Mitofilin and titin as target antigens in melanoma-associated retinopathy

Authors


Abstract

Melanoma-associated retinopathy (MAR) is a rare paraneoplastic syndrome in patients with melanoma. Since the onset of MAR symptoms is often associated with tumor progression or recrudescence of metastases, MAR-related symptoms are prognostic relevant. The pathomechanism underlying MAR is supposed to result from antibody production against yet unknown melanoma-associated antigens that are also expressed in retinal tissue, leading to the destruction of retinal cells and resulting in defective signal transduction. Only a 35 kDa protein in Müller glial cells, a 22 kDa neuronal antigen and retinal transducin have been identified as MAR-associated antigens to date. To identify additional antigens potentially involved in the pathogenesis of MAR, we screened a retina cDNA phage library for reactivity with antibodies in the sera from 9 patients with MAR or subclinical MAR using the serological analysis of recombinantly expressed clones (SEREX) approach. Six sera from melanoma patients without evidence of MAR and 10 sera from healthy donors served as controls. Mitofilin and titin were identified as antigens against which antibodies were found exclusively in sera of MAR patients, but not in the sera of MM patients without MAR or healthy donors. This is the first study to demonstrate that titin is highly expressed from retinal tissue and melanoma. The fact that none of the MAR-associated antigens detected to date by their capacity to elicit a humoral immune response is located on the cell surface questions a major pathogenetic role of the respective antibodies and suggests that cellular, rather than humoral mechanisms are operative in the primary immune attack against the retina in MAR. © 2006 Wiley-Liss, Inc.

Melanoma-associated retinopathy (MAR) is a rare paraneoplastic syndrome in melanoma patients. The pathomechanism underlying this syndrome is believed to be mediated by antibodies against antigens shared by malignant melanoma cells and cells in the retina.1, 2, 3, 4 Up to now only about 90 cases with MAR have been reported in the literature, but a recently performed screening study in melanoma patients without any signs or symptoms of MAR revealed that the presence of antiretinal antibodies in melanoma patients without ocular symptoms is more common than previously suspected.2, 3, 4 Ladewig et al. found in 53 out of 77 (68.8%) serum samples from melanoma patients antibodies reactive with various components of the retina. This phenomenon was melanoma stage-dependant and a higher antibody activity was found in advanced stages of disease.3 Moreover, in two previous studies we demonstrated that the melanoma patients present more frequently with subclinical signs and symptoms suspicious of MAR than formerly thought.5, 6 In a collective of 28 melanoma patients without vision problems 7 patients had clinical signs and symptoms consistent with MAR, 18 had some indications, while the remaining 3 had none.5, 6

Until now, it is still unclear which retinal antigens are involved in this paraneoplastic syndrome. In the last decade, investigations in single MAR patients identified a 35 kDa protein in Müller glial cells,7 a 22 kDa neuronal antigen8 and retinal transducin9 as antigens possibly involved in the development of MAR. With Western blot analyses being little informative in the past,2, 4 the use of serological identification of antigens by recombinant expression cloning (serological analysis of recombinantly expressed clones, SEREX) held promise to better define antigen(s) associated with the MAR syndrome. Using SEREX, Hartmann et al. screened sera from 3 MAR patients and 22 melanoma patients without visual symptoms10 for antibodies reactive with the expressed clones derived from a retina and a melanoma cDNA phage library. They identified rhodopsin and T-cell activation protein phosphorylase as antigens expressed in the retina library. Furthermore, they identified 20 different antigens that were recognized by the sera used in their study but no further work was done to show the relationship to MAR. Hartmann et al. considered rhodopsin and visual arrestin as the most interesting proteins based on the association of these proteins with human retina. They could show their expression in melanoma cell lines and in several control tissues. However, no evidence was provided that these retina-associated structures are also expressed by melanomas. We now report the identification of antigens recognized by antibodies in the sera of MAR patients and demonstrate by immunohistochemistry that these molecules are expressed in both the retina and in the patients' melanoma, suggesting a causative role of these antigens in the development of paraneoplastic MAR.

