Background
Anti-Hu (antineuronal nuclear antibody 1; ANNA1) and anti-Ri (antineuronal nuclear antibody 2; ANNA2) antibodies are paraneoplastic autoantibodies which react with nuclear and cytoplasmic proteins of neurons throughout the central nervous system. Anti-Hu antibody is closely associated with small cell carcinoma of the lung and has been reported in patients with a wide spectrum of paraneoplastic neurological disorders, including limbic encephalitis, cerebellar degeneration, brainstem encephalitis, dorsal sensory neuronopathy, motor neuron disease, and gastroparesis [
1],[
2]. Anti-Ri antibody, found predominantly in patients with breast adenocarcinomas and small cell carcinoma of the lung, has been associated with syndromes of opsoclonus/ataxia, cerebellar degeneration, limbic or brainstem encephalitis, ophthalmoplegia, laryngeal dystonia, and gastroparesis [
3]-[
12]. Although anti-Hu and anti-Ri antibodies produce essentially identical patterns of immunohistological labeling in human or rodent brain sections [
13], the antigens recognized by the two antibodies are distinct and are encoded by different genes [
14]-[
16]. Anti-Hu antibody recognizes proteins of 35-42 kDa in Western blots of neuronal lysates, whereas anti-Ri antibody recognizes antigens of 55 kDa [
1],[
5]. Paraneoplastic neurological syndromes associated with anti-Hu antibodies are accompanied by neuronal destruction and usually show no clinical improvement following immunosuppressive treatment or tumor removal [
17]-[
22]. Although some degree of neuronal destruction has been described in cases of anti-Ri encephalomyelitis, findings have been more variable [
3],[
4],[
23],[
24], and clinical improvement has been reported in some patients with anti-Ri antibody following immunosuppressive therapy or successful treatment of the underlying malignancy [
5],[
6],[
18],[
23]-[
28].
Despite repeated detection of anti-Hu and anti-Ri antibodies in sera and cerebrospinal fluid of affected patients, a direct role for either antibody in the pathogenesis of paraneoplastic neurological injury has been considered unlikely [
29],[
30]. Viable neurons have been widely believed to exclude immunoglobulin (Ig)G [
29],[
30]. Because both anti-Hu and anti-Ri antibodies recognize intracellular antigens, it has been thought that neither antibody would be able to react with its target antigens
in vivo to induce disease [
29],[
30]. We and others have previously reported destruction of neurons by anti-Hu antibody in dispersed cell cultures [
31],[
32]; however, the relevance of these findings to events occurring
in vivo has been uncertain, and attempts by others to produce neurological injury in experimental animals by immunization with recombinant Hu antigens have been unsuccessful [
33].
To study the interaction of antineuronal antibodies with neurons, we have established a brain slice (organotypic) culture system which preserves anatomical relationships present
in vivo and allows exposure of neurons to antibodies without interposition of the blood-brain barrier [
34],[
35]. We have previously demonstrated that living Purkinje cells in cerebellar slice cultures incorporated and subsequently cleared normal IgG. Although the intracellular presence of normal IgG
per se did not affect Purkinje cell viability, incubation of cultures with an IgG-daunorubicin immunotoxin resulted in Purkinje cell uptake of the immunotoxin and targeted Purkinje cell death [
35]. We have subsequently demonstrated that the paraneoplastic autoantibody anti-Yo, associated with cerebellar degeneration, was taken up by Purkinje cells, and that intracellular accumulation of anti-Yo antibody resulted in non-apoptotic Purkinje cell death [
34].
In the present study, using rat cerebellar and hippocampal slice cultures, we examined whether anti-Hu and anti-Ri antibodies might also be taken up by Purkinje cells or other neuronal populations and whether uptake of either antibody was associated with neuronal death. We also assessed whether specific binding of anti-Hu antibodies to intracellular Hu antigen was required for antibody-mediated cytotoxicity. We found that both antibodies were taken up by neurons and that binding of anti-Hu antibody to its intracellular target antigens induced neuronal death over time, whereas anti-Ri antibody did not induce cell death during the period of study.
Discussion
The present study, using organotypic cultures of cerebellum and hippocampus, demonstrated that IgGs containing anti-Hu or anti-Ri were readily taken up by multiple populations of neurons and accumulated intracellularly. Demonstration of antibody uptake by neurons in real time ruled out the possibility that detection of intraneuronal IgG following incubation with either antibody could have been an artifact of fixation and processing. Detection of intracellular IgG in neurons excluding SYTOX cell death dyes confirms that antibody uptake had occurred in viable neurons. Our findings in slice cultures thus demonstrated that IgG antibodies could be taken up by living neurons and bind to their intracellular antigenic targets. Although it has been held that viable neurons exclude IgG, the ability of antibodies to enter neurons and bind to their intracellular antigens has recently been reported by Congdon and colleagues in studies employing monoclonal antibodies reactive with tau protein in a mouse model of Alzheimer’s disease [
43],[
44]. Our observations, along with those of Congdon and colleagues, indicate that antibodies are capable of entering neurons and that antibodies could thus interact with intraneuronal antigens in other paraneoplastic or autoimmune disorders.
