Recoverin as a cancer-retina antigen

Abstract

In photoreceptor cells the Ca²⁺-binding protein recoverin controls phosphorylation of the visual receptor rhodopsin by inhibiting rhodopsin kinase (GRK-1). It can also serve as a paraneoplastic antigen in the development of retinal degeneration in some patients with cancer. The aberrant expression of recoverin in cancer cells and the presence of autoantibodies against recoverin are essential for the occurrence of cancer-associated retinopathy, which finally results in the apoptosis of photoreceptor cells. Noteworthy in cancer patients, the aberrant recoverin expression and the appearance of autoantibodies against recoverin are more frequent than paraneoplastic syndromes. We suggest the term “cancer-retina antigens” for this kind of proteins like recoverin that are solely expressed in retina and tumor tissues and evoke antibodies and/or T cells in patients with cancer. The rare development of a paraneoplastic syndrome is possibly caused by this immune response and probably depends on further events allowing to overcome the blood–retina barrier and the immune privileged status of the retina. It is still unknown whether aberrantly expressed recoverin could have a specific function in cancer cells, though it is suggested that it can be functionally associated with G-protein-coupled receptor kinases. This paper reviews the present knowledge on paraneoplastic syndromes associated with the aberrant expression of recoverin. A possible application of recoverin as a potential target for immunotherapy of cancer is discussed.

Paraneoplastic neurological syndromes associated with loss of vision

Paraneoplastic neurological syndromes (PNS) are rare neurological disorders that are related to cancer, even though the primary tumor and its metastases have not invaded the nervous system. PNS are thought to be autoimmune diseases that develop when malignant tumors express proteins, paraneoplastic antigens (PNA), which are normally present only in neurons. PNA expression outside the nervous system triggers the host’s immune response toward the corresponding PNA or its epitopes present in neurons, resulting in the development of PNS [31]. A pivotal point in pathogenesis of the disorders was the recognition by J. Posner and coworkers that patients with PNS harbor high-titer antibodies both in their serum and spinal fluid that recognize apparently identical antigens in normal brain and tumor tissues as shown by Western blot analysis [59]. Retinopathies as paraneoplastic syndromes are also thought to be induced by antibodies or cytotoxic T cells of the patient to antigens expressed both in the retina and the tumor (Fig. 1).

Fig. 1.

Hypothesis of the induction of retinopathies associated with cancers. Dendritic cells take up necrotic or apoptotic tumor cells and present paraneoplastic antigens to the immune system. This induces the production of specific antibodies and/or CD8 positive, cytotoxic T cells (CTLs). These CTLs might play a role in tumor defense. Antibodies and/or CTLs might eventually cross the blood–retina barrier and subsequently damage either photoreceptor cells (PhR) or bipolar cells (BC). GC ganglion cells

Cancer-associated retinopathy (CAR) is a very rare disease with only 55 cases published so far. An interesting feature of CAR is that it can be observed several weeks or even months before the underlying tumor is detected [16]. Indications are blurred vision, complaints of flashing lights, loss of peripheral and color vision, and night blindness [36]. The concept that cancer can promote visual dysfunction was proposed for the first time by Sawyer et al. in 1976 [63]. Seven years later, Keltner et al. postulated that autoimmunity plays a crucial role in retinopathy mediated by cancer [39]. The autoimmunity theory was supported by experimental data showing that serum antibodies from lung cancer patients with CAR can recognize certain cell types in retina sections [30, 42, 43]. The first indication that a retinal protein with an apparent molecular weight of 23 kDa could be “a CAR antigen” was obtained by Thirkill and coworkers [69]. Recoverin as a Ca²⁺-binding protein in photoreceptor cells was parallelly discovered by several groups [21, 22, 44]. Subsequently, recoverin was identified as being the CAR antigen [57, 71].