Material and methods

Patients

Sera from 9 patients with MAR (2/9) or subclinical findings suspicious of MAR in combination with antiretinal serum antibodies (7/9) were investigated (for details see Table I). Sera from 6 melanoma patients without evidence of MAR or subclinical MAR were used as controls. Patients' characteristics are summarized in Table I.

Table I. Characteristics of Patients Enrolled in the Study
NumberSexTumor classification1Tumor type2Evidence for MAR3
  • 1

    Staging according to the American Joint Committee on Cancer (AJCC 2002) tumor classification.

  • 2

    Tumor type: SSM, superficial spreading melanoma; NM, nodular melanoma; ALM, acrolentiginous melanoma; UK, unknown tumor type.

  • 3

    Evidence for MAR: 0, melanoma patients without any evidence for MAR; 1,complete MAR syndrome with typical vision deficiencies and antiretinal antibodies; 2, incomplete MAR syndrome with subclinical signs and symptoms consistent with MAR in addition with antiretinal antibodies.

1FemaleIV, pT2a pN2b pM1cNM1
2MaleIV, pT4b pN3 M1cNM2
3FemaleIIIC, pT4a rpN3 M0ALM1
4MaleIV, pT3a pN3 M1bNM2
5MaleIV, rpTx pN0 M1cUK2
6FemaleIV, pT3b pN2c M1cNM2
7MaleIIIC, pTx pN3 M0ALM2
8FemaleIV, pT3b pN2 M1cNM2
9FemaleIV, pT1a N0 rpM1cSSM2
10FemaleIIIC, rpT4b pN2b M0NM0
11FemaleIV, pT3a pN1a M1aALM0
12MaleIIIC, pT2a pN3 M0SSM0
13MaleIIIB, pT2a pN2 M0SSM0
14FemaleIV, Tx N0 M1cUK0
15MaleIV, Tx N0 rpM1aUK0

Sera

Sera were collected during routine diagnostic procedures. The study was approved by local ethic committee (Ethikkommission der Ärztekammer des Saarlandes) and was performed in compliance with the Helsinki declaration. All patients and controls gave written informed consent. Control sera were collected from healthy donors of our research group.

Retinal tissue

Retinal tissue for immunohistochemistry assays was obtained from freshly slaughtered pigs. The whole pig eyes were fixed for 48 hr in 4% formaldehyde solution. After this fixation process the eyes were cut in halves and underwent automate-assisted paraffinization.

Construction of a retinal cDNA phage library

Human retinal total RNA was purchased from Clontech (BD Biosciences, Palo Alto, CA), representing a mixture of 25 individual healthy retinae. PolyA+-RNA was isolated by Dynabeads mRNA purification kit (Dynal, Oslo, Norway) following the instructions of the supplier. Library construction was done starting with 50 ng polyA+-RNA and using the SMART cDNA library construction kit (BD Biosciences) as described in the kit manual. The primary library consists of 4 × 107 λTriplEx2 clones and was amplified once.

Immunoscreening of phages libraries

Sera were preabsorbed against E. coli and phage proteins as described.11 SEREX immunoscreening was performed with increasing concentrations of patient's sera as described before.12 In total, more than 2 × 106 clones were screened for immunoreactivity. Positive clones were subcloned to monoclonality and submitted to conversion of pTriplEx2 plasmid clones. The nucleotide sequence of cDNA inserts was determined using an Excel cycle sequencing kit (Epicentre Technologies, Madison, WI) on a LICOR automatic sequencer. Sequencing was performed according to the manufacturer's instructions starting with the vector-specific primers. Insert-specific primers were designed as the sequencing proceeded. Sequence alignments were performed with DNASIS (Pharmacia Biotech, Freiburg, Germany) and BLAST software on EMBL, Genbank and PROSITE databases.

Detection of antibodies against defined antigens

Monoclonalized phages from clones reactive with patient serum were mixed with nonreactive phages of the cDNA library as internal negative controls at a ratio of 1:10 and used to transfect bacteria. IgG antibodies in the 1:100 diluted E. coli-preabsorbed sera from other patients and healthy controls were tested with the above-described phage assay to assess the antibody responses against the respective antigen.

