#09) in the ERG,while a histological examination did not reveal any pathologicalchanges in the retinae of these mice (Table I and Fig. 3b). Onemouse (TgN#04) showed pathological changes in the ERG (Fig.4b), which correlated with histological detectable retinal degener-ation (Table I and Fig. 3b).

In contrast, tumor-bearing transgenic mice displayed drasticchanges in their ophthalmological analysis (Table II and Fig. 4c).Histological analysis of mouse eyes revealed only 2 mice withoutobvious retinal disturbances (TgT#01 and TgT#04), whereas 8mice had pathological changes in 1 and 5 mice in both eyes (Table

142 BAZHIN ET AL.

II). Notably, in 11/13 mice both the photoreceptor layer and thebipolar or other layers were histologically altered. These findingscorrelated well with aberrations in the ERG of the respective eye(Table II).

The observed histological data indicated strong morphologicalchanges in the retinae of tumor-bearing mice. Disorganization ofthe inner nuclear layer and neuroepithelial degeneration were seenin mice TgT#02, #03, #05, #06, #08 to #10, #12 and #13 to #15,and a fragmentation of the rod and cone layer in mice TgT#06,#09, #10 and #12 to #15 (examples in Fig. 3c). In addition, retinaldetachment (mice TgT#06, #09, #10 and #13) and degeneration ofthe pigment epithelium (mice TgT#06, #08, #09 and #10) wereobserved in some mice (Fig. 3c).

Since the ERG correlated well with the histological data, weperformed only histological examination of eye balls in additional6 transgenic tumor-bearing mice. All these mice had signs of reti-nal degeneration in the photoreceptor layer and 4 additionally inother layers (Table III).

Retinal degeneration is not mediated by tumorinvasion or the influence of ret transgene

Transgenic mice with periocular tumor were excluded from theophthalmological analysis described earlier in order to eliminateany influence by tumor cell invasion. Using microscopical analy-sis, we found that all investigated mice did not have any mela-noma cells in their ocular cups.

Since we found some mice with signs of retinal degeneration inthe group of transgenic mice without tumor, we wanted to clarifywhether the ret transgene could directly induce this retinal degen-

eration. We performed an immunohistological analysis of Retexpression in the retinae of 5 ret transgenic mice without tumorsand 5 tumor-bearing mice. Ret expression was observed in the tu-mor, but never in the retinae of these mice (Fig. 5).

Transfer of spleen cells and serum from tumor-bearing rettransgenic mice induced retinal degenerationin normal C57Bl/6 recipients

To test the role of the immune system in the observed retinopa-thy, C57Bl/6 wild-type mice were injected with spleen cells andserum from tumor-bearing transgenic mice. Four out of 16 miceshowed obvious signs of retinopathy 3 weeks after the transfer(Fig. 3d): Disorganization of the inner nuclear layer, fragmenta-tion of the rod and cone layers, retinal detachment and neuroepi-thelial degeneration were observed very prominently in 2 mice.The retina of the third mouse displayed similar but less profoundchanges. We also found retinal detachment and signs of pigmentepithelium degradation in the eyes of the forth animal. In contrast,all 6 wild-type mice, which received the same quantity of spleencells and serum obtained from wild-type mice, had no visiblechanges in their retinae.

MAR mice express CRA in their tumors and have AAagainst CRA in their sera

Next, we wanted to check whether tumor-bearing mice withsigns of retinopathy (MAR mice) express CRA in their tumors andcan produce AA against these CRA. Using RT-PCR (Table IV),we detected specific mRNA for Rho (3/7 mice), Tr (4/7), PDE6(6/7), GC1 (6/7), Rec (4/7) and Arr (2/7). Expression at the protein

FIGURE 1 – mRNA and protein expression of visual antigens in a stable melanoma cell line obtained from a patient with melanoma-associatedretinopathy when compared with antigens from human retina (first and second column). First column (gel-electrophoresis of PCR products): firstlane, retina; second lane, cell line of MAR1; house-keeping gene: GAPDH. Second column Western blotting with polyclonal, monospecific anti-bodies against Rho, Tr, PDE6, GC1, Rec and Arr: first lane shows the corresponding proteins or retina extract in case of GC1, second lane, cellline of MAR1; house-keeping protein: b-actin. The third column shows results of autoantibodies analyzed from serum of the same patient. Thepresence of AA in the serum of patient MAR1 was tested by Western blotting using Rho, Tr, PDE6, Rec and Arr proteins as antigens.