Melanoma-associated retinopathy (MAR) is distinguished from CAR by clinical differences. Today, the ophthalmological components of both CAR and MAR are well studied. In CAR patients, full field flush electroretinograms (ERG) show a reduction of the a-wave indicating the degeneration of photoreceptor cells (Fig. 2). Flush ERG of MAR patients show often a reduction of the b-wave reflecting a dysfunction of bipolar cells, which create Na⁺-currents through the Muller glia [12, 54, 73]. CAR patients’ sera strongly stain photoreceptor cells and weakly stain bipolar cells, while MAR sera have been reported to exhibit strong immunoreactivity in bipolar cells [12, 54]. The histological examination of the retina from MAR patients also indicated degeneration of the bipolar cells [27, 70]. This explains why MAR patients suffer from sudden visual loss and night blindness, whereas their color vision is often normal. However, in some MAR patients, the a-wave in the ERG can be reduced, degenerative regions are detected in retina and pigment epithelium, and sera recognize photoreceptor cells [13, 40]. These data suggest that in addition to bipolar cells, photoreceptor cells can also be affected in MAR patients.

Fig. 2.

Normal and pathological electroretinogram (ERG). (a) The ERG is a field potential in the whole eye normally composed of an a-wave and a b-wave. Photoreceptor degenerations manifest in a reduced or changed a-wave (b), while malfunction in the bipolar cells are seen as altered b-waves (c)

While the mechanism of paraneoplastic retina degeneration in CAR is rather well investigated, the mechanism of retinopathy in MAR is still unknown. The molecular mechanism of CAR is the apoptotic death of the photoreceptor cells attributed to autoantibodies reacting to the photoreceptor protein recoverin [4, 6]. Lei and colleagues have shown that serum autoantibodies of MAR patients injected in monkeys induce the degeneration of depolarized bipolar cells [45]. Antigenic mimicry has been implicated in the autoimmune response in MAR; cancer cells may have epitopes, which are similar to those in bipolar cells [60]. Still, at present there are no data about specific target(s) for autoantibodies in bipolar cells. In contrast, Potter et al. showed that photoreceptors can also be damaged in MAR patients [60]. The authors detected autoantibodies against the photoreceptor protein transducin, which is part of the visual signal transduction, in the serum of one MAR patient.

Recoverin as a paraneoplastic antigen in CAR

Recoverin is a Ca²⁺-binding protein present only in vertebrate photoreceptors, in certain other retinal neurons, and in pineal glands [53]. It represents a family of neuron-specific Ca²⁺-binding proteins, the so called neuronal calcium sensors [15], containing Ca²⁺-binding sites of an EF-hand type. Four potential Ca²⁺-binding sites of such a type are present in the recoverin molecule of which only two (the second and the third) are capable of binding calcium ions [25]. In vitro recoverin can form a complex with rhodopsin kinase (GRK-1) [17, 28] thus inhibiting phosphorylation of the visual receptor rhodopsin in a Ca²⁺-dependent manner [41, 64, 65]. It has been confirmed recently that recoverin serves as a Ca²⁺-sensor of rhodopsin phosphorylation under physiological conditions [50].

To date, many authors have described the presence of autoantibodies against recoverin in the sera of CAR patients with different types of malignancies located outside the nervous system as summarized in Table 1. The frequency of such cases is very low. Only about 25 cases have been described so far. The majority of the autoantibody-positive CAR cases have been detected in patients with small cell lung cancer, the most malignant lung tumor of neuroendocrine origin.

Table 1.