Expression of recombinant proteins

Immunoreactive phages were converted to plasmids as described in the manufactorers manual. To subclone all possible reading frames of the cDNAs for expression in pGEX4T1 (Pharmacia, Uppsala, Sweden) the Clontech in-fusion cloning kit (Clontech, Mountain View, CA) was used. (Primers LR1_s 5′-TGG TTC CGC GTG GAT CCC CGC GCG CCA TTG TGT TGG TAC C-3′, LR2_s 5′-TGG TTC CGC GTG GAT CCC CCG CGC CAT TGT GTT GGT ACC-3′, LR3_s 5′-TGG TTC CGC GTG GAT CCC CGC GCC ATT GTG TTG GTA CC-3′, LR_as 5′-ACG ATG CGG CCG CTC GAG TCT AGA GTC GAC TG-3′, PCR 30 cycles of 94°C 1 min, 50°C 1min, 72°C 2 min). Correct clones were transformed in E. coli BL21 and expression was done as described in the GST gene fusion system handbook of the manufacturer (1 mM IPTG at OD600 = 0.5, 4 hr, 37°C). For analysis of the expression products protein gel and Western blot confirmation were done using the individual patient serum. E. coli cells were lysed by sonification in PBS/Triton X100 (1% v/v) and the recombinant protein was purified by affinity column chromatography using gluthathione-agarose columns (Pharmacia Biotech, Freiburg, Germany). Elution was done with 10 mM glutathione (reduced form) followed by dialysis against PBS.

Depletion and affinity purification of patients' serum

Recombinant GST protein was immobilized on GST agarose following a published procedure (www.flemingtonlab.com). Patients' serum (100 μl) was diluted 1:10 (v/v) in PBS and purified by affinity chromatography using the corresponding immobilized recombinant GST protein. Washing was performed by 20 column volumes of PBS. Elution was done with 0.2 M glycine pH 3.0, followed by neutralization and dialysis against PBS. Purification was checked by protein gel electrophoresis and Western blot analysis.

Immunohistochemistry

Paraffin-embedded sections (4–6 μm) were deparaffinized through xylene and alcohol series, and pretreated in a microwave oven in citrate buffer for 14 min at 600 W, then cooled in PBS at room temperature. Sections were first blocked with 1:20 goat serum (Dako, Glostrup, Denmark) for 20 min at room temperature, then washed in PBS and incubated with normal human serum or patient serum at a dilution of 1:100 overnight in a moist chamber at 4–6°C. Slides were washed twice in PBS for 10 min and incubated for 2 hr in a dark humidified chamber at room temperature with the appropriate fluorescein isothiocyanate (FITC)-conjugated goat anti-human antibody (Dako) at a dilution of 1:30. Following a thorough wash in PBS, the slides were incubated with a polyclonal rabbit anti-FITC antibody conjugated with horseradish peroxidase (HRP) (Dako) at a dilution of 1:100 for 30 min at room temperature. Slides were washed again in PBS for 10 min. Localized antibody activity was visualized with 10% 3-amino-9-ethylcarbazole (AEC) solution in acetate buffer containing H2O2 (10 μl/ml) and stopped in aqua dest. Haematoxylin 5% was used as a counterstain. The sections were examined and photographed by conventional microscopy.

Immunohistochemistry with mitofilin and titin antibodies was performed using a different protocol. Paraffin-embedded sections (4–6 μm) were deparaffinized through xylene and alcohol series and pretreated in a water bath in citrate buffer for 30 min at 98°C, then cooled at room temperature and washed in aqua dest. Peroxidase was blocked with H2O2 (Fischer, Saarbrücken, Germany) (6 ml 30% H2O2 /250 ml aqua dest) for 25 min at room temperature. The slides were washed in PBS and incubated with mitofilin or titin antibodies at a dilution of 1:50–1:100 with Dako Cytomation Antibody Diluent with Background Reducing Components (Dako) for 1 hr in a moist chamber at 37°C. After washing in PBS, the slides were incubated with Dako Cytomation EnVision + R System Labeled HRP anti-rabbit (Dako) for 30 min at room temperature, washed in PBS and detected with 10% AEC solution in acetate buffer containing H2O2 (10 μl/ml). Counterstaining, examination and photography were performed as described earlier. Sera from healthy donors and PBS served as negative controls.