143CANCER-RETINA ANTIGENS AND MAR

level was tested by Western blot analysis for all CRA except GC1in the same 7 tumor samples. One tumor was positive for Rho, 1for Tr, 3 for PDE6, 3 for Rec and 2 for Arr. CRA proteins weredetected only in those tumor samples, which were tested positivefor the corresponding mRNA. In addition, we found AA againstArr in the sera of 2 mice, whereas AA against Rho, Tr and Reccould not be detected (Table IV). None of 10 control mice and 15transgenic mice without tumor showed any AA in their sera.

Discussion

In this study, we addressed the question, which antigens mightbe involved in the development of paraneoplastic MAR. Whilecancer-associated retinopathy is presumably caused by an AA19,20

or CTL response against Rec,21–23 MAR is most likely induced byan immune response against multiple retinal antigens16: (i) diverseophthalmological symptoms have been described in MAR, (ii) dif-

FIGURE 2 – CRA immunoreactivity in the lymph node melanoma metastasis of patient MAR1. (a) No Rho-immunoreactivity was detected.(b) Antibodies against Rec resulted in a spot-like pattern, which was abolished when the antibodies were preabsorbed with added Rec (c). (d)Higher magnification of (b). (e, f) Individual tumor cells were stained by the Tr antibody. (g–i) Arr was detected in individual tumor cells andselected stromal cells (i). All scale bars: 100 lm.

144 BAZHIN ET AL.

FIGURE 3 – Histological analysis of mouse eye balls and retinae. (a) Eye ball and retina obtained from a wild-type mouse without any ophthal-mological defects. (b) Two transgenic mice which did not exhibit any visible tumors. The retina of mouse TgN#02 was unchanged, while mouseTgN#04 had slight signs of retinal detachment (open arrows) and a weak degeneration of pigment epithelium (open arrowhead). (c) Eye ball andretinae obtained from 3 transgenic tumor-bearing mice (TgT#15, TgT#10 and TgT#09) as representative examples showing retinal degeneration:disorganization of the inner nuclear layer (arrows), fragmentation of the rod and cone layer (arrowheads), retinal detachment (open arrows) anddegeneration of pigment epithelium (open arrowhead). (d) Wild-type mouse after transfer of splenocytes and serum from transgenic tumor-bear-ing mouse exhibited strong signs of retinal and pigment epithelium degeneration (arrows and arrowheads as in c). GCL, ganglion cell layer; IPL,inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; IS (inner segments) and OS (outer seg-ments) of photoreceptors; RPE, retinal pigment epithelium.

145CANCER-RETINA ANTIGENS AND MAR

ferent MAR patients have degenerations in distinct retinal celltypes, (iii) sera of MAR patients stain different layers of the retina,and (iv) AA against 2 photoreceptor proteins, namely Rho18 andTr,17 have been described in MAR patients. On the basis of theseobservations, we hypothesize that CRA may also be paraneoplas-tic antigens in MAR. We have recently reported that these andother photoreceptor proteins could serve as CRA in mela-noma.24,25 On the basis of these observations, we investigatedhere whether CRA may also play a role as paraneoplastic antigensin MAR.

First, we tested CRA expression in the cell line and tumor tissueobtained from a melanoma patient suffering from MAR. We foundCRA expression both in the cell line and in the tumor at mRNAand protein levels. In addition, AA against Tr were found in theserum of this patient. AA against Tr have previously been shownto be present in the serum of a patient with MAR.17 Furthermore,we succeeded in finding AA against Arr in the serum of 2 addi-tional MAR patients; one of them also had AA against Tr. To ourknowledge, this is the first description of AA against Arr in MARpatients. Unfortunately, we did not have the possibility to check

TABLE I – OPHTHALMOLOGICAL EXAMINATION OF Ret-TRANSGENIC MICE WITHOUT TUMORS

Mouse Tg#

Funduscopy ERG1 Retina histology

Left Righta-Wave 500 cd/m2

(>22.5)b-Wave 500 cd/m2

(<109)a-Wave 12,500 cd/m2

(>223.8)b-Wave 12,500 cd/m2

(<121)