Although both anti-Hu and anti-Ri IgGs were incorporated into hippocampal and cerebellar neurons, the effect of these two antibodies on neuronal viability was markedly different. Anti-Hu antibody caused death of multiple neuronal populations in both cerebellar and hippocampal cultures. Neuronal death in cultures incubated with anti-Hu IgG was abolished following adsorption with the HuD antigen. These observations indicate that the cytotoxic effect of the anti-Hu sera and IgGs used in our studies was due to antibody interaction with Hu antigens and was not caused by other antibodies or factors present in patient sera. Our data indicate that anti-Hu antibodies were cytotoxic and may have a direct role in the pathogenesis of paraneoplastic neuronal injury. Although these observations do not exclude a concomitant role for T cell-mediated immunity, our study also showed that anti-Hu antibody could cause neuronal death in the absence of T cell-mediated immune response or antibody-dependent cellular cytotoxicity. The patient-to-patient variation in cell killing by individual sera containing anti-Hu antibodies may have been due in part to differences in antibody titer, and because clinical history was not available from most of the patients studied, we do not know the rapidity with which neuronal cytotoxicity might have occurred. However, this variation is also consistent with our prior observations that paraneoplastic autoantibodies strongly reactive with neurons in human brain sections may vary considerably in their reactivity with nonhuman neuronal tissue such as the rat tissue used in the present experiments [
45]. This finding also parallels the differences in killing by different sera - despite similar antibody titers against frozen sections of human brain - which we have previously observed in dispersed rat cerebellar granule cell cultures [
31].
Our finding that anti-Hu antibody induced neuronal cell death is in contrast to observations reported by Sillevis Smitt and colleagues [
33]. These investigators were unable to produce neurological injury in mice, rats, and guinea pigs by immunization with species-specific HuD antigen or, in mice, by intravenous infusion of human anti-Hu IgG [
33]. However, their study differed from ours in three important aspects. First, unlike our studies in slice cultures, their investigations were carried out in animals whose blood-brain barriers appeared to be intact, possibly limiting entry of IgG into brains and neurons. Failure of IgG to spread into brain parenchyma and interact with neurons in their study was evidenced by the absence of extravascular or intraneuronal IgG in brains when animals employed in their experiments were perfused to remove intravascular blood and antibody. Second, mice in their study were harvested no later than 48 hours after receiving intravenous human anti-Hu IgG. In the present study, slice cultures studied at 48 hours showed IgG uptake but no evidence of cell death above levels seen in controls. Third, cell death in their study was identified only by morphological methods, whereas our study used markers for both cell death and apoptosis, allowing for actual quantitation of neuronal destruction.
In previous studies, we have demonstrated that anti-Yo antibodies, associated with paraneoplastic cerebellar degeneration, accumulated intracellularly within Purkinje cells and induced non-apoptotic Purkinje cell death [
34]. In the present study, uptake and cytotoxic effect of anti-Hu antibody on rat brain slice cultures differed from that of anti-Yo antibody in two aspects. First, anti-Hu antibody accumulated in and induced death of multiple neuronal populations, whereas anti-Yo antibody-associated cytotoxicity involved only Purkinje cells [
34]. Second, although anti-Yo antibodies mediated non-apoptotic cell death, anti-Hu antibodies appeared to cause cell death, at least in part, by apoptosis. Our findings are consistent with those of Furneaux and Wong, who demonstrated apoptosis in BE2-N neuroblastoma cells incubated with anti-Hu antibodies [
46], and by De Giorgio and colleagues, who reported production of apoptosis by anti-Hu antibodies in cultures of SH-Sy5y neuroblastoma cells and in primary cultures of guinea pig mesenteric ganglion neurons [
47].