Other cDNAs

cDNAs that were not identified by SEREX screening during this work, but had been described by Hartmann et al.,10 were obtained as full-length clones from the German Resource Center (RZPD, Berlin, Germany).

PCR and subcloning in phages

PCR was performed using synthetic sequence-specific primers (MWG Biotech, Ebersberg, Germany), the cDNAs mentioned above and AmpliTaq gold polymerase (Perkin Elmer, Weiterstadt, Germany). Primers and annealing temperature are listed in Table II. PCR conditions were as follows: 94°C 12 min initial denaturation, followed by 30 cycles 94°C 1 min, annealing temperature 1 min, 72°C 2 min. Products were purified and subcloned in λTriplEx2 (Clontech, Mountain View, CA). After packaging into phage particles, these full-length cDNA expression clones were used to transfect E. coli. After amplification these phage particles were used for serum testing. Cloning was verified by sequencing.

Table II. Primers and PCR Conditions for Amplifying cDNAs from Hartman et al.10
NameRZPD IDPrimers and restriction sitesAnnealing (°C)
AdamtsIRATp970E0849D6START-EcoRI5′-GAATTCCATGGAATGCTGCCGTCGGGCA-3′66
STOP-XhoI5′-CTCGAGATGAACAAAGAAAGAAGGCAC-3′
 BBS1IRAKp961G1990Q2START-EcoRI5′-GAATTCCATGAGCCCCGGACCCCAGCTC-3′66
STOP-SalI5′-GTCGACCCTTCCACGTGGTACAAGTGC-3′
 GAPDHIRATp970C0537D6START-EcoRI5′-GAATTCCATGGGGAAGGTGAAGGTCGGA-3′60
STOP-XhoI5′-CTCGAGCTCGGGGAGGGTGATGCTGGG-3′
 GADD45IRAUp969E1148D6START-EcoRI5′-GAATTCTATGACTCTGGAAGAAGTCCGC-3′66
STOP-XhoI5′-CTCGAGCTCGGGGAGGGTGATGCTGGG-3′
 Peroxiredoxin1IRAUp969H0668D6START-EcoRI5′-GAATTCGATGTCTTCAGGAAATGCTAAA-3′55
STOP-XhoI5′-CTCGAGCTTCTGCTTGGAGAAATATTC-3′
 SRPXIRAUp969F0487D6START-MunI5′-CAATTGCATGGGGAGCCCCGCACATCGG-3′66
STOP-XhoI5′-CTCGAGGGTGTTACAGGTCTGGCTCAT-3′
 RhodopsinDKFZp686B2082Q2START-EcoR5′-GAATTCCATGAATGGCACAGAAGGCCCT-3′60
STOP-XhoI5′-CTCGAGGGCCGGGGCCACCTGGCTCGT-3′

Results

Summary of SEREX screening

More than 2 × 106 recombinant clones represented in the human retina-derived cDNA expression library were screened by the SEREX approach using diluted sera from 9 patients with MAR or subclinical MAR. The patients from whom the sera had been taken had no obvious disease other than melanoma or MAR; in particular, the sera were negative for antinuclear antibodies. Using patient's sera at a 1:10.000 dilution, we detected one immunoreactive clone whose cDNA insert was homologous to human mitofilin (Genbank AccNo NM_006839.1). The sera from 2 of the 9 MAR patients (22 %) investigated in our study were reactive against this phage clone at a 1:10.000 dilution, while no reactivity was found in the sera from the melanoma patients control group (0/6) or healthy serum (0/10), even at dilution of 1:100 (Table III).