Photoreceptorlayer2 Other layers

Left Right Left Right

01 n.d. n.d. 222 164 256 185 w.p. w.p. w.p. w.p.02 n.d. n.d. 213 217 226 163 w.p. w.p. w.p. w.p.03 w.p. w.p. 23 104 238 223 w.p. w.p. w.p. w.p.04 w.p. w.p. 214 102 219 116 1 1 w.p. w.p.05 w.p. w.p. 215 124 229 124 w.p. w.p. w.p. w.p.06 w.p. w.p. 26 133 232 147 w.p. w.p. w.p. w.p.07 w.p. w.p. 24 203 249 203 w.p. w.p. w.p. w.p.08 w.p. w.p. 215 75 226 86 w.p. w.p. w.p. w.p.09 w.p. w.p. 22 113 219 112 w.p. w.p. w.p. w.p.10 Bright spots 21 149 225 125 w.p. w.p. w.p. w.p.

Transgenic mice without tumors (TgN#01 to 10) were analyzed by funduscopy, ERG and histology. Values and observations differing fromnormal are bold face. n.d., not done; w.p., without pathology.

1ERG cutoff values are given in brackets.–2Photoreceptor layer: 1–3 gives slight to drastic changes.

FIGURE 4 – Representative ERGs from a wild-type mouse (a), 2 transgenic mice without (b) and 3 with tumors (c). Mouse TgN#04 showed areduced b-wave in contrast to the normal response in mouse TgN#02. The tumor-bearing mice showed a reduced b-wave. In addition, miceTgT#10 and TgT#09 showed no a-waves.

146 BAZHIN ET AL.

the expression of this CRA in the corresponding tumors. In con-trast to cancer-associated retinopathy,30 MAR can occur withoutinvolvement of Rec as a main paraneoplastic antigen responsiblefor the development of this pathology. Furthermore, we show herefor the first time the presence of AA against 2 different CRA inthe same patient.

Since MAR is described as a very rare event with only about 70published cases,12,16 we used here the ret transgenic mouse mela-noma model for a detailed investigation of MAR. We found thatmelanoma-bearing mice showed similar signs of retina degenera-tion as reported for MAR patients, including the reduction of a-and b-waves, degeneration of photoreceptor cells, bipolar cellsand pigment epithelium.8–13 Since this retina degeneration in tu-mor-bearing mice appeared to be a paraneoplastic event, whichwas independent from any ocular tumor invasion and from overex-pression of the ret transgene, we suggested that ret transgenicmice could be a good animal model of MAR.

A disputed question in MAR is the type of cells targeted duringretina degeneration.16 Our histological findings presented hereindicated the degeneration of photoreceptor cells and other retinalcells, as well as changes in the pigment epithelium, which was nota part of the neuronal retina but was also of neuroectodermal ori-gin. This suggested that not only proteins of bipolar cells but alsophotoreceptor proteins and possibly proteins of pigment epithe-lium could function as paraneoplastic antigens in MAR.

While MAR is a very rare syndrome in melanoma patients, thequestion arises, why we found retinal degenerations in almost alltumor-bearing animals investigated. It has recently been publishedthat subclinical manifestations of MAR may appear more oftenthan previously suspected in melanoma patients.31 The authorshave investigated about 30 patients with malignant melanoma byERG, static and kinetic perimetry, and nyctometry. SubclinicalMAR symptoms were found in 90% of the patients. This fre-quency correlates well with our data obtained in melanoma-bear-ing ret transgenic mice. Thus, retinal degeneration may be a fre-quent symptom in melanoma patients remaining, however, at asubclinical level. Pf€ohler et al.31 have shown that these subclinicalmanifestations of MAR are more frequent in advance stages ofmelanoma. Therefore, an ophthalmological examination may notbe suitable for an early melanoma diagnosis, although this needsto be tested by further clinical studies in melanoma patients and/orin mice during melanoma development.