The exact mechanisms by which anti-Hu antibodies might cause neuronal death are currently unknown. Hu proteins are a group of RNA-binding proteins which share homology with the drosophila protein groups Elav and sex-lethal [
48]. Four closely related Hu antigens have been cloned and sequenced: HuA (also termed HuR), HuB (Hel-N1), HuC, and HuD [
48]. All of these except HuA are expressed specifically in neurons and are involved in neuronal development [
48]. In older mice and rats, Hu proteins have also been shown to be important in neuronal plasticity and memory [
48]. Work by Ince-Dunn and colleagues has demonstrated changes in levels of glutamate transmitter and neuronal excitability in brains of HuC
-/- mice (Elav13
-/-) [
49]. Although Hu proteins have been thought to affect multiple aspects of the post-transcriptional RNA function, the present study is the first to provide evidence that interaction of anti-Hu antibody with the HuD protein has a direct effect upon neuronal viability. There is extensive sequence identity between HuB, HuC, and HuD [
48], and patient anti-Hu antibodies reactive with HuD have been shown to react with all four Hu antigens [
50]. As such, antibodies binding to HuD could potentially bind to and interfere with the functions of other Hu antigens. The failure of anti-Ri antibodies to produce neuronal death over a more prolonged time period, despite extensive neuronal uptake, indicates that anti-Hu-mediated neuronal death is not simply a consequence of intracellular antibody accumulation.
Although we tested a smaller number of anti-Ri samples in the current study, our findings that anti-Ri antibodies did not cause neuronal cytotoxicity during the period of observation were consistent among the cultures and were in keeping with our earlier report that anti-Ri antibody did not produce death in dispersed cultures of cerebellar granule cells [
51]. Our current data neither prove nor disprove a role for anti-Ri antibody in the pathogenesis of paraneoplastic neurological disease. Absence of detectable neuronal death in cultures incubated with anti-Ri antibody could possibly indicate that anti-Ri antibody is nonpathogenic. However, inability to induce cell death could also have been due to technical factors such as differences in antibody titer as compared with anti-Hu, diminished reactivity of the antibodies against nonhuman neuronal antigens as compared with those present in human neurons, or a requirement for extended exposure beyond the period of observation than that used in the present study. The major neuronal antigens recognized by anti-Ri antibodies, Nova1 (neuro-oncological ventral antigen-1) and Nova2 (neuro-oncological ventral antigen-2) represent neuron-specific RNA processing factors which control the alternative splicing of a wide array of transcripts important for synaptic activity [
52],[
53]. Buckanovich and colleagues have demonstrated that anti-Ri antibodies interfered with the binding of the Nova-1 protein to RNA
in vitro[
54], and it is possible that the antibody could exert a similar effect in living cells.
Patients with anti-Ri antibody response differ from those with anti-Hu antibody response in that clinical improvement following immunosuppression and/or tumor removal has been reported in a number of patients [
5],[
6],[
18],[
23]-[
28]. Only four patients with anti-Ri antibody have been studied at autopsy, however, and only two of these had exhibited a response to treatment [
3],[
4],[
23],[
24]. All four patients had varying degrees of neuronal loss in cerebellum, brainstem, or spinal cord, but brainstem findings were minimal in the two patients whose opsoclonus and other symptoms improved following treatment [
23],[
24]. Our finding that the interaction of anti-Ri antibody with its target antigen was nonlethal over the time period studied thus raises the possibility that anti-Ri antibody might, at least initially, be capable of impairing neuronal function without causing cell death. This is in contrast to the irreversible clinical outcome seen in virtually all patients with anti-Hu antibody, where our current experiments suggest that neurological deficits may result from antibody-mediated neuronal destruction.
Acknowledgements
The authors would like to thank Jeffrey Rothstein, MD, PhD, Dan Gincel, PhD, and Carol F Coccia, the Johns Hopkins University School of Medicine, for their assistance in establishing the cerebellar slice culture system used in these experiments. The authors thank Dr John W Rose, Dr James B Burns, and Dr Robert Fujinami for their helpful comments concerning the manuscript. This work was supported by a Merit Review award from the United States Department of Veterans Affairs and the Western Institute for Biomedical Research (JEG). IT is supported by grant R21NS059724 from the National Institute of Neurological Disorders and Stroke of the National Institutes of Health and by COBRE Center grant 8P20GM103433 from the National Institute of General Medical Sciences of the NIH.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JEG and NGC conceived of the study, participated in its design and coordination and helped to draft the manuscript. IT and SLC participated in design of this study, helped to draft the manuscript and/or provided critical review of its revision. TDJ carried out analysis of key patient materials used in the study and provided critical revision of the manuscript. SAC planned and carried out the tissue culture studies required for the study and provided critical revision of the manuscript. KEH and BW planned, carried out, and conducted the portion of the research involving confocal microscopy and provided critical revision of the manuscript. All authors have given final approval of the version to be published, and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Each author has participated sufficiently in the work to take public responsibility for appropriate portions of the content.