Table III. Antigens Detected by IgG Antibodies in the Sera from Patients with MAR
ClonePatient #Serum dilutionHomologyAccession number
151:10.000MitofilinNM_006839.1
811:10.000Mitofilin 
15811:3.500Mitofilin 
16111:3.500Mitofilin 
1834, 7, 81:500HemoglobinNM_000518.4
1907, 131:500Cytochromec oxidaseDQ246833
2077, 91:500TitinAF525413.1

Using serum dilutions at a range between 1:5,000 and 1:100, additional clones were detected. Most of them were not specifically associated with MAR, because they were also recognized by antibodies in the sera from melanoma patients without MAR or healthy donors, thus representing common autoantigens. At 1:100, another clone, coding for a part of titin (Genbank AccNo AF525413.1) was detected. This clone appeared to be also MAR associated because sera from 2/9 (22%) MAR patients, but not from melanoma patients without MAR (0/6) or from healthy donors (0/10) showed serum reactivity against it (Table IV). At 1:500, one clone coding for the last 40 amino acids of the mitochondrial cytochrome c oxidase subunit was detected. If this clone is also MAR specific remains unclear, because a serum screening revealed that 1/9 (11%) MAR patients but also 1/6 melanoma control patients had serum reactivity against it. Healthy donors were negative for antibody reactivities against this clone (Table IV).

Table IV. Antibody Reactivity in the Sera from MAR Patients, Melanoma Patients and Healthy Controls Against Antigens Detected by High-Titered IgG Antibodies in the Sera from MAR Patients
CloneHomologyMAR sera poolMM sera poolHealthy donor pool
1Mitofilin2/90/60/10
190Cytochrome c oxidase1/91/60/10
207Titin2/90/60/10
183Hemoglobin3/90/60/10

In addition, at a dilution of 1:500 a clone coding for haemoglobin was detected. The sera from 3 MAR patients reacted with this clone, whereas neither the sera from melanoma controls nor healthy donors showed serum reactivity against it (Table IV).

In the λTriplEx2 system every cDNA inserted into the MCS is expressed in all reading frames. To verify our findings we had to express the clones for mitofilin and titin in all 3 possible reading frames as GST fusion proteins. Clones showing an expression product of correct size and sequence were tested with the individual patient serum. These results are in agreement with our screening data showing that mitofilin and titin are responsible for the immunoreaction and not an unknown protein encoding by another reading frame.

Large-scale expression of positive clones was done and the GST fusion proteins were affinity-purified. Yield was about 50 mg GST-mitofilin respective 10 mg GST-titin when started with 1 l of expression culture. Protein gel electrophoresis and Western blotting were done to analyze the expression products (Fig. 1a). To show the specificity of the immunoreaction in retina tissue these recombinant GST-fusion proteins were used to isolate the specific antibodies from patients' serum (Fig. 1b).

Figure 1.

Expression of GST-mitofilin and purification of mitofilin-specific antibodies from patients' serum: (a) expression and purification of GST-mitofilin, denat. SDS-PAGE, Commassie stain, M marker lane (30 kDa, 50 kDa indicated), 1 crude extract, 2 affinity purification flow-through, 3 eluate; (b) Purification of mitofilin-specific antibodies from patients' serum, denat. SDS-PAGE, silver stain, M marker lane (30 kDa, 50 kDa indicated), 1 patients' serum, 2 affinity purification flow-through, 3 eluate.

Mitofilin

To provide further evidence that mitofilin, the target antigen of antibodies specifically found in patients with MAR (patient no. 1) or subclinical MAR (patient no. 5) is indeed expressed at the protein level in retina, immunohistochemistry assays were performed. Figure 2 shows different retina stainings that confirm the SEREX results for mitofilin: part A shows specific staining of different retinal layers by using the whole serum of patient no. 1. Part B also shows specific retinal staining of the same serum after affinity purification for mitofilin, while the flow-through from the affinity purification procedure showed no specific staining (Part C). Parts D and E served as negatives controls. Figure 3 demonstrates the presence of mitofilin in the autologous melanoma metastasis from patient number 1, supporting mitofilin as an antigenic target both in the retina and in the autologous tumor.

Figure 2.

Mitofilin expression in retinal tissue. Juxtaposition of different immunohistochemical stainings on retina: (a) serum from patient number 1, undiluted, specific AEC-staining (red) of different retinal layers; (b) serum from the same patient after affinity purification for mitofilin, dilution 1:100; weaker staining of the same retinal layers; (c) flow-through liquid obtained from affinity purification from the same patient, dilution 1:100, no specific staining; (d) negative control with serum from a melanoma patient without evidence of clinical or subclinical MAR, undiluted, no specific staining; (e) negative control with serum from a healthy donor, dilution 1:100, no specific staining. PE, pigment epithelium; PR, photoreceptor cells; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; NFL, nerve fiber layer.