Retinal degeneration in melanoma-bearing mice was found tobe associated with the expression of several CRA at mRNA andprotein levels in their tumors and the presence of serum AAagainst Arr. These data confirmed our previous observations ofCRA expression and circulation of serum AA against Arr, Tr andRec in some ret transgenic mice and HGF transgenic Ink4a knock-out mice bearing melanoma.24 In the present study, we alsodetected AA against Arr and Tr in the sera of MAR patients. Inaddition, we found neither tumor invasion nor ret overexpressionin the mouse retina.

All these findings allowed us to propose that immune cells and/or AA could be responsible for the observed retina degenerationsimilar to other paraneoplastic syndromes.2 To prove this hypothe-sis, we isolated splenocytes (as a source of both T and B cells) andserum (as a source of immunoglobulins) from tumor-bearing miceand adoptively transferred them into healthy mice. One quarter ofthe recipients showed retinal degenerations. The observed effectwas specific, since the transfer of splenocytes and serum fromwild-type mice induced no changes in the recipients’ retina. Inaddition to circulating AA, melanoma-bearing mice may possessboth CD8 and CD4 T cells specific for CRA. It was demonstratedthat the peripheral blood of CAR patients frequently containedRec-specific CTLs.21–23

It is interesting to note that 1 ret transgenic mouse without clini-cally visible tumor also showed moderate signs of retina degenera-tion. Immunohistological analysis of such tumor-free micerevealed microscopic melanoma lesions in their lymph nodes

TABLE II – OPHTHALMOLOGICAL EXAMINATION OF Ret-TRANSGENIC TUMOR-BEARING MICE

MouseTgT#

Funduscopy ERG1 Retina histology

Left Righta-Wave 500 cd/m2

(>22.5)b-Wave 500 cd/m2

(<109)a-Wave 12,500 cd/m2

(>223.8)b-Wave 12,500 cd/m2

(<121)

Photoreceptorlayer2 Other layers3

Left Right Left Right

01 w.p. w.p. 22/2 201/174 274/240 339/298 w.p. w.p. w.p. w.p.02 w.p. Turbid 212/23 148/88 235/219 200/130 1 1 1 103 w.p. Turbid 219/211 216/268 242/231 196/331 1 w.p. w.p. w.p.04 Turbid w.p. 22/24 60/56 223/254 48/79 w.p. w.p. w.p. w.p.05 Turbid Bright spots 214/226 77/161 217/264 35/190 2 w.p. 1 w.p.06 n.d. n.d. 23 100 226 171 2 3 1 307 n.d. n.d. 210 138 214 202 1 1 w.p. 108 Turbid w.p. 212/212 84/152 21/238 48/180 1 1 w.p. w.p.09 Turbid Turbid 214 8 4 22 3 3 1 310 Turbid Turbid 5/3 184/62 260/22 184/31 w.p. 3 w.p. 311 w.p. w.p. 3 155 223 265 w.p. 1 w.p. w.p.12 Turbid w.p. 211/213 82/156 25/228 86/146 w.p. 3 w.p. 313 Turbid w.p. 4/4 40/138 29/265 106/281 3 w.p. 3 w.p.14 Turbid Bright spots 3/2 88/107 222/286 145/266 3 w.p. 3 w.p.15 w.p. Turbid 238/222 288/27 261/214 303/87 1 3 w.p. 3

Transgenic mice with tumors (TgT#01 to #15) were analyzed by funduscopy, ERG and histology. Values and observations differing from nor-mal are bold face. n.d., not done; w.p., without pathology.

1ERG cutoff values are given in brackets. If two values are given in the table, they refer to the left and right eye, respectively.–2Photoreceptorlayer: 1–3 gives slight to drastic changes.–3Other layers: 1, all layers are present, but not in the normal thickness and structure; 2, slight changes;3, drastically changes in the layers and/or absent of layers.

TABLE III – HISTOLOGY OF RETINAE OF ADDITIONAL Ret-tgTUMOR-BEARING MICE

Mouse TgT#

Retina histology

Photoreceptor layer1 Other layers2

Left Right Left Right

16 2 2 1 117 2 1 w.p. 118 3 1 1 119 3 3 3 320 3 3 3 321 1 1 w.p. 1

Transgenic mice with tumors (TgT#16 to #21) were analyzed byhistology. w.p., without pathology.