Published cases of patients with CAR and anti-recoverin autoantibodies

Tumor type	Number of patients	Reference
Small cell lung cancer	2	[36]
	5	[7]
	1	[58]
	1	[14]
	1	[37]
	1	[51]
Non-small cell lung cancer	1	[62]
Endometrial carcinoma	1	[5]
	1	[7]
Ovarian carcinoma	1	[33]
Cervical cancer	1	[32]
Uterine sarcoma	1	[24]
	1	[67]
Breast cancer	1	[67]
	2	[35]
	1	[7]
Colon carcinoma	1	[7]
Skin squamous cell carcinoma	1	[7]
Invasive thymoma	1	[38]

A rat model system has been established to elucidate the contribution of recoverin autoantibodies to the development of retinopathy [3, 26]. The injection of recoverin led to the induction of uveitis and retina degeneration. In a guinea pig model, animals were sensitized with small cell lung cancer cell lines that induced production of anti-recoverin autoantibodies in the animals and caused retinopathy in them [68]. Based on the models described, it was concluded that autoantibodies against recoverin can be a principle cause of retina degeneration in patients with CAR.

How can autoantibodies reach the retinal cells and how do they induce cell death? The first hurdle is the blood–retina barrier (see also Fig. 1). It is still unclear why this barrier does not prevent autoantibodies from penetrating into the retina. But function or malfunction of the blood–ocular barrier may account for the low frequency of CAR (and possibly also MAR). Even when this first obstacle is overcome, autoantibodies have to enter the cell, as recoverin is located intracellularly. This issue was addressed by experiments incubating immortalized rat retinal cells with sera either from CAR patients or from animals immunized with recoverin [4]. It could be shown that the autoantibodies were internalized by the cells and produced their destruction in a dose-dependent and time-dependent manner. The authors suggested that internalization may occur through nonspecific endocytosis, as specific and unspecific IgGs were taken up similarly. The observed cell death was characterized by apoptotic features, which is in line with experiments of the same group showing induction of apoptosis in photoreceptor and bipolar cells in vivo by antibodies against recoverin [6].

The next question is as to how these internalized autoantibodies can account for apoptosis. Adamus and colleagues speculate in the same paper [6] that blocking of intracellular recoverin by the antibodies may lead to an increase in cytoplasmic free calcium and subsequently to the activation of a nuclear endonuclease, which finally results in DNA fragmentation. This assumption is supported by the observation that conformational changes of recoverin induced by calcium enhance binding of antibodies to the sequence within residues 64–70 of the second calcium-binding domain which is found to be the major antigenic region of recoverin molecule [1, 2].

Recoverin autoantibodies can be detected in CAR patients and can induce photoreceptor apoptosis. But is the autoantibody response against recoverin indeed triggered by recoverin expressed in the tumor cells? In fact, it has been shown that tumor samples and one cell line from CAR patients are recoverin positive [52, 58, 68].

Recoverin as a paraneoplastic antigen without PNS

From the above one might assume that autoantibodies against recoverin should be detected only in CAR patients. But this is not the case! Many PNA have the capacity to induce low titer autoantibodies without any signs of paraneoplastic syndromes as, for example, those described for the Hu-antigen [18, 29]. In the process of raising antibodies against recoverin, we found that immunized rabbits developed an immune response to recoverin at variable titers. Investigations of the eye bottoms and light microscopy of retina slices detected retinal degeneration in rabbits with high titers of the antibody, while the eye characteristics of rabbits with low titers did not differ from those of control animals [9]. These findings stimulated us to screen serum samples of patients with lung cancers irrespective of the presence of CAR. Indeed, we have revealed 15 patients with SCLC of the 99 individuals investigated (15%) and 9 of 44 patients with non-small cell lung cancers (20%) with relatively low titers of the autoantibodies in their sera, but without the manifestation of CAR at the time of the serum sampling [8, 11]. Thus, we have concluded that autoantibodies against recoverin can be detected in sera of patients with lung cancer without manifestation of paraneoplastic syndrome.

What is the reason for the development of retinopathy in some patients with autoantibodies against PNA but not in others? One can suggest the following explanations: (1) To overcome the blood–ocular barrier, the titer of the autoantibodies in the blood stream of patients should exceed a certain level. Otherwise the autoantibodies cannot penetrate into the cells, get the corresponding intracellular antigen and initiate apoptosis of the cells. The same might be true for the hemato–encephalic barrier in case of other PNS. (2) The PNS development might depend on the actual epitopes recognized by the autoantibodies. In the case of recoverin, autoantibodies from CAR patients mostly bind to residues 64–70 representing the EF-hand 2 domain and initiate apoptosis by blocking Ca²⁺-binding [2, 58]. It would be interesting to identify the recoverin epitopes against which the autoantibodies in patients without CAR are directed. (3) The manifestation of CAR might well depend on a second event allowing the transmissibility of the blood–ocular barrier.