Figure 3.

Mitofilin expression in melanoma: (a) immunohistochemical staining with the whole serum from patient number 1 on a lymph node metastasis from melanoma, positive staining of a single large melanoma cell; (b) serum from the same patient after affinity purification for mitofilin, positive staining of several melanoma cells; (c) staining of the same melanoma metastasis using S-100β-antibodies; (d) staining of the same metastasis using mitofilin antibody T386, positive staining of all melanoma cells; (e) negative control with serum from a healthy donor; (f) negative control performed by omitting of the primary antibody. All primary antibodies were used in a dilution of 1:100. Detection was performed with AEC.

Titin

Similar to mitofilin, expression of titin protein was demonstrated in the retina by immunohistochemistry (Fig. 4): the unfractionated serum from patient no. 7 (part a), as well as the affinity purified serum (part b) generated specific staining in the retina, while the flow-through (part c) and serum from a melanoma patient without evidence of clinical or subclincal MAR (part d) were negative. In Figure 4 a set of retinal stainings using titin-reactive antibodies is shown. A strong expression of titin in retinal tissue with accentuation of different layers, depending of the subtype of the titin antibody used was demonstrated. The fact that titin is expressed by retinal tissue was formerly not known (personal communication with Prof. Labeit, Mannheim, Germany). Autologous tumor tissue from the titin-positive patients was not available, therefore demonstration of cross-reactivity of the antibodies with the target antigen titin in the autologous tumor was not possible. Instead, we used tissue from an allogenic melanoma metastasis. Figure 5 confirms a strong titin expression from melanoma.

Figure 4.

Titin expression in retinal tissue. Juxtaposition of different immunhistochemical stainings on retina: (a) immunohistochemical staining with the whole serum of patient number 7, dilution 1:100, weak but specific AEC-staining of different retinal layers especially the photoreceptor cell layer; (b) serum from the same patient after affinity purification for titin, undiluted, stronger and even more specific staining of different retinal layers; (c) flow-through liquid obtained from affinity purification from the same patient, undiluted, no specific staining; (d) negative control with serum from a melanoma patient without evidence of clinical or subclinical MAR, dilution 1:100, no specific staining; (e) negative control with serum from a healthy donor, dilution 1:100, no specific staining; (f) titin antibody 9D10, undiluted, strong staining of all retinal layers with accentuation of photoreceptor cells; (g) titin antibody N2A, dilution 1:50, strong staining of all retinal layers with accentuation of photoreceptors and the nerve fiber layer; (h) Titin antibody X351-352, dilution 1:50, strong staining of the whole retina with enhancement in the photoreceptor cell layer and the inner nuclear layer; (i) titin antibody MIR-MGT 30, dilution 1:100, homogeneous and strong staining of all retinal layers. PE, pigment epithelium; PR, photoreceptor cells; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; NFL, nerve fiber layer.

Figure 5.

Titin expression in melanoma: (a) immunohistochemical staining of melanoma cells in a subcutaneous metastasis using a S100 antibody, dilution 1:1,500; (b) negative control performed by omitting of the primary antibody; (c) immunohistochemical staining of melanoma cells with titin antibody MIR-MGT 30, dilution 1:50; (d) immunohistochemical staining of melanoma cells with titin antibody X351-352, dilution 1:50. (ad) are consecutive sections of the same metastasis. Detection was performed with AEC.

Antigenicity of previously reported MAR-associated antigens

None of our serum samples investigated reacted with the clones identified by Hartmann et al.10 We focused on cDNAs that are specifically expressed in melanoma or retina and not in a wide spectrum of different tissues. We tested clones coding for BBS1, GADD45, PRDX1, ADAMTS, SRPX, GAPDH and rhodopsin using all MAR sera described in Table I at a dilution of 1:100. No immunoreaction was detected.