1Photoreceptor layer: 1–3 gives slight to drastic changes.–2Otherlayers: 1, all layers are present, but not in the normal thickness andstructure; 2, slight changes; 3, drastically changes in the layers and/orabsent of layers.

147CANCER-RETINA ANTIGENS AND MAR

(Umansky, unpublished data), suggesting thereby that even smalltumor load could already stimulate immunological reactions lead-ing to pathological changes in the retina. We are currently investi-gating which cellular and/or humoral components of immune sys-tems play a major role in the observed MAR.

Another important question is related with the role of theblood–ocular barrier in the development of this paraneoplasticsyndrome. It is still poorly understood how AA and/or T cells canpenetrate the ocular capillary endothelium forming the blood–ocu-lar barrier. We suppose that occasional dysfunction of this barrierin tumor-bearing mice may allow the internalization of cellularand/or humoral components of the immune system that mayfinally result in the development of MAR. A similar dysfunctionof the blood–ocular barrier could also occur in MAR patients.16

Moreover, the difference in the level of this dysfunction couldexplain why retinal degenerations are not always found in botheyes of MAR patients.10,12 Using our mouse melanoma model, wealso found that only 62% of affected animals had retinopathy inboth eyes.

Even when the first hurdle (blood–retina barrier) is overcome,AA have to enter the retinal cells, as many CRA are located intra-cellulary. This issue was precisely investigated on Rec antibodiesin the works of Adamus’ group.19,20 The authors have shown anAA internalization by the retinal cells and a destruction of photo-receptor and bipolar cells by apoptosis.

In summary, we suppose that—similar to cancer-associated reti-nopathy—MAR is most likely induced by an immune responseoriginally mediated against the tumor, but targeting retinal compo-nents. We propose that a broad range of antigens, most promi-

nently CRA, are the proteins against which the specific immuneresponse is directed, which is supported by our transfer experi-ments. Future experiments should clarify, whether AA or T cellsor both are the cause of MAR, which epitopes are targeted inMAR and in antitumor responses, and which molecular mecha-nisms are responsible for MAR.

Acknowledgements

We gratefully acknowledge Dr. K. Alexander (Department ofOphthalmology and Visual Sciences, University of Illinois, Chi-cago) and Dr. M. Potter (Department of Ophthalmology, Univer-sity of British Columbia, Vancouver, Canada) for providing seraof patients with MAR, Dr. Nakashima (Nagoya, Japan) for initialproviding the ret transgenic mice, and Prof. K.-W. Koch (Olden-burg, Germany) for the antibodies against guanylyl cyclase 1. Wealso thank Dr. Ekkehard Sch€affer and Dr. Julia Calzada-Wack(Institute of Pathology, GSF, Neuherberg, Germany) for assessingthe mouse eye histological sections. Ms. Anita Heinzelmann, Ms.Elke Dickes, Ms. Kathrin Frank, Ms. Mareike Maurer, and Ms.Maria Kugler are acknowledged for their excellent technical assis-tance and Dr. Rogers (DKFZ) for manuscript correction. Thiswork was supported in parts by the DKFZ intramural program toA.V.B., the Russian Foundation for Basic Research (06-04-48018)to P.P.P., the Swiss National Foundation (310040-103671) and theGottfried and Julia Bangerter Rhyner Stiftung to R.D., the GermanNational Genome Research Network (NGFN; 01GR0430) to J.G.,the Dr. Mildred Scheel Foundation for Cancer Research (106096)to V.U., and the Tumorzentrum Heidelberg/Mannheim to S.E.

FIGURE 5 – Histological analysis of Ret protein expression in the tumor (a) and retina (b) of ret transgenic mice. (a) Ret-positive membranestaining of a tumor (cryosection), obtained from a tumor-bearing transgenic mouse (positive control). (b) Retina obtained from a tumor-bearingmouse is negative for Ret. All scale bars: 100 lm. INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; IS (inner seg-ments) and OS (outer segments) of photoreceptors; RPE, retinal pigment epithelium.