Recoverin is a cancer-retina antigen

Investigation of recoverin expression in cancer tissues and cell lines irrespective of the CAR-syndrome was started by Maeda et al. [46]. The authors analyzed 33 cell lines of various types of cancer and detected recoverin expression in about half of them (Table 2). Our own studies revealed recoverin expression in 68% of small cell lung cancers and in 85% of non-small cell lung cancers [11], which is more frequent than many other tumor-associated antigens [23]. Furthermore, we detected recoverin expression in tumors and/or cell lines of melanoma, head and neck cancer as well as in colon cancer cell lines (Bazhin et al. in preparation; Table 2).

Table 2.

Recoverin expression in tumors and cell lines (positive/total)

		mRNA	Protein
Small cell lung carcinoma	Cell lines	0/7a 1/4e	0/16a 1/4e
Non-small cell lung carcinoma	Cell lines	0/2a	0/9a
	Tissues		25/29b
Lung adenocarcinoma	Cell lines	0/1a 2/5e	0/3a 0/5e
	Tissues		9/11b
Lung cancer (without typing)	Cell lines	10/14c
	Tissues	18/27c
Melanoma	Cell lines	20/32d	11/21d
	Tissues	37/47d	11/30d
Endometrium adenocarcinoma	Cell lines	3/3e	2/3e
Breast adenocarcinoma	Cell lines	9/12c 3/3e	3/3e
	Tissues	8/14c
Pancreas adenocarcinoma	Cell lines	0/3e	0/3e
Cervix adenocarcinoma	Cell lines	2/2e	2/2e
Squamous cell carcinoma of the cervix	Cell lines	1/2e	1/2e
Squamous cell carcinoma of the oral cavity	Cell lines	3/3e	3/3e
Squamous cell carcinoma of the esophagus	Cell lines	1/1e	1/1e
Squamous cell carcinoma of the stomach	Cell lines	3/3e	2/3e
Colon (without typing)	Cell lines	1/5d
Squamous cell carcinoma of head and neck	Tissues	0/5d
Leukemia	Cell lines	0/3d 0/1e	0/1e
B-cell lymphoma	Cell lines	1/1e	1/1e

References: ^a[10], ^b[11], ^cBazhin et al. unpublished data, ^dBazhin et al. in preparation, ^e[46]

Thus, in addition to the retina, recoverin can also be aberrantly expressed in various cancer cells at high frequencies. Its aberrant expression can induce autoantibody and T cell responses [47]. Importantly, recoverin is not the only photoreceptor protein with these characteristics. Recently, we detected rhodopsin and arrestin in some melanoma cells and found autoantibodies in the serum of melanoma patients against these proteins, though these patients did not suffer from any paraneoplastic retinal degenerations [34].

Therefore, by analogy with cancer-testis antigens [56], we suggest the expression cancer-retina antigens for this type of proteins for the following reasons (Fig. 3): (1) Normal expression is restricted to immunoprivileged tissues (retina/testis), (2) proteins are aberrantly expressed in tumor cells, (3) and show subsequently high antigenicity (autoantibodies and/or T cells); (4) the term “cancer-retina antigens” does not reduce this kind of proteins to the rare events of CAR or MAR as does “paraneoplastic antigens”. Interestingly, there is evidence that the aberrant expression of both, cancer-testis antigens and cancer-retina antigens (at least recoverin), might be due to promoter demethylation [10, 20].

Fig. 3.