Discussion

The MAR syndrome has been suggested to be a paraneoplastic retinal disorder in melanoma, which might be caused by antitumoral antibodies induced by melanoma antigens that recognize shared antigen(s) in the retina. Binding of these antibodies against their antigens in retinal cells could lead to the destruction of retinal tissue resulting in a diversity of subclinical or clinical ophthalmological symptoms. It is still unclear if T-cell mediated processes play an important role in the genesis of MAR. Focusing on the antibody-mediated pathomechanisms in our study, we investigated 9 sera from melanoma patients with MAR or subclinical MAR and compared their antibody reactivity with sera from 6 melanoma patients without evidence of (sub)clinical MAR. Applying the SEREX approach to a retina cDNA library we identified mitofilin and titin as potential targets in the pathogenesis of MAR that elicit antibody responses specifically in MAR patients. In addition to these MAR-associated antigens, antibodies were detected against haemoglobin and a mitochondrial sequence coding for about 40 amino acids at the c-terminal end of cytochrome C oxidase subunit I (Acc No DQ246833), which are not MAR-associated, but are also found in non-MAR sera.

Little is known to date about mitofilin, which was recently identified as a mitochondrial protein with impact on the morphological structure and presumptive influence on mitochondrial function.13 Downregulation of mitofilin results in abnormal mitochondrial structure and function. Treatment of HeLa cells with specific siRNA leads to decreased cellular proliferation and increased apoptosis.13 Mitofilin is anchored in the inner mitochondrial membrane and was recently demonstrated in lipid rafts of B-cells, which serve as platforms for B-cell receptor signal transduction.14, 15, 16 However, nothing is known about the function of mitofilin in B lymphocytes or other cells.

Mitochondrial cytopathies are a heterogenous group of hereditary diseases with abnormal mitochondrial function or structure. For the Kearns-Sayre syndrome, a rare genetic abnormality with disturbance of mitochondrial DNA, ophthalmological symptoms have also been reported.17 Patients suffering from Kearns-Sayre syndrome develop chronic progressive ophthalmoplegia and retinal degeneration before the age of 20.17 In addition, pathological morphology of mitochondrial cristae has been reported in a variety of other diseases such as myopathy, cardiomyopathy, rhabdomyosarcoma or Warthin's tumor.18 However, this is not surprising because the central nervous system and muscles belong to the tissues with the highest content of mitochondria in the human body. The two patients investigated in our study with antimitofilin antibodies in their serum showed no evidence for Kearns-Sayre syndrome, as both were much older than 20 years and ophthalmological examination confirmed subclinical or clinical MAR in both cases. Furthermore, none of these two patients suffered from neuromuscular diseases, heart diseases or any kind of a second malignancy.

The giant protein titin (4.200 kDa) forms the third myofilament of the striated muscle and is expressed in heart and skeletal muscles.19 It was first discovered in the seventies of the last century and its gene contains 363 exons within a 280 kb genomic sequence. Because of the organization of the gene, a diversity of differential splicing variants can be created.20 Altered function of titin is well investigated in patients with cardiomyopathy and diastolic heart failure.21, 22 In addition, titin mutations have been shown to be etiological for several types of hereditary muscle dystrophies.23, 24 Antibodies against titin were found in patients with different autoimmune diseases. For instance, in patients with myasthenia gravis associated with thymus neoplasia, who often develop autoantibodies to the acetylcholine receptor in the motor endplate, autoantibodies against the main immunogenic region (MIR) of titin have been observed.25, 26 Machado and coworkers reported titin autoantibodies in sera from patients with systemic sclerosis.26 In addition, they showed that titin is uniformly distributed along condensed chromosomes. In analogy to the role of titin in muscles the authors suppose an important role of titin for the elasticity of chromosomes that protects against chromosome breakage during mitosis.26 None of our two titin-positive patients had evidence for heart disease, neuromuscular disease or any other kind of autoimmune disease; in particular we had no clue for concomitant myasthenia gravis or systemic sclerosis. Hence, we suppose that titin antibodies in 2 of 9 MAR patients investigated in our study are also MAR-associated.

The work presented here is the first to demonstrate that titin is expressed in retinal tissue and melanoma. On that account, the role of titin in the retina as well as in melanoma must presently remain unknown and has to be investigated in further studies.