TABLE IV – EXPRESSION OF CRA IN THE TUMOR AND AA AGAINST CRA IN SERA OF TUMOR-BEARING MICE WITH RETINA DEGENERATION

Mouse TgT#Rho Tr PDE6 GC1 Rec Arr

RNA Protein AA RNA Protein AA RNA Protein RNA RNA Protein AA RNA Protein AA

1 2 2 2 2 2 2 2 2 1 2 2 2 2 2 22 1 1 2 1 2 2 1 1 1 1 2 2 1 1 13 2 2 2 1 1 2 1 1 1 1 1 2 1 1 14 2 2 2 2 2 2 1 2 1 2 2 2 2 2 25 2 2 2 2 2 2 1 2 2 2 2 2 2 2 26 1 2 n.d. 1 2 n.d. 1 2 1 1 1 n.d. 2 2 n.d.7 1 2 n.d. 1 2 n.d. 1 1 1 1 1 n.d. 2 2 n.d.

Mouse number (TgT#) corresponds to the numbers in Table II. n.d., not done.

148 BAZHIN ET AL.

References

1. Darnell RB. Onconeural antigens and the paraneoplastic neurologicdisorders: at the intersection of cancer, immunity, and the brain. ProcNatl Acad Sci USA 1996;93:4529–36.

2. Eichm€uller SB, Bazhin AV. Onconeural versus paraneoplastic anti-gens? Curr Med Chem 2007;14:2489–94.

3. Posner JB, Dalmau JO. Paraneoplastic syndromes affecting the centralnervous system. Annu Rev Med 1997;48:157–66.

4. Gure AO, Stockert E, Scanlan MJ, Keresztes RS, Jager D, AltorkiNK, Old LJ, Chen YT. Serological identification of embryonic neuralproteins as highly immunogenic tumor antigens in small cell lung can-cer. Proc Natl Acad Sci USA 2000;97:4198–203.

5. Darnell RB, Posner JB. Paraneoplastic syndromes involving the nerv-ous system. N Engl J Med 2003;349:1543–54.

6. Honnorat J, Cartalat-Carel S. Advances in paraneoplastic neurologicalsyndromes. Curr Opin Oncol 2004;16:614–20.

7. Roberts WK, Darnell RB. Neuroimmunology of the paraneoplasticneurological degenerations. Curr Opin Immunol 2004;16:616–22.

8. Milam AH, Saari JC, Jacobson SG, Lubinski WP, Feun LG,Alexander KR. Autoantibodies against retinal bipolar cells in cutane-ous melanoma-associated retinopathy. Invest Ophthalmol Vis Sci1993;34:91–100.

9. Weinstein JM, Kelman SE, Bresnick GH, Kornguth SE. Paraneoplas-tic retinopathy associated with antiretinal bipolar cell antibodies in cu-taneous malignant melanoma. Ophthalmology 1994;101:1236–43.

10. Boeck K, Hofmann S, Klopfer M, Ian U, Schmidt T, Engst R, ThirkillCE, Ring J. Melanoma-associated paraneoplastic retinopathy: casereport and review of the literature. Br J Dermatol 1997;137:457–60.

11. Gittinger JW Jr, Smith TW. Cutaneous melanoma-associated paraneo-plastic retinopathy: histopathologic observations. Am J Ophthalmol1999;127:612–14.

12. Keltner JL, Thirkill CE, Yip PT. Clinical and immunologic character-istics of melanoma-associated retinopathy syndrome: eleven newcases and a review of 51 previously published cases. J Neuroophthal-mol 2001;21:173–87.

13. Zacks DN, Pinnolis MK, Berson EL, Gragoudas ES. Melanoma-associ-ated retinopathy and recurrent exudative retinal detachments in apatient with choroidal melanoma. Am J Ophthalmol 2001;132:578–81.

14. Myers DA, Bird BR, Ryan SM, Tormey P, McKenna P, S OR, Breath-nach OS. Unusual aspects of melanoma. Case 3. Melanoma-associ-ated retinopathy presenting with night blindness. J Clin Oncol2004;22:746–8.

15. Thirkill CE, Roth AM, Keltner JL. Cancer-associated retinopathy.Arch Ophthalmol 1987;105:372–5.

16. Bazhin AV, Schadendorf D, Philippov PP, Eichm€uller SB. Recoverin asa cancer-retina antigen. Cancer Immunol Immunother 2007;56:110–16.