Comparison of cancer-testis antigens and cancer-retina antigens and the features characterizing these groups. Frequencies of expressions of and responses from MAGE-A1 and recoverin are taken from [11] and [72], respectively

Does recoverin have a function in tumor cells?

It is known that in photoreceptor cells, recoverin serves as a Ca²⁺-dependent regulator of rhodopsin kinase (GRK 1) [66]. Thus, one may speculate that in tumor cells, recoverin might also be responsible for regulation of certain GRKs. Miyagawa et al. [55] have shown that recoverin can functionally be associated with GRKs through calveolin-1 in cancer cell lines. If so, recoverin may play a role in tumor progression, metastasis and drug resistance Also, Maeda et al. transfected the lung adenocarcinoma cell line A549 with recoverin cDNA and showed a reduction in cell proliferation [48]. However, it cannot be excluded that the aberrant expression of recoverin is merely a consequence of disturbance in regulation of gene expression in cancer cells and thus has no function at all in these. In any case further investigations are needed to answer these questions.

Could recoverin be suitable for cancer diagnosis and therapy?

The presence of autoantibodies against recoverin in the serum of lung cancer patients is highly specific (98%), but only moderately sensitive (20%) [11]. This makes recoverin autoantibodies unattractive for consideration as a single markers. Although considering that the autoantibodies appear well before the cancer can be diagnosed by the usual diagnostic techniques, these could be used in combination with other markers of cancer. The recoverin protein is expressed at rather high frequencies in lung cancer, ranging from 68 to 85% depending on the tumor subtype [11]. In non-small cell lung cancer, the grade of differentiation is inversely correlated to the frequency of recoverin positive tumor cells. This qualifies recoverin as a possible prognostic marker, which should be further analyzed.

It was suggested that SCLC patients with CAR have a better prognosis than those without CAR [19]. Maeda et al. hypothesized that this benefit might be due to the presence of recoverin-specific cytotoxic T lymphocytes (CTL) [47], which could attack the tumor cells (see also Fig. 1). In fact, they described recoverin-specific CTL in two HLA-A24⁺ CAR-positive patients and in three of ten CAR-negative patients, but unfortunately also in two of six healthy individuals. The CTL precursor frequency of CAR-positive patients and that of CAR-negative patients was higher than that of healthy donors. Moreover, the authors identified three recoverin peptides which can induce peptide-specific CTL. Rong et al. [61] used tetramers with HLA-A24 dependent peptides of recoverin and showed recoverin-specific CTL precursors in colon, stomach, breast cancer patients and in some healthy individuals. The tetramer analysis had a good correlation with cytotoxic response.

In an experimental in BALB/c mouse model Maeda et al. could induce specific CTLs in response to the peptide R64 (AYQHVFRSF) and observed a tumor-preventive effect by the vaccination [49]. Though, this peptide also induced the generation of antibodies and retinal degenerations.

Thus, recoverin can be targeted both by antibodies and T cells. The circumstances which do or do not lead subsequently to a PNS still have to be clarified before recoverin can be considered as a target in therapy.

Conclusions

The expression and antigenicity of recoverin distinguish this molecule as a cancer-retina antigen. Immune responses against recoverin can have beneficial effects for the tumor patient, but might also induce paraneoplastic syndromes. The induction of the latter seems to depend on the presence of autoantibodies directed against a certain functional domain of the molecule. The known function of recoverin in photoreceptor cells did not give any hints as to its function in tumor cells, though an involvement in proliferation might be speculated. From the therapeutic point of view, it seems to be pivotal to unravel differences in antibody responses against different epitopes of the protein. The proper selection of specificities might be responsible for whether or not CAR/MAR is induced during tumor defense via recoverin.

Abbreviations

CAR

Cancer-associated retinopathy

CTL

Cytotoxic T cell(s)

MAR

Melanoma-associated retinopathy

PNA

Paraneoplastic antigen(s)

PNS

Paraneoplastic neurological syndrome(s)

SCLC

Small cell lung cancer