The impact of antibodies against cytochrome c oxidase (COX) in the development of MAR is more speculative. The function of mitochondria as intracellular organelles is mainly the production of ATP through oxidative phosphorylation.27, 28 This process depends on the mitochondrial respiratory chain that comprehends different enzyme complexes. These complexes are either encoded by nuclear or by mitochondrial DNA (mtDNA). COX is an important enzyme in the mitochondrial respiratory chain that is exclusively encoded by mtDNA.27, 28 Barron and coworkers showed that defects of enzymes of the mitochondrial respiratory chain may lead to morphological changes in the retina, especially involving photoreceptors and the retinal pigment epithelium. They could prove that these enzyme defects contribute to changes observed in the ageing eye or in age-related maculopathy.27 The authors further speculate that accumulation of mtDNA damage in photoreceptors plays an important role in the induction of apoptosis in the cells affected.27 With reference to our findings, we suppose that the induction of antibodies against COX theoretically may also lead to defects in the retina, resulting in ophthalmological symptoms as seen in MAR. However, antibodies against COX are possibly not MAR specific, because we likewise found these antibodies in a melanoma patient without evidence of MAR.

The role of hemoglobin antibodies found in 3/9 MAR patients must remain unclear. Our attempts in the search for the incidence and role of haemoglobin antibodies in tumor patients in the relevant literature were without success and less informative. Perhaps, transfusion of erythrocyte concentrates performed in those 3 patients due to tumor anemia in the past played an important role in the induction of these antibodies. But this must remain a speculation. Nevertheless, we found these antibodies exclusively in MAR patients and neither in melanoma nor healthy controls.

In conclusion, mitofilin and titin were identified as potential antigens underlying the paraneoplastic syndrome of MAR. However, neither mitofilin nor titin, nor COX (which has previously been reported to be an antigen expressed in a variety of malignant tumors and polymyalgia rheumatica) represent antigens located on the cell surface. To become immunogenic, these cytoplasmic and nuclear antigens must be released from the antigen-expressing cells by secretion, shedding or tumor lysis in order to be captured by antigen-specific surface immunoglobulins on B-cells that then develop with specific T-cell help into antibody-secreting plasma cells. Therefore, a primary role of the respective antibodies in the pathogenesis of MAR is unlikely; rather, it suggests that a cytotoxic T-cell response against cells presenting antigenic peptides derived from the antigen molecules in a MHC-restricted fashion on their surface represents the primary step of the immune attack in MAR. Possibly, this cytotoxic T-cell reaction is initiated because melanoma cells present the respective antigenic peptides in the context of danger29 and then the resulting cytotoxic T-cells cross-react with retinal cells that present the same MHC-peptide complexes on their surface. In this context of cellular mechanism playing the major role in the primary immune damage, the antibody response against the respective antigens shared by melanoma and retinal cells would rather be an epiphenomenon or by-product without a significant role in the pathogenesis of MAR. However, with the knowledge of the molecular structure of the target antigens gained by identifying these antigens on the basis of their capacity to elicit specific antibody responses, the role of such cellular mechanisms can now be specifically addressed. The approach of “reverse T-cell immunology,” which previously has been shown to be instrumental in the definition and analysis of T-cell epitopes of antigens originally identified by their reactivity with antibodies,30, 31, 32, 33 should also be helpful in dissecting the role of cytotoxic T-cell responses against antigenic peptides presented by both melanoma and retinal cells in MAR. With the sequence of the MAR-associated target antigens at hand, MHC-restricted epitopes eliciting cytotoxic CD8+ or CD4+ T-cell responses can now be defined and the presence of T-cells with specificity to the respective MHC-peptide complexes can then be investigated and dissected in melanoma patients with and without MAR.

Acknowledgements

Mitofilin antibodies T3866 and T3867 were kindly provided from Dr. Jiping Zha, Department of Pathology, University of Texas Southwestern Medical Center at Dallas, TX, USA. Titin antibodies 9D10, X351-352, X105-106, anti-N2A and MIR-MGT 30 were kindly provided from Prof. Dietmar Labeit, Anästhesiologie und Operative Intensivmedizin, Universitätsklinikum Mannheim, Mannheim, Germany. C.P., M.P. and W.T. are supported by a grant of Deutsche Krebshilfe (German Cancer Aid).

Ancillary