17. Potter MJ, Adamus G, Szabo SM, Lee R, Mohaseb K, Behn D. Auto-antibodies to transducin in a patient with melanoma-associated reti-nopathy. Am J Ophthalmol 2002;134:128–30.

18. Hartmann TB, Bazhin AV, Schadendorf D, Eichm€uller SB. SEREXidentification of new tumor antigens linked to melanoma-associatedretinopathy. Int J Cancer 2005;114:88–93.

19. Adamus G, Machnicki M, Seigel GM. Apoptotic retinal cell deathinduced by antirecoverin autoantibodies of cancer-associated retinop-athy. Invest Ophthalmol Vis Sci 1997;38:283–91.

20. Adamus G, Machnicki M, Elerding H, Sugden B, Blocker YS, FoxDA. Antibodies to recoverin induce apoptosis of photoreceptor andbipolar cells in vivo. J Autoimmun 1998;11:523–33.

21. Maeda A, Ohguro H, Nabeta Y, Hirohashi Y, Sahara H, Maeda T,Wada Y, Sato T, Yun C, Nishimura Y, Torigoe T, Kuroki Y, et al.Identification of human antitumor cytotoxic T lymphocytes epitopesof recoverin, a cancer-associated retinopathy antigen, possibly relatedwith a better prognosis in a paraneoplastic syndrome. Eur J Immunol2001;31:563–72.

22. Maeda A, Maeda T, Ohguro H, Palczewski K, Sato N. Vaccinationwith recoverin, a cancer-associated retinopathy antigen, induces auto-immune retinal dysfunction and tumor cell regression in mice. Eur JImmunol 2002;32:2300–7.

23. Rong S, Ikeda H, Sato Y, Hirohashi Y, Takamura Y, Sahara H, SatoT, Maeda A, Ohguro H, Sato N. Frequent detection of anti-recoverincytotoxic T-lymphocyte precursors in peripheral blood of cancerpatients by using an HLA-A24-recoverin tetramer. Cancer ImmunolImmunother 2002;51:282–90.

24. Bazhin AV, Schadendorf D, Willner N, De Smet C, Heinzelmann A,Tikhomirova NK, Umansky V, Philippov PP, Eichm€uller SB. Photo-receptor proteins as cancer-retina antigens. Int J Cancer 2007;120:1268–76.

25. Bazhin AV, Schadendorf D, Owen RW, Zernii EY, Philippov PP,Eichm€uller SB. Visible light modulates the expression of cancer-ret-ina antigens. Mol Cancer Res 2008;6:110–18.

26. Alexander KR, Barnes CS, Fishman GA, Milam AH. Nature of thecone ON-pathway dysfunction in melanoma-associated retinopathy.Invest Ophthalmol Vis Sci 2002;43:1189–97.

27. Geertsen RC, Hofbauer GF, Yue FY, Manolio S, Burg G, Dummer R.Higher frequency of selective losses of HLA-A and -B allospecific-ities in metastasis than in primary melanoma lesions. J Invest Derma-tol 1998;111:497–502.

28. Kato M, Takahashi M, Akhand AA, Liu W, Dai Y, Shimizu S, Iwa-moto T, Suzuki H, Nakashima I. Transgenic mouse model for skinmalignant melanoma. Oncogene 1998;17:1885–8.

29. Dalke C, Loster J, Fuchs H, Gailus-Durner V, Soewarto D, Favor J,Neuhauser-Klaus A, Pretsch W, Gekeler F, Shinoda K, Zrenner E,Meitinger T, et al. Electroretinography as a screening method formutations causing retinal dysfunction in mice. Invest Ophthalmol VisSci 2004;45:601–9.

30. Polans AS, Witkowska D, Haley TL, Amundson D, Baizer L, AdamusG. Recoverin, a photoreceptor-specific calcium-binding protein, isexpressed by the tumor of a patient with cancer-associated retinopa-thy. Proc Natl Acad Sci USA 1995;92:9176–80.

31. Pf€ohler C, Haus A, Palmowski A, Ugurel S, Ruprecht KW, ThirkillCE, Tilgen W, Reinhold U. Melanoma-associated retinopathy: highfrequency of subclinical findings in patients with melanoma. Br J Der-matol 2003;149:74–8.

149CANCER-RETINA ANTIGENS AND MAR